Plant problems
Where Yield Goes Missing: A Stage-by-Stage Guide for Process Manufacturers
Yield rarely disappears in one place. It leaks at the gate, in the tank, in the lab register and on the packing line. Here is how to find each leak.
Trikaan Research & Engineering5 min read
KEY TAKEAWAYS
- Yield loss is almost never one big leak. It is five or six small ones that nobody owns end to end.
- Without a stage-wise material balance, you cannot tell process loss from measurement error.
- The fastest wins usually sit at the gate (moisture, impurities, weighing) and at packing (overfill, rework).
- Capture input and output at every stage, daily, with a named owner — then let the numbers argue.
Ask a plant head what the yield was last month and you will usually get a number. Ask where the missing two or three percent went and the room goes quiet. Not because nobody cares, but because the data to answer the question does not exist in one place.
This guide walks through a process plant the way the material moves — from the weighbridge to the dispatch bay — and shows where yield typically leaks, how to measure each leak, and what to fix first. The examples lean on edible oil and food processing, but the method works for chemicals, pharma intermediates and most batch industries.
Why overall yield hides the problem
Overall yield is a ratio of two big numbers: what came in and what went out. It is useful for a board report and almost useless for fixing anything. A one percent drop could be wetter raw material, a leaking valve, a lab sampling habit, a calibration drift on the packing line, or simply a stock count that was done on a different day.
The first step is to stop treating yield as one number. Break it down by stage, by product and by batch. Once you see yield at each stage, the conversation changes from opinion to evidence.
Stage 1: The gate — where losses are often booked as purchases
Material is weighed at the weighbridge, sampled for moisture and impurities, and accepted. If the sample is taken after unloading, or the moisture result is recorded hours later, you may be paying for water and dirt and then discovering the shortfall inside the process.
Check three things: when and where the sample is taken, whether quality deductions are applied to the purchase price, and whether the weighbridge reading flows directly into the system or is re-typed.
- Tie moisture and impurity results to the specific truck and lot, not to the day.
- Apply agreed quality deductions automatically at goods receipt.
- Calibrate and log the weighbridge on a schedule, and review tare weight patterns by vehicle.
Stage 2: Storage and handling
Silos, tanks and godowns lose material quietly: spillage, moisture change, pest damage, and the gap between book stock and physical stock. If physical stock is only checked at month-end, every handling loss gets dumped into ‘process loss’.
A simple discipline helps: periodic dip or level readings for tanks, cycle counts for bagged stock, and a reason code whenever book stock is adjusted.
Stage 3: Processing — measure in and out at each step
Inside the process, each stage should have its own material balance: input, main output, by-products, and loss. In an oil refinery that might be degumming, neutralisation, bleaching, deodorisation and filtration. In a flour mill it is cleaning, conditioning, milling and sifting.
When the stage balance is captured per batch, patterns appear quickly: one shift with higher loss, one machine that drifts after maintenance, one raw-material grade that never reaches target.
| Stage | What to record | Typical hidden loss |
|---|---|---|
| Intake | Gross, tare, moisture, impurity per lot | Water and dirt paid as material |
| Pre-treatment | Input qty, chemicals, output qty | Over-dosing, entrainment |
| Main process | Batch input/output, by-products, time | Process losses, recycle not recorded |
| Lab & QC | Samples drawn, retained, rejected | Sample quantities never booked |
| Packing | Filled qty, rejects, rework | Overfill and repacking |
| Dispatch | Weighed out vs invoiced | Short or excess shipments |
Stage 4: The lab register nobody reconciles
Quality samples, retained samples and rejected material add up over a month. If they are written in a lab register and never posted to stock, they show up as unexplained loss. The same is true for material held for re-testing and later reprocessed.
Book every sample and every rejection as a stock movement. It feels like paperwork until you realise how much of your ‘process loss’ was sitting in the QC room.
Stage 5: Packing — the most expensive place to lose yield
By the time product reaches packing, it carries the full cost of raw material, energy and labour. Overfilling by a small margin on every pack, or reworking packs because of label or seal issues, destroys margin quickly.
Track declared versus actual fill weight by line and shift, and record rework with a reason. Many plants find that a single filler needs recalibration more often than the maintenance schedule assumes.
Separating real loss from measurement error
Not every gap is physical loss. Instruments drift, stock counts are taken at different times, and units get mixed (litres versus kilograms is a classic in oil plants). Before launching a loss-reduction project, make sure the measurement system is trustworthy: calibrated scales and flow meters, consistent density factors, and stock counts aligned to the same cut-off time.
A 30-day plan to recover yield
You do not need a big software project to start. You need disciplined data for a month and a weekly review.
- Week 1: Define stages, units and density factors. Assign one owner per stage.
- Week 2: Capture input, output, by-product and loss per batch at every stage — on tablets or even structured sheets.
- Week 3: Review the variance by stage, shift, raw-material grade and machine. Pick the two biggest gaps.
- Week 4: Run small fixes (sampling point, calibration, fill setting) and compare against the first two weeks.
Where software helps — and where it does not
Software cannot reduce loss by itself. What it does well is remove the reconciliation work: weighbridge readings flow in automatically, stage balances calculate themselves, and the morning report shows yesterday’s yield by stage before the production meeting. AI can go further — suggesting process parameters for a given raw-material quality based on past batches — but only once stage-level data is reliable.
In ASTR Production we built stage-wise yield capture first and recipe suggestions second, for exactly this reason: an AI that learns from bad data simply automates the confusion.
Frequently asked questions
What is a normal yield loss in process manufacturing?
It depends heavily on the product, raw-material quality and process. Rather than comparing with an industry figure, compare each stage with its own best batches — that gap is your realistic improvement potential.
How do you calculate yield by stage?
For each stage, yield equals main output divided by input, with by-products and recorded losses listed separately. All quantities should be in the same unit, at the same cut-off time.
Is yield loss the same as process loss?
No. Process loss is one part of it. Yield loss also includes handling, sampling, rework, measurement error and quality deductions at the gate.
Can AI improve yield?
AI can recommend process settings and flag drifting batches, but only after stage-level input and output data is captured consistently. Data discipline comes first.
- #yield loss in manufacturing
- #process yield improvement
- #material balance
- #yield variance analysis
- #edible oil yield
Trikaan Research & Engineering
Trikaan
Written by the researchers and engineers who build Anvaya, ASTR and Seed Track, with input from Trikaan's delivery and engineering teams.
