Thermal Properties of Wood and Fire Behaviour

Why this matters in the sawmill

Every kiln schedule, every steaming or heat-treatment cycle and every fire-safety conversation with a customer rests on a few thermal properties of wood: how fast heat moves through it, how much heat it stores, how much it expands and what happens when it is exposed to flame. A sawmill that understands these properties can explain why a 100 mm plank needs a very different schedule from a 25 mm board, why a timber beam can outlast an unprotected steel beam in a fire, and why a "fire-rated" label on a cladding board depends on thickness, density and mounting rather than on species alone.

This article collects measured values with their sources and marks which are European and which are North American. Where a primary document could not be opened, this is stated. Regulatory statements carry the date they were checked (26 October 2026). Numbers below are illustrative of the order of magnitude; they are not a substitute for the species-specific data in your product standard or kiln supplier's documentation.

Thermal conductivity: why timber insulates

Thermal conductivity (lambda, W/(m K)) tells you how much heat flows through a 1 m thick slab per square metre for each kelvin of temperature difference. Wood is a porous cellular solid: the cell wall substance conducts reasonably well, but most of the volume of dry softwood is air-filled lumen, and still air conducts very little. Conductivity therefore rises with density (more wall substance, fewer air gaps) and with moisture (liquid water conducts far better than air).

Measured European values (the draft attributed them to Slovak laboratory work; that origin was not re-confirmed from the chapter text), reproduced in an open-access book chapter, are a useful anchor. For beech (703.9 kg/m3) the chapter states 12 % moisture and 20 degrees C and gives 0.38 W/(m K) along the grain, 0.23 radially and 0.16 tangentially. Fir (414.5 kg/m3) has 0.36, 0.17 and 0.13 W/(m K); the chapter does not explicitly state the same conditions for fir, so treat the two species as not strictly comparable. Even dense beech is a poor conductor, and across the grain the lighter fir is better still.

Three patterns are worth remembering:

• Direction. Along the fibres heat follows continuous cell walls; across the grain it crosses walls and lumina repeatedly. Steinhagen's 1977 Forest Products Laboratory literature review (a US compilation of older studies, so not purely North American data) lists longitudinal-to-radial ratios of about 1.75 for ash, 1.85 for mahogany and spruce and 2.25 for walnut (air-dry, about 20 degrees C; Griffiths and Kaye, 1923), and 2.25 to 2.75 for Douglas-fir and red oak at 6 to 15 % moisture and about 30 degrees C (MacLean, 1941). Tangential conductivity was about 0.9 to 0.95 of radial in three hardwoods and one softwood. By our own calculation the beech and fir figures above give longitudinal-to-radial ratios of about 1.7 and 2.1, in the same range.
• Density. The review notes conductivity rises roughly linearly with specific gravity near room temperature, so spruce, fir and poplar insulate better than oak, beech or ash.
• Moisture. Conductivity increases as wood wets, and the direction ratios shift with moisture. The review uses different adjustment lines below and above 40 % moisture (its adjustment factor for conductivity is given graphically); no moisture coefficient is given here.

This is why timber suits walls and roofs: heat usually crosses the grain, where conductivity is lowest, and the material is light. It also feels warm to the hand, which is governed by thermal effusivity, the square root of conductivity times density times specific heat. The chapter lists 460 to 710 for beech and 300 to 500 for fir (W s^0.5/(m2 K), tangential to longitudinal). Low effusivity means the contact surface stays near hand temperature. No verified figure for metals was used, so that comparison stays qualitative.

Heat storage, diffusivity and thermal expansion

Specific heat (J/(kg K)) is the energy needed to warm one kilogram by one kelvin. For wood it is close to independent of species and direction, so it is treated as a scalar. The same European chapter gives 1900 J/(kg K) for beech and 1700 for fir. It cites older formulae for oven-dry wood over 0 to 100 degrees C, 1.571 + 0.00277 theta (Perelygin, via Pozgaj et al. 1993) and 1.1136 + 0.004856 theta (Dunlop, via Radmanovic et al. 2014), in kJ/(kg K) with theta in degrees C, so specific heat climbs with temperature. It also rises with moisture (general physics, not quantified in the sources used), because water holds far more heat per kilogram than dry wood, which is why a wet log takes much more energy to warm than a dry board.

How fast a piece warms depends on thermal diffusivity, alpha = lambda / (density x specific heat), in m2/s. The chapter gives 2.9, 1.7 and 1.2 x 10^-7 m2/s (longitudinal, radial, tangential) for beech and 5.2, 2.4 and 1.8 x 10^-7 for fir. Check for fir tangential: 0.13 / (414.5 x 1700) = 1.8 x 10^-7. The lighter fir has the higher diffusivity: it conducts less but stores far less heat per cubic metre, so it equalises faster.

Moisture changes the picture. Steinhagen quotes longitudinal-to-radial diffusivity ratios of about 4 to 5.5 at 20 % moisture and about 60 degrees C, but only about 2 at 90 % moisture or above (from Maku, 1954, who attributed the high dry-side ratio to moisture moving down the temperature gradient), so green wood is more isotropic. He also notes that free water freezes slightly below 0 degrees C, usually between -0.1 and -2 degrees C depending on dissolved sugars. The latent heat of free water freezing is quoted at 80 kcal/kg (Chudinov); our inference is that this makes frozen logs slow to thaw.

Thermal expansion (per degree C, at zero moisture) is small along the grain and moderate across it. For fir: 3.9 x 10^-6 longitudinal, 21.7 x 10^-6 radial, 31.6 x 10^-6 tangential. For beech: 3.1 to 4.5 x 10^-6 longitudinal, 31.8 x 10^-6 radial, 40.3 x 10^-6 tangential (Kollmann and Cote 1968 for fir; Glass and Zelinka 2010, i.e. the US Wood Handbook, for beech, as cited in the chapter; the chapter does not give moisture or temperature conditions for these). By our own calculation, a 50 K rise expands a tangential fir dimension by about 0.16 %, or 1.6 mm per metre. That is usually overshadowed by moisture movement, and along the grain it matters mainly where wood meets steel, concrete or glass.

Heat flow in kilns and why thick pieces dry slowly

In a kiln, three things happen at once: the board warms up (heat conduction), water moves toward the surface (mainly diffusion in the cell wall below fibre saturation, and liquid or vapour flow above it) and water evaporates, consuming heat at the surface. Because evaporation takes energy, the surface is cooler than the air, and the air layer next to the board becomes more humid than the bulk kiln air. Forest Products Laboratory report R1264 (sugar maple, a North American hardwood; dated April 1951 and marked as reviewed and reaffirmed at an unstated later date) documents this: for a given drying rate at 60 % moisture content, 1600 feet per minute air velocity needed only 64 % relative humidity to give 2 % moisture loss per hour, while at 400 feet per minute the relative humidity had to fall to 43 % for the same rate. Air speed controls how much the boundary layer saturates, so it controls drying rate.

Why do thick pieces dry slowly? Diffusion distance. Time to move moisture (or heat) a distance L scales with L squared divided by the relevant diffusivity. The FPL report says that for wide stock drying time is more nearly proportional to the square of the thickness; relative to 1 inch stock it gives factors of about 4 for 2 inch, 9 for 3 inch and 16 for 4 inch stock (the scanned table is partly garbled, so these values should be checked against the original). The draft's statement that narrower widths show smaller factors because of edge drying was not confirmed in the copy read and is treated as unverified. Real schedules deviate from a pure square law, but it explains why doubling thickness often means quadrupling drying time.

A similar scale applies to pure heating. A rough order-of-magnitude estimate that we calculate (not a published figure): with half-thickness 25 mm and a tangential diffusivity of 1.8 x 10^-7 m2/s, L2/alpha is about 3,500 s, around an hour; with half-thickness 50 mm it is about 14,000 s, nearly four hours. That is for dry wood; wet wood stores more heat, so its response differs. A board warm at the surface can still be cold in the core, and fast surface drying against a cold core builds gradients and stress.

Fire behaviour: ignition, charring and why large sections are predictable

Wood does not burn as a solid. Heat breaks it down (pyrolysis) into combustible gases that ignite, leaving a carbon-rich char. A 2013 open-access review (Lowden and Hull) gives decomposition ranges of 180 to 350 degrees C for hemicellulose, 275 to 350 for cellulose and 250 to 500 for lignin (citing Kim et al. 2006), with main wood pyrolysis beginning above about 225 degrees C and ending below 500 degrees C; charred surfaces can reach about 800 degrees C.

Ignition is better described by heat flux than by one "ignition temperature". A review for tall timber construction (Fire Technology, 2018; the author names were not visible in the page text read) gives about 10 to 13 kW/m2 for piloted ignition, 25 to 33 kW/m2 for unpiloted ignition and 5 to 10 kW/m2 for smouldering, typically near 200 degrees C surface temperature. It states that reported auto-ignition temperatures vary by more than 150 degrees C with test conditions. Moisture delays ignition: a cited relation implies dry wood ignites in about half the time of wood at 12 % moisture. Heat of combustion is about 17.5 MJ/kg (15 to 20).

The charred layer has about 20 % of the virgin wood density and an emissivity near 0.95 against 0.7 for wood. It insulates: about 6 mm below the char line temperatures are around 180 degrees C, so the wood behind it is largely intact. The review states that the char front is usually taken as the 300 degrees C isotherm (288 degrees C in North America).

Charring rates, mostly for standard-fire exposure:

• About 0.8 mm/min for light, dry softwoods down to 0.4 to 0.5 mm/min for dense, moist hardwoods, averaging about 0.6 mm/min (Fire Technology review; Lowden and Hull also give about 0.6 for softwoods).
• Eurocode 5 (EN 1995-1-2) uses a one-dimensional design charring rate beta_0 of 0.65 mm/min for solid softwood of at least 290 kg/m3 characteristic density; this is confirmed in a conference paper from Hasselt University quoting the standard and, as a 2010 secondary summary (Cachim and Franssen), also for glued laminated softwood (0.65) and for solid or glulam hardwood (0.65 at 290 kg/m3, 0.50 at 450 kg/m3 or more). The standard text itself was not opened, so these remain second-hand. The notional charring rate beta_n (which includes corner rounding and fissures) and the LVL values are not verified here and are left out. Another open paper states that beta_0 = 0.65 mm/min was adopted for softwood analytical calculations.
• In small-scale tests at 50 kW/m2, charring was 1.07 mm/min at 0 % moisture, 0.80 at 10 % and 0.64 at 20 %. Density and moisture together change rates by up to about 18 % over design ranges; incident heat flux matters roughly ten times more.
• The initial peak can be four to five times the steady rate, reached after about 20 minutes. One CLT study found 0.80 mm/min unprotected against 0.44 mm/min with a gypsum layer.

Richter and co-authors (Fire Technology, 2020) modelled 95 mm spruce slabs (430 kg/m3, 12 % moisture): EN 1995-1-2 gives 58.5 mm of char after 90 minutes of standard fire against 53.2 mm simulated. Current practice adds a fixed 7 mm zero-strength layer to the char depth; the authors argue this may not be conservative for non-uniform fires (they found layers up to 12 mm at 12 % moisture), and in their simulations at 5 % moisture the layer could be 16 mm.

Why predictable? In a large section charring is a surface process advancing at a nearly constant rate, with the char shielding the core, which keeps most of its strength. A section exposed on three sides for 60 minutes loses roughly 36 to 39 mm per exposed face at 0.6 to 0.65 mm/min (our own arithmetic, ignoring corner rounding). Small sections lose a much larger fraction, and a protective layer falling away accelerates charring.

Fire retardant treatments and reaction-to-fire classes

Reaction to fire, meaning how a material contributes to fire growth, is classified in Europe by EN 13501-1 into A1, A2, B, C, D, E and F, with smoke sub-classes s1 to s3 and burning droplet sub-classes d0 to d2. The main test is the Single Burning Item test (EN 13823). Untreated wood and wood-based panels typically fall in class D (Ostman and Mikkola, Tratek report, 2004: main class D verified for most wood-based panels; low-density products fall to E, OSB is listed as D-s2,d2). The FIGRA (fire growth rate) limits quoted in the open literature are below 750 W/s for D, below 250 for C and below 120 for B (Lowden and Hull, 2013). Cross-laminated timber is classified D-s2,d0 for walls and ceilings without further testing under Commission Delegated Regulation (EU) 2017/2293 (of 3 August 2017) when overall thickness is at least 54 mm, each layer at least 18 mm and mean density at least 350 kg/m3 (conditioned per EN 13238); for flooring the same CLT is class DFL-s1 (pine surface layers need 430 kg/m3, spruce 400 kg/m3). The same regulation covers laminated veneer lumber. A separate delegated regulation, 2024/1399 of 10 November 2023, covers solid wood panelling and cladding and entered into force on 20 August 2024 (EUR-Lex metadata). Its annex, as seen in a secondary copy, gives for untreated pieces of at least 390 kg/m3 mean density: D-s2,d2 at 9/6 mm total/minimum thickness and D-s2,d0 at 12/8 mm with no air gap or a closed air gap behind, D-s2,d0 at 9/6 mm with an open air gap of up to 20 mm, and D-s2,d0 at 18/12 mm for no or open air gap. Check the official annex before quoting any cladding class. Status as of 26 October 2026.

Fire retardant treatment (impregnation or surface coating) can lift wood into class B or C (Lowden and Hull; Ostman and Mikkola note that fire retardant treated wood products always have to be tested and classified separately) but cannot make it non-combustible. EN 16755:2017 classifies how long fire retardant treated wood keeps its reaction-to-fire performance in interior and exterior end uses (BSI listing; it excludes wood-based panels, which fall under EN 13986). The claim in the draft that performance can degrade outdoors as chemicals leach is plausible and is the reason for that standard, but the Ostman et al. (2018) paper that was cited could not be opened, so that attribution is unverified. Never assume a class for treated timber from its species.

Practical rules of thumb for sawmills

• Treat the heat-up time as a thickness-squared effect: doubling thickness roughly quadruples the time for core temperatures to follow the air.
• Do not set kiln schedules on air temperature alone; core temperature lags, and wet cores hold more heat.
• Higher air speed lowers boundary-layer humidity and speeds drying; check airflow before raising temperature.
• Expect stronger insulation across the grain than along it, and from light species than from heavy ones.
• Never quote a single ignition temperature to a customer; use the heat-flux concept and refer to the test standard.
• For structural fire calculations, refer customers to EN 1995-1-2 and the national annex.
• For fire class claims, quote only certified product documents.

What is uncertain or varies

Thermal property data vary by species, density, moisture, temperature, extractives and test method, so published values for one species can differ between sources. The US Wood Handbook thermal chapter and EN 1995-1-2 itself could not be opened, so it is not known which Handbook edition any coefficient comes from; North American values come second-hand via the Steinhagen review (1977) and the FPL report (1951). Beta_n values and LVL charring rates are left out because they could not be verified. The Ostman et al. (2018) paper could not be opened (access denied), so statements based on it were replaced by verified ones where possible. Euroclass regulations are updated regularly; the statements above were checked on 26 October 2026, so verify the current legal text.

Key points

• Wood insulates because it is porous; conductivity is lowest across the grain and in light species, and rises with moisture and density.
• Diffusivity (about 1 to 5 x 10^-7 m2/s in the data above) governs how quickly heat penetrates; time scales with thickness squared.
• Thermal expansion across the grain is roughly 20 to 40 x 10^-6 per degree C; along the grain about 3 to 4.5 x 10^-6.
• Charring of about 0.65 mm/min (EN 1995-1-2 beta_0, solid softwood of at least 290 kg/m3; Europe) and an insulating char layer make large sections predictable.
• Ignition depends on heat flux, moisture and test method; untreated wood is typically Euroclass D; fire retardant treatment needs certification and durability evidence.