What 60 electric kilns tell you about power

If you ask how much power a kiln needs, the trade answers in two voices. One voice quotes a volume rule: about a watt per cubic inch of interior, give or take. The other quotes a surface rule: about five to seven watts per square inch of interior wall. Both voices are Frederick Olsen’s, from The Kiln Book, Chapter 9. Olsen also said something the rules of thumb tend to hide. Smaller kilns work with a higher output per cubic foot than larger kilns. The variable can run as much as 25% either way. Experience matters.

This post does two things. It walks through the kiln builder, the calculator that turns a wall stackup into a steady-state heat loss at a target cone. And it puts Olsen’s two formulas on top of a 2025 dataset of 60 production kilns, to see how the rules of thumb fare against catalog reality fifteen years after Olsen compiled his own table.

Olsen’s two formulas

Both formulas come out of Chapter 9 of The Kiln Book (4th edition, 2014). Olsen states them as design starting points.

Olsen’s rule	Range	Olsen’s stated range (imperial)	Olsen’s caveat
Volume basis	61–73 W/L	1.0–1.2 W/in³	±25% variation
Surface basis	0.78–1.09 W/cm²	5–7 W/in² (averaging 6)	“You can’t determine the precise figure”

Olsen’s volume example: a 2 × 2 × 2 ft kiln is 8 ft³, and 8 × 1.5 kW/ft³ gives 12 kW. Olsen’s surface example: a 24 × 24 in front-loader has four walls of 576 in² each, 2304 in² total, and 2304 × 5 gives 11.5 kW. The two numbers (12 and 11.5) are deliberately close, because the underlying physics is the same.

Olsen prefers the surface rule, and Olsen is right to. Heat leaves through the surface, so a surface-based rule is less sensitive to geometry. The next two sections explain the calculator’s underlying physics, then put both rules on the data.

What the calculator computes

A kiln at temperature is a steady-state heat balance. Power in equals power out. Power out is the sum of three paths from the outer shell to the room.

Conduction through the wall. The driver is the temperature drop across the wall divided by the wall’s thermal resistance:

Q_cond = ΔT · A / R, where R = thickness / k, summed in series across layers.

Convection off the outer shell. Air rising past a hot vertical surface carries heat with it:

Q_conv = h · A · (T_s − T_a), with h ≈ 4 to 7 W/m²K for kiln shells in still studio air.

Radiation off the outer shell. Stefan-Boltzmann, with an emissivity around 0.9 for painted or oxidised steel:

Q_rad = ε · σ · A · (T_s⁴ − T_a⁴), σ = 5.67 × 10⁻⁸ W/m²K⁴, ε ≈ 0.9.

The trick is that conduction depends on the outer shell temperature, and so do convection and radiation, but the shell temperature is exactly what you are solving for. The tool finds it by Newton-Raphson iteration with damping. Convergence usually arrives in well under a hundred iterations.

One more wrinkle. Thermal conductivity of insulating firebrick and ceramic fiber is itself a function of temperature. K-23 brick reads about 0.13 W/mK at 260 °C mean and about 0.27 at 1260 °C. Ignore that and you will be off by a large fraction in wall heat flux at cone temperatures. The tool runs an outer loop: solve the shell temperature with a trial k, recompute k at the new mean wall temperature, repeat until both stop moving. The k(T) tables come from Morgan Thermal Ceramics datasheets (ASTM C-182 for firebrick, C-201 for ceramic fiber).

The survey

I gathered specifications for 60 electric kilns from Skutt, L&L, Olympic, Cress, Paragon, Jen-Ken, Evenheat, Cone Art, Bailey, AMACO, Nabertherm, Rohde, and Kittec. For each I recorded interior dimensions, watts, voltage, wall thickness, brick grade, and any extra backup insulation. Where the spreadsheet disagreed with the manufacturer’s own spec page, I took the manufacturer’s value. From dimensions I computed interior volume and interior surface area, and from those, the two power densities that matter: W per litre of interior, and W per square centimetre of interior wall.

The data

Sixty kilns, sorted by interior volume.

#	Brand	Model	Vol (L)	Watts	W/L	W/cm²	W/in²	Wall (in)	Construction
1	Skutt	FireBox 8×4	5	1725	367	0.99	6.4	2	—
2	Skutt	FireBox 8×6	6	1800	287	0.87	5.6	2	—
3	Olympic	HB89E	9	1800	191	0.67	4.3	2	—
4	L&L	Plug-n-Fire	9	1500	159	0.56	3.6	2	—
5	L&L	DLH11-DX	14	2800	200	0.87	5.6	2.5	K23
6	Nabertherm	Top 16/R	15	2600	171	0.76	4.9	2.5	5-side heating
7	Cress	ET911	18	1500	84	0.36	2.4	2.5	K23
8	Skutt	KM-614-3	21	2300	110	0.54	3.5	2.5	K23
9	Evenheat	Studio Pro 17	33	2160	65	0.32	2.0	2	Fiber lid
10	Jen-Ken	AF 1513	42	4800	116	0.67	4.3	2.5	K23
11	L&L	e14S-3	49	4980	102	0.67	4.3	3	K23
12	Evenheat	HF 1813	53	5040	94	0.64	4.1	2.5	K23
13	Rohde	Ecotop 43	54	2900	54	0.37	2.4	1.4	Microporous
14	Cone Art	1813D	54	4500	83	0.56	3.6	2.5	K23, 1” block
15	Cress	E18	56	5300	94	0.65	4.2	2.5	K23
16	L&L	e18S-3	63	5740	91	0.65	4.2	3	K23
17	Jen-Ken	AF 1815	66	6240	94	0.63	4.1	2.5	K23
18	Kittec	CB 70 S	70	2800	40	0.30	1.9	2.8	Microporous
19	Olympic	1818E	71	5040	71	0.53	3.4	2.5	K23
20	Skutt	KM-818	71	6660	94	0.70	4.5	2.5	K23
21	Paragon	TNF823	78	7200	92	0.71	4.6	3	K23
22	Rohde	Ecotop 60	80	3600	45	0.35	2.3	1.4	Microporous
23	AMACO	EX-232SF	87	6500	75	0.59	3.8	2.5	K23
24	Cress	E23	92	8600	94	0.74	4.8	2.5	K23
25	L&L	e18T-3	95	8400	89	0.71	4.6	3	K23
26	Kittec	CB 100 S	96	8600	90	0.74	4.8	2.8	—
27	Nabertherm	N 100	112	9000	80	0.64	4.2	3	—
28	Skutt	KM-1018-3	113	8400	75	0.64	4.1	3	K23
29	L&L	e23S-3	115	8640	75	0.66	4.2	3	K23
30	Cone Art	2318D	123	8400	69	0.61	3.9	2.5	K23, 1” block
31	Jen-Ken	AF3C 1822	143	6240	44	0.38	2.4	2.5	K23
32	AMACO	HF-101	143	9000	63	0.54	3.5	3	K23, 5” combined
33	Olympic	2518HE	145	10080	70	0.65	4.2	3	K23
34	AMACO	EX-257SF	156	10800	69	0.67	4.3	2.5	K23
35	L&L	e23T-3	173	11520	67	0.67	4.3	3	K23
36	Skutt	KM-1027-3	174	11520	66	0.67	4.3	2.5	K23
37	Cress	E24HP	178	11500	65	0.66	4.2	3	K23
38	L&L	e28S-3	182	11500	63	0.63	4.1	3	K23
39	Olympic	2023HE	189	11520	61	0.62	4.0	3	K23
40	Evenheat	HF 2327	192	11520	60	0.62	4.0	2.5	K23
41	Nabertherm	N 200	197	15000	76	0.74	4.8	3	5-side heating
42	Evenheat	RM 2329	206	10800	52	0.56	3.6	2.5	K23
43	Jen-Ken	AF 2822	222	10800	49	0.49	3.1	3	K23
44	L&L	eQ2327-3	234	13440	57	0.59	3.8	3	K23, 4-zone
45	Paragon	Dragon 24	255	16500	65	0.68	4.4	3	K23, 1” fiber
46	Bailey	TL-4222	262	18000	69	0.75	4.8	3.5	K24, 1” fiber
47	Bailey	TL-2327-10	272	11520	42	0.50	3.2	3.5	K24, 1” fiber
48	Bailey	TL-2827	272	15120	55	0.65	4.2	3.5	K24, 1” fiber
49	L&L	e28T-3	272	16620	61	0.71	4.6	3	K23, 3-zone
50	Skutt	KM-1227-3	275	11520	42	0.49	3.2	3	K23, Energy Saver
51	Olympic	2827HE	277	13440	48	0.57	3.7	3	K23
52	AMACO	HF-105	278	12500	45	0.53	3.4	3	K23, 5” combined
53	Cress	FXC30FH	283	18000	64	0.69	4.5	4	K23, 2.5” backup
54	Olympic	3027HE	313	14400	46	0.56	3.6	3	K23
55	Jen-Ken	AF JK² 29”	347	13200	38	0.45	2.9	3	K23
56	Skutt	KM-1227-3PK	366	14300	39	0.50	3.2	3	K23
57	Paragon	GL64	369	16500	45	0.46	3.0	2	Fiber, glass
58	L&L	T2327-D	413	19500	47	0.58	3.7	3.5	K23, sectional
59	Paragon	Super Dragon	430	21600	50	0.63	4.0	3	K23, 1” fiber
60	L&L	T2336-D	551	26000	47	0.64	4.1	3.5	K23, sectional

Headline statistics

The same 60 kilns described by two yardsticks.

Statistic	W/L	W/cm²	W/in²
Minimum	38.1	0.30	1.92
25th percentile	51.9	0.54	3.50
Median	67.6	0.635	4.10
75th percentile	91.3	0.67	4.33
Maximum	367.2	0.99	6.36
Mean	83.6	0.61	3.93
Coefficient of variation	69%	22%	22%

The volume rule has a coefficient of variation of 69%. The surface rule has a coefficient of variation of 22%. Both yardsticks describe the same kilns; the surface rule describes them with a band three times tighter.

Olsen’s volume range converts to 61 to 73 W/L (1.0 to 1.2 W/in³). The survey median is 67.6. Right inside Olsen’s nominal band, with no thumb on the scale. Olsen’s surface range of 0.78 to 1.09 W/cm² (5 to 7 W/in²) runs slightly high of the survey median of 0.635 W/cm². That gap is real and not a measurement artefact: it is what fifteen years of catalog evolution toward better insulation looks like.

Power density by size class

Olsen’s caveat (smaller kilns higher, larger kilns lower) shows up cleanly when the data is split by interior volume.

Size class	n	Median W/L	Median W/cm²
Tabletop, under 30 L	8	180.9	0.72
Small studio, 30 to 100 L	18	90.5	0.64
Medium, 100 to 200 L	15	66.7	0.64
Large, 200 L and up	19	48.5	0.57

The volume figure swings by more than threefold across the four buckets, from 181 W/L at the tabletop end to 49 at the commercial end. The surface figure barely moves. A small test kiln does not really need 181 W per litre any more than a Jen-Ken AF JK² needs 38; both need roughly the same watts per square centimetre of wall, because the wall is what loses heat.

Why the volume rule swings

It is a geometry argument, and Olsen named it. Heat leaves through the surface. Surface area scales as length squared. Volume scales as length cubed. Two kilns of similar proportions but different sizes have surface-to-volume ratios in the inverse ratio of their linear dimensions. The smaller one has more surface per litre, so it loses more heat per litre, so it needs more watts per litre to hold temperature. A volume rule has to climb as kilns get smaller. The survey reproduces this prediction across two and a half decades of interior volume.

The power-law fit

If watts scaled exactly with surface area, watts would scale with V^2/3 for kilns of similar shape. Fitting watts against interior volume on log-log axes gives:

W = 405 · V_L^0.630, n = 60, R² = 0.90.

The exponent is 0.630. The geometric prediction is 2/3 ≈ 0.667. They are close, and they should be: kilns are not all the same shape (cylinders, rectangles, ovals, front-loaders), and they do not all run to the same maximum temperature, but they cluster around the prediction that watts follow surface. Olsen said it in words. The fit puts a number on it.

Patterns in the data

A few correlations from the same dataset, with the standard caveats: small subgroup samples, observational, confounded.

Wall thickness and rated W/cm²

A reasonable prior: thicker insulation should let you spec less power per square centimetre, because steady-state loss falls. The data say barely.

r(wall, W/cm²) = +0.17, r(wall, W/L) = −0.40.

Catalog wall thickness has almost no correlation with rated W/cm². The moderate negative correlation against W/L mostly reflects that bigger kilns get thicker walls and bigger kilns have lower W/L by geometry.

The explanation is that rated wattage on a production kiln is not set by steady-state loss. It is set by the promised heat-up rate. To get from cone 06 bisque to cone 6 in eight hours you need power in the kilowatts; to hold a finished kiln at cone 6 you need a fraction of it. Catalog wattage is the heat-up number. Steady-state loss runs beneath it. Two kilns of the same interior with different wall thicknesses can land at the same rated power if their makers picked the same heat-up profile.

This also answers a question I get from people new to electric kilns. Their digital controller calls for 50 to 60% power at cruising temperature. The other 40 to 50% is the gap between steady-state loss and rated power. The kiln is not under-built; the controller is doing what it should.

Brick grade

K-23 is the standard grade across the field. The three Bailey units use K-24 (similar density, similar k, slightly higher rated service temperature). They group at a median 55.5 W/L versus 66.5 W/L for K-23. Most of that gap is size, not grade: the Bailey models are large, and large kilns sit lower on the volume rule by geometry. Inside W/cm² the K-23 and K-24 groups overlap.

Microporous construction

Three kilns in the survey use microporous insulation board: two Rohde Ecotops and the Kittec CB 70 S. They sit at a median 0.351 W/cm². The other 57 sit at 0.638. A 45% reduction in spec’d surface power for the only construction that genuinely changes the wall’s R-value. Microporous board runs roughly k = 0.025 W/mK, five times better than dense firebrick at the same thickness. Three kilns is a small sample. The direction is unsurprising.

What I’d still like to learn

Some questions the survey cannot answer without more data:

• Element life vs surface power density. Catalog watts is one thing; watts per element, watts per linear centimetre of element groove, and the duty cycle the controller runs are the actual drivers of element life. Manufacturers do not publish those, and they cannot be backed out from spec sheets.
• Heat-up curves at real load. A kiln loaded to half capacity with mass-heavy pieces draws differently from an empty kiln. The thermal-mass term grows; the steady-state term does not.
• Insulation degradation with age. A ten-year-old K-23 wall pulls more current than a fresh one because the brick has gained density from thermal cycling. That needs field data.

The kiln builder will get those when I have the data. For now it tells you what fresh insulation costs you to hold cone six, and the survey tells you where production kilns sit. If you find a catalog spec that disagrees with both, write to me.

TL;DR

The three numbers worth carrying out of this:

Question	Quick rule	Source
How many watts for a kiln of volume V?	61 to 73 W/L (1.0 to 1.2 W/in³)	Olsen, The Kiln Book, Ch. 9
How many watts for interior surface A?	0.78 to 1.09 W/cm² (5 to 7 W/in²)	Olsen, The Kiln Book, Ch. 9
Cross-check against 2025 production	0.55 to 0.70 W/cm² (3.5 to 4.5 W/in²)	This survey, n = 60

The surface rule beats the volume rule on every test the data supports, exactly as Olsen recommended. Sanity-check any volume calculation against W/cm². When the two disagree, trust the surface figure. Below the survey’s 0.55 W/cm² floor you find commercial-duty units with very heavy insulation or microporous small kilns; above its 0.70 W/cm² ceiling you find tabletops paying the surface-to-volume penalty.

For a new build aimed at a working studio (50 to 300 L), settle on 0.60 to 0.70 W/cm² of interior wall, then run the geometry through the kiln builder to see what the steady-state loss actually is at your target cone. If the calculator says less than 60% of your spec’d power, the controller has headroom to spare.

Bibliography

Only sources actually drawn from. The kiln builder and this post use four references; the rest of the kiln literature is good background but did not contribute specific values or formulas to either.

• Olsen, Frederick L. (2014). The Kiln Book (4th ed.). Krause. Chapter 9 is the source for both quick formulas (volume rule, surface rule) and the electrical specifications table the calculator matches against.
• Morgan Thermal Ceramics, K-Brick and Kaowool Cer-Wool product datasheets (ASTM C-182, C-201). Source for the k(T) interpolation tables in the calculator’s outer loop.
• Edward Orton Jr. Ceramic Foundation. Pyrometric cone temperature chart (self-supporting, 60 °C/hr). Drives the calculator’s target-temperature selector.
• Schonert, M. (2016). “Heat transfer calculations,” RepKiln project log, hackaday.io. The Newton-Raphson plus radiation and convection structure of the solver follows this writeup.