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?

What a passing control does and does not tell you

You made a 20 kg batch and it failed. You then made a fresh 1 kg batch from what you believe are the same raw materials, and it works. That result is valuable: it shows the formula, in those materials, can still make a good candle. The natural conclusion is that something about the large batch — how it was heated, scented, mixed, held or poured — caused the failure.

That conclusion is usually right. But it rests on an assumption that is easy to get wrong: that the control and the batch differ only in scale and process. If they also differ in inputs or in how they were tested, the control could be passing for a reason that has nothing to do with the process.

So the page's answer comes in two parts. First, make sure the comparison is fair. Second, use further small runs to find the exact process step that failed, rather than stopping at ‘it's the process’.

It is worth being precise about the strength of the evidence. A control that passes rules out a formula that has stopped working, and it makes a bad wax or oil lot unlikely if the lots genuinely match. It does not tell you whether the fault lies in heating, oil addition, mixing, holding or pouring, and it cannot catch a fault that only exists at depth, such as incomplete mixing through a full tank. Treat it as the end of one line of enquiry and the start of the next.

Five hidden differences to rule out first

Before concluding, check each of these. Any one of them can make a control pass while a batch fails, without the process being at fault.

The weighing check deserves special attention, because a scaling error changes the formula without anyone intending to. Take the worked example: Luxury Soy Wax Chunks with Eucalyptus Spearmint at 9% of total candle weight, as a 20 kg batch.

If the batch record reveals an error like this, the failure is a formula difference created during scaling. The control has in fact shown that the approved formula works; the batch simply was not that formula.

For cost context, the failed batch held 18.2 kg of chunks at ₹599.90 per kg (the 10 kg pack at ₹5,999.00): 18.2 × ₹599.90 = ₹10,918.18. Its 1.8 kg of Eucalyptus Spearmint at ₹3,068.00 per kg came to 1.8 × ₹3,068.00 = ₹5,522.40. That is ₹16,440.58 of wax and oil on hold.

Bridge runs: walk the control towards the batch

Once the comparison is fair and the weighing checks out, the process is the prime suspect. But ‘the process’ is several steps, and you need to know which one. The cleanest way is to make further 1 kg runs, each identical to the passing control except for one condition taken from the failed batch. Each run is a bridge between the control and the batch.

Choose the order from your batch record. If it shows oil went in when the melter display read high, start with run D. If the pour ran unusually long, start with B and C. Each run costs about the same as the control: 910 g of chunks from the 1 kg pack at ₹599.00 is 0.91 × ₹599.00 = ₹545.09, and 90 g of Eucalyptus Spearmint from the 100 g pack at ₹354.00 is 90 × ₹3.54 = ₹318.60, so ₹863.69 per run. Three bridge runs cost 3 × ₹863.69 = ₹2,591.07 — a fraction of the held batch.

Cure every bridge run alongside the control and retained batch candles, so all are compared at the same age. The product page states Eucalyptus Spearmint is rated ‘moderate to strong’; a difference between runs should still be clear when judged blind.

Reading the bridge results

The first bridge run that reproduces the failure identifies the step. If run B — the long hold — comes out weak while the control and run A pass, the batch's time at heat is the cause, and the fix is to shorten it: scent in smaller portions nearer to pouring, or pour with two people. If run A fails, the production melter is doing something the small method does not; compare its display with a separate thermometer, since the page states a ±10°C deviation.

If every bridge run passes, the cause is something a 1 kg run cannot recreate: mixing in a deep volume. Oil that is not fully distributed through 20 kg leaves some jars rich and some thin. Retained candles from the start, middle and end of the pour will usually show it, with throw varying by position rather than being uniformly weak.

The held batch and the fix that prevents a repeat

Keep the failed 20 kg batch on hold by pour position while the bridge runs cure. Once the cause is known, decide per position: candles that match the control after cure can be released; candles that do not should be kept back for rework or testers. Do not rework before the cause is known, or you may repeat it.

Then write the fix into the batch sheet as a fixed control: the maximum hold time, the oil addition temperature read on a thermometer, the tared-container weighing rule, or the mixing routine. Make the next batch with a same-day control beside it to confirm.

During every run, add Eucalyptus Spearmint below its published 199°C flashpoint and never with a melter at maximum. Supervise every comparison burn for its full duration, and extinguish any candle that smokes or develops a tall flame.

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

Frequently asked questions