Filling Accuracy & Production Speed

At a glance

Swipe to compare all columns.

Define an acceptable pack before setting a speed target

Filling accuracy and production speed measure two different things. A filler can run at a high cycle rate and still turn out packs that fail your weight, seal or presentation checks. Or it can dose consistently at a rate that misses your production schedule. Decide what counts as an acceptable finished pack, and how many you need over a given production period, before you start comparing machines.

Write down the product, its temperature or material condition, the packaging format, the target quantity and how you'll measure it. Then list the checks that decide whether a finished pack passes or fails. A sauce sachet might need net content, seal condition and coding checked; a powder stick pack should also get a look at product around the seal area. Keep this in one place so a faster setting gets judged against the same criteria.

Distinguish target error from variation between fills

Ask our engineering team what their quoted accuracy figure measures. It could be the gap between a sample average and the target, a range seen during a demo, or a tolerance that only holds under specific conditions. Bias is how far the average fill sits from the target; variation is how much individual fills scatter around their own average. Find out whether the number is in grams, milliliters or a percentage, and if it's a percentage, what it's a percentage of. A figure with no target dose, no product, no sample count and no operating conditions tells you little.

For an assumed example, picture fills clustering tightly around 10.2 grams against a 10.0 gram target: they agree with each other but show a positive bias. Another assumed set can average 10.0 grams while individual fills vary widely. Investigate average offset separately from spread; neither pattern alone diagnoses a component fault or defines an acceptance limit.

Agree how net quantity will be measured

Pick a measurement method that fits the dose and the purpose of the check. If you're using a balance, record its ID, readability and how it gets verified, as agreed with your quality owner. Decide how package tare will be determined and how samples get handled. If the target is stated in volume but you're checking by mass, agree the conversion for that product at that temperature before you start. Don't assume one gram equals one milliliter; for most products it doesn't.

Put the method in writing before the trial starts. Cover how samples are identified, the order you weigh them in, the calculation used and what happens if packaging or product is lost or damaged. Label the units in every results column. If the packaging varies in weight, think through how that affects a gross-weight check. A repeatable method lets you tell a real process change from a change in how it was measured.

Sample every active lane and more than one moment

On a multi-lane machine, check each lane on its own, not just the combined output. Mark every sample so you can trace it back to its lane and the moment it was taken. A single overall average can hide the truth, one lane running high while another runs low and the two canceling out. Work out a sampling plan that covers steady running as well as events like startup, a hopper refill or a restart. How many samples you take, and when, should match what you're actually trying to find out.

Keep a simple results sheet with columns for lane, time, target, measured quantity, material condition and what you observed. Note an adjustment right when it happens; don't lump readings from different settings together without saying so. If a lane gets switched off, say so when reporting total output. After any change, repeat the same checks so you're comparing like with like. That record lets you and our engineering team dig into a problem, or sign off on the next trial.

Use one row per observation. Record tare separately from gross weight and calculate net quantity in the same unit. Compare before and after a refill without pooling the two sets; opposite lane errors can disappear inside one overall average.

Hypothetical trial-record example only: the timestamps, readings and possible cause in the following rows are invented to show how to record an event. They are not factory measurements, and an observed offset alone does not diagnose hopper pressure or another component.

Record the material and feeding conditions

A dosing result only means something alongside the conditions it was measured under. For liquids and sauces, record the temperature, any visible foaming or separation, and whether there are pieces in the product. Note how product reaches the hopper and what happens during a refill. For powders, note the batch, how it's flowing, any dust and whether it's settling. For dry foods, note mixed particle sizes, stickiness and breakage. These notes let you and our engineering team figure out which variable moved when results change.

Record the starting condition, change the one variable you agreed to test, then collect another labeled sample set. Investigating a result that showed up after a hopper refill? Write down exactly when the refill happened and compare readings before and after it. Let the evidence point to the next step; telling the operator to slow the machine down can mask a feeding problem that hasn't been fixed.

Use the powder feeding review for blend condition, the liquid filling review for foam and transfer, the paste filling review for cutoff and particles, and the granule weighing review for mixed pieces and breakage.

Separate cycle rate, gross output and accepted output

Ask where any output number was measured. Cycles per minute describe a repeating machine motion; packs per minute describe units coming off the line. On a multi-lane system, active lane count changes the conversion. Gross output counts everything made at the chosen measurement point. Accepted output is no greater than gross output: it excludes anything that fails your checks. With no rejected packs, the two counts are equal at the same counting point and over the same period. A downstream station can cost you more packs after the filler lets them go.

A hypothetical four-lane system running at 30 cycles per minute gives a gross reference rate of 120 packs per minute. That doesn't mean the line delivers 7,200 acceptable packs every hour; stops, slower running and rejects eat into that number. State the start and end of your observation period, count output from one defined point, and keep the rate you observed separate from any theoretical reference rate.

For a current example of why the comparison denominator matters, read the SIG Neo output baseline analysis before using a capacity-growth headline in a purchasing case.

Use OEE to separate three types of production loss

Food Engineering interviewed an OEE consultant who breaks overall equipment effectiveness into three parts: availability, performance and quality. For a filling line, that maps to three separate problems: lost running time, running slower than it should, and output that gets rejected. Before comparing two results, agree the planned production window, the reference rate and the counting rules; otherwise two similar-looking percentages might describe very different operating conditions.

Take the hypothetical reference rate of 120 packs per minute from above and assume 85% availability, 90% performance and 98% quality over that same period. Multiply those three together and you get 74.97% OEE. Multiply 120 by 0.7497 and that works out to about 90 acceptable packs per minute averaged over the period. That's our illustrative arithmetic, not a measured result or an industry target. The same overall percentage can point to very different problems, depending on which of the three is dragging it down.

Use the same counting point throughout: availability = running minutes / planned production minutes; performance = gross packs / (reference packs per minute × running minutes); quality = accepted packs / gross packs. OEE is their product. If a denominator is zero or the counting boundaries differ, leave the result undefined and resolve the record before comparing percentages.

Look beyond the dosing unit for bottlenecks

A packaging line is more than the filler. Product supply, bag opening or film handling, sealing, coding, inspection and downstream collection can each stop the line. Watch for where packs pile up and where machines sit waiting. Record the reason and length of every interruption, using categories your operators recognize. Don't blame every lost minute on the dosing unit because its screen shows the line's speed setting.

Walk through the sequence with whoever runs the connected equipment. A transfer point might need a timing tweak, a guide adjustment or a procedure change; shortages that keep recurring need a different replenishment plan. Once an agreed change is made, measure the same output point and quality criteria again, tied to the exact product and format tested. Otherwise an improvement seen on one pack turns into a promise that can't be backed up for every product the line might run.

Write an acceptance trial that can be repeated

Before running a demonstration, agree on the product, packaging and machine configuration you'll use. Agree on the trial period, target dose, operating rate, sampling method, the pack checks required and how results get reported. Decide who's supplying the material and who's recording the results, and how startup, interruptions, adjustments and rejected packs will be documented. A short video can show what the machine looks like running, but it can't substitute for the measurements needed to make a real acceptance decision.

When the demonstration wraps up, hang on to the readings, counts and event log alongside the configuration record. Note any substitutions or untested formats and flag what's still open. If the results miss the agreed criteria, write down the proposed fix and what it would take to run a repeat trial. This is a technical review of this product and this setup, not a stand-in for regulatory compliance, shelf-life testing or proof the configuration works for a different formulation.

Keep the operating record useful after handover

Put together a short, clear routine for startup, product changes and in-process checks. Write down the approved settings, the product and pack they go with, how to measure, and what to do when a reading falls out of range. Keep these records where your operators can get to them. When parts, recipes or materials change, hang on to enough detail to compare later results against the original operating condition.

Reaching out about equipment? Send us the target fill quantity, how you measured it, the acceptable-output number you need to hit, and any trial data you have. Troubleshooting a line that's running? Send lane-specific results and a plain description of what interrupted it, not just the fastest speed that flashed on the screen. That's what lets us tell you what a proposed configuration should be tested against before you commit to it.

Keep the measured giveaway and reject records when comparing operating costs in the filling machine price guide.

When these calculations do not settle acceptance

A reference rate is an input for comparison, not a demonstrated delivery rate. The equations need a nonzero observation period, consistent units and one counting boundary. Sample records apply to the product and setup observed; regulatory compliance, shelf life and other recipes require their own evidence. Acceptance limits are agreed separately from the illustrative numbers in this guide.

Keep quantity inspection separate from a statistical design target

NIST Handbook 133 describes net-content inspection procedures, including the relevant average, individual-package and sampling requirements. A proposed mean equal to label quantity plus three standard deviations is an assumed normal-distribution design model; it is not a universal legal compliance rule. The applicable nominal quantity, MAV, market rules, sampling procedure and measurement method must be identified separately.

Common questions