Motionwell Automation designs filling lines in Singapore where the format change is engineered at the same time as the fill, not bolted on after the machine is already running. On the GMP liquid filling and sealing platform delivered under projects P23005 and P25026, container format parts swap on tool-free quick-change nest plates in under 5 minutes, and the food-grade tray filling platform running 20 to 60 units per minute uses a continuous dual-tray feed so the machine keeps filling through a tray change instead of stopping for it. Those two mechanisms answer the two halves of the problem: format changeover, where geometry has to move, and material changeover, where nothing does.
This page sets out what changeover time is actually made of on a filling line, which parts of a format change belong in a recipe and which stay in the toolbox, how much of a GMP changeover is cleaning rather than mechanics, and why the changeover concept has to be settled before the layout is frozen. The machines themselves are described on the custom filling machines page, and the qualification work that a format change drags behind it on a regulated line sits on the equipment qualification and validation page. If you already have an SKU list and a changeover time you are trying to hit, send it to an engineer.
What Actually Consumes the Changeover Clock on a Filling Line?
Measure it from the last good unit of the old job to the first good unit of the new one. Any other definition flatters the number, usually by parking the slow parts outside the measurement. Under that definition a filling line changeover is five separate pieces of work with different owners, different failure modes and different fixes.
| Block | What happens | Who owns it | What machine design can remove |
|---|---|---|---|
| Strip and remove | Product-contact parts come off, the wet path drains, old format parts and remaining product are cleared | Production | Tool-free fasteners, captive hardware, drain points sloped so the path empties without being tipped |
| Clean | Wash, rinse, dry, then swab or visual check to whatever standard your product demands | Quality and production | CIP spray ball coverage, no dead legs, surface finish that releases residue instead of holding it |
| Refit and adjust | New nest plates, guides, chucks and nozzles go on; heights, pitches, volumes and torques get set | Production, sometimes engineering | Keyed and dowelled mounting so a part fits one way only, motorised axes so settings arrive from a recipe |
| First-article confirmation | Fill volume, torque, seal, code and label checked and accepted before the batch runs | QA | Inline verification that gives the answer in the first few units rather than at a bench twenty minutes later |
| Documentation and release | Line clearance, recipe approval, batch record start | QA | Signature points in the machine record instead of paperwork walking to an office |
Most changeover projects spend their budget on block three. Refit is the visible part, the part with hardware in it, so it gets the attention. On a validated line blocks two and five are usually larger, and neither one is a mechanical problem. The honest way to find out which block is costing you the week is to log all five separately for a month; availability losses filed into a single bucket called setup hide exactly the thing you are trying to fix, an argument set out in more detail in our note on measuring OEE properly.
Which Format Changes Are Tool-Free on a Motionwell Filling Machine?
Tool-free is a specific claim, so it is worth stating what it covers. It means the operator removes and refits the part by hand. It means the part locates on dowels or a keyed interface rather than against a scribed line, so nothing needs re-teaching afterwards because position is mechanical rather than taught. It does not mean the part is light, and it does not mean the wet path behind it is clean.
| Delivered system | What changes | Mechanism | Figure |
|---|---|---|---|
| GMP liquid filling and sealing (P23005, P25026) | Container format nest and tray | Tool-free quick-change nest plates | Under 5 minutes |
| Food-grade tray filling platform | Incoming tray of containers | Continuous dual-tray feed | No stop at all |
| SCARA panel assembly line (P23045) | Panel variant, more than 15 on one line | HMI recipe, no mechanical adjustment | Under 3 minutes |
| Cleanroom test cell | Test fixture per connector type | Quick-change mount with dowel-pin alignment | Under 2 minutes |
| Carton palletizing cell | Layer pattern, up to 50 stored | Recipe called by barcode scan or MES command | No mechanical adjustment |
| EV battery module dismantling cell | Module fixture for a different pack format | Repositionable clamps plus a clamp-position recipe | About 30 minutes, two technicians |
The last row is in the table on purpose. A reconfigurable clamp grid was the right answer for a recycler who cannot control what arrives at the door, and the price of that flexibility was a 30 minute changeover with two technicians and a clamp-position recipe that has to be right, because a mislocated module is a mechanical interference on a live pack. Flexibility is not free. A machine built to accept anything usually changes over more slowly than one built around a defined envelope, and pretending otherwise is how a specification goes wrong at the start.
How Does Recipe Management Turn Machine Setting Into Recipe Selection?
An axis enters the recipe only if a motor or a software parameter controls it. That is the whole rule, and it is decided in the mechanical design, not in the PLC. A hand-wheel with a scale on it is not a recipe entry regardless of what the HMI displays next to it.
On the delivered filling platforms, four things are parameters rather than adjustments:
- Fill volume. Each fill head carries its own servo motor and ceramic-lined piston pump, so volume is a number in a recipe and calibration is held per nozzle rather than per machine. The pharmaceutical platform holds plus or minus 0.5 percent of target volume across the 5 to 500 mL range on that basis.
- Closure torque. Rotary capping heads run ROPP aluminium caps and screw closures with torque programmable from 0.5 to 5.0 Nm, and the torque curve is logged for every closure. The closure specification becomes a parameter, and its evidence becomes automatic.
- Code and artwork. Cognex DataMan readers and Domino thermal inkjet printers handle 2D DataMatrix marking and verification at up to 150 units per minute. Making the artwork variant part of the recipe is what catches the classic changeover failure of running yesterday’s label on today’s product, described further on our inline vision inspection page.
- Pattern and pitch. On collation and end-of-line stations, pick pitch runs on motorised axes and layer patterns are stored as recipes, so an SKU change is a selection rather than a re-teach.
What this buys is not only speed. It removes the operator’s memory from the setup. Startup scrap after a changeover is a quality loss with an availability cause, and stored parameters plus a first-article check built into the sequence are what shorten it.
On a regulated line the recipe is a controlled object rather than a convenience. Who may edit it, who may only select it, and what the audit trail records when it changes are design decisions with a rule behind them, 21 CFR Part 11 for electronic records and signatures, and how we build that split into the machine is covered on our electronic batch record page. A machine that lets any operator type in a new fill volume at three in the morning has a faster changeover and a deviation waiting.
Where Does the Recipe Stop and the Toolbox Start?
This is the question that decides whether a quoted changeover time will survive contact with your SKU list.
| What changes between jobs | Where it lives | Why |
|---|---|---|
| Fill volume inside the pump displacement range | Recipe | Servo stroke is a number, and per-nozzle calibration already exists |
| Fill volume outside the pump displacement range | Change part | A different cylinder or pump size, followed by recalibration |
| Container height | Recipe, if a motorised height axis was specified | A hand-set column is an adjustment, not a recipe entry |
| Container diameter or shape outside the machine envelope | Change part set | Star wheel, container guides and capping chuck geometry are all cut around one container envelope |
| Closure type, ROPP to screw | Change part | Chuck and head geometry differ |
| Capping torque within 0.5 to 5.0 Nm | Recipe | Programmable and logged per closure |
| Label and printed code artwork | Recipe | Print file plus the matching verification routine for that variant |
| Tray or nest layout | Change part, tool-free | Quick-change nest plate on a keyed mount |
| Product with different viscosity or foaming behaviour | Recipe plus a trial | Nozzle and speed profile may move with it, and that has to be proven rather than assumed |
| A new product sharing the line | Cleaning and qualification | Not a machine setting at all |
One sentence is worth taking into any supplier meeting: which axis of variation is in the recipe, and which is in the toolbox, because only the first one is measured in minutes. We wrote that lesson down after the SCARA panel assembly line, where more than 15 variants change in under 3 minutes as long as the difference between them is expressible in robot paths, grip profiles and vision parameters. A variant needing different gripper geometry or a different tray format still needs hardware, and hardware is a different number.
How Much of a GMP Changeover Is Cleaning and Verification?
More than the mechanics, usually. It is also the part a machine builder can influence but cannot decide.
We cannot tell you how long your clean takes. Residue limits, hold times, swab locations and acceptance criteria come from your cleaning validation and your quality unit, and we will not quote a changeover time that quietly assumes they move. What the machine contributes is whether the clean is repeatable and whether it can start immediately.
On the GMP platform that contribution is specific. Product-contact surfaces are SUS316L electropolished to Ra 0.4 um, welds ground flush and passivated to ASME BPE practice, and gaskets at product-contact sealing surfaces are FDA 21 CFR 177.2600 compliant. The fluid path carries no dead zones, clean-in-place spray ball ports cover every dead leg, and drain points are sloped at a minimum of 3 degrees so the path empties on its own rather than waiting for someone to tip it. Product-contact parts come off without tools, so cleaning does not queue behind a technician looking for a spanner.
The largest single reduction available on a GMP changeover is usually not mechanical at all. It is a second set of product-contact parts, cleaned and staged before the line stops, which takes the entire cleaning block off the critical path. What it costs is a duplicate part set, storage space, and the same identification and cleaning-status control as the first set, which is a quality system commitment rather than a purchase.
Performance qualification on the delivered filling platforms includes cleaning and changeover performed by the customer’s own staff following their own written procedure. That is the moment an optimistic changeover figure either survives or does not, and how the evidence gets structured is described on our computer system validation page.
Can a Filling Line Change Material Without Stopping?
Not every stop is a format change. A line that halts to load the next tray of containers, the next reel of labels or the next box of caps is losing availability to material changeover, which is a different problem with a cheaper fix: buffer it, or feed from two sources so one can be reloaded while the other runs.
The food-grade tray filling platform uses a continuous dual-tray feed for exactly this reason. The machine draws from one tray position while the other is reloaded, so a tray change does not stop filling. What decides whether that mechanism pays for itself is how many containers a tray holds against the rate the line runs at, because those two numbers set how often a reload comes round. On the GMP filling and sealing platform the tray input carries 500 to 1,000 pieces per load depending on container size, which is the kind of figure worth establishing before anyone specifies a second feed position.
The same reasoning applies to every consumable on the line. Whether a second reel position or a splice is worth its mechanism depends on how many times per shift you actually run out, and that is a measurement rather than an opinion. Accumulation upstream and downstream does the softer version of the same job: a buffer does not remove a stop, it converts a stop into a slow-down, which is often enough.
How Does SMED Land on a Filling Line in Practice?
SMED is one idea wearing a lot of packaging. Work done while the machine is stopped is internal, work that could be done while it runs is external, and most of the gain comes from moving work between those two categories before anyone tries to make any of it faster.
| Work that is usually internal | How it becomes external | What it costs |
|---|---|---|
| Fetching format parts from a store room | Kit the set on a station cart before the current run ends | A cart, storage at the station, part identification |
| Cleaning the parts that just came off | Stage a second cleaned set before the stop | Duplicate part set, storage, cleaning-status control |
| Setting fill volume by trial and error | Stored recipe with per-nozzle calibration | Motorised axes specified at concept, not later |
| Waiting for QA to attend a first article | Inline verification answering in the first few units | Checkweigh, fill-level vision, torque log, cap inspection |
| Preparing new labels and printed codes | Staged with the job order, verified by the reader on unit one | Artwork variant held in the vision recipe |
| Line clearance paperwork walking to an office | Signature points inside the machine record | Access control and audit trail designed in |
| One operator working through every step in sequence | Two operators with defined roles working in parallel | Access designed for two people, and a written sequence |
Two cautions worth stating plainly. The single-digit-minute target that gives SMED its name was set on press dies, not on validated pharmaceutical lines, and a line with a mandatory clean and a documented clearance will not reach it. Aim at your own previous number rather than at somebody else’s benchmark. And parallel work only pays if the machine was laid out so two people can reach it at once without one waiting on the other, which is a layout decision taken at concept and effectively impossible to add later.
Why Must Changeover Design Be Fixed at Concept Stage, Not Added Later?
Because most of what decides changeover time is geometry, and geometry is settled early.
The clearest example is on the delivered pharmaceutical platform. An 8-head rotary configuration is what makes 120 bottles per minute possible at a 100 mL fill. What it costs is format range: the star wheel, the container guides and the capping chuck geometry are all cut around a container envelope, so a container outside that envelope is a change part set rather than a recipe. An inline single-head machine is slower by roughly the head count and far more tolerant of odd containers. That is a concept-stage trade between rate and flexibility, and no amount of later effort converts one into the other.
Retrofitting quick change onto a machine that was designed for one format means re-machining mounting interfaces to accept a datum, re-cutting access panels so a part can come out by hand, and adding motorised axes where a hand-set stop was considered good enough. On a validated line it means more. After PQ, adding a change part set is a change control with an impact assessment and re-qualification of everything it touches, and the hardware is rarely the expensive half of that.
The place to fix all of this is the user requirement specification, and the test is whether a line in it could fail. A requirement stating that the machine shall be easy to change over cannot be qualified. A requirement stating that product-contact parts shall be removable without tools, and that a format change from last good unit to first good unit shall be completed within a defined window by two trained operators, is a test with a pass and a fail.
Motionwell runs 16 to 24 weeks from concept approval to factory acceptance testing on a standard build, and 24 to 32 weeks where cleanroom compatibility or full GMP validation applies. The changeover requirement has to be in the room during week one of that schedule, because by the time the frame is welded the envelope is decided. Machines are designed, built and tested at our Woodlands Link facility, which is also the reason the factory acceptance test is a chance to run an actual changeover with your own people rather than read a description of one. The company has delivered more than 150 special purpose machines since 2014 and holds ISO 9001:2015 and bizSAFE Level 3.
Is a Faster Changeover Worth What It Costs?
Sometimes not, and the answer arrives sooner than most specifications assume.
The arithmetic is short. Changeovers per week, multiplied by minutes lost each time, multiplied by what an hour on that line is worth, plus the startup scrap that follows each one and the labour standing around during it. Put a duplicate part set, a motorised height axis or a second tray position against that number and the decision usually makes itself.
We frame the same trade on the palletizing side as one combination tool versus a quick-change tool rack, and frequency decides it: four format changes a shift argues for one tool that compromises on every format, while weekly changes argue for the rack. The hardware differs on a filling line but the argument does not, and the end-of-line palletizing page works through that version of it.
If your line runs one product for three weeks at a stretch, spend the money on rate and verification instead. Changeover engineering pays back on high-mix lines, and on those it pays back quickly, because a high-mix line loses more of its week to switching jobs than to cycle time. That is the whole argument behind how we scope co-packing and high-mix consumer lines.
What Does Motionwell Not Do About Changeover?
- We are a machine builder and integrator, not a lean consultancy. We do not run kaizen events, time-study your existing crews or write your SOPs. What we do is design the machine so that a time study has something to find.
- We do not shorten your cleaning validation. Residue limits, hold times and acceptance criteria belong to your quality unit. We will not quote a changeover time that assumes they change.
- We do not sell recipe management as a software product. Recipe handling ships as part of the machine control system, built on whichever platform your site already runs, whether that is Allen-Bradley, Siemens, Mitsubishi, Omron or Beckhoff.
- We are not a notified body and we do not issue CE certificates. We deliver the technical file and the testing that supports the declaration. The declaration is yours to sign.
- Our delivered changeover figures come from small-container, tray and fixture platforms. We have not shipped a drum filler. Changeover on a bulk line is a different problem, with the container itself becoming the handling constraint, and it is discussed on the drum and pail filling page.
- We do not publish a changeover time for a machine we have not scoped. Under 5 minutes is what a delivered nest-plate tray change measures. It is not a promise about your container.
Frequently Asked Questions
How long does a format changeover take on a Motionwell filling machine?
On the delivered GMP filling and sealing platform, a container format change on tool-free quick-change nest plates runs under 5 minutes. That figure covers the mechanical swap only. On a GMP line the wet-path clean and the line clearance usually take longer than the mechanics, and those times come from your cleaning validation rather than from the machine. For comparison, the SCARA panel assembly line built under P23045 switches between more than 15 variants in under 3 minutes because the difference between variants is entirely in the recipe. We quote a changeover target only after seeing your SKU list.
What is the difference between a recipe change and a change part?
A recipe change is any setting the machine moves under software control: fill volume within the pump displacement range, capping torque anywhere in the 0.5 to 5.0 Nm band, pick pitch on a motorised axis, vision and code parameters. A change part is geometry: star wheels, container guides, capping chucks, nest plates. Recipes take seconds and leave nothing to lose or misfit. Change parts take minutes and need storage, identification and a check that the right one went on. Ask any supplier which axis of variation sits where.
Can changeover time be reduced on a filling machine we already own?
Sometimes, and less than people hope. The available wins are converting hand-set stops to keyed or dowelled positions, kitting format parts on a cart at the station, moving cleaning and first-article checks off the critical path, and putting settings into HMI recipes where the axes are already motorised. What cannot be retrofitted cheaply is a container envelope. If a new format falls outside the star wheel and capping chuck geometry the machine was built around, that is a new change part set, and on a validated line it is also a change control.