The practical glossary for packaging, automation & CAPEX

Every scan shows the hall on exactly one day. If a pipe is laid or a rack moved afterwards, the point cloud knows nothing about it. So have the scan done as close to detailed planning as possible, and agree beforehand what will be delivered, in which format and to what accuracy. Whatever the scanner cannot see is missing. Behind control cabinets and above suspended ceilings there will be gaps, and it will not find cables in the floor at all. If forklifts and people walk through the picture during the scan, you get ghost points in the cloud afterwards. The scan really pays for itself when moving equipment into the building and in clash checks. You then see on screen that a sprinkler pipe hangs exactly at palletiser height, instead of finding out during installation. Also check who in your organisation can actually open these volumes of data.

Most chains stop too early, and they stop at a person. Operator not paying attention, action training, case closed. A few weeks later the same thing happens on the other shift. At that point keep asking why the machine lets the error happen. A typical chain looks like this. Cartons open up at the palletiser because the glue does not hold. It does not hold because too little glue is applied. Too little comes out because a nozzle is partly blocked. It is blocked because it is only cleaned after a fault, and that in turn because the maintenance plan has no interval for it. Only there can something be changed for good. Do the analysis at the machine and not in the meeting room. Around a table, guesses quickly turn into facts because nobody contradicts them.

The tidy-up day is the easy part. Trouble starts afterwards, because within a few weeks everything lies wherever it happens to fit. Usually nobody owns the area and there is no fixed date for the audit. On packaging lines, a change-parts board with shadow outlines right at the machine pays off. If a part is missing, everyone sees it at a glance, and nobody searches for ages before finding out the stop is still in the workshop. I have little time for walk-rounds with a scoring table when nothing happens afterwards. If your changeovers involve a lot of searching, start with the change-parts trolley, because the result shows fastest there.

An 8D report from a supplier often reads like this: cause operator error, action operator retrained. That closes the matter for the supplier, but not for you by a long way. Check in D1 whether the team includes someone from the line where the defect arose. In D3 it pays to look closely at which goods were really blocked and sorted, including those in transit and in the supplier’s warehouse. For D4, ask for two causes, why the defect could occur and why it was not detected. The second is usually the more interesting one. Have the effectiveness proven with inspection results over several weeks, because a signature alone is not enough. The report is only closed once those results are in. If your own customer asks you for an 8D report, by the way, your D4 will be read just as critically.

A good acceptance certificate refers to the performance tests, meaning test records and endurance runs, and lists every deviation with an owner and deadline. It also sets out which portion of payment is withheld until resolved. Photos and HMI screenshots belong as an attachment. Never sign subject to reservation without concrete points. A blanket reservation carries little weight in court and still blocks cooperation. Go through the outstanding items with the supplier at the machine before signing, not at a desk. How the certificate works legally is something to clarify beforehand with your lawyer, ideally already at the contract stage.

Acceptance criteria without a measurement method are worthless. High availability cannot be accepted, but a fixed percentage of machine-related availability over three shifts with two named formats can. Set the criteria out in the user requirements specification, repeat them in the contract and verify them at FAT and SAT using the same protocol. Clarify beforehand what counts as a machine-related fault and what does not, otherwise every stop gets argued over at acceptance. I also have the product and packaging material written down that will be used for measuring. If you run with the best carton and fresh film, you reach values that nobody sees in daily operation later.

Pay the down payment only once the original guarantee is in your hands. That sounds obvious and is still often done the other way round, because the supplier is pushing to buy material. Check the amount, order number and beneficiary on the document. A guarantor from the same group of companies helps little if the whole group gets into trouble. The tricky part is a fixed expiry date. If delivery slips, the security lapses while your money is still with the supplier. Note the date and ask for an extension in good time. The original belongs in the accounts department’s safe, not in the project file. After delivery, return the document promptly. Until then the supplier pays commission on it, and sooner or later that turns up in his prices.

Classic AGVs follow fixed installed tracks and suit stable processes with a few well-defined routes, such as pallet transport from the end of the line to the warehouse. They are reliable and predictable, but layout changes take effort because the track has to be rebuilt. Economic viability comes from forklift trips saved in multi-shift operation. The Shopfloor Analysis provides the transport volumes per route as the basis for the calculation. Ask the supplier about the handover points as well. The interface to the line or the racking is usually where things stick first, especially when a pallet arrives crooked.

The benefit shows on the night shift, when no specialist is reachable and the fault is on page 340 of the manual. Whether the assistant finds it depends on the documents. That means well-maintained documents, a fault history with the solutions and not just the messages, and an interface that works with gloves on. Involve the line leads early, or the tablet stays in the cupboard. Let the shift staff ask real questions from their daily work, including the clumsily worded ones. And measure whether downtime actually gets shorter or whether the search has merely become more convenient.

The lubricator is a relic that refuses to go away. Current valves and cylinders come with a factory lubrication and need no oil in the air. Once you start oiling, though, you have to keep going, because the oil washes out the factory lubrication. With new machines I therefore ask whether the lubricator is really needed, and if so, which H1 oil it is filled with. Look at the condensate drain too. A drain that only opens when the pressure is switched off does little on a machine that runs round the clock. Where the unit is mounted matters as well. If it sits low at the back between cable duct and wall, nobody will change the filter element. Operators like to turn up the regulator when a cylinder seems sluggish. A lockable regulator saves a fair amount of discussion.

After an emergency stop the alarm list fills up, and the cause sits somewhere far down, usually between lots of follow-up messages. Require an alarm list with text, priority and operator guidance in the user requirements specification, together with first-out annunciation and follow-up alarm suppression. A good sign at acceptance: an operator finds the triggered sensor without the programmer. I have the alarm texts shown to me beforehand, because a message that only quotes a number helps nobody at three in the morning. Standing alarms get ignored sooner or later, and then the one important alarm is lost along with them. Either you fix the cause or the alarm goes.

On packaging machines, allergen changeovers are mainly a problem of accessibility. Product residue sits in vibratory troughs, under belts, in the multihead weigher and in the sealing jaws, in other words wherever you cannot get at it without tools. When buying, check whether all product-contact parts can be removed without tools or at least inspected. Plan the production sequence from allergen-free to allergen-containing, and the changeover gets shorter. Validate the cleaning with tests. A visual check alone is not enough, because you only see flour dust in a trough when it is too late. Put these requirements in the URS. Retrofitting later is expensive and brings downtime with it.

Because everything bolts onto profiles so quickly, some machines grow a structure of brackets, angles and extra plates over the years that no longer appears on any drawing. In wet areas the slots collect product residue and cleaning agent, and the cover strips are gone after a few modifications anyway. Settle on one system and one profile size for the whole plant. T-slot nuts and connectors from different manufacturers often almost fit, and on a guard door almost means it will work loose at some point. For fixed guard fencing the Machinery Directive requires fixings that can only be undone with a tool. Check this after modifications, before somebody fastens the fence with wing screws.

Andon only works with a response chain. It must be settled who arrives within what time when the light is red. Without a defined escalation the board becomes decoration. Digital Andon systems link the alarm to MES data and send faults to the maintenance team’s phones, but that only works when the fault causes are coded cleanly. At the start, a signal tower per line and a board with the five most frequent faults of the week are often enough. Fix the colours in writing and put their meaning up at the line, otherwise every shift reads the light differently. Also check that the tower can be seen from the aisle when the forklift is parked in front of it.

Anomaly detection reports deviations, not causes. After start-up, many alerts turn out to be changeovers, cleaning or a new packaging-material supplier. Without event context, the system learns these states as faults, or the team switches the alerts off. Then the project is dead before it has found anything. Plan a training-in phase in which the maintenance team evaluates the alerts. Give the system events as context, meaning format, order and cleaning. Define beforehand which alert triggers which response and who receives it. An alert that belongs to nobody simply stays unread, however clever the analysis behind it.

An anti-slip sheet only works if it actually lies between the layers. The palletiser needs a slip-sheet inserter with a reliable sheet gripper for that, otherwise the sheet is missing on every third layer. Check the coefficient of friction against your carton surface, because glossy laminated lids behave differently from rough testliner. For cold chains the coating must be moisture-resistant, otherwise the sheet softens. With export shipments you should document the proof of load securing with a tilt test. Mind how the sheets are stored too: if they warp in damp air, the inserter starts to jam.

On a walk through the hall the iPad is the most practical tool. You photograph the spot, write the comment straight onto it and sketch directly over the layout. Machine models can be placed into the hall as an AR view, see USDZ. That shows you before the purchase whether the plant fits the space intended for it. For an Efficiency Day the documentation then emerges during the walk-through and not later at a desk, where nobody remembers which photo belongs to which spot. Name the photos on site with line and station, otherwise you spend the evening sorting pictures that nobody can assign any more.

Getting onto a supplier list at a large corporation often takes weeks. It starts with registration in the procurement portal, then come proof of insurance, certificates and compliance, sometimes audits as well. Anyone who needs a new specialist supplier on a project starts onboarding in parallel with the technical clarification and does not wait for the award stage. Otherwise the preferred supplier is in the quotation but cannot be ordered from, and the schedule has a hole. Ask procurement early which suppliers are already approved and what lead times onboarding has. Smaller suppliers should prepare the typical documentation as a ready package.

Plenty of packaging specifications state an AQL but no defect classes. Goods receiving and the supplier then argue over whether a smeared print is a major or a minor defect, on precisely the day the line is waiting for the material. Define the defect classes with examples, ideally with photos and boundary samples. Be careful with defects that can lead to a recall, such as a wrong barcode or a missing allergen on the label. A sample is the wrong tool for those, they need a label check at the line that inspects every pack. Look at how the sample is drawn as well. Ten folding cartons from the top layer of the first pallet are not a random sample. And an accepted lot only tells you that the sample was fine.

For palletising and case packing, the articulated-arm robot is the all-rounder: long reach, high payload, any pack pattern. Its limits are at very high cycle rates, where delta or gantry robots take over, and in space. The safety fence and restricted area need floor space. Include the cell size in the layout, including access for pallet changes, because that is where many retrofit projects in existing buildings fall down. Check the factory floor as well, since the robot puts dynamic loads into the foundation. And count the payload honestly. Gripper, layer package and an allowance for the worst case all count.

Artwork is created in marketing, but it has to run on the line. Something tends to get lost in between. The print mark moves a few millimetres because it spoils the layout, or the blank area for the best-before date suddenly sits where the printer on the machine cannot reach. On folding cartons the glue flap gets varnished as well, and the hotmelt no longer holds. That is why I have every artwork approval countersigned by someone from engineering, even if supposedly only some text has changed. A different varnish alone changes the coefficient of friction, and the film suddenly behaves differently on the forming shoulder. Set up a short sheet of technical requirements for each line, covering position and contrast of the print mark, blank areas and varnish-free zones. Attach that sheet to every brief sent to the agency.

The control cabinet holds the drawing with the red marks from commissioning. On the server sits the PDF from the FAT. And nobody remembers which jumper went onto which terminal, or why. That gets expensive at the first weekend breakdown, if not before. Specify in the contract when the as-built documentation is due and which software version it must reflect. If part of the final payment depends on it, it usually arrives more punctually. Then spot-check the documents on site. Pick three terminals, a sensor and a pipe and compare them with the drawings. If those do not match, the rest usually does not either. The red-marked drawings stay in the cabinet until the clean set has been checked.

POWERJO manages project tasks and the open items list, LOP for short, in Asana. Everyone on the team can see it, and every item has a due date and an owner. That replaces the Monday morning question of who was actually going to do what. Anyone who keeps a shared task list with the supplier on a client project saves the weekly back-and-forth of comparing email statuses. The question of which version of the Excel list is the current one disappears as well. It takes discipline: an item without an owner and a due date is not an item but a wish. And close tasks only when the result has been checked, not as soon as somebody says they took care of it.

Aseptic filling is a process project and cannot be handled as a machine purchase. Product preparation, sterilisation, filler, packaging and cleaning chemistry must be validated together, and the operating team needs different knowledge from a standard filling line. Anyone with a sterility problem seldom finds the cause at the filler alone. It can lie with the packaging, a seal, the sterile air or a door that was opened too early. Do not underestimate the time needed for validation and regulatory coordination. And cost honestly into the operating budget the sterilisation time before every production start and the consumption of chemicals and sterile air.

In packaging areas, Ex zones often arise unnoticed, at bagging plants, flour silos, icing sugar dusters or sugar conveyors. Icing sugar in the packing hall is regularly overlooked, because supposedly nothing but packing happens there. The operator must designate the zones in the explosion protection document, and the machine supplier provides equipment of the appropriate category. To do that, the supplier needs the zone for each installation location in writing, which is why it belongs in the URS. If it is missing, the machine gets retrofitted. ATEX-compliant motors, sensors and extraction systems are considerably more expensive than standard components and have longer lead times. Also think of dust deposits on machine parts, since whirled-up dust can create a zone in the first place. Settle the classification early with the explosion protection specialist.

An audit trail is only as good as the login in front of it. If all operators share one password at the HMI, the log ends up saying operator and nothing more, and that helps nobody. With new machines, clarify which parameters are recorded, how long the data is kept, where it is stored and who evaluates it. At acceptance, try changing a setpoint and then look for the entry. Sometimes it is simply not there. An audit trail nobody ever reads only proves its worth after a complaint, but then it does so very well. A glance now and then at changed sealing temperatures, limits and recipes costs little.

Almost every machine supplier delivers layout drawings as DWG or DXF. Explicitly request the installation drawing in the exchange format in the tender, and not just as a PDF. Only then can the machine be placed cleanly in your hall layout and checked against material flow. A PDF can be looked at but not pushed into place. Ask for media connections, maintenance areas and door swing areas to be in the drawing as well, otherwise the machine fits on paper and no longer once the control cabinet door opens. Also specify which coordinate system and unit apply, otherwise drawings arrive at the wrong scale or rotated.

An AS/RS pays off with many small items and high access frequency on limited floor space, typically spare parts stores, small parts or e-commerce. The interface to production makes the difference. How containers reach the line, how they get back and who sets the priority are points that some projects leave open while the store itself is planned in detail. A Shopfloor Analysis often shows that it is not the warehouse but the routes between warehouse and line that set the takt. Also ask what happens when a device stops. Without emergency operation or a way round, the whole aisle stays out of reach.

AMRs pay off with changing routes, many handover points and layouts that change often. You pay for that with traffic management for many vehicles, WLAN coverage throughout the hall and clear rules for mixed traffic with forklifts and people. The dead spot behind the racking is the classic. Then the vehicle stops and waits for a job that arrived long ago. Also check how the vehicle behaves at handover points when the floor is dirty or pallets sit crooked. Route analysis comes before the investment. Without solid transport volumes per route, any vehicle count stays a guess. Also ask who clears a stuck vehicle, and how fast.

An availability promise is only as good as the measuring rule behind it. Write into the contract which stoppages count: machine, packaging material, product, operator. Also define the period of measurement and the formats it covers. One good day says little about a month with three format changes and film from a poor batch. The supplier usually guarantees only machine-related availability. Bad cups, torn film and the missing forklift driver are on your account, and the line stands anyway. So insist on proper downtime recording with cause codes from acceptance onwards. Only once the coding of the stoppages is settled does the percentage figure in the contract mean anything.

The closure must suit the contents, the bag quality and the transport, whether it is sealed, sewn or clipped. Common faults come from product caught in the seam and from uneven bag feeding. A sensor for seam inspection and clean extraction before closing save a lot of rework. Have the machine show you how the bag mouth is pulled smooth before it is closed, because a wrinkle in the seam often shows up as a leak only at the customer. Do not judge the seam by eye alone. At acceptance, ask for a leak test on bags from the running line, and use real product, because a trial run with clean goods hides the product caught in the seam.

A ball screw copes badly with dirt and side loads. The nut is built for axial force. If it also has to guide the carriage because the linear guide is misaligned or undersized, the screw will not last long. At acceptance, check that screw and guide run truly parallel and that a bellows protects the screw from product residue. On vertical axes the brake belongs in every design review, because without it the load simply runs down when the power fails. Screw and nut are matched as a pair at the factory. So ask whether nuts are available on their own or only as a complete assembly. A nut is never wound off the screw without a mounting sleeve. Otherwise the balls end up on the shop floor and the nut goes back to the manufacturer.

Paper bands save plastic and are considered easy to recycle, which makes them a topic in every multipack project with the EU Packaging and Packaging Waste Regulation, the PPWR, in mind. The limit is stackability and transport load. A multipack with a band tolerates less handling than one in shrink film, and when the pallet shifts in the lorry the bundle is the first to show it. Band width and the gluing or sealing method count in the design. Also tell the supplier whether the bundle travels in a carton or loose on a tray, as that changes the required holding force considerably. Have bands checked with a transport simulation on the finished pallet before the changeover is released.

Coatings are increasingly replacing aluminium foil and multilayer films so that packaging remains recyclable. For the line this means that sealing windows and friction values change. A film that looks the same on the data sheet runs differently. It seals at a slightly different temperature, slides differently over the forming shoulder and may stick to the jaws when warm. So test new coated materials on the machine at nominal output and under the climate conditions of your plant, in summer as well as in winter. A coating that is tight in the laboratory can leak after creasing or at the seal seam. Check the finished pack, not just the reel.

The most honest view of a plant is the load profile of a Sunday night. Nobody is producing, yet the factory often draws a surprising amount of power. Usually a compressor keeps running because the network is hissing in many places. Then there are machines that never get switched off completely, because heating up on Monday takes too long. Sometimes that is justified, often it is just habit. For new machines, ask for an energy saving state that lowers the heaters, switches off vacuum and drives and comes back up at the push of a button or on a timer. Put it in the specification, or it will not come. And settle who shuts off the compressed air to a hall on Friday. A solenoid valve in the supply line that closes on a timer does not forget.

Basic engineering produces the layout, material-flow concept, capacity balance, main component list and a cost estimate with a stated accuracy. That is the basis for the approval request and the tender. Detail engineering follows after award with foundation drawings, utility connections, interface lists and the safety concept. Keeping the two stages cleanly separated avoids detail questions being decided within the tender that would restrict competition. Mix them and you get quotations built on half-finished planning, and you pay for the clarification later as a variation order. Define in the basic stage what the supplier may design freely and what not.

Buyers like to compare only grammage and price per square metre. That is misleading, because a lighter construction using better papers often delivers more edge crush resistance than a heavier one made from recovered paper. Buy strength values, not weight. Conversely, reducing basis weight only saves money if the stacking factor in the warehouse still holds and the cartons do not buckle on the line. In a quote comparison, have the constructions disclosed: paper grades, flute type and basis weight per ply. Two offers with the same grammage can differ clearly in a stacking test. At goods-in, a spot check of basis weight and moisture pays off. A delivery at the lower end of the tolerance behaves differently on the line than one at the upper end.

ISA-88 pays off once several products run on the same plant and recipes are changed frequently. The effort lies in a clean breakdown of the phases: phases that are too coarse are inflexible and ones that are too fine become unmanageable. Clarify who is allowed to maintain recipes and how batch reports are generated. Without recipe discipline and training, even a batch control system is quickly bypassed with manual intervention again. Demand that phases can be tested individually and run in manual mode, because at two in the morning only the programmer will help you otherwise. Have recipe changes logged with version and user, otherwise nobody knows after the next complaint who changed what.

The gaps in batch management usually sit where material moves on by hand. A silo filled from three deliveries, a tub of rework that goes back into the mixer the next day, a reel of film that runs across two orders. If nobody books anything there, traceability ends at exactly that point. In an emergency you then block a whole week of product instead of one shift. Run through it once with a real batch and time how long it takes until you know which raw materials went in and which customers received it. The IFS Food or BRCGS auditors want to see such a test anyway. On new lines, the question of where batches change and where they are recorded belongs in the user requirement specification. Later it tends to be dismissed as a pure software topic, and by then the line is already built.

Batches are getting smaller, which is why changeover time often counts for more than cycle rate on new lines. A machine with a high nominal output helps little if it is rebuilt for a long time after every order. Before you enlarge batches and build up stock, cut the changeover time, for example with SMED. Then small batches become economical. Scheduling should sequence formats with similar settings back to back, so all the pouches in one format first, then the changeover. At acceptance, ask for the changeover time between your most frequent and your rarest format. The fastest variant alone says little.

Many variations arise in the no-man’s-land between two trades. That is the last metre of conveyor belt, the cable from the control cabinet to the machine, the compressed-air shut-off valve or the emergency-stop chain across the line. Mark every limit with a handover point and responsibility on the layout. With three companies involved, an interface overview sheet belongs with every order. Otherwise, when something stops, everyone points at someone else. Go through the limits at the kick-off with all trades together on the layout, and each one sees what is still missing from their quotation. Read the quotations also for what is missing: if no supplier mentions cable trays or fixing material, you pay for them later.

Everyone likes to turn the tensioning screw on a belt conveyor when the belt runs to one side. After a few weeks it is tensioned crooked and wanders off even more. Decide who adjusts the tracking and mark the basic setting. For cleaning, a quick-release tensioner is worth its weight in gold. Without one the belt only gets wiped from above, and underneath grows whatever should not grow there. Fabric belts with open edges soak up moisture, which is a hygiene problem in wet areas. When ordering, clarify how the belt is joined into a loop. A welded endless belt only fits a conveyor that can be opened sideways or taken apart for it. Otherwise the splice has to be made on site with a press and a specialist. Keep a ready-made spare belt of exactly the right length in stock.

I always ask for the BOM with the manufacturer’s designation and the manufacturer’s item number. A list that only knows the machine builder’s internal number ties you to them for every order, and even for a standard part you then have to ask the machine builder instead of the dealer round the corner. Check whether the bought-in parts comply with your plant standard, meaning drives, sensors and pneumatics. Otherwise you have one more make in the store. An electronic BOM that can be imported into the maintenance system saves more time than any discount negotiation. Watch that the status matches the plant: after commissioning, the BOM has to be brought to the as-built status.

Bin picking is mature for metal parts and stable cartons, but remains difficult for pouches, films and transparent items. On a project, the recognition rate with a nearly empty container and the time for image capture plus path calculation decide the outcome, because that time counts towards the cycle. Clarify what happens with parts that cannot be gripped, how the container is shaken or exchanged and who brings the reference parts for the test. Insist on a test with parts from running production instead of clean samples, because dirt, deformation and batch change the recognition. Define the acceptance criteria beforehand, including what counts as a failed pick.

Bio-based coatings are a topic for the future, but not yet mature for every application. Barrier values, sealing behaviour and availability vary, from batch to batch and from supplier to supplier. Ask for values at your own humidity rather than at standard climate, because many barriers lose effect in damp conditions. Pilot with one product and one line, and document machine parameters and shelf-life tests before you convert the range. Also clarify with the recycler whether the coated material belongs in the waste paper stream or the plastics stream before you market it as recyclable.

The blank is the packaging material with the greatest influence on the fault rate at the end of the line. Damp, warped sheets or inconsistent scoring jam in the magazine and tear when folded, and the machine gets the blame. Dust and burrs from the die-cut edge also build up in the magazine and at the pick-off points. Fix tolerances, storage conditions and inspection criteria in the specification, and trial every new supplier on the machine before switching. Blanks from cold or damp storage should be given time to adjust in the hall before processing. Check a sample at goods receipt for scoring and flatness, otherwise the shift hunts for the fault at the machine.

Many cartoners stop not because of the machine but because the magazine is empty or two blanks are picked at once. Check the magazine capacity against your output and work out how many minutes the machine runs without refilling. Refilling once an hour hardly anyone notices, but every ten minutes ties up one person at the machine. Watch for an ergonomic refill height, because a magazine that has to be loaded above head height will not be loaded carefully on the night shift. Warped or damp board will not run through any magazine, so the storage conditions of the blanks matter too. Double-blank detection and a level sensor belong in the basic specification.

With blister machines the costs sit in the format tooling and in film consumption. A format change usually means a new tool set, from the forming die to the punch. So check tool change time, tool price per format and lead time. A new product without a tool waits for weeks, and the tool often decides the start date. Ask about punching waste as well. Punching leaves a skeleton of film that counts as loss and has to go somewhere, and the arrangement of the cavities on the web decides how big that skeleton is. For food products, also look at the hygienic design and the cleanability of the forming station.

Blisters offer product protection and visibility but face recycling pressure because of the material mix of plastic and board. Check alternatives such as paperboard skin packs or mono-material blisters before investing in new tooling. The PPWR is changing the framework here. Tooling is tied to the format, so a wrong decision stays with you for a long time. Have samples in the new material run on your machine before you order a tool. Ask your recycler as well whether the composite can be separated. Mono-material often seals in a narrower window, and board takes up moisture when it stands in the warehouse.

Block stacking only works to first in, first out with discipline and clear block formation per item and batch, ideally with floor markings. If you mix batches in one lane, the first retrieval has you moving half the store while the shift stands by and watches. Stack height and pallet stability limit its use, and a good pack pattern pays off doubly here. Check what the bottom pallet bears once the carton has absorbed moisture. The aisle width for the forklift needs planning too, otherwise the block corners take the damage. Before building racking, ask first whether a better organised block store is enough.

Unscramblers are sensitive to bottle geometry. Light, tall or asymmetric bottles tip over, oval bottles jam in the pockets. Every bottle shape is its own format with its own pockets. Have every format tested at the manufacturer with your own bottles, including bottles from different deliveries so that the real tolerances show up. Then check hopper capacity and how the hopper can be emptied at changeover. Leftover bottles in the hopper cost time, and light bottles tend to pick up a static charge and cling to the guides. The brochure output usually applies to the easy bottle. What the unscrambler still manages with your smallest and lightest bottle is not stated there.

An Efficiency Day begins with the search for the bottleneck. You look at where products back up, where the next machine waits and which station stops the line most often. Downtime data from the MES are often enough, otherwise a stopwatch and one shift of patience will do. Then the bottleneck is relieved, right where it sits. Check the buffers before and after it, keep changeovers at the bottleneck short, and arrange breaks and shift handover so that it never runs empty. A minute lost at the bottleneck is a minute lost for the whole line. Then measure again, because the bottleneck moves. I look first at where goods are waiting and only then at the evaluation. A queue often says more than a chart.

The BCT value becomes an issue the moment you change carton dimensions or pack pattern. Taller stacks or flatter cartons increase the load on the bottom layer, and a pallet in the high-bay rack may suddenly carry a load nobody calculated. Have the value confirmed by the carton supplier, and remember that hand holes weaken the carton further. The lab value applies to standard climate and a short test time. Humidity, long storage, pallet overhang and misaligned cartons take a good deal off it, and in the warehouse much less is usually left. I therefore always ask for the value at the planned storage time and calculate with a stacking factor.

For machine projects, the requirements for foreign-body control, cleanability, the glass and hard plastic register and documentation count. Involve the quality manager early, ideally as early as the URS. If the auditor first stands in front of the new machine during the walk-round, it is too late to talk about covers. Retrofitting covers, materials or inspection points afterwards is expensive and delays acceptance. Auditors like to look at loose parts, glass and lamps, cavities that cannot be cleaned and cables above open product. Have the supplier give you the materials in the product area and a list of all glass and hard plastic items, and the register is quickly updated.

Brownfield projects need a conversion concept that records which plant keeps running, what conversion windows exist, such as weekends or factory shutdowns, and which temporary workarounds bridge the gap. Relocations, temporary conveyor technology and night work often cause the largest share of the costs, more than the new machine. On top of that, existing drawings often differ from what actually stands in the hall. A survey or 3D laser scan before planning costs less than a machine that does not fit through the door. Greenfield is simpler to plan but demands early decisions on utilities, foundations and building interfaces that can hardly be changed later.

A budget quotation clarifies, for example, whether the line costs closer to 300,000 or to 800,000 euros, and it can be obtained within a few days. It contains no binding performance data and often excludes the periphery, meaning conveyor technology, safety technology and installation. In the approval request, add a contingency to the budget quotation and agree its size with controlling. List ancillary costs such as installation, utilities and training as separate items. Get at least two budget quotations, otherwise you have one number and no range. Ask the supplier what is not included before you write the figure into the request.

Buffers are the line’s shock absorbers. They absorb short stoppages and keep the bottleneck running, on many lines the filler. Size and position should come from the downtime analysis and not from guesswork. Before any buffer discussion I look at the downtime data first: how often, how long and on which machine. The most common mistake is a buffer in the wrong place that does not protect the bottleneck at all. Depending on the line, the buffer has to protect the bottleneck from starving or from blocking. A spiral conveyor needs height, an accumulation table needs floor space, and both are settled early in the layout. Also check product protection and hygiene. A buffer that is hard to clean becomes a problem at audit.

Burst strength tells you whether a carton survives handling and the parcel carrier. It does not tell you whether the carton is stackable on a pallet. Many specifications still carry burst values from old grade tables, while the supplier has long since switched to ECT-optimised papers. Then the test certificate and the specification no longer match, and the argument about the supposed deviation starts. Decide by the load case. Assess parcel shipping and heavy single items by burst strength, pallet storage by ECT. And write into the specification which value applies when both are demanded, otherwise you get whatever suits the supplier at the time.

Most business cases contain assumptions that cause trouble later: capacity usage that is never reached, staff savings nobody implements, a ramp-up curve planned too steeply. Make every assumption explicit and name who is responsible for it. Whoever approves the request should be able to see which two assumptions move the result most. Solid actual data from the existing line, for example from an Efficiency Day, is the best basis. Calculate the do-nothing option honestly, including downtime costs and spare parts risk of the old plant. I like to have a cautious variant calculated as well and see whether the request still holds up.

Cables at handover points between trades often end up as nobody’s responsibility, and then they lie loose on the floor. Fix the routes in the layout and clarify who supplies them. In wet areas, open, sloped cable trays with wall clearance work better than enclosed ducts, where water and product can collect. For drag chains, watch the bend radius of the cables and keep spare capacity. A cable that has to be pulled in later finds no room in a packed chain and ends up lying next to it. Also clarify who documents the cables after a modification, otherwise the wiring diagram shows something different from the wall.

In an audit the first thing anyone spots is the calliper with an expired label. The real problem lies elsewhere. If calibration finds a deviation, you have to assess everything that was measured and released with that instrument since its last calibration. With a yearly interval that is a lot of product. Equipment that comes with a new machine is just as awkward: test weights for the checkweigher, test pieces for the metal detector, sometimes a torque meter for the closures. It ends up in the control cabinet or with the line leader and appears on no list. At acceptance I have all supplied test equipment handed over with its calibration certificates and enter it in the register straight away. And the fitter’s private calliper from his own toolbox has no business at a measuring point used for release decisions.

The cap sorter causes downtime more often than the capper itself. Typical triggers are caps with flash, changeovers with many adjustment points, sorting chutes that jam, and closures damaged because they circulate too long. Ask about sorting output with a margin over line output and about the number of change parts. Take caps from different batches to the FAT, including the poorer ones, because injection moulded parts vary. Clarify how and how often the caps are refilled, because an empty hopper is the most banal stoppage. Hygienically, cap feed is critical. An enclosed design and UV sterilisation of the chute are common for sensitive products.

A CAPA list tells you a lot about a plant. If nearly every action reads staff retrained, hardly anyone looked for the real cause. And without an effectiveness check nobody knows whether the training achieved anything. Many actions are technical. A sensor is to be retrofitted, a guard modified, a chute redesigned. Items like these land with maintenance or the project team, often with no budget and no date, and get extended again at the next audit. Settle early who pays for them. Before writing the user requirement specification for a replacement line, I always read the CAPA list of the old line. It states fairly precisely what the new machine has to do better, and no supplier knows that list unless you hand it over.

The CAPEX process begins long before ordering. First the need is justified, then options are compared, budget quotations obtained, the business case calculated and approval given. Tender, award, execution and acceptance follow. What is often missing is the post-completion review, meaning a look at whether the promised savings were actually achieved. Anyone who measures this plans the next project better. Build the review into the approval request with a date and an owner, otherwise nobody looks after it once the machine is accepted. POWERJO takes on CAPEX projects as project management on demand from needs clarification through to acceptance.

A request with an unclear benefit stays on the pile, even if the price is right. The word efficiency improvement convinces nobody. A list of the biggest causes of downtime with the hours per year they cost does. Back up the figures with the downtime analysis and a clean quote comparison. Agree the assumptions with controlling beforehand, otherwise the committee takes the request apart over the calculation and not over the content. And write down what happens if the investment is not made. With an old machine that often means: spare parts get scarce, repairs more frequent, and one day the line stands with no plan B.

Critical are torque control, cap feed and changeover. Ask whether the machine rejects containers with the wrong torque automatically or merely displays a figure. With screw caps, torque is a quality characteristic: too tight and the customer cannot open it, too loose and the pack leaks. Plastic caps keep losing torque after closing. So measure directly at the capper and once more later in the store. If the cap supplier changes, the torque needed can change too, even with the same datasheet. Test equipment and test interval belong in the quality plan. I have the test method demonstrated by the person who will use it day to day.

In many plants the quality data are scattered across lab computers, the checkweigher and a spreadsheet that only one colleague really understands. A CAQ system brings it together, but only if the measured values flow in automatically. If operators copy values off a display every two hours and type them in, you get pretty charts with suspiciously round numbers. On new machines, check whether the inspection devices on the line can hand over their results through an interface, and put it in the user requirement specification. Watch the return path too. If the CAQ blocks a batch, the ERP has to know before the lorry is loaded. And keep the inspection plans lean. An inspection plan the shift cannot manage still gets filled in, just not with measurements.

On packaging lines the big adjustments sit in material use with film and carton, in the energy for shrink tunnels and compressed air, and in transport. Transport depends on how well the pallet is filled. If more packs fit on the pallet or into the truck, fewer trips are needed. Such a volume optimisation reduces carbon and cost at the same time. It is usually the most convincing starting point for management, because the saving on the invoice convinces even without the climate argument. Take compressed air seriously too. It is expensive electricity, and anyone who leaves leaks in the network pays for them around the clock, weekends included, when nothing is being produced.

On a cartoner, success depends on the product infeed. If the products do not arrive clean, on cycle and correctly oriented, even the best machine produces faults. Clarify the infeed, the inserts and the format range in the Lastenheft, meaning the user requirements specification. Inserts get forgotten until a leaflet arrives folded differently and stops running. With many formats, tool-less changeover with counting scales is worth more than a higher rated speed. I always watch a complete format changeover at acceptance, with a stopwatch and with the operator who will also be standing there at night. The manufacturer’s technician is always faster.

When a cartridge heater burns out, the bore is often to blame. If the heater sits loosely in a worn or scaled bore, heat builds up in the sheath and the coil gives up. So when replacing one, take a quick look with the calipers. Heaters that have baked in can often only be driven out, so a through bore is a good idea on new sealing jaws. The second weak point is the lead. It lies against the hot jaw, gets flexed with every movement and eventually breaks where it leaves the sheath. On moving jaws, ask for heaters with protected leads or an angled outlet. The heaters in stock should be exactly the ones that are fitted. Same diameter and same wattage are not enough if the voltage or the unheated zone differs.

Poor case quality shows up first at the case erector. Damp or warped blanks jam in the magazine, the score lines crack, the bottom does not hold. Before adjusting the machine, check how the blanks are stored and what the specification says. Blanks from a damp corner of the warehouse behave differently from those used in the trial run. When buying, what counts is format range, magazine capacity, changeover time and the closing method, hot melt or tape, not the rated speed in the brochure. With hot melt, the heat-up time of the unit at the start of a shift also counts, and it is often missing from the capacity planning. I ask how long the operator needs to refill the magazine and whether that works while the machine runs. If not, the line stops with every stack.

The design follows the product. Pouches and soft products are usually loaded from above, stable cartons and bottles are pushed in from the side. When comparing options, look at the product feed, because that is where most faults occur. Ask how many formats the machine handles and how long changeover takes, how carton blanks are stored in the magazine and how far the magazine runs without refilling. A small magazine sounds harmless until an operator on the night shift has to refill every half hour. And have the supplier show what happens when a carton arrives crooked, whether the machine stops or pushes the product against the flap.

Tape or hot melt is rarely a question of price. It depends on the load unit, the climate and whether the cases are reused. Hot melt closes cleanly and tamper-evidently but needs care at the nozzles and filter. Tape is more forgiving when case quality varies. The most common fault in operation is a side guide set wrongly for slightly varying case dimensions, so record change parts and tolerances in the URS. How well the closure holds often shows only at the loading ramp or in the cold store. Try it at acceptance with cold, slightly damp cases, and then pull at the flap yourself. Knives and pressure rollers of the taping heads are wear parts and belong in stock.

The spend plan usually gets interesting in autumn. That is when controlling asks why so much of the annual budget is left and whether the money has to be carried over into next year. The answer is often the same. Final payments depend on acceptance, and acceptance depends on the punch list. Allow for that from the start and schedule the final instalment on a realistic date instead of the date you wish for. Also ask your suppliers when they intend to send which invoice. Some invoice a milestone payment only weeks later, and your plan is off again. Be wary of the temptation to bring down payments forward in December just to use up the budget. Without an advance payment guarantee your money then sits with the supplier, and the machine is still far from finished.

Ten to fifteen criteria are weighted, ranging from performance and changeover through service and spare parts to references, price and lead time. Each quotation gets points per criterion, and you agree the weighting with production, maintenance and procurement before the first quotation is on the table. Otherwise somebody tweaks the weights in the end until the favourite supplier is in front. The result is documented in a traceable way and helps when the cheapest supplier does not win the award and management asks why. As a check, I always rerun the calculation with different weights. If the ranking flips with small changes, the quotations are too close together for scoring alone.

For line projects, the contract should state who declares conformity of the overall plant: the integrator, a main contractor or the operator itself. If that line is missing, in the end nobody does, and everyone points at everyone else. This includes a risk assessment of the linkage, an evaluation of the interfaces and the safety control system, and a joint operating manual. The assessment must be repeated after major modifications, even if only one machine was replaced. For projects delivered from 2027, the new Machinery Regulation applies. Clarify early which version applies to your project, and discuss doubtful cases with the safety specialist or your lawyer.

Automatic lubrication tempts people to stop looking. That is exactly the mistake. Lines get kinked or torn off during modifications, distributors clog, and the reservoir runs empty because refilling is not in any schedule. The bearings then run dry while everyone believes they are being supplied. Ask to be shown how the system reports a fault, and whether the message appears on the HMI or only as a flashing light on the pump. With progressive distributors, a cycle switch belongs in the monitoring. Fill only the approved lubricant, and do it through the fill connection with its filter. With an open lid and a spatula, dirt goes straight into the metering pistons. Anyone adding new lubrication points should first check whether pump and distributors can still cope.

Change parts should be clearly marked, assigned to the format via recipe, and stored in a fixed location. RFID identification prevents mix-ups that would otherwise only show up as scrap. When buying, clarify what an additional format set costs and how long it takes to arrive. Also ask whether change parts are included in the spare parts package and whether you may make simple parts yourself. Guides from a 3D printer are realistic today, but only where the part has no product contact or the material is approved for it. Heavy parts need a lifting aid, otherwise every changeover becomes a test for the back. I always include the change parts in the total cost of the machine.

A change parts trolley per format, labelled and complete, shortens the changeover immediately and needs no investment in the machine. Combined with digital position scales on the adjustment axes and a checklist on the HMI, this is the first step of any SMED implementation. Missing parts show up at the empty slot straight away, ideally on return and not at the next changeover at two in the morning. Cleaned parts go back on the trolley and not onto the hall floor. The trolley has a fixed place that everyone knows, otherwise the shift suddenly looks for the trolley instead of the parts. When buying, also check that it fits through the doors and gets right up to the machine.

Changes happen on every project. The only question is whether they are documented. A simple form with what, why, cost, schedule and approval prevents disputes at final invoicing. Collect small changes and decide on them together in batches, because every change approved individually by email becomes hard to trace later. The fitter who is asked on site whether the light barrier can sit a bit higher happily says yes, and exactly that shows up later as an item on the invoice. I write every promise made verbally at the machine into a change request on the same day.

A changeover is measured honestly from the last good part to the first good part at target speed. Anything else flatters the numbers, even if the machine is blinking again after the rebuild. Improvements come from three directions: preparation, technology and organisation. Change parts and packaging materials are staged while the machine is still running. Position indicators and quick-release clamps save steps, and servo adjustment by recipe saves them altogether. A changeover matrix helps to run similar products back to back, and well-drilled teams beat any technology. A changeover assistant on the HMI guides new staff through the steps.

In many plants the changeover matrix has existed for years, only it lives in the shift supervisor’s head. He knows you do not simply run the allergen-free variety after the nut variant, and he plans the week accordingly. When he is on holiday, production planning schedules by delivery date and is then surprised by half a day of cleaning. So write the matrix down, using measured times from production data acquisition. The target values in the machine supplier’s quotation usually assume a practised fitter with a prepared format parts trolley. In the matrix, keep the conversion time at the machine separate from the cleaning, since different people are often responsible for each. And expect sales to break up the finest sequence with a rush order on Wednesday.

Changeover time is the most underestimated capacity reserve. With three changeovers a day at 45 minutes each, over two hours are lost, every day. SMED principles can often cut the time clearly without investment, including external preparation, change parts on a trolley and counting scales instead of trial and error. When buying new, tool-free changeover and stored format recipes belong in the user requirements specification. In food production the time also depends on the order of the products. A changeover matrix that puts products without allergen before those with allergen saves cleaning time without anybody working faster. Time one changeover completely, from the last good part to the first, and hold it against the figure in people’s heads.

For project managers it pays off on routine tasks. It puts meeting minutes into shape, checks a chapter of the URS for gaps and sets two quotation texts side by side. Two rules always apply. Never enter confidential customer data into public AI services, and that includes drawings and prices. And always have technical statements checked by a specialist, because the result sounds convincing even when it is wrong. Anonymise what you paste in beforehand, state what the result should look like, and check figures and references to standards against the original source. Otherwise you get a lovely text that is of no use to you.

The checkweigher watches over compliance with the Prepackaged Goods Regulation and also delivers the data against give-away. What counts is weighing accuracy at line speed, confirmed rejection and a clean evaluation of the averages. The pack must sit entirely on the belt while it is weighed, and at high rates with short packs that gets tight. Draughts and vibration from the neighbouring machine also falsify the readings. I have the accuracy demonstrated at acceptance with test packs at full speed, not at standstill. Anyone using the scale only as a reject station wastes its potential. The trend curve tells you when the filler needs readjusting before scrap occurs.

Child-resistant closures demand more from the capping machine than torque. The closure must latch on application without damage to the safety feature, and the torque window is narrower. Two-part caps have to stay properly aligned in the feed, otherwise the inner cap sits crooked and the safety feature does not engage correctly. Switching to a different closure supplier means the child-resistance test must be passed again, because it applies to the specific container and closure combination. I therefore get the torque and application force in writing from the closure supplier up front. Allow weeks in the schedule for panel testing.

Cooling water by customer is quickly written into a quotation and just as quickly forgotten. The existing chiller was sized for the machines that were there at the time, and every new one takes a little more. Before ordering, clarify cooling capacity, supply temperature, flow rate and water quality with the supplier and with building services. Glycol content and water treatment belong on the same list, otherwise the fine cooling channels in the forming tool clog up. If the supply temperature is below the dew point of the hall, uninsulated pipes sweat and drip onto machine and product. A classic is the air-cooled condenser on the roof. It gets dirty over the year, and on the hottest day the chiller trips on high pressure, right in the middle of the season.

CIP capability must be specified in the URS and the design. That includes pipe slope, dead-leg-free valves, spray balls with adequate coverage and a clean separation of product and cleaning paths. The cleaning cycle of pre-rinse, caustic, intermediate rinse, acid and final rinse costs time and utilities. Shorter cycles noticeably increase net production time, so this is where people like to cut. Then ask what it means for the cleaning result. Validate the cleaning with test methods before the line goes into regular operation, and record times, temperatures and concentrations. Otherwise nobody knows after the first failure what was actually run.

Require the circuit diagrams in native CAE format and also as PDF, with equipment labels that can be found again on cables, terminals and devices. A diagram whose numbers are not on the cabinet is worth little at three in the morning. At acceptance I always take a spot check: find a terminal in the diagram, then look in the control cabinet to see whether wire and label match. Changes made during commissioning must be updated with a date. Have the revision status shown on every sheet, so that it is clear which diagram is the valid one. The hand-corrected diagram in the cabinet is common practice, but it must also go into the file, otherwise the next electrician works with yesterday’s status.

The basic rules are quickly told. Notify every deviation in writing, document hindrances with date and impact, have extra costs priced before execution. On the operator side this means bundling change requests and formally instructing them instead of discussing them at the edge of the site. A shout to the fitter to just do it quickly turns into a disputed item later. Site diaries and dated meeting minutes cost little and help more in a dispute than anyone’s memory. A project manager who keeps changes cleanly documented often saves more in the end than they cost, and that holds on both sides of the contract.

Suppliers’ models are often simplified, and precisely in the places that will hurt later. The open guard door is missing, or the robot’s swing range, or the space for the change parts trolley. I have such areas explicitly added to the model as solid bodies, otherwise the software finds nothing there. Old halls have sprinkler pipes and ventilation ducts that appear on no drawing. Without a laser scan you find out about them while bringing the machines in. Look at the walkways as well. A gangway you have to squeeze through sideways clashes with nothing and is still useless. And the check is only valid for the data it was run on. If a supplier changes the design afterwards, the model has to go through again.

At POWERJO, Claude is the digital co-lead. It researches, creates documents, evaluates tender documents and automates office workflows. Decisions and customer contact remain with the person. The AI prepares, checks and documents, nothing more. Technical outputs are always cross-checked before they end up anywhere a customer reads them. That applies especially to figures and references to standards, because they look right even when they are not. Anyone using it themselves should work the same way and hand over the necessary documents with the task, for example the URS and the schedule, rather than a mere keyword.

The cleaning method must be fixed before purchase, because a machine built for dry cleaning will not survive a hose. Describe cleaning agents, water temperature, pressure and frequency in the user requirements specification and have the manufacturer confirm suitability. Alkaline cleaners attack aluminium, and some seals swell in the wrong agent. Work out the time needed. An hour of cleaning per shift is capacity that is not producing. Parts that come off without tools save more time than any increase in output. Also ask where residual water stays behind after cleaning, for example in hollow sections and corners, because that is where the trouble starts in the next shift.

Cleanroom packaging means defined air cleanliness, airlocks, clothing and material flow, and the machine is only one part of it. Fix the cleanroom class and the handover point early, and plan in the cleanability of all plant parts. A control cabinet with a fan inside the cleanroom is a classic planning mistake, because the fan stirs up air and particles. It is better to put the cabinet and the drives outside and let only what is necessary pass through the wall. Ask about abrasion too, since belts, sliding guides and the exhaust air of pneumatics generate particles. Also clarify how packaging materials get into the cleanroom. The carton does not simply walk through the door.

Cases that shine in the test lab under standard climate can fail in cold storage or in a sea container to South East Asia. Simulate the real supply chain. Corrugated board loses a large share of its crush resistance at high humidity, stretch film becomes brittle in the cold, hot melt lets go in the heat. The changeover is the nasty part: push a cold pallet from the cold store onto a warm ramp and within minutes there is condensation on case and label, and self-adhesive labels let go. Define the climate stages together with logistics and get sign-off under the worst-case climate, not the most comfortable one. Standing times on the way count too, in port or at the transhipment point, and the haulier knows them better than production does.

All of POWERJO’s on-demand specialists log their hours in Clockify, per client and project. This gives the client a traceable timesheet for the invoice, and deviations from planned effort show up early and not just at month end. For you as the client this means: ask before the project starts what the timesheet looks like and how often you will receive it. A short interval helps, because then the effort still matches your memory. Also agree which activities count as working time, travel time for example. That spares both sides the discussion about disputed hours when the invoice arrives.

Measure the opening torque regularly with a torque tester, long after commissioning as well. Capping clutches wear, and threads and closures vary from batch to batch. Check again after an hour’s storage, because the torque changes as the gasket settles. Agree tolerances with the closure supplier and record them in the packaging material specification and the test plan. When there are complaints about bottles that will not open or that drip, look at the measurement series first. Without it, all you have is opinion. I always ask the closure supplier which container the torque was measured on, because values without that information cannot be compared.

A CMMS is only as good as what the technicians write into it. If every order just says done, after a year you still do not know why the labeller keeps stopping. Keep the feedback simple, but insist on the cause and the parts replaced. On new plant the data transfer tends to be forgotten. The supplier sends the maintenance plan and spare parts list as PDFs, and then an apprentice spends weeks typing. Write into the user requirement specification that both must come as importable spreadsheets, with manufacturer part numbers and intervals. The asset hierarchy should match the equipment identification in the wiring diagrams, so the electrician finds the right component even at night. Break it down to assembly level. Nobody needs individual screws in the CMMS.

A low Cobb value protects against softening from condensation in a cold store and during breaks in the cold chain. In return, bonding with dispersion adhesive becomes more difficult, because the adhesive’s water cannot penetrate the surface. If glued joints open up after a paper change, ask about the Cobb value before you suspect the machine. With such complaints I look at paper and adhesive first and at the settings only afterwards. For humid environments the value belongs in the specification with a limit, otherwise every supplier delivers something different. State which side is measured, since the top and bottom of a web can differ.

Cobots are suitable for small palletising outputs, machine tending and small-parts handling. Realistically planned, they are a good entry-level automation for small lines. As output grows, the cobot can become the bottleneck, because force limiting makes it slower than an industrial robot behind a fence. The safety assessment looks at the whole application, since a gripper with sharp edges or heavy cartons can injure even with limited force. Often the cobot is combined with laser-scanner zones, fast with no person nearby and slow with a person present. I ask the supplier who signs the risk assessment for the whole cell, because the robot on its own is only partly completed machinery.

The inspection is only as good as the link behind it. The target value must come from the same order source as the printer content, otherwise the system checks against itself. Lighting, camera angle and reflections on glossy films determine read quality more than the camera does. Set the inspection scope, the response to errors and the reject confirmation in the URS, and test with real packs. At acceptance, build in faults on purpose, a swapped code, a smeared date, a missing batch number, and watch whether the pack really ends up in the reject bin. Also clarify what happens when the camera fails, because the line should not simply carry on.

Coding is where retail requirements and technology collide. The customer wants GS1-128 on the pallet, a best-before date and batch on the outer pack and a readable DataMatrix on the product, and all of it must run automatically through a changeover. Clarify early who maintains the master data and how printer, scale and ERP work together. A reader positioned after the printer stops unreadable labels leaving the plant. Plan the mounting position with care. The print only works on a calm, flat surface, and the operator has to reach the printer to change the ink. Also check what happens when the data connection fails. If the printer carries on with the old date, the goods are wrongly marked before anybody notices. In that case the line should stop or reject the packs.

Too low a coefficient of friction lets bags slide off the pallet and cartons shift within a layer. Too high brakes the film at the forming shoulder and causes drag marks and poor seals. Fresh rolls and rolls six months old run differently, because the slip additive only reaches the surface gradually. A cold roll from the store also runs differently from one at room temperature. Set a tolerance range and have it certified per batch. Ask for film against metal and film against film separately, because the first figure helps the form-fill-seal machine and the second helps the pallet. Check the coefficient again after any recipe change by the supplier.

Coextrusion delivers layer by layer in one pass. Thickness variation in individual layers is not visible on the film, though, and only shows up later as a barrier or sealing problem. Ask for the layer structure with nominal thicknesses and tolerances, and check the cross-section under a microscope in case of complaints. When changing supplier, same structure is no guarantee of the same machine behaviour, because formulations and additives differ. So I always run the new film on the machine before the old supplier is dropped. Coefficient of friction and seal initiation temperature can have shifted even if the data sheet promises the same.

The cold phase tends to get shortened because product is ready and everyone wants to see results. That backfires. Mechanical collisions and safety faults are cheaper to find without product, because a collision without product costs a bent bracket, and with product it costs the bracket and the batch. For the hot phase you need enough packaging material of every format and released product of production quality. Lab samples behave differently from line output, and you see it in the seal seam and in how the pack slides. Schedule both phases with owners in the project timeline, and settle who delivers more goods in the hot phase when the stock runs out sooner than expected.

Cold seal film ages. Over time, and faster in warm storage, the bond strength drops, and the seals then open in the shop rather than in the machine. Keep an eye on the production date of the reels in the film store, and store them cool. The adhesive is only where sealing is supposed to happen. If the print drifts out of register, the jaws press onto bare film next to the adhesive strip, and the seal looks closed but does not hold. A properly set print mark sensor is mandatory here. Check the profile of the sealing jaws as well. Worn serrations and uneven pressure across the width create channels in the seal through which air and moisture get in. Have the film supplier state storage conditions and shelf life in writing. Otherwise a reel that spent too long in a warm hall is your problem.

Collision detection is often set so sensitively that the robot stops at every jolt. Then the operator turns it off, and the gripper is wrecked at the next crash. The threshold should be set per program section, sensitive when approaching and less sensitive when accelerating with a load. Correct load data and a properly measured gripper in the robot are a prerequisite. Count every trip and show it in the shift report, or the cause stays hidden. For cases where the detection reacts too late, there is a mechanical crash protection at the flange that gives way on impact and trips a switch. It costs little compared with the gripper it saves. Ask whether it is in the quotation.

Plan weeks, not days, for commissioning a linked line. Break it into milestones for individual functions, linking, performance run and operator training. Clarify beforehand who supplies product, packaging materials, utilities and staff, because hardly anything is more expensive than a commissioning engineer waiting for cartons. A short site meeting every morning prevents idle time and shows who needs whom. Expect teething problems at the handovers between machines that each ran cleanly on their own. Commissioning ends with the trial run and the documented performance test, not with the first carton running.

In the schedule, commissioning is often one single long bar. That is not enough. By the first trial with product at the latest, the packaging material is missing, or the warehouse only has the old print version. Clarify early who supplies how much product and packaging for the trials and where the rejects go. With food, that quickly adds up to a lot of goods that may not be sold. If several suppliers are involved, one of them has to lead line integration. Otherwise each tests its own machine and nobody checks the signals between the machines. Schedule the safety function tests before the product trials, because no operators from production should work on the line while the guard doors still have open points. And write down when a stage is finished. Without that criterion everything hangs at nearly done.

The film will not run on the night shift, the crew fights with it until morning, and in the end the reel goes into the skip because the space at the machine is needed. By the next day someone wants to file a complaint, and there is neither a sample nor a reel label. So secure the evidence immediately. The label from the reel or core, a few metres of material, photos of the defect and the machine parameters are usually enough. A box for complaint samples at every line costs next to nothing. On costs, do not expect much. Many terms of delivery limit compensation to the material, and you carry the downtime yourself. That is worth knowing before you order and worth looking at with a lawyer. And evaluate complaints regularly. Sometimes the suppliers with the most cases are not the ones people grumble about loudest in the morning meeting.

A composite film gives you barrier, sealability and printability, but it is hard to recycle, which is why the PPWR puts it under pressure. When switching to mono-material, sealing window, stiffness and running behaviour all change. If the outside is the same plastic as the inside, the film sticks more easily to the sealing jaws, because the outer layer no longer gives any heat protection. Plan the switch as a project with machine trials and not as a purchasing decision, and involve the film supplier early. I insist on trials with a whole reel and not a metre-long sample, because problems in web running and sealing often show up only after a while.

In many plants leak detection runs as a campaign. Once a year someone comes round with the ultrasonic detector and hangs a tag on every leak, and six months later the tags are still hanging there. Finding leaks is the easy part. It gets expensive because repairs often need a stoppage and nobody schedules the work. So enter leaks as normal maintenance orders, with priority and due date. Searching is easiest during a break or at the weekend, when the network is under pressure but no blow-off nozzles or valves are working. Many leaks sit in the machines themselves, on hoses in drag chains or on valve terminals. There it helps to shut off the machine’s compressed air automatically during standstill.

Compressed air networks grow with the plant, and rarely to plan. Every new line gets hooked on somewhere until the pressure at the far end of the hall is no longer enough. The usual answer is to turn up the pressure at the compressor for the whole plant. That costs energy every day and raises leakage losses, and nothing is solved. On new lines I have a shut-off valve, a pressure gauge and a flow meter fitted at the transfer point. Then you can see whether the problem lies in the machine or in the network, and you can isolate the line for work without shutting down the neighbours as well. Check the cross-section of the supply line before a large consumer is added. A bottleneck in the last length of pipe cannot be fixed by any compressor in the world.

Poor compressed air shows up as water in the valve island, gummed-up valves and oil on the product. Ask the machine supplier for the required class and compare it with what your network actually supplies, including summer values at high humidity. Dryers and filters belong in the project budget, condensate drainage in the maintenance plan. Treat product-contact air separately. On a walk-round, look into the condensate drains and filter bowls. If you find a cloudy brew there, either the treatment is wrong or nobody takes care of it. When a single machine keeps having trouble, check the last stretch of pipe between the network and the machine before you touch the whole network.

Oil-free or oil-lubricated is rarely a purely technical question in food plants. Quality assurance wants to know which class to ISO 8573-1 arrives at the point of product contact, and that depends on air treatment and its maintenance as much as on the compressor itself. So for compressors I always ask for a quotation for the complete station, with dryer, filters and controller. Two speed-controlled compressors without a common controller like to fight each other and both run at part load. Idle time costs electricity without delivering air. Take a look at the compressor room too. If it is hot and dusty in summer, the machines draw in warm, dirty air and trip. And plan for redundancy. If your whole production hangs on a single compressor, you usually discover the downside on the night shift.

Sensible starting points are heavily loaded drives, bearings at sealing stations, vacuum pumps and compressors. Many sensors already deliver diagnostic data via IO-Link anyway, and the skill lies in setting limits and alarms. A simple trend dashboard with current draw and temperature per unit often reveals wear early enough to move the replacement into a planned shutdown window. Decide beforehand who receives a warning and what they do then. A message that nobody processes is just one more alarm. Only once data and response processes are in place does the step to predictive maintenance pay off. Start with a few measuring points whose failure is really expensive.

Connected loads in quotations are cautious figures. The supplier adds up everything in the cabinet and rounds up. If the electrical planner then totals every machine in a new hall, the result is a transformer that will never be fully loaded. The opposite case costs more. Your new line just fits into the sub-distribution board, but the plant is already at the limit of its grid connection, and an upgrade by the grid operator takes time the schedule does not allow for. So at the FAT I have the supplier show me the actual power drawn in production mode, in addition to the figure on the rating plate. Check early how much reserve the main distribution board and the transformer still have. That question belongs in layout planning. In the week before installation it is too late.

On packaging lines consignment stock is mainly interesting for wear parts and change parts. The parts are within reach, and you do not have to advance any money for them. The small print sits in the contract. Settle how withdrawals are reported, who carries out the stocktaking and when, who is liable if a part is damaged or goes missing, and what happens to the remaining stock when a line is shut down. Also decide who sets the minimum stock. Otherwise the shelf is empty when a sealing jaw fails at night, and nobody feels responsible. And keep access tight: whoever may go to the rack also records what they take out.

Suppliers’ connection values are rarely wrong, but often not comparable. One gives the peak, another the average at rated output, a third the sum of all fuses. Simply add them up and you end up with a transformer that is far too large. So ask for nominal consumption and peak demand separately, and for compressed air always together with the pressure the volume refers to. States outside normal production tend to be forgotten. The shrink tunnel draws the most power while heating up, and cleaning needs water in quantities that do not occur in normal operation. Get the list early. Compressors and supply lines have their own lead times, and connecting them often needs a shutdown in the plant. Otherwise the new line stands there finished, waiting for its compressor.

Contactors switch away quietly for years until the contacts are burnt or one welds shut. Wear depends on the number of operations and on the load. A motor that is jogged under load all day wears out its contactor far faster than one in continuous duty. If a main contact sticks, the drive keeps running although the control has switched off. That is why safety functions of higher categories use two contactors in series, with their mirror contacts wired into a feedback circuit. For a replacement, coil voltage, frame size and auxiliary contacts all count. One with the same rating but an ordinary NC contact instead of a mirror contact appears to work perfectly. The safety function, however, is no longer the one that was validated. And a contactor between inverter and motor should only switch while the inverter is not putting out voltage.

A reserve hidden within the budget is used up at the first extra. Show it openly and set out in the project manual who releases it. In packaging projects it tends to flow into the same few holes: the interface to existing conveyor technology, additional formats, floor remediation under the old machine, a longer commissioning period. I have the amount derived from the risk register, each risk with its probability and cost impact, and only then look at the total. A flat percentage does not survive the first question from controlling. Whatever is left unspent at the end, show openly in the closing report.

A CIJ printer is undemanding while running and troublesome when stopped. Clogged nozzles after longer downtime, solvent consumption in warm halls and condensation on chilled bottles are the usual issues. Check ink adhesion on your actual surface, especially on damp containers. The print position must be reproducible at changeover, and a fixed print-head holder with a scale saves arguments. Count ink and solvent as running costs, and have the flushing routine for the weekend explained to you, because a printer that is clogged on Monday holds up the early shift.

The contour check ahead of the palletiser or stretch wrapper stops a shifted layer from blocking the plant or jamming in the racking. A light curtain or camera checks overhang and height. The measurement is the easy part. Responding to a deviation is harder, because you have to define who stops, who corrects and how the pallet is recorded. Without a defined procedure, faulty pallets simply get acknowledged and carry on. At acceptance, test with a deliberately crooked pallet to see whether the plant really stops. Also check whether hanging film or a loose strap can fool the light curtain.

Under a contract for work you buy a result, namely the running plant, under a contract for services effort, namely an expert’s hours. For interim experts the contract for services is usual, for machine supply it is the contract for work with acceptance. It gets awkward when an expert comes in under a contract for services and is then managed like an employee, with a fixed place in the shift plan and instructions from the foreman. Then paper and practice no longer match. A clear task description, own work equipment and no integration into shift operations help. The contract has to say what actually happens on the shop floor. Discuss it with your lawyer before the first expert starts.

During a modification in a running plant there are soon more companies on site than anyone has on their list. The machine supplier brings a subcontractor for the pipework, who in turn brings a colleague for the insulation. That is why I ask every contractor in advance for the names of all fitters, including those of their subcontractors. Anyone not on the list and not inducted does not get onto the premises. The induction has to be given in a language people understand. A polite nod during a briefing in German does not count. What has proved its worth is a short round at the site every morning, in which each company says where it is working that day. Then you notice in good time that someone is welding up top while someone else is storing film underneath. Tell the shift management who is on site, too. The people on the line usually meet the fitters before the project manager does.

Control cabinets mostly cause trouble through temperature. The typical picture is the frequency inverter in an unventilated cabinet right next to the shrink tunnel. Ask for a heat dissipation calculation and air conditioning that suits the shop-floor environment. Dust, heat and the way the hall is cleaned belong in that picture. Demand spare space for future expansion and enough free terminals. Make sure the labelling and equipment identification in the cabinet match the circuit diagram, otherwise troubleshooting at night turns into guesswork. Look at the mounting location before you sign. The door has to open during operation, even when the next cabinet already stands beside it. In hygienic design the cabinet belongs outside the wet area, otherwise the cleaning works against the electrics.

Sooner or later in summer, many plants have a cabinet door standing open with a pedestal fan in front of it. That cools for a while, but it pulls dust, flour or sugar dust into the cabinet, and the IP rating is gone. Filter fans need clean mats, otherwise they only run for the noise. In wet areas with foam cleaning they have no business at all. With cooling units, keep an eye on condensate. If the setpoint is too low, the cabinet gets damp because the air is cooled below its dew point. A door switch that stops the unit when the door is open is therefore always part of the package. Ask the supplier for the heat balance of the cabinet and the ambient temperature it was designed for. The figure often suits his assembly hall but not a platform right under the hall roof.

Anyone who does not write the reserve into the tender gets the tightest cabinet that just meets the standard. I therefore state in the enquiry how much free mounting area, free terminals and digital and analogue channels must remain, plus a thermal reserve. Check this at the FAT. Reserve on paper is often already claimed by change order number three. Besides space, cooling capacity counts. One extra frequency inverter can quickly push a tightly packed cabinet to its thermal limit. Look at the power supply and incoming feed as well, because free terminals help little when there is no power left for the next consumer.

Often a backup lies somewhere, only not of the current state, not of every component and not tested. Without saved parameters, someone sits at the machine with the manual after an inverter swap and guesses values. Define who backs up after every change, where the backup is kept and who may restore it. The restore should be rehearsed once a year, ideally during a planned shutdown. I have every supplier hand over the state after acceptance and note in the protocol where it is kept. Also check whether the backup can be restored without the supplier’s laptop, because otherwise passwords and programming software exist only there.

Rarely is a controls migration a straight like-for-like swap. The old programme has grown over years, is hardly documented and is full of special cases known only to the night-shift supervisor. So talk to the people who stand at the machine every day before anyone writes a quotation. Plan a software inventory before quoting, a simulation test before the downtime window, and a fallback plan with the old CPU and its programme state in stock. Have the I/O list rebuilt before anything is wired, because deviations between drawing and cabinet otherwise show up only at start-up. Do not carry over safety technology unchecked. That is usually the point where a migration turns into a retrofit, with a new risk assessment and everything that comes with it.

Designing conveyor technology is about product care, accumulation behaviour, cleanability and modularity. Accumulation roller conveyors decouple without pressure on the product, belt conveyors are cheap, modular belts are hygienic. Buffers are sized in seconds of run time. What counts is how long the downstream machine may stop before the upstream machine has to stop. That figure belongs in the user requirements specification. I ask for the buffer time at each handover in a table and hold it against the usual stoppage duration of the downstream machine. Plan the cleaning in from the start, because a belt that cannot be removed never gets properly clean in a hygiene zone.

Tracking eliminates stoppers and singulation stations and increases output, but requires clean mounting of the encoder and an accurate calibration. Slip between belt and product, for example from condensate or oil, leads to missed picks. A stretched belt or a shifted encoder also moves the hits slowly, until someone finally notices that the products are no longer picked in the middle. Document the hit rate at maximum belt speed at acceptance. I have the test run with a belt stop and restart in the middle of operation as well, because in daily life the belt stands still from time to time and starts up again.

The corona treater itself usually sits at the film manufacturer. You still feel its effect as soon as your coder or thermal transfer printer prints on PE and the code can be wiped off with a thumb. Then it pays to look in the warehouse. Rolls that spent a long time at the back of the rack have lost part of their treatment, and FIFO is more than a formality here. Also check whether the treated side actually faces the printhead. If the roll arrives with the wrong winding direction, you are printing on the side that was never treated. A set of test inks costs little and settles such questions at goods receipt within minutes. Write the required minimum value into the packaging specification, together with the point in time at which it applies. Otherwise the supplier measures ex works and you measure after months in storage.

Corrugated board is the workhorse of secondary packaging. Flute type, grammage and paper quality determine stacking strength, machinability and price, and the supplier likes to adjust one of these three when the price is under pressure. Damp storage reduces strength significantly, which you notice particularly in cold stores or at the loading ramp in autumn. When changing supplier or quality, always test samples on the case erector. The same FEFCO number does not mean the same running behaviour. Flat sheets, clean scoring and constant thickness only show in a long run, not on the data sheet. Run at least one full pallet in one go before you release.

The grade designation is a minimum requirement and not a description of the material. Two suppliers can meet the same grade with completely different papers and area weights, and the cartons then run differently on the erector and capper. A lower price often comes from one supplier hitting the minimum values narrowly while the other delivers with reserve. In the quotation comparison matrix that looks identical and is not. I therefore ask for the paper combination in every quotation. Alongside the grade, fix the flute type, total area weight and paper combination, and require first samples for a machine trial with every supplier change.

For acceptance of filling and dosing processes, Cpk is calculated from a sample fixed in advance. Write into the contract how large it is and for which product and format the measurement is taken. A Cpk below 1.0 means regular tolerance violations, so reject or give-away. Above 1.33 the process is considered under control. The value depends on the product. Powders, granulates and liquids behave differently, so the check per product group belongs in the acceptance. Measure only once the process has settled, not straight after start-up, otherwise the spread looks worse than it is in continuous running.

Anyone sourcing granulate, powder or raw materials from the chemical industry receives CP pallets, and they rarely fit racking and conveyor technology planned for the Euro format. Clarify early which pallet types goods-in, storage and depalletisers can handle, and plan for runner orientation and deck-board spacing. The problem usually shows first at goods-in, when the forks do not fit the entry openings or a deck board gives way. Also include disposal or return in the calculation, because CP pallets do not run in an exchange pool. Otherwise that item appears in no cost sheet, and the disposal firm sends its invoice anyway.

In packaging projects the critical path almost always runs through the lead time of the main machine, the change parts and the release of packaging materials. Assembly on site rarely does. Supplying packaging samples, product samples and format lists early keeps the path short, because without them the supplier can neither design nor order change parts. Check the supplier’s schedule for float. A plan without float is more of a wish. After every schedule change, check whether the path has moved. The activity that still had float yesterday may be the tightest today. Ask which bought-in parts have the longest lead times and when they are ordered. The order date says more than the delivery date.

A cup filler’s output stands or falls on destacking. Cups that have gone oval in storage or transport, or that stick together, cause most of the stoppages. Have cups and lidding foil from the actual supplier run at the FAT, not sample material, which is usually hand-picked. Also check whether the sealing head matches the foil thickness and the cup rim area, and how product splashes in the sealing zone are prevented, because a single drop on the rim makes a leaking seam. And ask which parts are swapped and which are only adjusted when the cup shape changes.

The line’s cycle time is set by the slowest link, usually the filler or the cartoner. All other machines should be able to run faster so that they can catch up after stops, otherwise product backs up at every small halt. When designing new lines, rated cycle, continuous output and peak output must be kept clearly apart. A machine that only reaches its rated cycle for one format has been wrongly specified. Ask the supplier for which format and which packaging material the cycle applies, and have the other formats confirmed one by one. The catalogue cycle likes to stand large on the first page, with the format in the small print.

When the print mark wanders and the bag length varies, people tend to look at the print mark sensor. Often the cause sits a few rollers further upstream, at the dancer. A stiff bearing on the pivot arm, a pressure regulator that every shift turns a little, or a worn position sensor is enough to make the web tension pump. Watch the dancer while the machine is running. Calm in the middle is good. Constant swinging from one end stop to the other means the control cannot keep up. Thin film tolerates less tension than thick film, and too much tension stretches it until print and cut no longer match. So write the cylinder pressure for each film into the recipe, or at least clearly onto the regulator. During maintenance, take the dancer in your hand once. It has to move easily and without jerking.

Under DAP the manufacturer’s delivery obligation ends at the named place, usually in front of the hall. After that the work only starts for you. Unloading with a crane or heavy-duty forklift, bringing the machine in through doors and aisles, and aligning it on the foundation must be agreed separately, often as a distinct line item in the quotation. Also clarify transport insurance up to handover and who bears damage during unloading. Check door height, floor load capacity and access before the lorry sets off. Settle who guides the delivery on site, because the driver does not know your premises. A low-loader stuck in front of the factory gate costs waiting time and nerves.

Without a data model, the cause of a stoppage is called error code 17 on machine A, film fault on machine B, and nothing at all on machine C. Require a signal list in the URS with name, unit, cycle and meaning, based on PackML or the Weihenstephan standards. This costs the supplier little and saves you the translation work for every dashboard later. Agree a common catalogue of stoppage reasons right away, so that equal causes get equal names. And make sure units and timestamps are uniform. A speed in revolutions per minute on one machine and per second on another falsifies any analysis.

Under the GS1 Sunrise 2027 programme, tills worldwide are to be able to read 2D codes, so the switch from the EAN barcode to a 2D code on the pack is foreseeable. For the line, this means printers with sufficient resolution, camera verification of code quality and a data connection for variable content such as batch and best-before date. Anyone buying coding systems today should require 2D capability and grading to ISO/IEC 15415. Find out whether the system builds the GS1 data structure with its separators correctly by itself or whether the code has to be assembled by hand. The error otherwise only shows when the scanner at the customer cannot separate the fields.

Decide who on your side is allowed to order daywork. Otherwise the shift supervisor asks the fitter whether he could just rebuild the chute quickly, and at the end of the month it turns up on the time sheet. Sign time sheets daily or at least weekly, while you still know who was there and when. A sheet that says eight hours installation helps nobody. Insist that the work is described so that a third party can follow it. Waiting time is a frequent cost driver. If the fitter is waiting for power, compressed air or the scaffolding, the clock is running at your expense. So prepare the site before the people arrive. Agree beforehand how travel time and waiting time are charged. That is often in the small print of the quotation.

If you want a decision, make it easy for the committee. Put the question and your recommendation at the very top of the first page, worded so that it can be answered with yes or no. The reasoning comes afterwards. Many papers do it the other way round and hide the recommendation on page seven. Offer genuine alternatives. Any management team spots one good solution next to two obviously unusable ones straight away, and then the discussion turns to your credibility instead of the matter itself. The do-nothing option belongs in there as well, together with its consequences. Write down what a postponement costs, too. If the order goes out a few weeks later, delivery often slips too, and the conversion no longer fits into the planned shutdown window.

The Declaration of Conformity is not a form the supplier just includes. It is the legal outcome of the risk assessment. Require it in the URS as a condition of acceptance, together with operating instructions in German and the technical documentation. I always compare type and serial number in the declaration with the nameplate, and I check that the standards listed fit the machine and do not come from the template of a completely different one. For lines made up of several machines, it must be clarified who declares overall conformity, the integrator or the operator. This question is easily overlooked. After a rebuild, the old declaration does not automatically stay valid.

Deep learning pays off where classic machine vision fails: with natural products, glossy surfaces and packaging with changing print artwork. The problem is the fault images. There is no shortage of good parts, while the rare faults turn up every few weeks at best. Ask the supplier who collects and labels the images, how retraining works and who approves a change. I also ask to see what the system does with an image it has never seen. For dimensional checks and presence checks, a fixed rule is usually simpler, and it is easier to explain when quality management or an auditor asks questions.

A spare actuator on a cable tie in the control cabinet says more about the machine than about the operator. As a rule there is a fault behind it that can only be cleared with the door open, or the machine takes minutes to restart after every opening. Then the shift finds a way. A formal warning on its own achieves little while the reason remains. Go to the machine with the operator and have them show you what they need the door open for. Often there is no setting mode at reduced speed with a three-position switch, or a hatch is in the wrong place. With new equipment this belongs in the acceptance test. Have typical faults such as a film break or a toppled cup cleared in front of you, without any tricks. If that cannot be done, the machine is not finished.

People rarely argue about the defect. They argue about the letter in front of it. At the end of a FAT the supplier tends to see everything as C, while production sees everything as A the moment it gets annoying. That is why I write into the acceptance criteria what we mean by A, B and C, with a few examples from this very machine. Anything that touches safety is A, and that is not up for negotiation. A sensor that triggers a false stop every few hours looks harmless in the report and costs time on every shift. I do not grade that as C. Put a name and a date against every item, otherwise the list looks exactly the same six months later. And tie the final payment to the B items being closed. In my experience that speeds things up more than any email.

Delta robots achieve very high pick rates with light products, but need clean product feeding and good lighting for the vision system. Hygienic designs with IP69K, stainless steel and open construction are usually demanded in the food sector. Costs arise less in the robot than in the gripper, vision, conveyor integration and safety cell, so always compare the whole cell and not the robot price. The working envelope is limited, and the products have to pass through the right window. Draw the layout with the real belt widths and the real product spacings before you fix the frame size.

For depalletising, layer pattern, interlayers and product stability decide the technology: vacuum layer grippers for jars and cans, clamp grippers for cartons, robotics with vision for mixed pallets. Handling interlayers and empty pallets is a separate process step with its own space requirement. Anyone who forgets it in the layout ends up with forklifts in the line area later. Look at the pallets that really arrive. Crooked layers and slipped wrap from the supplier’s plant are everyday business, and a machine that only copes with sample pallets keeps stopping in its first month. Before ordering, I have photos of delivery pallets from recent weeks shown to me.

Approval does not mean taking on responsibility. Write into the contract that checking the documents does not limit the supplier’s warranty, and settle the wording with your lawyer. Focus your check on what you know better than the supplier: placement, utilities, operating side, hygiene zones and existing conveyor technology. I draw up a short checklist for this and do not go through every screw. Make your comments directly on the drawing, with the revision status, so that they can be found later and the supplier cannot overlook them. Also set a deadline for the check. An approval package that sits on a desk for weeks becomes your delay in the schedule.

A freeze is quickly declared and rarely lived. After the cut-off date the customer still sends new packaging samples, the supplier swaps a drive because of lead times, and nobody calls it a change. So before the freeze, check that all formats, packaging materials and interfaces have really been sampled. Without samples, the freeze is just a date on paper. Put the date in the schedule and announce it where people see little of the project, in marketing and procurement. Experience says the late wishes come from there. Have the supplier confirm in writing, too, that a substitute type chosen because of lead times counts as a change.

A design review on the 3D model saves steel. This is where you notice that the motor behind the guard cannot be removed, or that the change parts can only be reached with a ladder. On screen that costs a click, on the finished machine a rebuild. So invite operators and maintenance staff as well as the engineers. They also notice that the cleaning hose does not reach the rear corner and that the operating side faces the wrong wall of the hall. Keep every point as a list with a deadline and a name, and check at the next review what was implemented. Anyone who waits until the FAT finds the faults on a machine that is already built.

Even the neatest planning board usually lasts until the first serious breakdown. If the cartoner stops on Tuesday morning, the plan for the rest of the week is waste paper. Then you find out whether someone can rework it quickly or whether the shift leaders start replanning on their own. Plan with the output the line really achieves on average, not with the nominal output from the data sheet. Otherwise every plan is too ambitious from the start, and the gap shows up on Friday in the late shift. The changeover sequence often achieves more than any software. Light to dark, allergen-free before allergen-containing, every line leader knows that, and it still belongs in a changeover matrix the planner can see. If the entire schedule lives in a spreadsheet on one colleague’s computer, feel free to ask what happens during the holidays.

Abrasion and broken-off corners from scrapers, guide rails and screws end up in the product. A detectable grade means the inspection line finds them. Check, though, whether your detector actually recognises fragments of the relevant size, and include the parts in the test routine. Ask the supplier which test piece was used to prove detectability and in which product. Detectable plastics are mechanically somewhat weaker, so coordinate the design with the design engineer. Blue alone proves nothing. Standard plastics come in blue as well, and no detector will find them.

Whoever pays for the die-cutting tool should also own it, otherwise procurement gets stuck when changing supplier. How the contract words this is best clarified with a lawyer. Record tool number, revision status and storage location in the packaging material specification. After changes to scoring or dimensions, a new approval sample has to run on the machine. A worn cutting rule shows up first as fraying edges and dust in the erector’s magazine, not on the drawing. The ejection rubber ages too and eventually stops pushing the blanks out cleanly. I manage die-cutting tools like machine parts, with a number, and ask regularly where they are right now.

The digital twin makes sense in three cases. First, material flow simulation before layout decisions. Second, virtual commissioning of the control system on complex interlinked lines. Third, comparing variants for rebuilds during ongoing operation. POWERJO uses a material flow model in visTABLE in its Shopfloor Analysis to put numbers on routes, loading and floor space before anyone decides on conveyor technology or AGVs. Count the effort honestly. A model nobody maintains is waste paper after the first rebuild, and the results are only as good as the cycle times you feed in. Ask beforehand who owns the model and whether the supplier hands it over. For a single machine, a proper trial run at FAT is usually enough.

Worker guidance pays off with many variants, changing staff and manual assembly or packing steps: projection or a screen shows step by step, sensors verify. Its benefit stands or falls on maintaining the work instructions. Without an owner, the system goes out of date after the first product change. Define who changes and releases content and how quickly a change reaches the workstation. Pictures and short videos work better than long texts in mixed-language teams, because they need no translation. Also plan for failure, because if the screen stays dark at night, the shift needs a paper version. I involve the workers before the start, because instructions written past them tend to get ignored.

Direct thermal saves ribbon and changeover time but costs legibility. A pallet label that hangs for months in a high-bay warehouse or lies in the sun at the gate is often no longer scannable afterwards. For traceability over the storage period, use thermal transfer. Put a sample on a windowsill in the sun for a week and scan it afterwards, which tells you more than any datasheet. Also look for abrasion-resistant, surface-coated grades and for freedom from questionable colour developers such as bisphenol A, which is restricted in thermal paper in the EU. Ask the label supplier about it explicitly.

Displays have short runs, many variants and tight retail deadlines. They are often assembled by hand or semi-automatically, and automation only pays off for recurring base formats. Agree graphics and pallet dimensions with the retail customer before production runs. A display that looks good on the pallet but collapses when unpacked in the store, or that fails on the pallet height of the retailer’s warehouse, costs you the promotion. Allow time for print approval and building a sample, because that is where delays often arise. And settle who sets up the display in the store.

Documentation not specified in the purchase order does not arrive, or arrives late. Set out the language, file formats, number of sets and timing, plus the as built revision status. Ask for PDF and editable formats as well, otherwise you can neither extend the wiring diagrams later nor adapt them to your plant standard. Suppliers like to forget the PLC programme with source code and the drive parameters, until someone asks for them after a replacement. Withhold part of the payment until the documents have been checked. In the Tender Package, the scope belongs in as a separate annex, so that all bidders quote the same thing.

Manufacturer figures for dosing accuracy usually apply to ideal products at medium output. Agree in the functional specification the product, output rate, pack size and the test method used to verify accuracy at FAT and SAT. Your sticky or foaming product shows what the machine can do, not the supplier’s test product. Before ordering, work out what one gram of overfill costs per year at your volumes and raw material price. This often justifies a better dosing system more than any increase in speed. I always run that calculation before the first talk about machine output.

For special-purpose machines, 30 per cent down payment, 60 per cent on delivery or FAT and 10 per cent after acceptance is a common structure. Negotiation settles the split. The higher the down payment, the more important the advance payment guarantee, because if the manufacturer becomes insolvent the money is otherwise gone. Check the guarantee for amount, term and return conditions. A guarantee that expires before delivery does not help. Also check whether the down payment can be tied to a milestone with proof, for example approval of the functional specification or the material order. I do not transfer anything until I hold the original guarantee.

The downtime analysis is the starting point of every efficiency improvement. For two weeks every stop is recorded with duration and reason, then the list is sorted by frequency and duration. Mostly a few causes generate the bulk of the losses, and most of them can be fixed without investment. In the Efficiency Day, we do exactly this in concentrated form, on site, in a single day. The catch lies in the recording: where operators type reasons freely, the same thing ends up four times in four spellings. A short catalogue with fixed reasons solves that. And the operator must see that his entries achieve something, otherwise he soon types nothing but Other.

Anyone who knows their downtime cost per hour decides faster on spare parts, maintenance contracts and redundancy. Work out the figure honestly once and post it by the line. It ends many arguments about whether a spare motor in stock is too expensive. Calculate separately for the bottleneck and for the other machines, otherwise the figure is too high in one place and too low in the other. Count the time to full speed after the restart too, because the line does not run at full output straight away. I also always ask whether the customer threatens a penalty for late delivery. It appears in no cost centre of the line and still arrives.

For the line, the doypack is demanding. The pouch must be opened, filled, de-aerated and sealed cleanly, and a zipper or spout brings extra stations. Premade pouch machines are flexible on formats, while form-fill pouch machines that form from the reel are more material-efficient at high volumes. Check the recyclability of the pouch film early, because mono-material doypacks have narrower sealing windows. Product in the zipper is the classic: one grain of powder in the closing track and the pouch never closes properly again. Look at how clean the sealing zone stays after filling during the trial run.

The classic error: the machine is built to revision C, the documentation shows revision B, and years later maintenance orders a spare part from the wrong drawing. Then the shaft does not fit, and nobody knows why. At acceptance, require a drawing set as built with a revision list. Also require that the handwritten changes from commissioning on the wiring diagram are entered within a fixed deadline. At acceptance I check a few components in the machine against the drawing status. If the index is already wrong there, it will not get any better once the machine is in daily operation.

Drives are sized for what is in the user requirement specification. If a request for more cycles or a heavier format comes along later, the drives are often already at their limit, and nobody remembers how tight the calculation was. So I have the sizing reports for the axes handed over with the documentation. It costs the supplier nothing, since the calculation has been done anyway. Things that tend to be forgotten are the friction of product on an accumulation conveyor, the stiff grease in a chilled room and the modest efficiency of a worm gearbox. You will need the figures at the latest when replacing discontinued drives. Without them the new motor gets ordered one size bigger to be on the safe side and then no longer fits the space or overloads the old amplifier.

The drop test reveals weaknesses that are not visible on paper: insufficient corner cushioning, products that shift inside the carton, closures that pop open on an edge drop. Always test with the production product and production packaging, not hand samples. Anyone shipping via parcel networks should test with realistic drop heights and multiple drops, since a single flat drop says little about the last mile. Also try a carton from a cold store or from damp storage once, because it falls differently from a dry one from the office. Photograph every damage and record which packaging variant was tested, otherwise you start the test again from scratch at the next change of packaging material.

Displays on Düsseldorf pallets are popular with retail customers but a special case for the line. Palletisers, stretch wrappers and conveyor technology must handle the small format reliably, and two half pallets on one Euro pallet require precise positioning. Clarify height, weight and stability in store with the customer before the display is fixed. Otherwise a display looks fine on delivery and tips over in the shop at the first knock. Also check whether your warehouse can store half pallets at all or whether they belong on carrier pallets, because otherwise they do not sit securely on the beams.

Shipping packaging for e-commerce needs different strengths than retail packaging. Drop and climate tests count rather than shelf appeal, and close-fitting cartons rather than filler material. Automated carton-sizing systems pay off at high shipment volumes, below that a few standard sizes and a clean packing process help. Send trial parcels through the real parcel network and do not rely on the drop test in the laboratory alone, because a carton that passes the test is still handled differently in the sorting centre. Watch the ratio of air to goods in the carton, as you pay for the air too.

Early payment discounts on machine deliveries are rare but negotiable, especially on down payments. Weigh it against your financing costs. If two percent is granted for paying a few days earlier, the annualised rate is high, often more than the loan costs you. More important than the discount on investments, though, is the payment structure. A large balance held back until after acceptance protects you better than a small deduction. Also check when the period starts, at receipt of the invoice or at the invoice date. A discount that accounts payable keeps narrowly missing achieves nothing.

Earned value only calculates as well as the progress measurement behind it. With a few large suppliers, that measurement is the real work, because paid milestones are not progress as long as the machine is not running. Define measurable stages for each work package, for example drawing approved, material ordered, pre-assembled, FAT passed, and give each stage a fixed share of the value. The supplier’s feeling that they are nearly done does not count. That way you see early whether spend and performance are drifting apart, and do not find out with the final invoice. For large components, check on site whether the reported status is true.

Edge devices are installed quickly and then quickly forgotten. No updates, no backups, nobody responsible. After a few years there are dozens of unmaintained computers in the control cabinets, and nobody knows any more what runs on which. So define beforehand who operates the devices and how they are monitored. Ask about the heat in the cabinet, because a computer without airflow will sooner or later stop computing in the summer. Also keep it cleanly separated from the control network. With every edge solution I ask for a list of device, software version and contact person. Edge pays off where images or high-frequency signals are involved.

When pallets sag at the bottom in storage, it’s usually the case, moisture or overhang, rarely the pallet. Have the ECT verified per delivery in the test certificate, not just as a grade specification. The value applies to standard climate. In cold storage or tropical humidity, corrugated board loses considerable crush resistance, and that belongs in the stacking factor. Then there is time: a pallet that stands under load for weeks holds less than the short test value suggests. Handle and ventilation holes, creases and damaged edges also cost strength. The same grade designation can hide a different paper, as long as the minimum values are met.

Automatic edge-protector applicators on the wrapper pay off at high volumes. If the protectors are fitted by hand, someone will forget them sooner or later in shift operation, and the goods arrive with crushed corners. I always have the transport test check whether edge protection can cut the amount of film. The combination of edge protectors and fewer wraps is often cheaper than more film. Board profiles lose stiffness when damp, so keep them dry in stores and away from the ramp. Also match the profile length to the height of the load unit, otherwise the ends stick out or the top cases are left unprotected.

Motors with a better efficiency class are not automatically a one-to-one replacement for the old ones. They have less slip and turn slightly faster. On a pump or a fan the new motor then delivers more and eats up part of the saving straight away. When started direct on line, such motors often draw a higher starting current, and the old motor protection circuit breaker may trip on start-up. Check the overall length too. Some motors are longer in the same frame size and no longer fit under the guard. The better class really pays off on motors that run many hours a year under load, such as pumps, fans and compressors. A conveyor drive that stands still a lot between cycles recovers the extra cost much more slowly.

EHEDG guidelines cover design details that determine cleanability in everyday operation: weld seam quality, sealing materials, surface roughness, avoidance of gaps and dead spaces. For the tender, a reference to the relevant EHEDG documents and a requirement for certified components in the product area is often enough. The certificate applies to the component, not to the installation. A certified fitting that is welded into the pipeline with a gap or a pocket is still a cleaning problem. At acceptance, a cleanability test with the hygiene team is worthwhile, ideally with the real product.

Anyone replacing pneumatic cylinders with electric cylinders usually calculates with the compressed air saved. The sums are right, but the new cylinder behaves differently. A pneumatic cylinder can push against an end stop all day without getting warm. If an electric cylinder has to hold force continuously, for clamping or pressing, the motor heats up, and the sizing has to cover that. In a power failure it gives way under load unless it has a brake or a self-locking screw. Agree with the supplier how often the screw needs lubricating, and add that to your maintenance plan. The big advantage shows at format changeover. Stroke and force come from the recipe, and nobody has to move sensors or readjust end-position cushioning any more.

In a fleet analysis, electric pallet trucks often turn up in larger numbers than planned. Some sit broken in corners, others run routes for which conveyor technology would long since be more economical. Check battery management, statutory UVV inspection status and the usage profile of each unit. On the third shift the batteries are flat and the charging bays are full, and a unit wanders over from the neighbouring area. I therefore always count the charging bays as well. A unit that has stood in the same place for weeks is probably not needed. As a starting point, a list of location, year of manufacture, condition and typical route is enough, plus a photo of the rating plate. It is the basis for any decision on conveyor technology or AGVs.

Portable equipment is where you lose track. Cable reels and hand lamps wander round the plant, and some of them simply fail to turn up on inspection day. Think of contractors who bring their own tools as well. Those should be tested before anyone plugs them in on your premises. With fixed installations the problem is planning. The line has to be switched off for the measurements, so the inspection belongs in the shutdown window and not in some gap in the production plan. On machines with frequency converters and other electronics the inspector needs to know what to disconnect before the insulation test. Ask about it before he starts. New machines should also come with the manufacturer’s test report to EN 60204-1. Have it included in the delivery, though it does not cover the supply cable on site.

When switching to an eBR, people like to transfer the paper record page by page into screen forms. Then you have digital paper, and operators click through fields that nobody read before either. Use the changeover to clear things out, and have process values taken straight from the control system instead of being typed in. Allow for the validation effort. Every later change to the master template goes through change control, and that takes time. Also ask what happens when the network or the server goes down. Someone then has to be allowed to continue on paper, and that paper has to be entered properly afterwards. And try out the input device at the line, with gloves and wet fingers. More than one touchscreen has given up at that point.

Electronic cams turn a changeover into a recipe matter instead of a job for the spanner, provided the profiles are properly parameterised. Require the supplier to run the changeover through recipe parameters and not through a program change by a service technician. Limits are set by acceleration peaks. A profile that looks nicely smoothed on screen can drive the gearbox to its limit or let the motor run hot. Jerk limitation and the thermal load of the drives must be part of the sizing. Also check who is allowed to change the cams. If every operator turns the profile, two shifts run two different machines. A secured copy of the cams in the control backup belongs to the package.

EMC problems have one nasty habit. They come and go. Sometimes an analogue value jumps, sometimes a drive briefly loses the fieldbus, and in the workshop the removed part works flawlessly. With faults like these I look into the cable duct and at the shield connections first, before anyone starts swapping modules. A classic is the shield twisted into a thin pigtail and landed on a terminal. At high frequencies the shield then delivers only a fraction of its effect. Just as popular is the encoder cable running the full length of the duct right next to the motor cable. In retrofits the trouble often appears only when a new inverter is connected to the old unshielded motor cable. Plan the new cable in from the start. That is cheaper than weeks of fault finding.

Emergency stop buttons must be reachable at every operating point, including places operators run to when there is a fault, such as buffers or transfer points. On integrated lines, clarify which machines an emergency stop halts, only its own or the neighbours too. Either can be correct, but it must be stated in the risk assessment and the operating concept. After an emergency stop, restart must only be possible through a deliberate reset. At acceptance, walk the whole line and trigger every emergency stop once. That shows which buttons hang hidden and where the neighbouring machine simply keeps running. Have the emergency stop circuits tested regularly in maintenance too, because a button nobody has pressed for years is an untested component.

For returnables the inspection is mandatory, for new glass it protects the filler from glass breakage. Returnable bottles come back with things the consumer put inside, from cigarette ends to paint. Watch the reject rate. If it suddenly rises, the cause is usually the returnable supplier or the washer and not the inspection machine. Anyone who turns up the sensitivity to push the rate down just buys a different problem. Ask how the machine is taught and regularly tested with test bottles. An inspection that was calibrated once and has run unobserved since will end up inspecting for show.

Without an enabling switch nothing moves on a robot or palletiser during teaching, and that is how it should be. It releases motion only while the person deliberately holds it in the middle position. Anyone who lets go or pushes through in panic stops the machine. Every safety professional knows the classic: the switch is fixed with a cable tie or tape so that one hand stays free. That removes the whole protective effect. Look for tape on the pendant when you walk the plant, you find it again and again. Train the operators and explain why the switch is built the way it is. The pendant should also have a fixed place, so nobody has to hunt for it after every shift.

For changeovers and adjustment axes, absolute encoders are the right choice. The axis knows its position immediately, and recipe changes run without a homing move. On retrofits of older machines, incremental encoders can often be replaced cheaply with absolute encoders, which lays the groundwork for automated format adjustment. In wet areas the weak points are ingress protection rating and the cable routing at the encoder. Have the supplier show you what the machine does when an encoder fails or its cable breaks, and whether the fault message points at the encoder or at the motor. That takes five minutes at acceptance and saves the search on night shift.

The end of the line, meaning case packer, palletiser and wrapper, is the area where automation in mid-sized manufacturing often pays off fastest. Work there is heavy and needs many people. Check pallet logistics and floor space first, then the technology. A palletiser without a pallet buffer stands idle more often than it runs. Pallet quality counts too, because a warped wooden pallet can stop any palletiser. Ask where the empty pallets come from and how the full ones leave. That is where the plant hangs in the third shift, when the forklift is missing. I draw the routes of forklift and people into the layout before I look at the machines.

Audit reports quite often end up in a binder, and the binder in a cupboard. That is often down to the report itself. If the auditor spends a day walking through the plant and then recommends LED lighting, leak detection and new heating pumps, nobody needed an audit to know that. Ask for the large consumers to be logged with measuring equipment for a week, meaning the compressors, the refrigeration plant and the lines with heating zones. Take someone from production along on the site visit who knows when each machine runs and why. And have the measures prepared so that they fit straight into an investment request, with cost, savings and payback period. Then the report stands a chance of turning up again in the next budget.

A certificate to ISO 50001 is soon hanging on the wall. Things get harder when someone wants to know how much electricity and compressed air packaging line 3 uses per shift. Many plants have one main meter and then nothing for a long way. Without submeters the indicators remain estimates, and the management review turns into a debate about assumptions. For projects this means that meters for electricity, compressed air, steam and cooling belong in the specification of every new line. During installation they cost little. Retrofitting them needs a shutdown and an electrician who happens to have time. Bring production on board as well. The energy manager sees the consumption, but only the shift knows why the line ran through the weekend.

Where energy is lost shows up best during a break, at weekends and during downtime. Then the compressed air escapes through leaks, and heaters hold temperature for nothing. A building meter alone is not enough, so measure per line and relate consumption to machine state. For new machines, require a standby mode and consumption data via the interface. Decide who looks at the values regularly, otherwise the system just collects data. When a new meter goes live, I first look at what it shows on Sunday morning. If the value is low, the plant is switched off properly. A high value gives you the list of suspects immediately.

Changes get expensive after design approval. That is when marketing turns up with a new format size, the packaging material changes supplier, or a connection has to move a metre to the left. Each of these needs a request with cost and schedule impact, and the client has to decide, not the build team. The quick phone call is the most expensive route, because someone changes a drawing and the control programmer never hears about it. Keep a change list as a contract appendix. Then there is no argument about extras at final invoicing, and at the next rebuild you still know why the machine looks the way it does.

Sustainability in packaging is rarely a material swap alone. Less material, a recyclable structure, better volume use and less scrap work together. The line has to cope with the change, which is why production needs to be at the table early and not just once purchasing has placed the order. Paper-based solutions run stiffer and seal differently from plastic film, and thinner films tear more readily. Run real machine trials and count the scrap of the start-up phase. Whatever environmental promise appears on the pack, you should be able to prove it, and the wording is best cleared with a lawyer.

Without a specification, every supplier on a line brings its own designation system. Motor M1 then exists several times over, and the maintenance technician searches the wrong conveyor in the middle of the night. So define the scheme in the specification, ideally with the plant and location levels used in your factory. The fault message on the HMI should name the component’s designation. A message that just says infeed sensor is no help to anyone standing in front of a row of identical sensors. At acceptance, look at where the labels are stuck. On a removable cover they end up on the wrong component after the first cleaning, and on the sensor itself they disappear with the first replacement. A label on the bracket or mounting is the better choice. The bracket stays when the sensor is changed.

When the machine tips over during positioning, suddenly everybody was standing somewhere else. Up to acceptance, the risk of damage to the plant often sits in a grey area between supplier, operator and haulier. Erection all risks insurance covers a drop during positioning, water damage or mishandling during trial operation. Clarify contractually who takes it out, and have the policy shown to you before the first crane arrives. Also ask whether your existing plant and the building around it are covered, or whether your own insurance steps in, and discuss that with your insurance broker. On large projects the premium is cheaper than a single crane accident.

Usually the ERP system only shows up in a project at handover, and by then it is late. The new equipment needs an asset number, equipment master data, maintenance plans and above all material numbers for the spare parts. Without them maintenance cannot order anything at the first breakdown, and the supplier’s technician has already gone home. So ask for the spare parts list early and in a format that can be imported, with the manufacturer’s designation and part number. Allow time for creating the master data. Nobody does that on the side. The same applies to new formats or products. Without a bill of materials and routing in the ERP there is no production order, and without a production order the line can run, but nothing it makes gets booked.

The interface tends to be treated as an IT matter until the line stops. If the ERP system cannot be reached at night and no order arrives in the MES, the line has no batch number or best before date for the printer, and the shift waits. So agree an emergency mode, such as a stock of released orders held in the MES or manual order entry with later posting. Units are a classic. The ERP counts cartons, the MES counts packs, and at the end of the month the stock does not match. I have it written down for every field which system leads and in which unit it is transferred. Test the interface before the SAT with real orders, including the awkward cases such as order cancellation, partial confirmation and reversal. Those come up in operation sooner than anyone expects.

An escalation rule is only worth something if it was on the table beforehand. Both contracting parties must know it and see the step as an agreed tool, otherwise even the first escalation counts as an attack. The deadlines should fit the schedule. If each stage takes several days and the critical path has hardly any float, the decision comes too late. Enter the contacts at the upper stages by name in the contract, or the search only begins once there is a dispute. Prepare the report on a single sheet, with status, commitments made so far and what you need from the upper level. External project managers use the rule deliberately to break deadlocks without personal conflict, because the demand then comes from the level above and not from the person who sits at the table with the supplier every day.

The most common self-deception: looking only at the invoices already paid and forgetting the commitments. Because the invoice only arrives next quarter, it does not show up now. Keep the three columns separate, paid, ordered and still to be ordered. At every report, ask what additional work has become necessary on site, for example changes to the conveyor technology, extra safety fencing or overtime during commissioning. Items like these are usually missing from the first forecast. Include variations that are still under discussion as well, with an honest view of whether they will come. Report extra costs only at the end, and you have no options left.

EtherCAT has the edge on machines with many servo axes and fast control loops. PROFINET fits where Siemens standards and in-house maintenance line up. Mixed landscapes run via gateways but cost engineering effort and diagnostic convenience. That is why I put the preferred control system and bus into the user requirements specification and have any deviation justified. The supplier often has good reasons, but you need serviceability for years to come. Ask about the topology too, because in a plain line structure everything behind a pulled plug drops out. Have the network configuration handed over as a file as well, because the circuit diagram alone is not enough.

For most applications on the line the obligations are manageable, for instance document search or anomaly detection. Even a quality inspection without a safety function usually needs no more than transparency and training. It gets critical when the AI makes decisions with safety implications or assesses employees. So assess the risk class for each use case separately, and record the purpose, data and responsible person in writing. A bought-in assistant or a camera with built-in evaluation belongs on this list too, because as a deployer you have obligations of your own. If in doubt, take legal advice before the system runs and not after acceptance.

For automation, pallet quality counts. Broken boards, protruding nails and warped runners cause faults on conveyor technology, palletisers and in the warehouse. A pallet check at goods-in or before the high-bay warehouse saves downtime. H1 pallets are more expensive but dimensionally consistent and hygienic, and often the better choice for internal loops in food plants. Decide who sorts out a damaged exchange pallet and where it goes. Otherwise it runs through the plant again three days later, especially at night when nobody looks closely. Plastic slides differently from wood on roller tracks and at stops, which only shows in a trial run.

EVOH turns a PE film into a barrier film, but only while the layer stays dry. Moisture in the laminate or moisture contact during hot filling and sterilisation can temporarily degrade the oxygen barrier. Ask the supplier for barrier performance under realistic moisture and do not accept the value measured dry as the answer. In thermoforming the thin layer is stretched along with the film and gets thinner in corners and edges, so check the finished tray and not just the film. I also always ask for the EVOH share, because it helps decide the recyclability classification, and that is better settled before the order than after.

An expansion reserve costs money upfront and is therefore readily cut. Decide deliberately where future expansions are likely, for example new formats, higher output or an additional packaging step. Plan reserve there and nowhere else. Reserves in the control cabinet, the controller and the cable route are cheap. Hall space, conveyor conversions and utility lines become expensive retrofits. I have the quotation state how many inputs and outputs and how much cabinet space stay free, and I mark in the layout the area that explicitly stays empty. Otherwise a shelf stands there in two years.

Ask a supplier whether everything is on schedule and the answer tends to be yes, right up until shortly before the delivery date. So ask for evidence. Photos of the build status, order confirmations from sub-suppliers and confirmed dates for long-lead parts tell you more than any green traffic light in a status report. Things often get tight with bought-in components, such as the control cabinet or a robot that arrives later than expected. Go there in person when a lot depends on the date. If there is still no machine frame in the workshop a few weeks before the FAT, you know more than after ten phone calls. Record every promised date in writing, with the date it was given. Otherwise nobody will later remember who promised what and when.

EXW sounds cheap but shifts effort and risk entirely to the buyer. The buyer arranges freight forwarding, import customs clearance, transport insurance and unloading. For machines from abroad, customs formalities and input VAT questions come on top. To compare quotes, gross up EXW quotes to a DAP basis. Only then are they comparable with domestic quotes. Obtain freight forwarder quotations for this and ask who loads and secures the machine. Load securing and loading damage are easily overlooked. Clear the input VAT questions with your tax advisor before ordering and not after the customs notice arrives.

False rejects are a frequent reason why inspection systems get quietly detuned on the shop floor until they stop finding anything. Measure them separately from genuine rejects by counting how many of the ejected packs were actually fine. Someone has to go through the reject bin for that, and more than once. Agree an acceptable false reject rate with the supplier as an acceptance criterion. Test it with your own product over several batches and not with the sample set. I also have it recorded who may change thresholds, because otherwise every shift ends up detuning the inspection on its own.

The FAT should take place with the actual product, actual packaging materials and the future operating personnel. With dummy goods and test cartons, any machine runs. The test plan with performance run, changeovers, fault scenarios, safety functions and documentation check is agreed beforehand, not on the morning of the test. Have your operators do the changeovers themselves and trigger a fault on purpose, for example a packaging jam. Defects found go into a punch list with a deadline and a name. Release for shipment is not acceptance, even if the supplier has already booked the lorry. Anyone who just watches at the FAT gives away the cheapest testing opportunity in the whole project.

Most tickets read much the same. Machine down, please come. Then they get closed with the word done. Nobody can build a Pareto analysis from data like that, and a year later no one knows why the labeller stopped so often. Make reporting easy, with a QR code on the machine or a tablet at the line, and provide selection lists for fault pattern and cause. Free text on its own never gets analysed. Microstops hardly ever appear in tickets, because the operator clears them himself. They belong in machine data collection. On new equipment it pays to set up the equipment structure in the CMMS before the first fault arrives. Anyone who adds it months later has wasted the ramp-up phase, which is exactly when many faults appear for the first time.

A feasibility study clarifies, before the budget request, whether the project is viable technically, spatially and economically. The effort of a few days is well spent. That applies especially on brownfield projects, where building height, foundations or utility connections can hold nasty surprises, for instance when the machine fits into the layout but the compressor cannot cope with the extra load. Ask for the assumptions behind the rough costs of each option, otherwise you end up comparing figures that do not mean the same thing. The result is a basis for a decision, and the solution comes later. A no is a useful result too, if it comes early and only cost a few days.

The FEFCO number says how the case is built and nothing about how good it is. Machinability depends on scoring, basis weight, flute type and tolerances, and these belong in the packaging material specification. A supplier change with the same FEFCO number is a typical cause of sudden faults on a case erector. The same style then sits in the magazine, but the scoring is executed differently or the board is softer. Do not bring a new supplier on board until their case has passed on the running machine. Set minimum values for edge crush resistance so that equivalent can actually be tested. And put flute type and paper grade on the order along with the number.

Choose the fieldbus based on your maintenance capability. What your electricians know, and what spare parts you hold in-house, is the better standard, even if the supplier prefers another. On lines made up of several machines, the interface list defines which signals run over the bus and which are hard-wired. Emergency stop chains must be implemented as safety circuits. For every machine, ask which diagnostic tools are needed and whether the licence for them is in the scope of supply. At three in the morning a bus diagnosis only helps if somebody can open it. Also have the address list with all participants and station names handed over, otherwise every replacement starts with working out which device is which.

In the system everything runs on FEFO, at the loading dock on the shortest route. The forklift driver takes the pallet at the front, and in block storage the oldest one just happens to be at the back. If you are serious about FEFO, storage locations have to be assigned so that the right pallet is within reach, and removal should be checked by scanning. Think about packaging materials too. Cold seal films, labels and adhesives also age and run differently on the machine after long storage. For cold seal the storage life is on the data sheet, and the machine often notices before the system does. Also easily forgotten is the part-used reel that goes back to the warehouse without a label after a format change and drifts to the very back. Retail customers add a required remaining shelf life on delivery, which sales knows about and the warehouse often does not.

Choose the method based on container and product. For glass and clear liquids, camera or high-frequency inspection is often enough, while opaque containers and foam need X-ray or gamma, with the associated radiation protection requirements. Foam and condensation are the classic sources of error, and a misted container throws any camera off. Define how fill level is converted to volume, since the Prepackaged Goods Regulation asks about content and not millimetres. I have the measurement demonstrated with containers that are deliberately underfilled and overfilled, at operating temperature and not in the laboratory.

For trend control to work, the checkweigher must be able to assign each pack unambiguously to its filling station. On rotary machines this runs via the valve number, otherwise via sequence and spacing. A common mistake is control that is too aggressive, reacting to every scatter and making the fill quantity oscillate. Dead time adds to it, because between filler and scale sit packs that were filled with the old value. Anyone who ignores that always controls behind reality. Set the control band and step size together with the manufacturer and test the interface at FAT. The scale itself belongs in the test plan too, because with a shifted zero point it leads the filler reliably in the wrong direction.

With filling lines, filling accuracy decides give-away and compliance with the Prepackaged Goods Regulation, the German Fertigpackungsverordnung. At the performance test, have the mean and the standard deviation measured on your own products, not on water. Water flows, does not foam and has no lumps in it. Your sauce has all of that. Also clarify cleaning by CIP, changeover and the connection to capper and labeller before you assess the filler on its own. A filler that doses precisely but needs two hours of cleaning after every changeover no longer pays. And a capper that cannot follow the pace slows the whole line.

Criteria to look at are filling accuracy as the standard deviation at target quantity, product changeover and cleaning by CIP, the container format range, drip protection and dosing principle. Volumetric dosers are cheap, gravimetric ones more accurate with varying density, and mass flow meters work hygienically and need little maintenance. At the FAT the original product at operating temperature belongs in the machine, because viscosity affects accuracy. Also have it shown how the nozzle shuts off after filling. Drips and strings on pasty products are a typical nuisance at the container rim and later in the seal.

Calculate how many roll changes occur per shift and how long each takes. With small rolls and high output this is often the single biggest loss on the machine, which an Efficiency Day on the line quickly reveals. An automatic splicer then pays off, but it needs consistent roll quality, clean roll cores and trained staff. Badly wound rolls throw even the best splicer off its stride. You should also clarify what happens to the packs at the splice. If they are not rejected, they end up with the customer. A larger roll diameter or a second unwinder is often a good alternative to an automatic splicer.

A few fewer micrometres save a lot of money over a year, but the machine notices immediately. Web tension no longer fits, the sealing window narrows, the film wrinkles more. Ask the supplier for the thickness tolerance and the thickness profile across the web, since uneven thickness winds up into ridges and produces telescoping rolls and web-tracking problems. Anyone who buys by weight pays for every micrometre above the nominal thickness. Thinner film always belongs on the line first, as it reacts faster to the heat of the sealing jaws. I spot-check thickness at goods-in with a gauge and compare rolls of the same length and width: if their weights differ, something is wrong.

Fixed guard fences are cheap and stand up to a lot, but they hinder changeovers and fault clearance. Then doors get bridged permanently, and the fence protects only on paper. Plan access points where operators really have to go, with locked doors and safe restart, instead of fencing everywhere. At acceptance I walk through a changeover once with the maintenance technician and watch where they loosen screws, hold doors open or step over something. That quickly shows where the door is missing. Gaps under the fence and the distance to the conveyor belong in the walk-round as well, because that is where someone reaches in who should not.

What a fixed price is worth depends on the scope description behind it. Every gap in the specification gets filled by the supplier, either with a surcharge now or with a change order later. So read the assumptions in the quotation as carefully as the price. If it says utilities are to be brought up to the machine by the client, that belongs in your budget. Commissioning is where it gets interesting. If product or packaging material is missing, or the line stands because your own conveyors are not yet running, the waiting time usually comes back to you as a change order. A fixed price protects you from the supplier’s effort, not from your own. For small, clearly defined jobs it is convenient. On modifications to existing equipment whose condition nobody really knows, it quickly gets expensive, because the supplier prices in the unknown.

The flap gap is the first thing to look at when there is trouble at the case sealer. If the flaps gape, there is no contact for hot melt or tape. If they overlap, the machine crushes the case. Measure the gap on several cases from one delivery and compare it with the drawing. It is often the flute grade or unequal flap lengths, not the machine setting. Skewed cases after erection lead the eye on to the manufacturer’s joint. The tolerance belongs in the specification, otherwise you start the discussion from scratch with every complaint. A limit sample that shows both sides what is just acceptable saves a lot of phone calls. Note the humidity in the store when you measure, so that you can later tell whether the weather played a part.

FIBCs are the standard on the raw-material side of many lines, and they bring their own problems: poor stability on the pallet, dust escaping during discharge, residue in the liner. Plan discharge stations with a massaging device and dust extraction, and define the bag specification jointly with the raw-material supplier. Mind the lifting arrangement. The bag hangs on its loops, so the hoist and the installation height of the station have to match. For combustible dusts the bag type belongs in the explosion protection concept and is no mere purchasing decision. The wrong type only shows up when somebody reads the label.

Nobody asks the floor during layout planning. It only comes up when the new machine is to go onto the upper floor of an old building, or a high-bay rack onto a slab that was poured for manual workstations. On paper the distributed load is often fine, but the machine stands on six feet with small pads, and the load concentrates under each foot. Load spreading plates help if the structural engineer approves them. Think about the route there as well. During rigging the machine travels on forklift tines or on skates through corridors and over drain covers, and that route sometimes carries less than the installation site. Get the foot loads from the supplier early and send them to the structural engineer along with the layout. An assessment takes time, and it has no business being on the critical path.

In filling, flow measurement replaces weighing and makes fast, clean dosing possible. That only works if the meter suits the product. An electromagnetic meter fails on oil, a Coriolis meter costs more but knows the density. Air bubbles in the product falsify the reading, so the measuring point belongs in a full, calm pipe and not where air gets into the line. Calibration and temperature dependency must go into the test plan. When changing to a product with a different density or viscosity, the recipes need adjusting, otherwise the fill quantity is right on paper only. At acceptance, have the measurement checked against a checkweigher.

Pallet flow racks only work with good pallets. A broken bottom board, a different pallet design or an overhanging load, and the pallet stops halfway down the lane. Then comes the interesting part, getting it out again. Nobody should climb into the lane, even if it would be quicker on the night shift. Check with the supplier which pallet types and which orientation the rack has been tested with, and reject other pallets at goods receipt. Heavy and light pallets in the same lane start moving at different speeds. Brake rollers wear and are rarely checked until a pallet hits the front too hard. In deep-freeze storage, ice on the rollers adds to the trouble. For supplying a line with carton blanks or film reels, a short flow rack is still often the simplest solution.

When buying, what counts is film width, sealing method, format range, infeed and print mark control. Sealing can be hot, cold or ultrasonic, and for crumbly products like biscuits ultrasonic is often the answer, because it copes better with crumbs in the seam. Vulnerable areas are film tracking, web edge control and the cross-seal and cutting station. Then there is the film roll change: with small rolls the machine otherwise stops more often than the brochure suggests. With mono-material films the sealing window narrows, so temperature control and sealing jaw geometry must suit them. I therefore insist on a test with the actual film at FAT, ideally a long run with the real product.

Running behaviour and seal integrity depend on the film. Sealant layer thickness, slip properties and print register must match the machine. Have sample rolls run on the supplier’s machine before the FAT, preferably rolls from the production batch and not a hand-picked sample. At changeover, the forming shoulder and sealing tools are the time-eaters. Anyone running many formats should demand changeover times as a specified value in the tender. Also ask about the film reel change, meaning how long the machine stands when the roll runs out and whether an automatic splice is possible. See also VFFS machine.

Switching from B to C flute looks harmless on paper. The blank gets thicker, fewer sheets fit into the same magazine height, and the folding stations are set up for the old thickness. Often the machine then runs only after readjustment, or not at all. It just as easily happens that a supplier short of material quietly delivers the other flute and considers it equivalent. Fix the flute type in the packaging material specification and require it on the delivery note. At goods receipt, measure the thickness of one blank with a calliper, which takes a minute. You should also note the change parts for the magazine and folding stations in the machine recipe.

FMCG manufacturers live on speed, volume and retail requirements. Packaging projects must therefore deliver short changeovers, retail compliance and high availability, for example shelf-ready packaging and GS1 labelling without rework. Anyone managing projects here needs to understand listing deadlines and promotional goods. A retail deadline does not move because of a machine. Plan the ramp-up backwards from the delivery date and put the reserve before the deadline. Also check how the line runs at peak times, for example ahead of seasonal goods and promotions, because that is when it is expected to run longest.

An FMEA for a complete packaging line quickly becomes a mountain of paper. Focus on critical functions such as seal seam, labelling, checkweigher, rejection and handovers. Use fault data and complaints from existing lines as input, that is the most honest source. For this I put people around the table who have actually run the old line, so not a round of just the project manager and the quality manager. Give every action a name and a date. And make sure the actions land in the user requirements specification and the test planning, otherwise the FMEA stays a document for the auditor.

Switching from GC to GD saves money but costs bending stiffness at the same basis weight. High-speed case erectors thank you with erecting faults and open flaps. So specify basis weight, thickness and stiffness in both directions as well as the grade, and state the grain direction, because blanks with an unfavourable grain direction score and fold badly. With recycled board in food, the question of mineral oil migration also belongs on the table. Have every board change approved on the running machine before purchasing signs the framework contract. After that the price is fixed and the machine is not.

Board quality is a specification issue. Basis weight, scoring, moisture content and storage conditions determine how reliably a case erector erects the blank. Fluctuating supplied quality is among the most common causes of minor stoppages. If blanks go from a cold lorry straight into the magazine in winter, or sat pressed in the stack for a long time, they open less willingly and the shift blames the machine. Agree tolerances with the packaging material supplier and check the glued seam at goods-in, because an open or skewed glue flap stops any erector. I have the FAT run with original blanks from several batches and not with the smooth samples from the machine manufacturer, because those always run.

When food contact compliance comes up, most people think of the packaging first. But every part that touches the product is checked: conveyor belts, lubricants, seals, gripper surfaces. Require declarations of compliance for these materials from the supplier, and a list of the H1 lubricants used. Evidence for seals, hoses and wear strips goes missing most often, because they are ordered as small parts. This is often forgotten on retrofits, and then it shows up at the next IFS audit. A spare part from the odds-and-ends box, say a slide rail in an unknown plastic, is a finding as well. Decide who maintains the evidence on site, and replace wear parts only with parts that come with it.

The typical mistake isn’t the wrong oil in the gearbox, but the universal grease in the workshop cartridge that ends up everywhere. Agree a lubricant list per lubrication point with the supplier, reduce the number of grades, and colour-code lubrication points. Check whether gearboxes are supplied with H1 fill from the factory, because that is rarely stated in the quote by default. I also ask about compatibility whenever one grease meets another. Changing grades mixes residues, and not every combination gets along. Keep H1 stock visibly apart from everything else in the store and on the workshop trolley, so nobody reaches for the wrong cartridge.

The contract deals with three points: which events count, which legal consequences follow and which notification duties apply. Supply shortages of components in particular are a bone of contention, because suppliers like to file them under the clause. A supplier invoking a chip shortage as force majeure needs evidence. For an operator affected by it, the schedule needs alternatives. Ask your supplier early about its sub-suppliers and their lead times, then you notice a problem before it arrives as a letter. Also have your lawyer read the clause, because legal consequences and deadlines are legal questions.

In many plants the forklift fleet has grown historically: different manufacturers, unclear use, high damage costs to racking and doors. A fleet analysis with a trip profile shows which routes are driven regularly and which only occasionally. Regular main routes are candidates for conveyor technology or AGVs, the rest stays with forklifts, but with fewer units. Look at the traffic routes too. Where pedestrians and forklifts share the same route, separating them is cheaper than the next accident. Impact protection on rack feet and at doors costs little compared with the repair afterwards. The damage to racking and doors shows where trucks drive without anyone having planned it.

A forklift fleet management system allocates driving jobs by priority and vehicle position, which means fewer empty runs and no shouting across the floor. It pays off from several forklifts and many orders per hour upward. Beforehand, the route analysis should clarify whether the problem lies in dispatching or in the layout. A system that organises long routes well remains a system for long routes. Talk to the drivers and the works council early, because tracking the vehicles quickly feels like surveillance. And clarify what happens when a driver does not confirm a job or the terminal fails. Without maintained location data the system sends the forklift to places where nothing is standing.

FFS machines are flexible and material-efficient because the pack is formed only inside the machine. Output depends on film, product and doser and not on the brochure. When selecting, look at how quickly the forming shoulder or tooling can be changed and how easy the machine is to clean. Seal inspection belongs on the list, because crumbs or dust in the seam are the classic. Run the FAT with your own films and products, and long enough for variations to show up. A machine that runs on the sample reel says little about your rolls from the third delivery. The test also shows how long the machine takes after a film tear before good packs come out again.

The format list is the most important single piece of information in an enquiry, and the one most often incomplete. Include even the formats that run only twice a year, and mark the lead format with the largest share. Whatever is not on the list comes later as a change order with new change parts. Before sending an enquiry, I always look into the item master data and talk to production, ideally the shift leader. They know which special item runs once a quarter and which pack only works with manual help. That saves weeks. Also ask about formats planned for the next two or three years, otherwise the format range no longer fits at handover.

Buy the format range you will need over the next few years, not the widest one on the market. A machine that can do everything from the smallest to the largest format runs ideally for none of them. In the tender package, list all formats with volumes, and mark phase-out formats and planned new ones. Then the supplier can see where the machine has to earn its money. The variation of the product itself belongs on the list too, because a naturally varying product uses more of the range than the nominal dimensions suggest. Extreme formats at the edge of the range need to be proven at FAT, since output usually drops there. I have the most-run format and the two edge formats demonstrated and record the output per format in writing.

Wrinkled longitudinal seams, film tears and print image misregistration often start at the forming shoulder. Behind them are worn edges, a wrong position relative to the fill tube, or a film the shoulder was never designed for. Buy shoulders specified against film type and rated output and not just pouch width. A set of shoulder and fill tube for each pouch width costs money, so clarify early how many widths you really run, and ask about the lead time for a replacement. Treat them as precision parts. They belong in the change parts trolley and not on the machine frame. Check edges and radii regularly for scratches and dents. If a fingertip can feel something, the film feels it too.

Ask for the foundation plan early, not just a few weeks before delivery. Old building floors have ducts, reinforcement in unexpected places or a thin slab over basement rooms. The fall towards the floor drains comes on top, and it has to be allowed for when the machine is levelled. Palletisers, stretch wrappers and checkweighers react to uneven or vibrating floors, the checkweigher with unsteady readings. Have the floor surveyed and uncover what is under the old machine before the new one arrives. And clarify who drills the anchor holes and who closes the floor again afterwards.

FPY exposes rework that easily disappears inside OEE, because it is done by hand right at the line and nobody books it. Re-labelled cartons, re-weighed pouches and manually replaced products belong here. If FPY is low while the quality rate looks good, hidden manual work is behind it. It can be tracked via the rejects and the rework stations. Often a simple tally sheet per shift is enough to see the extent, and it costs nothing. Ask the people at the rework station what they correct most often. The answer usually comes faster than any evaluation and is surprisingly often right.

On recurring assignments a framework agreement with a provider of on-demand specialists or with a machine supplier saves negotiating a contract for every project. Terms, liability, confidentiality and billing are settled, and work is called off order by order. Make sure the framework agreement contains no obligation to place orders, otherwise you have committed to volumes you may never need. Have the rates reviewed every year. Define who may place call-offs as well. Otherwise one plant orders from the same provider on the strength of the framework and the others never hear about it. Price escalation and notice periods belong in the framework too.

Freezing tunnels are hungry for energy and space. Sizing by product mass, freezing time and belt loading decides whether the plant fits the line or becomes the bottleneck. Check the defrost cycles, because defrosting costs production time. A tunnel that delivers frozen products in clumps needs a singulator ahead of the packaging line, otherwise it jams at the handover. Also allow for condensation: when cold products enter the warm hall, the film mists up, labels stick less well and seals get damp. Cleaning the interior also raises the question of how long the plant needs afterwards to get back down to temperature.

With frequency inverters you run speeds by recipe, start softly and save energy at partial load. On retrofits they replace pole-changing motors and gearbox stages. Look at ingress protection and mounting location. In the cabinet the inverter is protected, while next to the motor it saves cable but needs a rating for the wet area, and long motor cables call for screened cable and sometimes a filter. Connect the inverter to the control system by fieldbus, so parameters and diagnostic data sit in one place. Ask about braking too, because a motor that brakes often usually needs a braking resistor.

Where the functional description is missing, how the machine should behave gets discussed at the control cabinet, at commissioning-day rates. Insist on an approved document before programming starts, one your line leaders have read for sequence steps and fault responses. I never hand it only to purchasing or engineering but also to maintenance, because they have to live with it later. Keep it short, with flow diagrams. Any change after approval is a change request, even a small one. A revision level on every page belongs to that. At acceptance, testing is against this description, not against verbal instructions. Anyone who says it was simply agreed in conversation has nothing to show.

I put the user requirements specification and the functional specification side by side and tick off every requirement individually. That shows what is missing, what is met only in essence, and where there are exclusions or customer-side works you had not planned for. Make sure performance figures come with a reference: for which product, which packaging material and which operating state they apply. Whatever is not in the functional specification comes back later as a variation order. The comparison is tedious, but once you have signed, every variation order costs more. Do not approve the document until everything is clarified, even if the schedule is pressing. By approving it you have made it your own, and the supplier will point to exactly that version in every dispute.

Choose a gantry robot when area and load become too large for an articulated arm, for example a cell palletising several lines, or with bulky products. At high cycle rates it quickly becomes sluggish, because long axes carry their mass with them. In an existing plant, look at the building height first. A gantry that does not fit under the crane runway is not an option. Next come the supports, which need floor space and a load-bearing floor. Also ask how you can reach lubrication points, belts and cable chains. Whatever hangs several metres above the line does not maintain itself, and without a lifting platform every replacement turns into a climbing exercise.

Geared motors are wear parts with a long service life if oil changes and loading are correct. Standardise sizes and manufacturers across your lines, then a few spare motors are enough. In hygienic areas, look for smooth housings, ingress protection rating and food-grade gear oils. I have the mounting position written into the datasheet for every geared motor. If it is later installed differently than planned, the oil quantity no longer fits, and the gearbox runs with too little oil or sits in too much. Oil grade, fill quantity and change interval per type also belong in the maintenance list, because that is where maintenance looks first.

At the Gemba you see what figures can only hint at. Why the line really stops, how long the operator spends searching for change parts, where material backs up. That is in no MES. An Efficiency Day is therefore a structured Gemba day. You observe over a full shift, log the stoppages, ask operators and maintenance technicians and put the MES data next to it. The actions come from what was seen, not from slides. Go in on the night shift once too, because the same machines often run differently then. Allow enough time. Anyone who leaves after a short walk-round has seen only the good side of the line.

With split contracting you take on interfaces, schedule coordination and integration risk. That pays off when an experienced project manager is available in-house, or brought in on demand. Without one, you pay back the saving on purchase during commissioning, when every supplier considers its machine finished and the line still does not run. A general contractor bundles the responsibility, but prices in mark-ups and risk for doing so. Check in the quotation what it delivers itself and what it merely passes on. A middle path is split contracting with an integrator for the control system and conveyor technology who answers for the line function. The contract then has to say so as well.

Product damage is hidden waste. It shows up as a complaint and not as downtime, so hardly anyone in production notices it. When selecting a machine, watch drop heights, transfer points, belt speeds and gripper forces. At FAT, run the most sensitive products and not the undemanding standard format. For that I like to be given the product the plant would rather leave out of the test. Run the samples warm and cold, because baked goods behave differently warm than cold, and so does chocolate. After the run, collect the breakage at the end of the line and weigh it, and you have a number instead of a feeling.

Many drawings carry geometric tolerances simply because the CAD system offers them. That gets expensive. Every tight tolerance costs time in machining and even more in inspection, especially if the supplier needs a coordinate measuring machine for it. Conversely, the callout tends to be missing exactly where it is needed, for example the alignment of two bearing seats in a welded frame. When requesting quotes for machined parts, I therefore go through the drawings with the design engineer and ask what each tight tolerance is for. Quite often it gets deleted afterwards. The same goes for incoming inspection. Measure the features that matter for function, and have the supplier deliver a measurement report for exactly those features.

Registration in the packaging register has to be in place before the first pack goes to market. Without it a distribution ban looms, and retail platforms block suppliers. Clarify in the project who the first placer on the market is. For contract filling and private labels this is often not the filler. Put the responsibility into the contract and have the classification checked if in doubt. Transport packaging such as stretch film and pallets is not subject to licensing but is subject to take-back obligations. Keep the reported quantities consistent with the consumption data from production. Anyone who orders film in one quantity and reports a very different one will be asked sooner or later where the rest went.

The supplier carries the CE marking, the operator carries the BetrSichV, and for as long as the plant runs. After every retrofit the risk assessment has to be updated, even if all that changed was a new controller or an extra light curtain. Decide who in the company schedules the inspections, otherwise everyone relies on someone else and the deadline passes. Keep the inspection records per plant item and not in one collective folder for the whole site. That saves discussions with the trade association and the insurer when somebody asks questions after an accident. The instruction of employees belongs in there too, with date and signature.

Give-away results from dosing variability and safety margin. A high standard deviation forces you to set the target weight higher to avoid underfilling, so every pack gives away a little. The approach therefore lies in the dosing technology and in control via the checkweigher. Check whether the set point has stood unchanged and high for years although the machine has long dosed better. That happens because nobody wants to touch the margin. An Efficiency Day calculates the annual value of give-away from raw material price and unit volume, often the fastest business case on a line. With an expensive raw material, fractions of a gram are enough to make the calculation interesting.

When buying new machines, auditors ask. They want to know whether the lamps are splinter-protected, the sight glasses are made of polycarbonate or shatterproof glass and the sensor windows are protected. The manufacturer should supply a list of all glass and hard plastic items so that it can be added to the register. These days I ask for this list with the quotation and not at delivery. In the event of breakage, a defined procedure with product hold applies, and this must also be set up for new equipment. Think of the small items such as indicator lamps, displays on the operator panel and camera housings, they are the first to fall through the grid when the list is written.

Glass requires careful handling. Broken glass means a line stop, cleaning, and a glass check on all packs in the surrounding area. So plan for empty-glass inspection, breakage detection and a documented procedure for glass breakage from the start, because this is standard in IFS and BRCGS audits. Glass against glass is the enemy: on conveyors and in accumulation the bottles rub against each other and become more fragile with every scratch, which is why plastic guides and a clean accumulation concept count. Metal detectors cannot see glass. Pallets of glass are heavy and separated layer by layer with interlayers, so check the floor loading too. On glass lines I ask first to see the glass and hard plastic register.

A badly glued case rarely stands out on the line. It comes apart weeks later in the warehouse or at the customer’s. Glue detection is meant to prevent that, but it only sees whether glue is present. How well it holds is another matter. If the hotmelt is too cold, the compression section too short or the case arrives ice cold from an outdoor store in winter, the bead sits in the right place and still bonds poorly. Test the system at acceptance in the simplest way, by deactivating one nozzle on the glue unit and watching whether the pack is really rejected. I also always ask whether the inspection can be switched off and whether that is logged. If the sensor throws out good product because of glue strings, it gets bypassed quickly at night. So add it to the cleaning schedule.

In a GMP environment the project timeline is extended by qualification steps: URS, DQ for design qualification, FAT, IQ, OQ and PQ, each with protocols and approvals. Machine suppliers must provide material certificates, calibrations and software documentation. The effort is considerable and belongs in the schedule and budget. I therefore plan qualification as a work package of its own with its own owner and not as an appendix to commissioning. Outside pharma, orientation to GMP principles without formal qualification is usually enough. Clarify early which documents the customer or the auditor really wants to see, otherwise you write protocols nobody reads.

If the grain direction of a blank is wrong relative to the score, the blank breaks at the score when folded or springs back open after erection. The cartoner then pushes the flaps instead of folding them. For folding cartons the grain direction belongs on the drawing, for corrugated board the flute direction does. Blanks that warp after a weather change in storage are almost always a matter of moisture and grain direction, not a machine fault. Ask the supplier how the blanks lie on the pallet and whether they arrive wrapped in film. Open stacks take up moisture through the edges and curl at the rim. A simple test helps on goods-in. Bend a strip lengthways and across, and in the grain direction it feels clearly stiffer.

Do not mix greases on the off chance. When two incompatible thickeners meet, the mixture can soften and run out of the bearing or harden and stop lubricating altogether. Before switching to a new plant standard grease, ask the manufacturer about compatibility or purge the old grease as far as possible. More does not help more either. An overfilled rolling bearing runs hot because the grease is churned, and the grease then ages faster. The fitter who gives a few extra strokes on every round means well and achieves the opposite. Relubrication quantity and interval belong in the lubrication schedule, and whoever does the greasing should know how much grease one stroke of the gun delivers. H1 grease and standard grease look the same in the cartridge. Mark them in colour and store them apart.

Grippers are special designs: vacuum for cartons and trays, clamp grippers for sacks, needle grippers for textiles, form grippers for bottles. At changeover the gripper is often the component that must be swapped or adjusted, and automatic gripper changers save time there. Require testing with all products and packaging materials at FAT, including the worst carton quality. A gripper that works with the sample from development does not have to work with the carton from the night shift when that one is damp or dusty. I always ask the supplier about wear parts. Suction cups are one, and they belong on the shelf. Have the drawing and parts list of the gripper handed over as well, so that it can be rebuilt in an emergency.

An automatic gripper changer sounds like flexibility, but it costs cycle time, installation space, payload and money. For two formats a combined gripper is often better. If you do change grippers, watch the repeatability of the coupling, correct-tool recognition in the program, and make sure the vacuum and compressed air feed-throughs stay leak-tight. Leaking couplings are a classic cause of sporadic gripping faults. People then search at the suction cup, at the hose, at the robot, and in the end it was the seal in the changer. The mating surface also gets dirty, and dirt interferes with the locking. Plan the docking station with room and access for the operator, otherwise it ends up where nobody can reach it.

When operators fit substitute actuators to the switches, malice is rarely the reason. Usually the machine stops at every small intervention, and the door sits exactly where people must go ten times a shift. A reprimand does not help there. Change the operating concept instead, for example with an access point for clearing faults that stops only the affected section. Walk the line and look at the switches. Fresh scratches, a cable tie or a second actuator in the control cabinet show where it hurts. At acceptance, have every door opened once while the machine runs. It must stay shut until the drive has stopped.

On the VFFS machine, gusseted bags need precise film guidance and stable change parts. If the film wanders, the gussets sit crooked and the bags no longer stand straight. For valves on coffee bags or for resealable closures, an applicator is added that limits output. When buying, clarify which bag shapes the machine can run without conversion. Ask about the effort of changing between pillow bag and gusseted bag, because the folding wedges are change parts and have to go with it. Check the seal seam at the crossing points in an endurance run. Where many layers come together, seams become leaky first.

For new machines I have it written into the user requirement specification that gearboxes and hydraulics leave the factory filled with H1 oil. Without that sentence, the machine arrives with the gearbox manufacturer’s standard fill and nobody notices. Converting afterwards is tedious. Some mineral oil always stays in the housing, which is why certain lubricant manufacturers recommend a second change after a short running period. Incidentally, the same ISO VG grade does not mean the same oil. A mineral VG 220 and an H1 VG 220 still do not belong in the same gearbox. Label containers, oil cans and filling points with the same colour or number as in the lubrication schedule. A can that holds hydraulic oil today and chain oil tomorrow makes the finest registration worthless.

HACCP becomes concrete at the control points of a machinery project. There the checks must be validated, limits defined and records unbroken. Metal detectors and X-ray need test-piece routines, interlocked rejection and alarming. The checkweigher has its test weights and a traceable rejection too. New equipment must be aligned with the HACCP team before the layout is fixed. A CCP in the wrong place is hard to move later, because it needs space, power and a reject point you can see. Put the reaction to an alarm into the plan as well and define who releases the held product and where it sits until then.

The FAT takes place at the supplier’s at a comfortable room temperature. In your plant the same machine stands next to the shrink tunnel in August or by the roller door in January, and suddenly it runs differently. Cartons from a cold outside store that go straight to the case erector fog up, go soft and jam in the magazine. Film charges up in dry winter air and clings to every guide. Hotmelt sets too fast in a cold draught and then does not hold. So write the climatic conditions at the installation site into the specification, summer and winter separately. Let packaging materials acclimatise in the hall before processing and plan floor space for that. Air conditioning the hall for the staff often takes care of the machine too, just not the corner right by the door.

Camera-guided robot cells rarely fail because of the camera, but because of the calibration. A slightly shifted camera mount, a new lens or thermal drift, and the pick position is off. Require a documented calibration routine that maintenance staff can run themselves in a few minutes, plus a verification routine with a reference part. Without this routine, every fault becomes a service call to the integrator. At acceptance I have the camera removed on purpose, refitted and recalibrated once. Then you see how long it really takes and whether the instructions are right. The calibration pattern belongs to the delivery and at the cell, not in the commissioning engineer’s drawer. After a crash, too, the check should run before the cell picks parts again.

Between acceptance and handover lies the most delicate phase. The project manager is already thinking about the next project, and production still sees the machine as a building site. Agree a supported start-up period with fixed presence times, so that the operators know whom to ask and when. Hand the punch list over to the line with names and dates. Sign the handover record only when the spare parts are in the store and the maintenance plan is in the maintenance system, not when both have merely been promised. Define who calls the supplier when something goes wrong, at night just as on Monday morning. Without a named person responsible for the equipment, every fault stays stuck with the project, and by then the project has long been disbanded.

When machines from two manufacturers come together, the handshake is the classic point of dispute. Each waits for the other’s signal. Before ordering, define which signals are exchanged in what logic, who supplies the connector and who tests the coupling. Simulate fault cases, for example a jam behind the labeller or an emergency stop on the neighbouring machine. A timing diagram in the interface list saves weeks. Insist that both manufacturers couple against a simulator or test box before delivery, otherwise the programs meet for the first time on your site. And agree on timeouts: if a reply fails to come, the machine should report whom it is waiting for instead of just standing still.

In packaging technology, a HAZOP pays off wherever media, pressure, temperature or gases are involved, so at shrink tunnels, MAP gas flushing, hot melt, CIP connections and inerting. Get the right people into the room. Operators and maintenance staff know what actually happens when a sensor fails, and that is on no flow diagram. Allow enough time, because one section after the other takes long, and after a long day without a break nobody says anything clever any more. Record the results as actions with owners and feed them into the risk register and the risk assessment.

Clear with your health and safety officer whether a machine rebuild in a live shop floor counts as a construction site under the ordinance, before the first outside fitter arrives. Whatever the answer, you need someone to coordinate the trades. An electrician under the platform and a fitter above it, both on the same day, is a classic accident scenario. A joint schedule in which the jobs do not overlap protects more here than any leaflet. Coordination also covers briefing the outside firms on escape routes, restricted areas and contact persons. The site manager and the SiGeKo may be the same person, provided they are qualified.

Compressor waste heat is the textbook case, and many plants still blow it out through the roof. In winter you would like to have it in the hall, in summer nobody wants it. The system has to be designed for that. Compressor cooling must never depend on someone taking the heat, otherwise the compressor trips on overtemperature on a warm day. An emergency cooler or a changeover damper is therefore always part of it. Preheating water for the CIP system or other cleaning processes works well, because the demand is there all year round. Clarify early who operates the system. The heat comes from building services, it is used in production, and when the pump fails nobody feels responsible.

Switching from film to paper sounds like a change of material. At the machine it is more like a small project. Paper barely stretches, it tears or creases at the forming shoulder, and the barrier layer cracks at sharp folds. The sealing station often needs different temperatures and longer dwell times, and the machine not infrequently runs slower than with film. So plan a machine trial with real material from several reels. The sample the paper maker sends is not enough for that. Repeat the trial when the climate in the hall is different. Paper reacts to moisture, and what runs cleanly in dry winter air may cockle in a humid summer. Store a few finished packs for several days as well, then check them again for leaks.

A high-bay warehouse is a construction decision that lasts for decades: fire protection, floor loading, structural design and permits run in parallel with the plant engineering. It pays off through space savings, headcount reduction and stock accuracy. The connection to production with pallet inspection, conveyor technology and WMS interface decides whether the investment saves time day to day or merely shifts the bottleneck. Also clarify how you reach the pallets in an aisle when a stacker crane fails and who is trained for it. Without emergency operation the stock sits there while production waits for its raw materials.

High speed is only valuable if infeed, packaging materials and the end of line can keep up. Many operators buy capacity they never use, because product separation hits its limit far below rated output. Calculate sustained output over a shift and not the peak figure in the brochure. Micro-stoppages add up quickly at a high cycle rate and appear in no report if nobody counts them. Ask for the rated output per format and for what the machine really achieves with your packaging material. Wear parts must be quickly accessible, because every minute of downtime costs many packs at a high cycle rate. A test of several hours at target output shows more than any promise.

A good HMI reduces operating errors and downtime: fault messages in plain language with an image of the fault location, a changeover assistant with a step sequence, recipe management with permissions, language switching for international teams. At the FAT, insist that your own staff operate the machine. That shows whether the interface is self-explanatory or only usable by the programmer. Watch for the small things that annoy in daily use. Buttons that miss with gloves on. A flood of messages in which the one important message disappears. I like to sit beside the operator at FAT and just watch where they hesitate. That is where the fault sits.

A holding brake rarely fails from too much holding. More often the switching sequence is to blame. If the brake opens before the amplifier has built up torque, the axis drops briefly. Should it close while the motor is still turning, it wears away a bit more lining every time. At acceptance, watch a vertical axis to see whether anything gives way at switch-on and at emergency stop. A centimetre of sag is not a cosmetic flaw, there might be a hand underneath one day. Also ask how the axis can be moved without power if someone is trapped or a part has jammed. Some brakes have a manual release for this, others need an auxiliary supply. Long brake cables with a thin cross-section cost voltage. The brake then only half releases, drags and gets hot.

Hood shrink wrappers are standard for beverages, building materials and outdoor storage, because the pallet becomes rain-tight and dimensionally stable. Drawbacks are energy demand, film consumption and fire protection requirements, since heat and gas burners need their own fire protection concept. Check whether stretch wrapping with a top sheet is sufficient for your transport chain. Often it is the more economical solution. With heat-sensitive products such as confectionery it pays to ask, because the pallet is briefly exposed to heat during shrinking and leaves the machine warm. Heat-up time and standby consumption belong in the calculation as well, since many plants do not run continuously.

Hot filling puts a load on the whole line. Containers must withstand the temperature without deforming, labels and adhesives must cope with cooling and condensation, and the closure must withstand the vacuum created on cooling. The cooling section costs space and energy. Cooled bottles arrive wet, and labellers and inkjet coders notice that first. The product also changes volume as it cools, so the fill level varies and can only be judged after cooling. With PET, the bottle’s vacuum panel design needs clarifying before purchase. Ask the container supplier for the permissible filling temperature and hold time, because hot-fillable is not a single value. If the product temperature drops too far during filling, the heat no longer sterilises the closure, so temperature monitoring belongs at the filler.

Adhesive cost is a running cost that can be calculated per case from application quantity, adhesive price and unit count. Dot application instead of a bead, the correct temperature and clean nozzles save material and prevent stringing. Adhesive that is too hot chars in the nozzle, adhesive that is too cold wets poorly. I like to check a bond by tearing the case open: if fibres stay stuck to the adhesive, it holds. On case sealers, it is worth looking at glue melters with automatic refilling. Manual refilling is a common cause of downtime on night shifts, and hot adhesive calls for gloves and goggles. If the cases go into cold storage, the temperature range of the adhesive belongs in the enquiry to the adhesive supplier.

For project work, consumer devices are not enough. CAD layouts and large assemblies need a machine with decent graphics performance, otherwise the screen stutters as soon as the whole hall is displayed. A document scanner with a feeder turns the pile of paper supplier documents into searchable PDF files and is therefore the foundation of any digital filing system. Plan for a printer for the large formats too. When a layout is agreed around the table, everyone wants a plan in their hand that they can scribble on. The scanner pays for itself the first time an inspection record or certificate is needed and is on screen within minutes.

At POWERJO, customer enquiries, appointment bookings and contact history run through HubSpot, including the forms on this website. The benefit for you: whoever enquires does not end up in one person’s inbox. Instead the enquiry becomes a case with an owner and a follow-up date. That way nothing gets lost because a colleague was on holiday or the email sat in the wrong folder. If you introduce a CRM yourself, keep the mandatory fields few. Whatever is annoying to enter stays empty, and sooner or later the database is full of gaps. Also decide who creates new contacts, otherwise the same company with the same contact person appears three times.

Dirt in the oil is the classic cause of hydraulic faults. Valves stick, pumps wear, and then the valve gets replaced while the clogged filter has long been running in bypass. At acceptance, check whether the filter contamination indicators report to the controller or just pop out a button on the filter housing. New oil is not automatically clean. Top up through a filter unit, not with a can and funnel. If the unit stands next to an oven in summer, the cooler has to be sized for that. Hoses age even when the machine stands still. The manufacturing date is on the fitting, and the DGUV, the German statutory accident insurance, publishes recommendations on service life. In food plants, every unit with lines near the product needs H1 oil, because sooner or later every hose drips.

For fitters, the hygiene station is often the first surprise on site. Nobody gets through the turnstile with a toolbox and a box of spare parts, and the route via the material lock has not been agreed with anyone. Clarify that with quality assurance before the first day of installation, including the rules on clothing, jewellery and mobile phones. Allow for the time, too. Anyone who switches between workshop and line several times a day loses a few minutes in the station every time, and with a larger installation crew that adds up over the week. In new builds or conversions, look at capacity. At shift change everyone wants through at once, and a station that copes in normal operation turns into a bottleneck.

Zoning determines layout, airlocks and conveyor technology. Cartons must not enter the high-risk area, and staff change clothing. Plan handover points and machine placement around the zones, or the line layout and hygiene concept will collide at the latest audit. Lines that cross a zone boundary are especially delicate. The wall opening must be cleanable, and for maintenance and fault clearance it must be clear who may reach the machine from which side. Also ask about the routes for spare parts and tools, because the technician who arrives with a spanner from the workshop otherwise turns the zone boundary into a line on paper.

Practical checkpoints when reviewing quotes are open frames instead of hollow sections, sloped surfaces, motors set back, enclosed cable routing, stainless steel 1.4301 or 1.4404 in the product area and IP65 to IP69K depending on the cleaning regime. When I receive quotes, I ask for the cleaning times on comparable installations. They are a real cost factor that does not show up in the purchase price, and they quickly show which supplier really designs for hygiene. At a reference plant, look at the places the cleaner touches every day: covers, guides, handovers. There you can see without a brochure how much time a changeover costs.

At POWERJO, IBScad is the tool for line layouts: machines, conveyor technology, escape routes, operating sides and maintenance areas in one scaled plan. A layout produced ahead of the tender prevents the classic surprises. The column is in the way, the door is too narrow, there is no room to work in front of the control cabinet. Draw in the swing of doors and the opening of guards, and keep the forklift routes clear, otherwise this only shows up during installation. Hall plans from the archive do not always match today’s state, so measure critical spots again. Attach the layout to the tender, then every supplier quotes for the same space.

At audit, what counts is what the new machine can do to your site. It puts the certificate at risk faster than people expect. The usual findings are glass inspection windows, cable ties that no detector can find, lubricants without food-grade approval and maintenance work without records. Bring the QA team into the URS and the FAT early, because they know the audit questions better than any design engineer. Have the supplier hand over a documentation package covering everything IFS touches: material certificates, lubricant list, list of glass and hard plastic parts, cleaning instructions. Anyone who asks for this only after delivery usually gets it with gaps.

IIoT rarely fails on the technology, mostly on purpose. Data gets collected without anyone having a question to ask of it. Start with a concrete benefit, for example downtime causes, compressed air consumption or the temperature trend in the sealing tool, and expand only afterwards. Clarify early who is responsible, maintenance or IT. Just as early, settle how the network is separated from the control level and who will maintain the sensors on existing equipment later on. Retrofitted sensors on old machines get knocked out of position or torn off sooner or later, and nobody notices. So plan a check that the readings still match reality.

IML delivers high-quality, wipe-resistant decoration directly during injection moulding or thermoforming, and there is no labeller in your line. In return the result depends on stable processes at the packaging supplier, because misplaced labels or bubbles only show up on the finished pot. Batch sizes must be sufficient, and changing the design is more expensive than with labels, because printing and cutting are set up for the motif. If the label is made of the same plastic as the pot, that fits well with a mono-material solution. Discuss this early with the supplier, before the pots are printed. Ask for the declaration of compliance covering the whole combination of pot, label and ink.

For machinery purchases DAP and DDP are the usual terms, and EXW mostly turns up with used machines. POWERJO’s tender package specifies the term so that all suppliers calculate on the same basis. I only weigh quotes against each other once the term is identical everywhere, otherwise apples sit next to oranges. The term alone does not settle it. Unloading, positioning in the hall, transport insurance and packaging should be written into the contract explicitly, because Incoterms do not cover them fully. Rely on the term alone and you end up at the ramp, looking at the low loader and wondering who ordered the crane.

An indexer is a piece of mechanics you rarely see and even more rarely touch. That is exactly why it gets forgotten when the machine is supposed to run faster. The cycle rate is quickly turned up at the drive motor, but the gearbox was designed for a certain indexing frequency and a certain mass moment of inertia. If the rotary table later gets heavier change parts and a higher speed at the same time, do the calculation first or let the manufacturer do it. Watch for play at the output. If the table wobbles during the dwell, the cam followers or the output bearing are usually worn, and positioning accuracy suffers at every station. Oil level and oil changes are in the lubrication schedule. They still get forgotten, because the gearbox sits deep in the machine base.

Seal quality depends on three things: the distance from the induction head to the closure, closure torque, and cleanliness of the container rim. Too little energy leaves leaks, too much burns the liner or deforms the rim. On changeovers, the head distance must be reproducibly adjustable, ideally with a scale and a recipe and not by feel. A seal looks the same from the outside whether it is good or not. So define a test method, for example peeling the foil and inspecting the seal ring, or a leak test, and record the test intervals in writing. At acceptance I have them show me how the test finds closures that were deliberately fitted crooked.

Anyone mixing industrial and Euro pallets needs palletisers, wrappers, conveyor technology and racking that can handle both formats. This is often forgotten in tenders. Define which formats need to run today and in five years, and have the pallet inspection check both formats. The changeover between formats is where it gets difficult. If an operator has to adjust guides by hand, that is skipped sooner or later, and the pallet enters the plant crooked. An automatic format change or a clean separation of the routes is better. Think about lorry loading as well, because the number of pallets per trailer changes with the format.

Gas is where people like to save money, until the complaint arrives. Calculate the consumption per pack at full output, not at half speed. A leaking seal seam makes the whole flush worthless, and you still pay for the gas. Spot-check the residual oxygen and record it per batch. Ask the supplier as well where the nitrogen is meant to come from: a tank, bottle bundles or a generator. That changes purity, pipe size and running costs. And look at how much air the product brings with it. Powders and loose bulk goods carry oxygen between the particles, and the flush has to drive it out first.

The cleaning method decides which IP rating is needed. Where high pressure and foam are used, IP54 components are not enough. Specify in the user requirements specification which areas are wet-cleaned and what IP rating motors, sensors and control cabinets need there. A control cabinet with an air conditioning unit in a wet area is a classic cause of consequential damage. The weakest link usually sits unobtrusively at the edge: a cable gland, a connector, the open cap on a sensor. Seals age too, and cleaning agents attack them. I ask for a list of components with their IP rating per installation location and check whether a hole drilled later has cost the rating somewhere.

Inspection plans grow. After every complaint a characteristic is added, and hardly anything is ever removed. At some point the line leader on the night shift is supposed to check more than can be done in an hour, and from then on boxes just get ticked. For every characteristic I ask what happens if the value is out of tolerance. If there is no clear answer, I put the characteristic up for discussion. With a new line it pays to revise the inspection plan together with the plant. What the checkweigher and vision system inspect on every pack does not also need weighing and checking by hand, as long as their function is tested regularly. Schedule the metal detector test at the start of the shift, at fixed intervals and after every format change. Hang the inspection plan where the testing happens. Nobody at the line reads it in the CAQ system.

The building site shows what the user requirements specification did not say. A door is too low, a floor anchor sits exactly under a machine foot, and the compressed air connection ends at the wrong end of the hall. Have the access route proven with dimensions before delivery, including the floor loading along the way. Clarify site rules and working hours, and who runs power, compressed air and network up to which point. Coordination of outside contractors is easily forgotten, and then the fitters stand at the gate making phone calls. Appoint a site supervisor who signs off daily what is finished. That saves a lot of work on the punch list, because nothing first surfaces on the last day.

The first question is trivial and still gets forgotten, namely what has to go into the hall first while the way is still clear. Once the conveyors are in, the big machine no longer reaches its spot, or only after partial dismantling. Ask every supplier for their own sequence and lay the plans side by side. You quickly see that three companies need the same corner on the same Tuesday. Also plan where the material will stand until it is installed. Storage space is tight in a running plant, and crates in an escape route cause trouble with the safety department. Crane times and closed walkways belong in the plan with dates, because production next door wants to keep running. And the electrician can only pull cables once the control cabinets are standing. Sounds obvious, but often it is written down nowhere.

On lines with several suppliers the interface list is the document that saves the most trouble. For every connection it states who supplies, who connects and who is responsible, whether it concerns mechanics, power, signals, utilities or software. Without it, the handover table, the start signal or the network port is missing during installation, and both sides are sure the other one was responsible. Have both suppliers countersign the list, because what only one side knows is not an interface but an assumption. Go through it before the order is placed and not just during installation. For each handshake, ask who comes to measure first when a signal fault occurs.

Interim project management brings experience into the business for exactly the duration of the project, without hiring and without a headhunter. That only works with a mandate. The project manager needs direct access to the decision-makers, clear budget limits and a fixed interface to purchasing and production. Otherwise they sit in a waiting loop and bill daily rates for meetings where nobody decides. A defined end belongs to it just as much. Whoever looks after the equipment after acceptance must be involved in the project from the start, so that the knowledge does not leave the building with the external person. Also define what must be on the table on the last day, such as documentation and the punch list. Have the contractual set-up checked in advance with purchasing and your lawyer.

Interlayers cost material, time in the palletising cycle and a magazine on the palletiser. First check whether a different pack pattern, an anti-slip coating or strapping can deliver the same stability. If you need the sheets, plan the magazine for a whole shift so that nobody has to keep topping it up. Automatic feeding needs a check that exactly one sheet is down, because two on top of each other or none at all let the pallet shift in transport. With reusable sheets, take-back comes on top. Who stacks them at the depalletiser and who drives them back to the supplier should be settled beforehand.

The choice depends on the required output and on the product. Delicate or hard-to-feed products often run more calmly on intermittent machines, because the transfer happens while stationary. Continuous machines need a synchronised infeed and products that can be reliably spaced. Intermittent mechanics accelerate and brake with every cycle, and the wear on the drives shows it later. Work out what net output your infeed really delivers before you buy a continuous machine that then runs indexed and never earns its premium. With printed film, the print mark detection has to cope with full speed. Try that at acceptance with your own film, not with the manufacturer’s sample reel.

The classic problem: the packaging developer calculates with internal dimensions, the logistics planner designs the pallet with the same figures, and the case ends up overhanging. Clarify with every drawing which dimension is meant, and measure the external dimension on the erected sample rather than calculating it. Double-wall board adds several millimetres per wall, which adds up over a pallet layer. Filled cases also bulge, so plan the pallet with the measured size of the filled pack. An overhang costs stacking strength, and the edges suffer in transport. Always state on the order whether internal or external dimensions are meant, so that the supplier does not have to guess.

Every three years Interpack gives you the quickest market overview you will get. Go with your URS and specific questions, not to be dazzled. Which manufacturers supply your format, what changeover times are realistic, what references exist in your industry. Book the important appointments beforehand, otherwise you queue at the stand and end up talking to the wrong person. Ask for the technician instead of the brochure. I like to have changeovers and cleaning access shown on the machine itself, because on slides everything looks equally quick. One day at the fair replaces weeks of research, but not an enquiry and not a reference visit.

Many special-purpose machine builders design in Inventor. As a customer, demand neutral 3D data alongside the drawings, for example in STEP format. Then your own designers or external experts can plan conversions and add-ons without being tied to the supplier’s tool. I also have the bills of materials supplied in editable form and write into the contract that the data is handed over in the state of the machine as delivered. Anyone who raises this only at acceptance may end up with PDF drawings alone, and every later change then starts with redrawing. Check the delivered files at handover by opening the STEP assembly once in your own system.

IO-Link pays off on the packaging line mainly for sensors that are frequently swapped or reparameterised. The parameter set sits in the master and a new sensor is written automatically, so no technician with a laptop is needed any more. The user requirements specification should state which sensors must be IO-Link capable and whether the diagnostic data should feed the control system or an MES. Otherwise you buy the capability and never use it. Many optical sensors report contamination as a diagnostic value before they fail. That only helps if the message reaches somebody. Try the real case at acceptance: fit a spare sensor from stock and see whether the line runs again without a laptop.

Outside pharma and medical devices, IQ, OQ and PQ are often overdone or not done at all. A lean middle path works: installation verification largely overlaps with the SAT protocol, functional verification with the acceptance criteria from the URS, and performance verification with the performance test. Merge the three documents beforehand and save yourself weeks of paperwork and arguments with the supplier. Give the supplier your templates early, ideally with the enquiry. Otherwise protocols arrive in his format and you rewrite everything. Watch the instruments. Without calibration certificates for the sensors used, any installation check can be challenged.

In a project, ISA-95 helps mainly as a common language between IT, automation and production. The levels clarify who is responsible for what, which data goes up and which instructions come down. That helps when, in an MES project, IT talks about orders, automation talks about signals, and both mean the same thing. Draw the level model once with the real systems in your plant. Then it quickly shows where data is maintained twice and where there are gaps. A standard does not replace deciding which KPIs you actually need. And it does not tell you which interface your supplier really delivers.

Many industrial customers expect suppliers and service providers to have quality management to ISO 9001, often already at the supplier assessment stage. As a buyer, look at the certificate itself. Check that site and scope cover the part of the business that handles your order, and that it is still valid. A certificate replaces neither a supplier audit nor a reference visit, it only says that a controlled system exists. POWERJO is preparing for certification: documented processes for enquiry handling, expert selection, project support and handover. For you, that means traceable, repeatable quality instead of a one-off.

The topic becomes tangible as soon as spare parts and change parts are made again. Anyone having parts manufactured to drawing needs fit and surface-finish specifications, or the shaft will bind or the bearing will creep. Require manufacturing drawings with full tolerances for wear parts. A drawing without tolerances invites trouble, because if nothing is stated, the general tolerance in the title block applies, and that is far too coarse for a bearing seat. For 3D-printed change parts, tolerances need to be assessed differently than for machined parts. Tell the supplier which surfaces get reworked, such as bores for bearings or pins.

On packaging lines, Kaizen rarely fails for lack of ideas. It fails for lack of capacity to implement them, because the operators’ suggestions land on a list that nobody works through. A fixed weekly slot with maintenance and the line lead has proved itself, in which three small things actually get done, such as a stop, a marking or a changed sequence during changeover. The operator who had the idea should see the result at his own machine. Then the next idea comes by itself. Clarify beforehand who may release small changes, otherwise every marking waits for a signature. For safety topics and changes to the control system, the normal change route applies.

A good kick-off takes half a day and clarifies three things: who decides what, which dates are fixed, and what counts as done. Production, maintenance, quality, procurement and the supplier sit at one table once, before the first drawing exists. I send out the agenda and the URS beforehand, so that nobody turns up unprepared and everyone knows why they are needed. Skip the kick-off, and you will clarify the same things later in escalation meetings, with the schedule slipping. Invite the people who will have to live with the result, meaning operators and shift leaders. They have no authority to decide, but they know what otherwise goes wrong on the line.

For chilled and export packaging, use kraftliner outer facings. For dry domestic distribution, testliner is usually sufficient. Suppliers are happy to quietly swap the facing grade for a cheaper one if only the corrugated board grade is named in the framework agreement. So specify the paper composition, the facing basis weight and the minimum ECT value, and spot-check on goods-in. Damp cold stores are a case of their own, because corrugated board quickly loses stiffness there. For new cases, have the board construction written into the specification and not just the grade name, since the same name can hide different papers.

Start with the question of which fault is really costly: the wrong label with a recall, the missing one with a complaint, or the crooked one that only hurts appearance. A barcode scanner after the labeller prevents the recall at low cost. Camera systems for position and wrinkling are more elaborate and need tolerances agreed with the customer, otherwise you generate false rejects. I would only choose the camera when the scanner is no longer enough. Check the setpoint at changeover, because otherwise the label for format A is mercilessly checked against the setpoint for format B. And clarify in advance who releases after a false alarm.

Every material has its use and its limits. Paper labels go soft and tear in cold storage, PP labels shrink on warm bottles, PE labels stretch during dispensing. Thermal paper fades with heat and light, and you notice at the latest when the barcode stops scanning after a few weeks. So test the label on the actual container under real conditions: wet, cold, after weeks. On tight radii a label that is too stiff lifts at the edges. Anyone changing the material has to adjust the dispensing edge, liner take-up, print head temperature and ribbon of the thermal transfer printer at the same time. Otherwise the labeller runs unsteadily afterwards.

When comparing machines, look at what the machine does with your actual container and not at the brochure value. Ask for the labelling accuracy in millimetres and have the roll change during operation shown to you, for example with a double unwinder. Camera-based label inspection should catch the wrong type and an offset label. Common causes of faults are static charge that makes labels dispense badly, rolls with the wrong winding direction or the wrong type, and containers that vary within their tolerance. If the bottles arrive from the filler wet or misted, even the best machine struggles. On roll-fed beverage lines, glue application and film quality decide the result.

The typical weak point of laminated structures is delamination. The layers come apart, usually at seal seams, at folds or after hot filling. Causes are too little curing time at the manufacturer, moisture or components from the printing inks. Ask for data on laminate bond strength and for food-contact approval of the adhesive. With solvent-based adhesives ask about residual solvent, because odour on the film ends up in the product. Freshly laminated rolls going straight from the lorry onto the machine are a warning sign. I always ask the film supplier how long a roll was stored before it left the plant. That belongs in the specification and not in the gut feeling of the dispatcher.

An LLM is useful when it can access your documents and cites its sources. Without that connection it invents convincing but wrong spare-part numbers, and they look just as believable on the screen as the right ones. Clarify before use what data the model may see, where it runs and who checks the answers. Design data and customer documents do not belong in a public service without asking first. I start with tasks where a mistake is cheap, so search and summarising, and not with changing machine parameters. Give your staff a clear brief on what the tool is meant for and what it is not.

A laser can only mark what reacts to it. On some films and inks the contrast stays weak, and only a laser-reactive field in the print design helps. Have all packaging materials sampled beforehand, including the variants from the second source. Extraction is mandatory and often forgotten, as is the protective enclosure matching the laser class. When switching from an inkjet printer, weigh the saved consumables against the higher investment. Also clarify how the laser switches off when the conveyor stops and who maintains the extraction filters. The enclosure doors should be interlocked, because a door that can be left open will be left open sooner or later.

A layer palletiser handles a lot, as long as the packs cooperate. The tricky moment is when the finished layer moves from the table onto the pallet. Slippery shrink packs or a layer pattern without proper interlocking come apart at that point, and the line stands until someone straightens the layer by hand. Ask before ordering who will be allowed to create new layer patterns later. If only the supplier can do it via remote access, every new item number costs you an invoice and a few days of waiting. People also tend to underestimate everything around it. An empty pallet magazine, a tier sheet dispenser and the conveyor run to the stretch wrapper all need space. And broken pallets from the exchange pool stop the line just as effectively as a defect in the palletiser itself.

Have the palletiser supplier draw the layer patterns for every format and get them signed off by logistics and the customer beforehand. Retail chains have their own requirements on pallet height, overhang and number of layers. An interlocked pattern is more stable, a column pattern is gentler on the packs. Both have consequences for the wrapper and for transport. If a layer pattern changes after commissioning, it costs programming effort and usually new gripper tooling. Also ask how the plant deals with an incomplete layer at the end of an order, because the last pallet of an order is rarely full. Pack dimensions vary in practice, so sign-off should be done with real samples and not just on screen.

Layer picking pays off with many mixed pallets built from whole layers, while for single cases other systems are more suitable. The stability of the layers, slip sheets and gripping technology for different case qualities are the difficult parts. A layer that holds on the source pallet sometimes comes loose when gripped, especially once the film is off. Check your order structure with real numbers before comparing technology. That includes how often whole layers are ordered and how often just a few cartons. Only then do you know whether a layer picker or a system for single cartons fits. Insist on a trial with your real cartons and pallets before you order.

Machines are the easy part of a layout. What lies between them is harder: buffers, access, pallet spaces, routes for material and people, cleaning and maintenance. Plan a day’s movements into the 2D layout, not just the machine outlines. Draw in where the change-part trolley stands during a changeover, where the forklift stops with the empty pallet and how the maintenance fitter reaches the motor. Go through the layout with a shift leader and a forklift driver before it is frozen, because they know the routes that appear on no plan. A layout that crosses forklift routes or hinders changeovers costs time every day, for years.

A quote shows a lead time, and what is often meant is: from order clarification. That takes a while if samples, format list and drawings are missing, and meanwhile the supplier’s clock is not running. Write into the contract when the period starts and what you have to supply by then, and keep to those dates yourself. Otherwise the supplier points at you the first time something is late. Also ask about lead times for change parts and critical spares, because they later decide how long a stoppage lasts. A motor is found quickly, a special gripper is not. Put the same question for bought-in items such as the control system and drives, since the machine can only be delivered as fast as the slowest of them.

Most leaks do not originate at the machine but from product in the seal seam, film wrinkles or damaged pack-material edges. A sample taken once an hour at the water bath finds systematic faults but not sporadic ones. If leak complaints are an issue, check the feed, the dosing and the seam area first, before investing in an inline test. You then have to manage its reject rate, and it only tells you that something is wrong, not why. Look at the seam under good light and with a magnifier, you can usually see the cause. Define who tests, how it is documented and what happens to the batch when a value goes out of range.

Leasing gets an investment going without a large cash outflow, but costs more over the term. Buying pays off with long-lived machines that simply keep running after depreciation. Many packaging machines do that if they are maintained properly. Run the numbers with your tax adviser before you commit, because the tax treatment can look quite different depending on the type of contract. Settle in the lease who bears maintenance, spare parts and retrofits. After a few years a control system replacement is often due, and by then it should be clear who pays for it and what happens to the modification on return. Nobody likes rebuilding a machine that belongs to someone else, so settle it before you sign.

Lessons learned only work if they happen shortly after the project and are honest. As blame they are useless, as a list for the next project they are very good. Write down what was missing from the URS, which interface was underestimated and which supplier kept its promises. If the knowledge is only collected after half a year, nobody remembers the reason for the late rebuild. Two pages are enough. File them in the project folder and read them out at the next kick-off, and they are worth more than any debrief meeting. Enter the points into the checklist straight away, otherwise it stays with good intentions.

If internal approval is still running but design work is meant to start, an LOI helps. The supplier gets planning certainty, and you get your delivery slot, which is scarce when other customers compete for the same production dates. Before signing I always ask what it will cost me if the project falls through after all. Write that in explicitly, for example cost coverage on cancellation and which stage of design is then paid for. Draft the LOI with procurement or a lawyer, because one that is too loose achieves nothing and one that is too strict is already a contract. Give it an expiry date as well. An LOI without one is still hanging on the project when the plans have long changed.

Foam, vapour and build-up throw level sensors off, and a filler has plenty of all three. Over a foaming product an ultrasonic sensor either loses its echo or measures the top of the foam, and vapour above hot product changes the speed of sound and with it the reading. On capacitive point level switches a film of product clings to the tip, and the sensor reports full although the vessel emptied a while ago. At acceptance, insist on a test with real product, after cleaning and with warm product as well. Water is no substitute for cream. Think about hygiene too. A sensor that reaches into the vessel from above goes through the cleaning cycle with it, and its process connection must be every bit as hygienic as the rest of the tank.

Light curtains guard handover points on palletisers and conveyor technology, where material must pass through but people must not. With muting, the protective field is briefly bridged for the pallet. Incorrectly positioned muting sensors are one of the most common causes of nuisance stops and of tampering by operators. Plan the pass-through together with the safety specialist, not just the electrician. Pay attention to the mounting as well. A light curtain knocked out of line by a forklift will, at best, announce itself through nuisance stops. Ask for sturdy brackets and impact protection. Build the regular test with a test rod through the whole protective field into the maintenance plan. It takes minutes.

Most vision problems in operation are lighting problems. Stray light from an open hall door, aged LEDs, dirty diffusers, glossy films. Require an enclosed lighting cabinet or at least stray-light shielding, strobed lighting and a reference test with a test sample at the start of each shift. LEDs get dimmer over the years without anyone noticing, until the reject station suddenly misbehaves. Plan a regular check of the brightness for that reason. I insist that the inspection station is designed with your product and your packaging materials, not with the integrator’s sample carton.

Most mislabelling comes from leftover stock. A partial roll left on the labeller from the day before, blanks in the magazine, caps in the chute, and the wrong print ends up on the wrong goods. Line clearance only works with a checklist per machine, clear areas for leftover material and sign-off by a second person. The printer belongs on that list. A best-before date from the last job goes unnoticed until the retailer rings. When buying a machine, make sure magazines, chutes and buffers are fully visible and can be emptied completely, otherwise the check becomes a formality. Build the time for it into the changeover, or it gets cut short on night shift.

Buy every machine at the same rated output and you have no line compensation. Then any stop anywhere stops the whole line, and the filler stands idle although nothing is wrong with it. The buffers need floor space and a control system that adjusts speed to fill level. Both belong in the layout early, because adding them later is tight and expensive. The filler is the pacemaker, and the rest follow it with headroom to spare. At the performance test, have it demonstrated that the line really works its buffers down again after a stop. If it runs flat out all the time, that reserve is missing from the design. Write the capacities of all machines in one table, and you will see before ordering where the V is missing.

The work in line integration lies in the interfaces: signals between machines, buffers, speed control, a shared operating concept, and the question of who stops on a fault. Clarify the interface list before machines are ordered, and name someone responsible for the whole line. Without that person, every supplier points at another at acceptance. I have the sequence of start-up, stop, fault and run-out drawn as a diagram before programming, because contradictions show up there that a list never reveals. Also ask for a line test with all machines running together, if possible with product, before the first supplier leaves site.

A good line lead delivers more output than some investments, if they have time for the line instead of paperwork. Involve line leads early in the URS, the FAT and commissioning, not just for the training. They know the faults that appear in no report, and they will be the ones running the new plant on shift afterwards. At the FAT, let them carry out the changeovers themselves instead of watching the supplier demonstrate them. Anyone who cannot get a changeover done briskly there will certainly not manage it at three in the morning. Ask, too, which routines on the old line exist only in somebody’s head before they retire with that person.

Start the format change for the whole line with one click, that is how line management systems like to be sold. It only works if every single machine actually masters the interface. If it is not in every purchase order, it comes later as a variation order, and from each supplier separately. Clarify early who keeps the signal list for the whole line and who answers in the end for everything starting up together. Format numbers are a fine classic. The filler knows the format as 12, the cartoner as 7, and the palletiser works with a name. Somewhere you need a mapping table, and someone has to maintain it for every new item. At acceptance, also test what happens when the line management system fails. The machines should then be able to carry on locally, otherwise one server stops the whole line.

Pallets of blanks, film reels and labels need more space at the line than the layout usually allows. So they end up in the aisle, and the forklift reverses between the operators. Work out how many pallets stand at the line per order and how often they are replenished, and draw those areas in before the layout is frozen. Changeovers show whether supply works. In the good case the material for the next order is ready when the last carton has gone through. Otherwise the whole line waits for the forklift. In changeover analyses I always ask about this waiting time, because it tends to be booked as changeover time and nobody looks for the cause. Remaining quantities need a fixed route back to the warehouse, with batch and quantity. If not, they pile up behind the machine, and one day someone runs the wrong film.

On the supplier’s layout a single little figure stands at the line. In operation it turns into three, because one fills the blank magazine, one fetches film reels and one empties the reject bin at the checkweigher. The machine runs automatically, but it is still supplied by hand. So at the FAT, ask how long the magazines last and who refills them. Have the operation demonstrated for an hour and count the hand movements. In the business case, look at the breaks. If the line runs through, it needs relief cover, and that appears in no calculation. Think about the night shift as well. If the experienced line leader only works the early shift, staffing at night is the same on paper but not in practice.

Linear guides die from dirt and lack of lubrication, not from load. In wet or dusty areas, wipers, covers and central lubrication matter more than a higher load rating. At the design review, check accessibility of the grease nipples and make sure guides don’t sit under zones where product can fall. Stiff format adjustments are almost always dirty guides. Before anyone turns at the drive, push the carriage by hand and feel whether it runs evenly. Rough spots or jerking point to wear or foreign bodies. Ask the supplier for the lubrication interval and the grease per guide, and whether carriages can be exchanged individually.

Load cells rarely fail on their own, but often because of their surroundings. These include vibration from neighbouring machines, draughts from ventilation systems, compressed-air blasts and full-jet cleaning hoses. Plan a decoupled mounting location and protect the cell mechanically against overload during changeovers, because a hammer blow or a change part set down on the platform can throw it off for good. Product residue under the weighing platform and a tightly stretched connecting cable also falsify the reading. Calibration intervals and test weights belong in the maintenance plan, otherwise the legal-for-trade scale eventually becomes just a display. At acceptance I like to put a test weight on, take it off again and check that the zero comes back.

Load securing starts on the pallet: pack pattern, slip sheets, stretch force and, where needed, strapping decide whether the goods arrive intact. Check pallet stability with a simple tilt test or transport trial before cutting back on film. If you reduce film thickness, you must reset the pre-stretch and the number of wraps, and align the load unit with the carrier. After a format change the wrap programme and pallet height often no longer match, because nobody carried the recipe over. Badly wrapped pallets usually give way first at the upper layers and at the corners. Complaints about shifted goods show most honestly whether the securing is enough.

The load unit is the interface to the carrier, warehouse and retail customers: height for racking and the lorry, weight for the forklift, overhang for aisle width. Define the load unit before packaging development. Many projects optimise the case and only later find that the pallet does not fit the racking. In the first conversation I therefore ask for the largest pallet that the most demanding customer will still accept, and calculate backwards. Then it is settled how high the stack goes before anybody touches a carton drawing. Include the return of empty pallets and interlayers in the calculation too, as it is easily forgotten.

Without a complete loop check, commissioning starts with troubleshooting instead of functional testing. Swapped sensor pairs, wrong addresses and wrong rotation directions cost days if they are only found once in automatic operation. A sensor only counts as being on the right input when someone operates it at the machine while a second person watches the screen. I have the list signed off jointly by the supplier and the plant electrician. After that the electrician knows every terminal, and that pays off when something does not run at night. Safety circuits such as e-stop and guard doors are checked separately and with a protocol.

LOTO is so tricky on packaging lines because machines sit close together and energy comes from several sources. Electricity, compressed air, hydraulics, stored energy in springs or raised loads. A procedure per machine, lockable main switches and compressed-air shut-offs belong in the URS, and so do bleed valves that really release the pressure. One lock per person and a multi-lock hasp for several fitters cost little and save a lot of argument. At acceptance I walk through the plant once with a lock and try whether everything can really be locked off. On retrofits, check whether new energy sources were added without updating the procedure.

For frozen products, films and adhesives must still seal and bond at deep-freeze temperature. Condensation when packing cold products is a classic cause of leaking seals and slipping labels. Climate control in the packing area and material testing under real conditions are therefore part of the job, with the real film, the real label and the product at storage temperature. Polypropylene can turn brittle in the cold, so have the cold suitability of the film grade confirmed. Check the route from the freezer to the machine as well. If goods thaw in between and refreeze, the label ends up on a damp pack.

Many lubrication schedules stay in the binder. They sit in the operating manual, are written in the manufacturer’s language and name a grease nobody in the plant has heard of. Before acceptance I therefore ask the supplier for the schedule as an editable spreadsheet, so it can go into the maintenance system without retyping. Compare the lubricants with what you already keep in stock. Nobody needs a new grease for every new machine, and each extra type raises the chance that the wrong cartridge ends up in the grease gun. Check whether the intervals suit your shift pattern too. If the schedule says weekly, ask how many shifts the manufacturer assumed. In three-shift operation a week adds up to several times the running hours.

You will usually be shown the Cmk at the FAT, and it almost always looks good. No surprise, since the machine has just been set up and is run by the supplier’s best fitter. Insist that the sample is produced with your product and your packaging material. Check the measuring equipment too. If the scale itself scatters or has too coarse a resolution, its error ends up in the calculated spread, and the index says little about the machine. A measurement system analysis beforehand saves you that argument. Besides the cover sheet with the green number, ask for the individual values. Do not be surprised if the Cpk turns out lower in production later. That is when changing batches, several shifts and a summer in the hall come into play.

The technical capture is the smaller part. Quality stands or falls with how downtime reasons are assigned. If operators have to choose from fifty reasons at every stop, they pick the first one in the list. Limit the list to a few clearly distinct reasons and let short stops be assigned automatically. Feed the analysis back to the shift, otherwise data discipline drops within weeks. At acceptance I look at what is in the system after a shift change. That shows whether stops are really given a reason or whether everything ends up under Other. Anyone who cannot hit the operator terminal with gloves on will not enter anything anyway.

The frame is an easy place to save money, because it has no rating plate and takes up one line in the quotation. You will notice later. If the frame vibrates, sensors drift out of adjustment, bolts work loose and the machine gets louder the faster it runs. Food plants add cleaning to the list. Hollow sections that are not sealed by welding take in water and release it for a long time, sometimes with a smell. Ask the supplier whether tubes are seal-welded or open and rinsable. Take a look at the feet too. A frame on a tiled floor with a fall to the drains needs levelling feet with enough travel, and the lock nuts must be tightened after alignment. Otherwise the machine is out of level after the first session with the pressure washer.

People like to argue at length about the costs in the numerator, yet the real assumption sits in the denominator. If controlling has calculated with three shifts and the line runs on two, the rate is too low and every calculation based on it looks better than it is. So ask about the planned hours before you take a rate into a business case. With investments the hourly rate can also mislead. A new, faster line usually has a higher rate than the old one. That is no argument against it. Compare the cost per unit, meaning the cost per pack, because the new line turns out more packs in the same hour. For make-or-buy decisions on change parts, take an honest look at your own workshop. The milling machine there often stands idle more than it runs, and with realistic hours your own production is no longer as cheap as it looked.

Send packaging material from current production to the trial, not hand-made samples from the sample department. The film from the sample book always runs. Your product should arrive the way it arrives at the line, with the temperature, humidity and shape it has in production. Chocolate at room temperature in a test centre behaves differently from chocolate straight out of the cooling tunnel. I prefer to go to these trials with someone from the line who handles the product every day. That person notices things a project engineer overlooks. Have every parameter written into the report and film the critical points. Watch the speed. A trial at half output proves that it works at half output. Also take samples home that you can later lay next to the samples from the FAT.

Under the Machinery Regulation, new rules apply from 2027, including on cybersecurity and digital instructions for use. For your project this means: clarify with the supplier under which legal basis the machine is placed on the market, and get that confirmed in writing. What counts is the date of placing on the market, not the date of the order. If delivery slips past the cut-off date, the legal basis changes, and with it the documents and evidence. For substantial modifications to existing machines, you yourself become the manufacturer, with all the duties that brings. Operators like to overlook that when rebuilding. Involve your safety officer before the rebuild order goes out.

A maintenance contract pays off for machines whose failure is costly and for which you have no in-house specialist knowledge. Look at the content: whether wear parts are included, whether remote support is available and how quickly a technician is actually at the machine, at night and at weekends too. Clarify what response time means in concrete terms. Sometimes someone merely picks up the phone, and the technician arrives at the machine the next day. I always have the consequences of non-compliance written into the quote. Then compare the contract cost against the downtime cost of one shift. The decision is then usually clear.

Manufacturer intervals are starting values, not the truth. The manufacturer knows neither your shift patterns nor your product, and a dusty product loads bearings and guideways differently from a clean one. Transfer the plan into your system before handover. During the first year, review every maintenance task for effort and findings, and adjust the intervals afterwards, upwards as well as downwards. A task that never finds anything may become less frequent, and one that finds something every time comes round more often. Statutory inspections are excluded from this. A classic failing is the maintenance that cannot take place in the shutdown window because the parts are missing. So ask for the parts list with the plan and put the wear parts in stock before the date arrives.

One strategy for the whole plant sounds tidy and almost never fits. The geared motor on the infeed conveyor, which can be swapped from stock in half an hour, can safely run until it fails. On the other hand, the filler that sets the pace of the whole line as its bottleneck deserves fixed intervals and regular inspection. That is why I first rank equipment by the consequences of failure, before anyone talks about intervals. Predictive maintenance sells well at every trade fair. It only pays off where a failure really hurts and the measurement reliably tells you something about wear. And servicing needs feedback. A maintenance plan that ticks off tasks but never records what was found learns nothing.

In-house hours are happily treated as free, because the people are there anyway. That changes once they are tied up in the project. Then they are missing from maintenance, and breakdowns on the other lines pile up. So cost your own work at an honest hourly rate and include whatever is left undone in that time. When you link up a line yourself, there is a second point. Whoever combines machines into an assembly usually takes on the manufacturer’s obligations for that assembly, including the risk assessment and CE marking. You have to be able and willing to do that. For change parts and spare parts, having them copied by the machine shop round the corner is often cheaper. Check beforehand whether drawings exist and whether you are allowed to copy the parts. And the machine supplier’s warranty should still stand afterwards.

A skewed manufacturer’s joint gives a case with a parallelogram cross-section that jams in the erector and tips over on the pallet. Check the sleeve dimensions diagonally on goods-in, not just length and width. Internal glue flaps reduce the clear internal dimension and cause trouble with tightly fitting products. Stitched joints are usually unwelcome in food plants, because metal staples are a foreign-body risk. Specify the design in the specification. When there is a complaint, what counts is how many cases on the pallet are affected. A single skewed case is bad luck, a pallet full of them is a production fault at the supplier.

Protective gas extends shelf life, but only in a tight pack. Residual oxygen has to be measured by spot checks, and leaking seals devalue the whole batch. A common cause of leaks is product in the seal, such as crumbs, fat or sauce at the edge. When buying the machine, look at gas consumption, flushing times and the integration of a leak test. Over the machine’s lifetime the gas costs are a noticeable item, and consumption depends on how cleanly the machine is set up. Also check who calibrates the residual oxygen analyser and where the readings are recorded. Without records the measurement is worthless in a complaint.

New items tend to arrive at the last minute, and the master data are then created by someone who has never seen the line. If the pallet factor in the system is wrong, the palletiser dutifully builds the wrong packing pattern, and the pallet label is wrong as well. When purchasing switches to a thinner film, the old packaging weight often stays in the master data, and the checkweigher calculates with the wrong tare. So on projects I ask early on who maintains which master data, who approves them and which values go straight to the machines. I rarely get a clear answer the first time. Settle this at the latest when the format list for a new line is drawn up. The same dimensions and weights turn up there again, just in a different file and at a different revision.

Material flow is about routes, stock levels and waiting times between goods-in, production, packaging and dispatch. A routing analysis makes it measurable. Clean layouts with short routes for packaging materials and finished goods often achieve more than any amount of automation. For this, follow a forklift for one shift and draw in the routes. Often the packaging material sits at the far end of the hall because that was where there was room during planning. Ride along on the third shift as well, because at night the flow looks different from the day. Count how often the same pallet is touched before it leaves the plant.

Before you argue with the supplier about a Cpk, agree what you are measuring with. Otherwise you end up discussing the scales rather than the machine. A simple trial is enough to start with. Ten packs, three people from different shifts, and each measures every pack twice without knowing the others’ readings. If the results are far apart, the measurement system has a problem, however expensive the instrument was. Often the method is to blame. One person squeezes the calliper hard on the carton, another only lightly, and both believe they are precise. So write the measuring instruction in a way that a new colleague on the night shift applies it exactly the same. Take care with destructive tests such as seal strength or opening torque. The same part cannot be measured twice, which calls for adapted methods and someone who knows them.

Cam mechanisms have a poor reputation they do not entirely deserve. A well designed cam runs at the same rhythm for many years, with no parameters, no software and no encoders to fail. The price is rigidity. Every new format with a different stroke needs a newly machined cam, and that takes time. With used machines and retrofits it therefore pays to look closely at the cam tracks. Worn spots or pitting show where the roller is hammering. Often it lifts off at high speed because the return spring has weakened, or a seized roller slides instead of rolling. Cam followers are wear parts and belong in the stores. If the old cam is to be replaced by a servo, ask who is going to measure the motion profile first. The cam drawing is missing surprisingly often.

An MES only pays off once the lines deliver clean data: downtime reasons, quantities, formats, batches. Many companies buy the software first and then find the machines do not provide usable signals. The better route is to measure on two lines first, manually or with simple downtime capture, draw conclusions, and then define the MES requirement, with an interface list in the user requirements specification. I ask who maintains the master data before any MES discussion. Articles, formats, bills of materials and fault reasons change all the time, and an MES with outdated master data produces figures nobody believes. Plan the upkeep with names and a time budget, not on the side.

Metal detector or X-ray is not a matter of belief but depends on the product and packaging. Aluminium laminate or metallised film blinds a metal detector, leaving X-ray as the only option. X-ray also finds glass, stone and dense plastics, but costs more to buy and run. Wet, salty or warm products disturb a metal detector through the product effect, which costs sensitivity, and you only notice it in a trial run with your real product. Put the unit as late in the line as possible, after the last point where anything can still get in. For IFS and BRCGS, documented test-piece trials and rejection with confirmation count. An alarm the line does not react to helps nobody on night shift.

Micro-stops can only be identified by measurement. Either the controller records every stop automatically, or you have a tally sheet kept at the line for a week. The causes mostly lie in the infeed, the packaging material quality and the sensors. Ten stops of 40 seconds an hour are almost seven minutes of downtime in every hour, and not one of them shows up on the fault list. The treacherous part is that the operator clears them so routinely that he hardly notices them any more. Ask him anyway where he intervenes most often. If the controller counts the stops, have the evaluation split by stop location and not just the total. Most of these stops can be fixed without any investment.

The film supplier’s migration test applies to their test conditions, not automatically to your product, your filling temperature and your shelf life. Check whether the simulant, contact time and temperature in the declaration of compliance cover your application, especially for fatty products and hot filling. If the pack is later heated in a microwave or oven, that needs its own assessment. Printing inks and adhesives on the outside can transfer to the inside through set-off in the reel, and that also needs assessing. If the supplier changes the film recipe, the ink or the adhesive, the old test no longer applies. Have him commit in the framework contract to reporting such changes beforehand.

The supplier has to pre-finance material and bought-in parts, and you want to keep some pressure on until the plant runs at your site. A milestone payment brings both together. The argument almost always comes over the final share. If it is too small, the punch list becomes a sideshow for the supplier. A payment on delivery also means paying for a machine that has not yet packed a single product in your plant. Write the criteria so that nothing is open to interpretation: a passed FAT with a signed protocol, not roughly finished. And have the deposit covered by a guarantee. Get the wording checked by your lawyer.

The MTA lives on honest forecasts. This pattern is typical: the delivery date stays flat for months and then jumps by weeks shortly before delivery. That means nobody followed up, they just copied the date forward. Ask the supplier for the facts behind the date, such as drawing status, order date for long-lead items and start of sub-assembly. A date that only moves when you ask says something too. Run the MTA in the project status report and not just in the scheduling tool, so that the steering committee sees the lines. Show the curves to the supplier as well, and the talk is about drawing status instead of blame.

The milk run replaces a forklift that runs on request with a fixed route on a fixed cycle, for example packaging materials to the lines and empties and waste back every 30 minutes. It works when staging locations are defined and quantities match the cycle. A routing analysis shows beforehand whether the route is even worthwhile. With three lines within 50 metres of each other, a tugger train is often overkill. I also look at what happens when a line needs something urgent after the round has passed. If there is no rule for that, somebody soon sets off on their own with the forklift again, and the milk run has only added cost.

POWERJO uses Miro for layout workshops, process capture and decision overviews. Everyone involved sees the same picture, even when the plant, head office and supplier are in three different locations. A photo of the shop floor plus sticky notes is often the fastest route to a shared understanding. In process capture one person writes while the production staff talk, and the group corrects as it goes. That saves the follow-up query after the meeting. Set up a structure before the workshop, for example columns for current state, problems and actions, otherwise it ends in sticky-note chaos. Transfer the decisions into minutes after the meeting, because hardly anyone reads a board full of sticky notes afterwards.

Cases are usually designed for the single-article pallet, with column or interlocked stacking and a BCT value to match. On a mixed pallet the same case suddenly sits over a gap or on two different neighbours and gives way. The complaint then arrives from the retailer’s distribution centre. With automated mixed palletising, the first thing I ask about is master data. If dimensions and weights in the system are wrong, the robot builds leaning towers, and does so very reliably. Have new articles measured and weighed before start-up. For hand-built pallets, a simple packing instruction with a picture at the picking line helps. Otherwise the heavy drinks cases end up on top of the crisps sooner or later. Plan the wrapping as well. A mixed pallet with steps in height holds less well with the same wrapping parameters than a flat one.

Setup mode is the most dangerous state of a plant because guards are partly bridged. Set rules for who gets the key and where it is kept, because a key that stays permanently in the switch defeats the whole concept. For a robot, setup mode allows a reduced speed of no more than 250 mm/s at the tool. Make sure switching modes is not possible in the middle of a motion. At acceptance I try out what happens when the mode is switched while the plant is running, and whether it then starts again by itself. It should stand still and wait for a deliberate acknowledgement of the new mode.

An AI model is not a test instrument that you calibrate once and then leave alone. Define metrics that are monitored continuously, such as rejection rate, share of uncertain decisions and false rejects. Take regular samples and have people assess them, because that is the only way to notice whether the model is still right. Define who decides when drift occurs and how retraining gets approved. Otherwise every setter turns the thresholds until the line runs again, and eventually nobody knows what was originally set. Tell the person who looks after the model in advance when packaging material or lighting is changed.

A modular belt forgives a lot, but not everything. The rods gradually work their way out sideways once their retention has worn, and a broken piece of module can end up in the product. That is why the food industry likes to use detectable plastics that the metal detector will find. When choosing a belt, look at the hinges. Dirt collects there, and open designs are far easier to clean than closed ones. At transfers, small or tall products like to topple over. Finger transfer plates or dead plates help, but they have to match the belt type. Keep modules and rods of exactly the installed type in stock. Series from different manufacturers look alike and still do not fit together.

Mono-material is the response to the recycling requirements of the PPWR, the EU packaging regulation, but it brings changes to the process. PE or PP mono-films seal within narrower temperature windows, run differently over the forming shoulder and often need adapted sealing tools. Always test new films on the actual machine at rated output, not just in the lab. Factor the changeover costs into the material switch. Ask the film supplier whether the film stays within the window when the thickness varies, because with narrow windows every variation shows. Whoever optimises the sealing parameters for the sample reel alone will have scrap again on the next delivery.

If a motor protection circuit breaker trips more often, sooner or later someone turns the setting up. After that all is quiet until the winding burns out. The trip is usually a symptom. Behind it there is a stiff chain, a dry bearing or a conveyor starting up against a backlog. At acceptance I have the supplier show me that the set values match the rating plates, and I put them into the documentation. If a motor with a different rating is fitted later, the breaker has to be adjusted as well, and that is easily forgotten. On motors fed by a frequency inverter, the inverter or a thermistor in the winding takes over motor protection. The breaker upstream then only sees the input current of the inverter and is set accordingly. Wiring the auxiliary contacts to the control system costs little. The fault message then tells you straight away which drive has tripped.

MQTT is simple, and that is exactly why topic names and units quickly get out of hand. One machine reports in degrees, the next in tenths of a degree, and everyone looking at the dashboard wonders about the curve. So define a topic structure and a data model before the first machine starts sending, and decide who may create new topics. Otherwise there will soon be three spellings for the same temperature. Also check whether the broker runs redundantly and who operates it, because if it fails, every dashboard goes blank. MQTT is not intended for hard real-time or safety functions. Fieldbus and safety networks remain responsible for those, and the specification should say so.

Excel is still found everywhere in project business, for budgets and schedules as much as for format lists. What matters more than the tool is discipline: one file, one version, one owner. As soon as the same list circulates in three emails with three file names, everyone calculates with different numbers. Keep the file in a shared location and send only the link to it, so nobody works on an old copy. Also name one person who enters changes, even if everyone may read along. POWERJO delivers tender documents and format lists as editable Office files so you can keep working on them yourself.

Manufacturers like to quote MTBF as a catalogue figure. In operation it is only as good as the downtime capture. You have to define whether micro-stops count or only faults lasting over five minutes. Fix the definition before the performance test, including whether operating time is counted in calendar hours or running hours, otherwise you are comparing two figures that have nothing to do with each other. For maintenance, MTBF per assembly is more interesting than the line-level figure, because it shows where spare parts and maintenance intervals really belong. I always ask the supplier on what data his value is based, field failures or a calculation from component data.

MTTR often has more influence on availability than fault frequency. A fault that takes 40 minutes because the spare part sits in the central warehouse, or only one colleague knows the machine, costs more than ten micro-stops. The adjustments are spares held at the line, good fault messages on the HMI, trained operators and remote support from the manufacturer. At acceptance, ask once about the fault that would take longest, and have the maintenance technician show how he finds and clears it. For the line, what counts is the time until the machine runs again, which includes the service technician’s journey. If the circuit diagram first has to be fetched from the office, you already know your MTTR.

Multi-carrier systems show their strength with many formats and small batch sizes, not on a single-product high-speed line. Pay attention to hygiene, meaning gap dimensions and cleaning of the track, to the number of movers as a potential bottleneck, to wear on the running rollers, and to dependency on the control system manufacturer. Ask for references with your product and have the behaviour during power loss and restart explained to you. Ask as well how a defective mover is replaced and how long the line stands for it. That answer says a lot about how well the supplier knows the system in operation.

On piece-form and free-flowing products the multihead weigher keeps give-away to a minimum, provided product infeed and vibratory feeders are set up correctly. Losses come from sticky products, incorrect filling of the pre-hoppers and too few heads for the required accuracy. At the performance test I always demand the real product, taken from the batch that arrives in daily operation. Put the target accuracy in the contract and not the catalogue figures. Ask about cleaning as well. Sticky or greasy products build up on the pans, and if cleaning only works with tools and plenty of time, it gets cut short during the shift.

Multipacks mean additional formats for the collation packer, labelling and palletising. Keep the number of variants small and check whether one packer can run several bundle sizes without a tool change. Labelling with the multipack’s own EAN rather than the individual pack’s is a common mistake with new launches, and it often shows up only at the till at the retailer. Make sure that only the right code can be read at the scanning point if the individual packs in the bundle carry their own barcode. In palletising, film bundles behave differently from cases and tend to slip, which belongs in the pack trial.

Muting bridges a light curtain so pallets or cases can pass without stopping the plant, which makes it a safety-critical point. Incorrect sensor arrangement can let a person in behind the pallet. Insist on a muting design engineered to standard, check it at acceptance, and watch for frequent nuisance stops. They are a sign of poorly set sensors and not of excessive safety. At acceptance, try what happens when a pallet stops in the pass-through or someone walks in behind the load. And ask for the time window of the bridging, because windows that are too long invite people to cheat the system.

Plant networks that have grown organically are often one single flat network in which the office printer talks to the safety PLC. Segmentation does not have to be expensive. Usually a cell firewall per line, a transition network to IT and documented remote-maintenance access that the plant releases will do. With every new machine I ask for a network diagram and a list of the connections required. Suppliers who want everything left open are a warning sign. Insist that remote access is active only on request and with your approval, and have the IP address ranges named in advance, otherwise there is an argument with IT on the day of commissioning.

POWERJO stores project data in a Nextcloud, that is CVs, drawings and quotations. For customers with strict data-residency requirements this is an argument in its favour: the storage location is known, permissions can be controlled by folder and by project, and revoked again once a project ends. Clarify at the start of a project who gets which access, and withdraw it at the end. Otherwise people still have access years later to layouts they stopped working on long ago. Also ask about backups. A storage system without a tested restore is just a server with hope. If you run it yourself, also clarify who in the company installs the updates.

Typical cases in plant engineering are quickly listed. The customer specifies a film that the supplier knows runs poorly on its machine. In another case the hall floor lacks the flatness stated in the specification. Or another supplier’s conveyor arrives at a height that does not match the transfer point. As the client, it is best to read such a notice like a free expert opinion. Then decide deliberately, and in writing, whether you stick to your specification. Some suppliers send notices of concern about anything and everything, especially towards the end of a project. That is usually cover for later discussions about defects. Go through each one anyway, because a genuine concern likes to hide among the precautionary ones. How far liability and warranty shift as a result of a notice depends on the contract and the individual case. If in doubt, clarify that with a lawyer.

On site it looks like this. The fitters arrive on the agreed day, but the foundation is not finished, the old machine is still standing and compressed air will only be available next week. An experienced supplier writes a notice of obstruction that same day. As the client, you should not just file it and forget it. Reply in writing and clarify whether the obstruction really lies within your sphere. Record as well when it ends. A site diary with dates and photos helps both sides more than anyone’s memory. It gets dangerous at the end of the project, when a pile of these notices comes back as a variation claim. Set out form and deadlines for such notices in the contract. And if it comes to a dispute, have the situation checked by a lawyer before you make any concessions.

At POWERJO, Notion is the logbook: what was decided, what went well, what went badly. For customer projects we recommend the same principle. One place for decisions and lessons learned means they do not have to be relearned on the next project. Write the entry while the project is still running. After acceptance nobody remembers why the more expensive drive was chosen. Keep the entries short, with date, decision, reason and the people involved. Agree on a fixed template. If everyone writes in their own format, the logbook turns into a box of loose notes in which nobody finds anything.

After many years the mechanics of a packaging line are still running, but the control system has long been outdated. The PLC, operator panel or inverter is no longer manufactured, and the spare part is only available second-hand. Keep program backups and documentation, otherwise even a matching replacement cannot be loaded. When buying, I always ask for the product lifetime of the electrical components and have the makers of the main components named. Check the manufacturers’ discontinuation notices for your most important parts once a year. A planned retrofit of the control system is cheaper than an unplanned stoppage without a spare part, and it can be put into a shutdown window.

OEE is only useful if everyone in the plant uses the same method of calculation. What counts as planned time, what counts as downtime, which rated output applies. Otherwise two lines compare apples with pears, and the discussion is about the calculation instead of the losses. On an Efficiency Day we calculate it honestly once, and the figure is often well below what management assumes. More than the number itself, the three factors count. Availability, performance rate and quality rate each point to different adjustments. I distrust an OEE that arrives on the table without downtime reasons, because then nobody knows where to start.

Offline programming saves downtime. Palletising patterns or robot paths are created and simulated on a computer while the line keeps running. Clarify whether you get the software and the rights to create new pack patterns yourself. Anyone who has to call the manufacturer for every new format pays more over the years than the licence would have cost. On palletisers there is often a pattern assistant right at the operator panel, and then no offline software is needed for new pack patterns. I still ask what happens if the licence or the computer fails, and where the project is backed up. A robot program that exists only on the commissioning engineer’s laptop does not belong in the plant.

With several suppliers on site at once, on-site management decides matters over weeks. The same questions come up every day: who gets the crane when, who gets the electrician, who gets the forklift. A short meeting with all fitters in the morning saves more time than any schedule, because that is where it emerges that two firms want to work in the same corner. Bringing in a temporary site manager who is on site daily and logs outstanding items immediately usually costs less than a week’s delay on a line that produces five-figure sums per shift. Add a daily list of open points with names and dates that everyone can see.

OPC UA is the vendor-neutral language between machines and IT, for example for MES, energy monitoring and analytics. Require an OPC UA server with a documented data model in the user requirements specification, ideally following PackML. Without this agreement, manufacturers often deliver proprietary interfaces that become expensive to connect later. Also write down which data you actually need, with name, unit and sampling time, otherwise you end up with a server that offers everything and nothing useful. At FAT, have a client read the data live, using a tool that does not come from the machine manufacturer. Do not rely on the statement that the interface is available. Available and switched on are two different things, and sometimes switching it on needs a licence that costs extra.

An open items list rarely dies from too few entries. Usually it grows to a few hundred lines and nobody reads it any more. Make sure every item has a name and a date. A company is not a responsible person, and until further notice is not a due date. In meetings I only go through the items that are due or overdue, and everyone reads the rest beforehand. An item is closed only when the person who raised it agrees. Otherwise supposedly settled topics suddenly resurface during commissioning. Keep a single list for the whole project. If electrical, mechanical and civil works each maintain their own list, exactly the items that sit at the interfaces fall through the cracks.

Require the operating manual in German, structured by operation, changeover, maintenance, fault clearance and spare parts. Ideal is a spare parts list with the manufacturer designations of bought-in parts, because nobody orders a sensor at three in the morning with the machine builder’s internal item number. Have an advance copy delivered at the FAT and review it before shipment, preferably together with the maintenance technician who will work with it later. Missing documentation, or documentation only in English, is an acceptance defect. Anyone who chases it only after final payment waits a long time. So have the contract state that complete documentation is a condition of acceptance, and settle the wording with your lawyer.

Training given during the hectic commissioning phase rarely sticks. Split it up: basics before delivery at the supplier’s site or at the FAT, hands-on practice on the running machine after the ramp-up phase, and maintenance training separately with wiring diagrams and drive parameterisation. Name a line owner per shift who receives in-depth training. Secure the right to refresher training if staff change in the first year. I also insist on training documents in the language of the staff and on an attendance list, otherwise nobody knows in a year who learned what.

You pay the purchase price once and the running costs every day. Over the life of a packaging machine they add up to a sum that can keep pace with the purchase price. Anyone who only lines up quoted prices has seen the start of the calculation. I therefore have consumption figures for power and compressed air per operating hour written into the quotation, plus the wear parts with service life and price. It is one sheet of paper, but it rarely turns up unasked. Then calculate with your own energy prices and shift pattern, not the figures from the brochure. Also ask what a service engineer costs per call-out before you need one.

In packaging plants, picking mostly concerns dispatching mixed pallets and supplying packaging material to the line. Both often run on a forklift and a shout across the floor. An Efficiency Day quickly shows how many kilometres and how much search time lie behind that. Before automation comes into play, fixed staging locations, clear kanban quantities and clean stock records in the system help. Fixed locations need a floor marking, otherwise everybody puts things down wherever there is room. I first check whether the system stock matches the shelf. If it does not, every picker keeps searching, and no robot turns that into order.

The Machinery Regulation and NIS2 place requirements on the cybersecurity of machines and operators. Your lawyer and your IT manager should clarify whether your company falls under NIS2. For packaging lines this means: authorise remote-maintenance access in a controlled way, separate the control network from the office network, assign user rights on the HMI, and back up the programs. Ask the supplier how remote maintenance technically works and who has access, and set it down in the maintenance contract. Typical gaps are default passwords on the panel and a USB stick that travels from machine to machine. Both can be checked quickly at acceptance.

The catch is usually in the in full part. A lorry can be in the yard on time, and if a single pallet is missing, the whole delivery counts as a miss. Inside the plant, the problem often ends up at the packaging line. If it fails during the night before loading, the lorry leaves half empty, and on Monday sales wants to know why the figure has collapsed. I like to apply the same yardstick to suppliers in a project. A machine that arrives on time but without its format parts or documentation has not been delivered as far as I am concerned. Have every crate checked against the packing list at goods receipt on site, and record shortages in writing. Otherwise you will be arguing weeks later about whether a part never arrived or is lying somewhere in the hall.

Brochure figures usually refer to the lightest format and to packaging materials as they arrive only in the ideal case. Put output into the URS, the user requirement specification or Lastenheft, as guaranteed continuous output with your real product, a named format and a fixed run time, for example 120 cycles per minute over eight hours at format X. Also clarify whether gross or net is meant. A supplier who runs 120 cycles and rejects a share of them has kept to the number and still delivered too few good packs. I have the measurement demonstrated at acceptance with my own product, not with the supplier’s test material.

An overhaul pays off when the frame and mechanics are good and the spare parts are still available. If the control system is the problem, you are looking at a retrofit, not an overhaul. Check beforehand which assemblies are actually worn. A downtime analysis often shows that a handful of assemblies cause most of the faults, and then you replace selectively instead of touching the whole machine. Clarify with the manufacturer whether the overhaul happens at their works or at your site. At their works you miss the machine in production, on site the fitters get in your way. Plan the downtime window realistically and with contingency, because dismantling almost always turns up something that was not on the list.

If a slip clutch slips more often, sooner or later someone tightens the nut. From then on it is a rigid connection with a friction lining, and at the next jam something more expensive breaks. So at acceptance I ask how the tripping torque has been set, and have the setting marked with paint and documented. Clutches that never slip for years can seize and then fail to release when it matters. An inspection interval in the maintenance plan helps here. Machines with mechanical cams need a synchronous re-engaging clutch. Otherwise, after a trip the motion timing no longer matches and tools meet where they were never meant to meet. Check too that the switch on the clutch is actually wired to the control system. Otherwise the motor cheerfully keeps running after the trip.

For modified atmosphere packs, shelf life depends heavily on OTR, but only if the seal seam and residual oxygen play along. A high-barrier film is no use if the seal leaks or gassing leaves too much residual oxygen. Note that EVOH and PA barriers deteriorate at high humidity. Demand measured values under real storage conditions and check the oxygen level in the pack over shelf life. The data sheet applies to flat film. Creases at folds and edges, especially in metallised layers, make the barrier of the finished pack worse. Ask the supplier whether he has measured this on folded samples.

An interlocked pack pattern with staggered layers is more stable, a column stack carries more load. Calculate pack patterns with software rather than working them out on a table, and check overhang, the box compression strength of the cartons and retailer requirements. Every new pattern must be taught on the palletiser and validated in a transport test. I therefore ask for a photo of the finished pallet and the wrapping parameters for every pattern. A pattern that only holds with hand adjustments is no good for the night shift. Also ask for some reserve in pallet height, because carton dimensions vary from batch to batch. A lot fits on paper that tips over on the line after the third layer.

Whether a pack is tight or not is decided by the seal seam, the material and the handling after sealing. A water bath and a vacuum test suit spot checks, while inline leak testing checks every pack. Set inspection intervals and write down what happens when a pack fails, that is stop the line, block the last packs, look for the cause. If fault rates rise, react at once. Usually a contaminated sealing tool or a changed film is behind it. After a film change an extra leak test is worthwhile, because the first seams of a new batch show early whether the settings and the new film suit each other.

For line planning the chain matters more than the definition. Every packaging level needs its own machine, its own format list and its own packaging material supplier. Changes to the primary packaging travel through the whole chain, one millimetre more bag length all the way to the layer pattern on the pallet, where suddenly the overhang no longer fits. That is why the packaging material specification belongs in phase two of the project, before a machine is selected. Whoever writes it only after the order ends up rebuilding the machine afterwards. Talk to the packaging supplier early, because their tolerances are your machine’s tolerances.

Packaging licensing scheme participation is often dismissed as admin, yet it is a real cost factor. Material type, weight and recyclability determine the fee per tonne, so every packaging change directly affects licensing costs. Factor these costs into material decisions, for example when switching from composite to mono-material. Make sure the reported quantities match the piece count and packaging weight from production, because audits start exactly there. Put purchasing, production and the environmental officer around one table for this. Often the data sits in different departments, and nobody has ever laid it side by side.

Most start-up problems after a packaging material change happen because the sample was never run on the machine. Insist on an approval run with production material, over at least one reel or pallet quantity, and keep a signed and dated approval sample on file. A hand-made sample almost always runs, the series later does not. A limit sample showing what is only just acceptable saves a lot of time in a dispute. A change of sub-supplier for board or film also triggers a new approval, otherwise you have no reference. Oblige the supplier to report changes to paper, film, adhesive or tooling beforehand.

The distinction sounds academic but is useful on a project. Packs and packaging aids have different suppliers, approval processes and cost centres, and the line depends on both. Tape, hot melt and labels are often forgotten in the URS, leaving no specification at acceptance for half of the faults that occur. Keep a complete list of packaging materials per product, including the packaging aids and their specifications. The consumption of tape and hot melt per case only becomes visible in operation, so add it after the ramp-up. One line per article with supplier, article number, dimensions and approval status is enough as long as it is up to date. It also helps when a supplier changes, because you can see what else is affected.

Without a packaging material specification, nobody is responsible when faults occur. The machine manufacturer points to the film, the film supplier points to the machine. Draw up the specification before ordering, together with both parties, with tolerances the machine can actually handle. Changes to the packaging material, even for cost reasons, must go through a change approval with a machine trial and not through a note on the purchase order. Add an approved sample and ask for the manufacturer’s inspection record of the critical dimensions with every delivery. Also define who maintains the specification and which version applies. In a dispute you otherwise get three versions on the table, and each side holds its own to be the valid one.

Mechanical engineering, materials science, logistics and retail requirements all meet in packaging technology. Hardly any investment touches so many departments at once. Projects rarely fail on the machine. What usually goes missing is coordination between production, purchasing, quality and sales. Anyone who buys a line made up of machines from several manufacturers soon has one more supplier who was not responsible when something goes wrong. An interface list and clear battery limits, meaning the exact boundaries of each supplier’s scope, spare you the discussion after acceptance. Experience at this interface is why interim project managers are so often in demand here.

On paper some machines have PackML, but inside the controller sits a cut-down version with a data block only the manufacturer understands. So write into the user requirements specification which states, operating modes and data you want. At acceptance, have someone walk you through every state, including Hold and Abort, and check that the downtime reasons arrive properly coded. Only then can you compare one line with another, and the MES does not have to be connected afresh for each machine. Ask for PackML from the start and it costs the manufacturer little. If it comes afterwards, you pay for a rebuild of the controls. I ask about it first on every new packaging machine.

An automatic pallet changer at the palletiser prevents the most common stoppage at the end of the line. The pallet is full and the forklift is not there. Alternatives are a palletiser with two positions or a roller conveyor buffer for two to three pallets. Which solution fits depends on cycle rate and forklift availability, and a downtime analysis at the line end supplies the figures. Ask the supplier how long the change takes and whether the machine keeps producing during that time or collects products in a buffer. A changer that briefly stops the line every time only moves the problem. When transferring from wood to hygiene pallets, the wooden pallets stay in front of the lock, and everyone on the shift should know that.

Pallet dispensers are small things that bring the palletiser to a standstill. The culprits are broken boards, protruding nails, warped pallets or mixed pallet types in the stack. Agree with purchasing on which pallet quality is delivered, and add an inspection station beforehand if pooled pallets are used. Dispenser capacity should cover at least one hour of run time, so that the forklift driver does not become the pacemaker. Also check the space for replenishment. If the forklift has to cross the packing zone to refill the magazine, trouble with personnel safety is guaranteed. Access from outside or a separate route avoids that.

Work out the pallet factor early. If a case becomes two centimetres wider, it can cost a whole row per layer and lower the factor noticeably. Logistics only notices once the freight quote is on the table. Conversely, a higher factor with a more stable layer pattern is often the cheapest gain in the whole project, because it saves journeys, storage positions and pallet changes. Also check that the pallet height suits the lorry and the rack bay, since one more layer is no use if it does not fit into the bay. State both the current and the target factor in the enquiry. If the figures are missing, every bidder calculates with their own layer pattern, and the quotes cannot be compared.

A damaged pallet blocks the equipment or topples in the racking, and both like to happen at night. For pooled reusable pallets, automatic inspection pays off from medium volumes upward, because exchange pallets arrive in very varied condition. With single-use pallets, incoming inspection at the supplier is usually enough, provided the specification has been agreed and is written into the order. Also define where the rejected pallet goes and who takes it off at night, otherwise the line stands because of one single pallet. Do not set the limits too tight either: an inspection that keeps triggering falsely will be disconnected sooner or later.

Retailers reject pallets whose label is missing or unreadable, and you often only hear about it with the complaint. A labeller at the line end, with a reader to verify the print and a link to the WMS, belongs on the line from medium volumes upward. The label sits on stretch film that is shiny and creased, and that is where some scanners at the ramp give up. Clarify with your largest customers where the label should go and on how many sides, two being the usual. I also ask to see how the numbering carries on after a controller failure or a software change. Nobody on the line notices a duplicate SSCC, but the customer’s goods-in system does at once.

Overhang costs stacking strength and cases, underhang costs freight space. Both happen when the case outer dimensions and the pack pattern do not match. Recheck the pack pattern with every case change, since double-wall board grades and moisture change the outer dimension. Filled cases also bulge, so measure on the full case and not on the empty blank. Retailers check overhang and height at goods-in and reject pallets. A contour check before the stretch wrapper catches outliers but does not replace proper planning. Agree with the customer how much overhang is still accepted before the first pallet comes back.

Pallet positions are the quiet reserve of every layout. Empty pallets, full pallets, layer pads, film and rejects all need floor space and clear routes. Anyone who plans the layout around the machine alone ends up putting pallets in the traffic path later. Budget for at least the positions for one shift without removal at each line end, plus space for change parts and cleaning equipment. The forklift also needs room to manoeuvre, and that room does not belong to a pallet position. Draw the floor markings into the layout from the start and keep to them after the move. After half a year the empty pallets will otherwise stand exactly where the escape routes should be.

Pallet trucks appear in no investment appraisal and still cost money: inspections, repairs, accidents, space. In the Shopfloor Analysis we count them too, because every manual pallet movement points to missing conveyor technology or an awkward layout. They usually stand where the last user finished, and with luck the next person finds them again. A rule such as pallet trucks only with an inspection sticker and a fixed parking spot is a simple first step towards order. On a walk-round you quickly see which units have stood unused for weeks. Units with an expired or missing sticker belong out of circulation at once, otherwise the next colleague uses them anyway.

Which technology fits, whether robot, gantry or layer palletiser, follows from the requirement profile: cases per minute, number of formats, pack patterns with layer pads, available floor space, staffing. A robot is flexible, a layer palletiser is fast, a cobot is affordable at low cycle rates. Have the pattern and stability demonstrated with your own cases, as this is the most common point of dispute at acceptance. I always have a finished pallet driven to the warehouse once, through the stretch wrapper and with the forklift, and not just left standing at the conveyor. Only there does it show whether the pattern holds. Also check how quickly an operator can reach the pallet during a fault. The longer the way through doors and safety fences, the longer every minor fault lasts.

Invoices, delivery notes, contracts: at POWERJO, everything lands in Paperless and is found again in seconds. The same approach is worthwhile for project departments too. Acceptance records, certificates and supplier documents are immediately at hand during an audit, instead of sitting in three folders and a mailbox. The filing only works with discipline. Scan a document when it arrives and not weeks later. Fix a small number of document types and stick to them, otherwise the tag list keeps growing until nobody can oversee it. Clarify with your tax adviser which paper originals you have to keep after scanning.

On packaging lines, a Pareto analysis of downtime almost always shows the same picture. A few causes produce a large share of the losses. Rank by total duration, not by count, because a rare but long stoppage is often more important than many micro-stops. Look at the biggest bar in the chart first. If it is called Other, the recording is not good enough yet, and you should first sharpen the reason code catalogue. The result is the priority list for the coming weeks. Repeat the evaluation after the first measures, and you will see whether the big bar has really become smaller.

Partly completed machinery such as robots without grippers or conveyors without a control system must not be operated on its own. Whoever integrates it into a line becomes the manufacturer of the complete machine and carries the CE responsibility. Many projects underestimate that until someone asks for the declaration of conformity for the line. Clarify early who takes on this role, the integrator or you yourself. Have the declaration of incorporation and the assembly instructions supplied with it, and actually read the assembly instructions. They contain the conditions under which the machine may be built in, for example on guards and ranges of movement. The layout has to stick to them.

Rust on stainless steel is often foreign rust. Someone deburred an edge with the angle grinder that had just been used on a steel beam, or cleaned the weld with an ordinary steel brush. The embedded iron particles rust, and from the outside it looks as if the stainless steel were poor quality. Put it in the site rules that only stainless steel tools are used on stainless steel, and check during installation that this actually happens. For pipework built on site, get written confirmation of whether and how it was pickled and passivated. Pickling paste is no toy. It usually contains hydrofluoric acid, needs protective equipment, and the rinse water must not simply go down the drain. Work with pickling paste therefore belongs in the work permit system.

Cycle time losses are often not in the gripping time but in poorly planned paths: too many exact intermediate points, unnecessary lift heights, returning by the same route. Have blending radii and acceleration ramps set up properly, but not right at the limit, because you pay for it later in gearbox wear and vibration at the gripper. For pick-and-place, it pays to look at the sequence of the placement positions. Also check that the load data of gripper and product are correct in the robot. If the payload is wrong, it calculates with the wrong masses, and the motion becomes less smooth or wear increases. A gripper that has become heavier after a modification often shows up first at the gearbox.

Many companies require a payback of under three years. Packaging machines run for 15 years or more, though, and a short deadline then mainly rewards the cheap offer. Supplement payback with TCO and risk. A machine with a four-year payback and half the operating costs beats the cheaper offer in the long run. Calculate with documented savings and not with brochure values. Also ask the supplier which assumptions its savings figure rests on. Find out where the saving comes from, whether output, scrap, staff or downtime, and who actually delivers it in operation. A saving nobody books pays back nothing.

A classic design error: the calculation uses the weight of the product, and the gripper with valves, hoses and camera is added later. Then the robot runs at its limit, accuracy suffers and the gears age faster. Ask for the payload diagram for your gripper length, because the permissible load moves with the centre of gravity. On palletisers there is the further point that a gripper holding several cartons at once carries more and pushes the centre of gravity further out. Plan in reserve for the next format and a heavier gripper. Enter the load data, with mass and centre of gravity, into the robot controller. Wrong values distort the collision detection, which then trips too early or too late.

Payment terms on machine supplies are a matter of negotiation and depend on the milestone structure. Tie payments to events you can verify, such as delivery, SAT or proof of performance, and not to calendar dates. A calendar date keeps running even when the machine is not. Make sure each milestone is described in the contract so that both sides understand the same thing by it. Otherwise you argue about the invoice before the machine is standing. For assignments of on-demand specialists, monthly invoicing with a payment term of a few weeks is usual. That is easy to plan on both sides.

Expensive peaks rarely arise in normal operation. They come on Monday morning, when all lines start up together after the weekend and the compressors refill the empty network. A quarter of an hour is enough, and the bill carries it around for a year. With new machines, look at the heating power. Sealing jaws, shrink tunnels and hotmelt units draw considerably more while heating up than in operation. So besides the rated power, I always ask the supplier whether the heating zones can be switched on with a time offset in the control system. That costs him a few lines of code. Talk to production about the start-up sequence after the weekend as well. Sometimes a plan on the wall is all the load management you need.

Peelability is a balancing act. If the seam is too strong, the consumer tears the tray open. If it is too weak, the pack opens on the shelf or in transit. Peel force depends on the tray, lidding film, sealing temperature and dwell time. Check it after storage and not just fresh off the machine, since it can change over time. For retailer complaints, only a documented peel-force specification with a tolerance band helps. Also ask how the seam behaves in the cold and in the warm, when the pack sits in the chilled cabinet and later lies in the customer’s car. A pull tab on the lidding film makes opening easier and belongs on the drawing.

A penalty for schedule slippage appears in many contracts and only works if the milestones are unambiguous. Write down which event counts as delay, for example delivery from the works or a passed factory acceptance. Define just as clearly what you have to contribute yourself. Otherwise the supplier points at late samples or missing approvals and works out the delay against you. A penalty clause is no substitute for project management. Anyone who reviews the schedule with the supplier every month rarely needs to invoke it. And whoever does invoke it usually has a project that was already running off course. Have procurement or a lawyer draft the clause, since amount and cap are negotiable.

What the bond covers is decided by its wording alone. The heading tells you little. Have someone check whether contractual penalties, defects before acceptance and repayment of overpayments are included. When it is exchanged after acceptance, keep to a simple order. First the new bond arrives in original, then you return the old one. Do it the other way round and you may be without security for a while. Think about the effect on the bidder list as well. Smaller special machine builders have limited guarantee facilities, and a large bond ties up room they need for other orders. Some would rather not bid at all. So ask in the enquiry whether the supplier can provide the required bond. Best agree the exact wording with your lawyer.

Performance rate reveals what downtime lists do not show. The line is running, but slower than planned, because of micro-stops, reduced speed after a fault, or because somebody throttled the machine out of caution. Compare actual output per shift with rated output, and ask the operators for which format or packaging material they prefer to run slower. That usually has a reason. There is often more potential here than in availability, and the fix costs nothing but adjustment and training. Look at the display on the night shift for this. The speed there is sometimes set to a value that was once safe and was never touched again.

The performance test is the moment the URS meets reality. Fix beforehand how long it runs, at least a full shift, better several days. Also settle the products and formats, who operates, how stoppages are attributed and from when the test counts as passed. A stop because the packaging material is out of tolerance is not a machine fault, a stop because the ejector jams certainly is. That must be settled in advance. Without these rules, every stop gets argued over, and acceptance drags on for months. Record counter readings and fault logs jointly, ideally signed by both sides.

Monday morning, the fitters are waiting at the gate, and the only person allowed to sign permits is stuck in a meeting. More than one site has started like that. Clarify beforehand who issues permits and who stands in for that person. And settle how early a request has to come in. Contractors should know the procedure before they travel, not learn about it at the gatehouse. In food plants, cardboard, film reels and dust are often closer to the welding point than anyone would like. That is why the fire watch is no formality. If a fire detector is switched off for the job, the permit has to say who switches it back on. Otherwise exactly that gets left undone on a Friday afternoon. As a rule, one permit covers one job in one place. Anyone who moves on from welding at the conveyor to the tank needs a new one.

PET trays are stiff and clear but cannot be closed without a seal layer. Tray and lidding film must be approved as a system, otherwise the seam peels uncontrollably. For recycled content, known as rPET, in food contact, require evidence of an approved recycling process. PET withstands heat better than PE, but thermoforming and shrink behaviour are sensitive to material changes. Every new recipe belongs in a trial run. With clear trays, product residue in the seal is visible at once, which makes the check at the conveyor easier. For chilled goods, clarify with the supplier whether the tray needs an anti-fog finish, because condensation on the lid hides the product.

A large share of micro-stops on packaging lines comes down to dirty, misaligned or wrongly chosen photoelectric sensors. On an Efficiency Day at the line, this is usually the first thing to show up. Shiny film, transparent bottles and white cases can fool simple diffuse sensors. Use background suppression or through-beam sensors at critical points, mount sensors securely and reproducibly, with a stop rather than a slotted hole, and add cleaning to the shift schedule. For film, a reflector with a polarising filter helps, for clear bottles often only a sensor made for transparent objects does. Mark the setting, for example with a paint pen on the screw, so you can see when somebody has been at it. A similar-looking sensor with different optics can react very differently to the same product.