I Increased My Batch Size from 1 Kg to 10 Kg and the Finished Candles No Longer Perform Exactly Like the Test Batch — What Changed?

Scale changes more than the quantity

A 1 kg test is quick and close: a double boiler you watch, a spoon that reaches everything, six jars poured in a few minutes, a thermometer in your hand. A 10 kg production run is a different event. The melter takes longer to bring the wax up. The jars are often not all ready when the wax is, so the scented wax waits at heat. A spoon or paddle that mixed 1 kg thoroughly may not reach the bottom corners of a big pot in the same time. Pouring 66 or more candles takes an hour, with the pitcher refilled from the pot two dozen times while the pot slowly changes. And somewhere along the way, the maths gets done at speed.

None of this is exotic. Each is a known tendency: long holding at heat can drive off lighter top notes and may darken some oils; short mixing in a large volume can leave the oil less evenly spread; a long pour session means the first and last candles leave the pot in different conditions. The work here is not to guess which one it was, but to find out from your own records. For the forward-looking version (what to plan before a scale-up), see the 1 kg to 25 kg planning post.

Start from the symptom: which rows could explain it

"Doesn't perform the same" covers several different faults, and each can only come from certain rows. Name the symptom first; it shortens the sheet. The ranking within each row is editorial judgement, not measured data.

If the production run was burn-tested at day 5 and the test batch at day 14, stop and wait: until the cure row matches, every other row is unreadable. Cure wording is per oil page, and 25 of 102 CSI oil pages state one in the text we captured (48–72 hours for soy where they do); many makers judge at day 7 and 14. The weaker-throw case has its own investigation order in the 20 kg weaker-throw post; this page is the sheet that serves every symptom.

The side-by-side batch sheet, rebuilt from records

Build it from what you actually recorded: the notebook, the order emails, the jar count, the dates on the labels, photos on your phone with timestamps. Where you can't reconstruct a row, write "unknown" rather than what you think you did. The worked example below is a 1 kg test batch and a 10 kg production batch of Autumn in CSI 464 at a planned 8% of total weight, poured into the 150 g Clear Glass Jar with Golden Lid.

The first three rows are the ones people are surprised by. 800 g of oil added to 10 kg of wax makes 10,800 g of candle mix at 7.41%: 0.59 percentage points below plan, 11.11 g of oil in each 150 g candle instead of 12 g, and 72 jars filled instead of 66. To hit 8% of total with all 10,000 g of wax you would have needed 869.57 g of oil, 69.57 g more (₹377.60 of Autumn at the 1 kg price); simply adding the missing 64 g to the 10,800 g mix would reach only 7.95%, because the total grows with the oil. The maths is covered properly in the 10 kg without the 10.8 kg trap post; here the point is diagnostic. If you filled more jars than the plan said, you have almost certainly found a basis slip.

The rest of the example sheet is typical: several rows differ, several are unknown, and only one matches. That is normal, and it is why the next step is the candles, not a guess.

Test candles against production candles

The sheet lists suspects; the retained candles decide between them. You need three groups, two candles each: T, retained test-batch candles; P-early, production candles from the first few jars poured; P-late, production candles from the last few. If your production jars aren't numbered by pour order, use whatever you have (the last tray filled, for instance) and note it.

No retained test candles? Make a fresh 1 kg batch the test way (80 g of Autumn and 920 g of CSI 464, 6 candles, 150 g jars and C1 wicks): ₹1,502.85 of materials, excluding labour, freight, packaging and GST. It will be younger than the production candles, so burn it at day 14 against production candles that are at least 14 days old.

Pinning the row, then proving it

When the pattern points at the batch-wide rows and the sheet shows several of them differ, test them in order of how cheap they are to prove. In the worked example, the jar count makes the basis slip the lead. You can prove it without waiting for a whole production run: make a reproduction batch at test scale with only that row changed, 80 g of oil onto 1 kg of wax as the production batch did, giving 7 candles at the same 7.41%, everything else the test way. If it burns like the production candles and unlike T, the row is confirmed. Materials: ₹1,643.74 (Autumn at the 100 g price, CSI 464 1 kg, seven jars, seven C1). If it burns like T, the basis wasn't enough on its own, and the next row on the list (the 45-minute hold, in the example) gets the same treatment.

Stakes put the effort in proportion. At the planned 66 candles, the example batch is ₹15,540.18 of materials (₹235.46 a candle, excluding labour, freight, packaging and GST). A reproduction batch is about a tenth of that. It is worth running before the next 10 kg, not after.

What not to buy after a disappointing scale-up

We sell every item on that list. None of them is the answer to "which row changed", and in our judgement most scale-up problems are answered by a pencil, a jar count and a thermometer reading.

Wax, melter, oil and the two tools

Five items from the worked example. Chosen on page statements, price per gram or unit, and stock on 26 September 2026; CSI publishes no measured throw data or lot data, so nothing is ranked on performance. The oil table shows every size, 150 g candles at 12 g, and whether a bottle covers a 10 kg batch at 8%.

Frequently asked questions