If a 25 Kg Batch Fails, Should I Immediately Change the Formula or First Reproduce the Original 1 Kg Control to Determine Whether the Process Caused the Failure?
Share
Why not just raise the fragrance load? If the process caused the failure, a higher load costs more oil and the next batch fails the same way. You also lose the approved formula as a reference.
What if the control fails too? Then the process is not the main suspect. Look at raw-material lots, the approved retained candle and the test method before the formula.
What happens to the failed batch meanwhile? Hold it, labelled by pour position. Its candles are the evidence you will compare against the control.
Why the formula should be the last suspect
When a 25 kg batch fails, the instinct is to reach for the recipe: more oil, a different wick, a switch of wax. It feels decisive. It is usually the wrong first move, for three reasons.
First, the formula has evidence behind it. It passed at 1 kg, under conditions you controlled. The failed batch differs from that test in many ways — batch size, melter, hold time, pour length, perhaps lots — and the formula is the one thing that did not change.
Second, changing it destroys your reference. Once the formula moves, you can no longer compare the new candles with the approved ones. If the next batch passes, you will not know whether the formula change fixed it or whether the process simply behaved better that day.
Third, it can hide a process fault that returns. If oil was added too hot, or the pour ran too long, a higher load might mask the symptom in the next batch and then fail again the first time conditions slip.
So the decision rule is simple: reproduce the original control before touching the formula.
The arithmetic of guessing versus testing
Take a real case. The approved formula is Luxury Soy Wax CSI 464 with Vanilla Bean at 8% of total weight in 200 g jars. The first 25 kg batch has noticeably weaker hot throw than the approved test candles.
The control costs less than a third of what the guess adds to a single batch, and it answers a question the guess never can. It also keeps Vanilla Bean at a load you already approved, rather than moving towards the top of the range with an oil whose page flags 2.5% vanillin and potential discolouration in lighter waxes.
Reproducing the control properly
A control is only useful if it is a faithful copy of the approved test. Anything that differs becomes a second variable and weakens the conclusion.
Reading the result
Burn the control, the original approved candle if you have it, and failed-batch candles from the beginning, middle and end of the pour, side by side in the same room. The product pages state a soy cure of 48–72 hours minimum; as a suggested starting point, compare again at about one week. Judge throw blind if someone else can label the jars. Keep watch over every burning candle and put out any that smokes or flares.
| Control today | Failed batch | Most likely meaning | Next step |
|---|---|---|---|
| Performs like the approved test | Weak at every position | Something in the production process changed the whole batch | Check oil addition temperature, weighing and time at heat; change one at a time |
| Performs like the approved test | Weak only at the end of the pour | Hold time or cooling during the pour | Shorten hold, scent smaller portions, split the pour |
| Weaker than the approved test | Weak | Materials or method have changed since approval | Compare wax and oil lots; check for a new carton or drum |
| Weaker, and the original candle is also weak today | Weak | Test conditions or perception, not the product | Re-test in a different room, with a second assessor, blind |
| Performs like the approved test | Matches control too | The ‘failure’ was not real | Check cure age and test conditions of the first assessment |
What to do with the 25 kg batch while you test
Keep the whole batch on hold. Store the candles by pour position if you numbered them; if you did not, keep the order in which trays were filled. Do not rework or remelt anything until the control result is in, because the failed candles are your evidence.
If the control shows a process cause that affected only part of the pour, candles from unaffected positions may be releasable after they pass against the control. If the whole batch is weak, rework options — remelting with a measured top-up of oil, blending into later batches, or using for testers — depend on the cause and are best decided once you know it.
Write down the hold properly: batch number, date poured, number of candles, where they are stored and why they are held. Put a sign on the shelf. It sounds excessive for one batch, but held candles have a way of being picked for an urgent order by someone who did not know about the problem.
Use the waiting days productively. Reconcile the batch record while memories are fresh: wax and oil weights actually used against the sheet, the temperature at oil addition, the time the pour started and finished, and the number of good candles against the 125 expected. Gaps in that record are themselves findings. If oil weight cannot be confirmed, or the temperature at addition was read from the melter display alone — the page states a ±10°C deviation — you already have a candidate cause to test the moment the control result comes back.
When a formula change is actually justified
Sometimes the control does fail, and the lots, method and test conditions all check out. Then, and only then, is the formula the suspect. Treat any change as a new formula: test it at 1 kg with its own control, then validate it at production size before it replaces the approved version.
Change one thing at a time — load, or wick, or wax — never several together. Keep Vanilla Bean within the stated load range on its product page, and remember the page's advice to wick-test per container and wax. A formula change made in this order improves the candle. A formula change made in a hurry after a failed batch usually just moves the problem.
Where this page fits
CSI already has guides that cover parts of this question. This page covers the failure investigation for the question above; these go deeper on the rest:
Three things to buy for this job
| Option | Price | Status on the September 2026 pull |
|---|---|---|
| 15 g | ₹130.00 | In stock |
| 50 g | ₹410.00 | In stock |
| 100 g | ₹790.00 | In stock |
| 500 g | ₹3,840.00 | In stock |
| 1 kg | ₹7,646.00 | In stock |
| Option | Price | Status on the September 2026 pull |
|---|---|---|
| 1 litre | ₹2,360.00 | In stock |
| Option | Price | Status on the September 2026 pull |
|---|---|---|
| Digital scale | ₹354.00 | In stock |
CANDLEMAKINGSUPPLIESINDIA supplies raw materials and production equipment, not finished candles. This guide separates general process physics from what CSI's own product pages state, and labels every mixing time, temperature window, hold limit, loss allowance or sample count as a starting point for your own validation.
Prices are live CSI prices from September 2026; product pages are authoritative. Where a spec is not published — such as the scale's capacity or the thermometer's range — this guide says so rather than filling the gap.
For bulk quotes, GST invoicing, IFRA and MSDS documents, message us on WhatsApp at +91 7397976926.
Frequently asked questions
- I Want My 50 Kg Production Batch to Perform Exactly Like My Original 1 Kg Formula — What Should I Validate Before Approving Full Commercial Production?
- My Large Batch Failed but a New 1 Kg Test Using the Same Raw Materials Still Works — Does That Strongly Suggest a Scale-up Process Problem Rather Than the Formula Itself?
- My Large Batch and Small Control Both Suddenly Fail — Should I Investigate Raw-Material Batch Changes Before Changing the Production Process?
- I Have 25 Kg of Scented Wax That Does Not Match My Approved Test — What Should I Investigate Before Attempting to Correct or Rework It?
- I Have Already Poured 200 Candles From a Questionable Batch — Should I Hold the Remaining Production Until Samples Have Been Evaluated?
- My First 100 Candles and Last 100 Candles From the Same Batch Perform Differently — How Should I Sample Retained Products to Identify Where the Process Drifted?
- Should Every Commercial Batch Have Retained Candles From the Beginning, Middle and End of the Production Run?
- How Can Beginning-Middle-End Retained Samples Help Identify Fragrance Distribution, Temperature Drift and Filling Problems in Large Batches?
- I’m Producing 5,000 Candles a Month — What Batch Records Should I Keep so I Can Reproduce Successful Production and Investigate Failures?
- My Small Test Batch Has Excellent Hot Throw but the 20 Kg Production Batch Smells Weaker — What Production Differences Should I Investigate Before Blaming the Fragrance Oil?
- How to Create a Standard Candle Formula for Commercial Production
Customer reviews: The reviews at the top are genuine, published Judge.me reviews — Arminder Kaur (5★, verified); Raajull Bhatt (5★, verified); Manasa GN (5★, verified); Arminder Kaur (4★, verified); Lalitha Jagan (5★, not recorded as verified); Arshad Shaikh (5★, verified). Names, ratings, dates and products as recorded by Judge.me.
Product facts (September 2026, CSI live store): Vanilla Bean Fragrance Oil: 15 g ₹130.00, 50 g ₹410.00, 100 g ₹790.00, 500 g ₹3,840.00, 1 kg ₹7,646.00 — flashpoint 203°C · vanillin 2.5% (page: potential discolouration, test with lighter waxes) · strength "moderate to strong" · page recommends wick testing per container and wax. Mini Electric Wax/Soap Melter (1 litre): 1 litre ₹2,360.00 — 300 W / 600 W modes · stainless steel liner · 1-year warranty. Candle Making Weighing Scale: Digital scale ₹354.00 — digital, grams or ounces · capacity and resolution not published.
Assumptions: Fragrance load is a percentage of total candle weight (200 g = 184 g wax + 16 g oil at 8%; a 25 kg batch at 8% = 23 kg wax + 2 kg oil). Melted soy wax is taken as roughly 0.9 g per ml for volume conversions, so the melter's stated 9.5 litres is about 8.5 kg — weigh one real fill. Mixing times, temperature windows, hold limits, loss allowances and sample counts are suggested starting points to validate, not measured CSI data. Example logs and readings are illustrations. Prices and availability change — product pages are authoritative.