Cone Calorimeter 101: HRR, THR, and MARHE Explained

A cone calorimeter burns a 100 mm square of material and hands you a fire-hazard profile in roughly fifteen minutes. Three numbers carry most of that story. Peak heat release rate, or pHRR, says how fiercely the sample burns. Total heat release, or THR, says how much fuel it really holds. MARHE — the maximum average rate of heat emission — says how bad the worst sustained burn gets. Rail, maritime, and building codes all lean on these three. Yet no single one of them tells you enough. So this post takes each apart: what it captures, how labs compute it, and where it misleads.

TL;DR

• A cone calorimeter turns a 100 × 100 mm sample into a full fire-hazard profile in roughly 15 minutes.
• The whole method rests on one fact: burning organic fuel yields 13.1 MJ of heat per kg of oxygen, to within ±5%.
• pHRR captures fire intensity. THR captures fuel load. MARHE folds peak, timing, and duration into one scalar.
• Vytenis Babrauskas built the first cone at the US National Bureau of Standards in 1982.
• ISO 5660-1, ASTM E1354, and NFPA 271 all describe the same bench.
• EN 45545-2 caps MARHE at 90 kW/m² for HL2 rail interiors, and at 60 kW/m² for HL3.
• pHRR at 50 kW/m² spans two orders of magnitude: 150 kW/m² for pine, 1,738 kW/m² for a carbon-fibre composite.
• One lab repeats pHRR to 5–10%. Across labs, TTI and THR scatter by 15–25%.

Who invented the cone calorimeter?

Vytenis Babrauskas built the first one at the US National Bureau of Standards, today NIST, in 1982. He wrote it up in NBSIR 82-2611.

The design solved a real frustration. Earlier heat-release benches leaned on sensible-enthalpy or substitution-burner tricks. They ran, but they drifted badly. As a result, no two labs could compare results with confidence.

Babrauskas took a different route. Rather than close a thermal energy balance, he tracked how much oxygen the burning specimen stripped from the exhaust. That single switch cut systematic error by an order of magnitude. In 1988 the bench won an R&D 100 Award — the first fire-testing tool ever honoured.

Three parallel documents now govern the instrument. ISO 5660-1 came first, in 1993; its third edition dates from 2015, with a 2019 amendment. ASTM E1354-23 covers North America. NFPA 271 serves the fire-protection trade. All three describe the same bench, so results travel between them.

How does a cone calorimeter test run?

A conical heater irradiates a 100 mm square sample from 25 mm above. A spark igniter then lights the volatiles. The whole run rarely exceeds twenty minutes.

That name comes from the heater — a truncated conical resistance element. Makers wind it so the smaller aperture faces down. The cone geometry then spreads radiant flux across the sample to within ±2%. Labs can dial that flux anywhere from 0 to 100 kW/m².

Four settings dominate practice: 25, 35, 50, and 75 kW/m². In order, they stand for an early fire, a developing compartment fire, the onset of flashover, and a severe post-flashover exposure. Most reaction-to-fire specifications name 50 kW/m².

Sample prep decides half the answer. Labs condition every specimen for at least 48 hours at 23 °C and 50% relative humidity. They then wrap it in aluminium foil. Next they back it with a ceramic fibre blanket, and seat it in a retainer frame. A 10 kV spark igniter hangs above. ASTM E1354 also allows a vertical set-up. That option suits only materials that will not melt and run.

Why does oxygen consumption measure heat?

Because almost every organic fuel gives off the same energy per kilogram of oxygen burned. What the fuel contains barely matters.

W. M. Thornton spotted the pattern back in 1917. Clayton Huggett (1980) then extended it to solids and pinned the number down: 13.1 MJ per kilogram of oxygen, with a standard deviation of just 0.35 MJ/kg. That works out to roughly ±5%.

That constancy carries the whole cone calorimeter method. Ignition chemistry, soot yield, and char fraction swing wildly from one fuel to the next. The per-oxygen energy release does not. So you can skip the thermal balance entirely. Instead, you watch oxygen vanish from the exhaust duct. Then you multiply by 13.1.

W. J. Parker turned that idea into algebra, first in a 1982 NBS report and again in a 1984 journal paper. His formula still runs inside the software of every modern cone calorimeter.

$\dot{q} = E \cdot \frac{\varphi}{1 + \varphi(\alpha - 1)} \cdot \dot{m}_a \, \frac{M_{\text{O}_2}}{M_a} \, X^{0}_{\text{O}_2}$

Read it term by term. $E$ carries the 13.1 MJ/kg-O₂ constant. The oxygen depletion factor $\varphi$ measures how much oxygen the flame stripped from incoming air. The expansion factor $\alpha \approx 1.105$ corrects for the volume change of combustion. Finally, $\dot{m}_a$ gives the incoming air mass flow. Then $X^{0}_{\text{O}_2}$ sets the ambient oxygen mole fraction.

Marc Janssens sharpened the formula in 1991. When the exhaust carries carbon monoxide, his version adds a second constant, $E_{CO} = 17.6$ MJ/kg-O₂. That figure covers the heat CO would have released had it burned all the way to CO₂. ISO 5660-1 and ASTM E1354 both run the Janssens form. Later, Yang (2019) re-derived the whole family for the Journal of Research of NIST. In 2016 the DiNenno Prize named oxygen-consumption calorimetry one of fire science’s most consequential ideas.

What hardware turns gas into a number?

Five sensors inside a cone calorimeter turn exhaust gas into a heat release curve. Each one feeds its own term of the equation above.

A paramagnetic analyser tracks oxygen with excellent linearity. Two NDIR sensors watch carbon monoxide and carbon dioxide. A strain-gauge load cell, resolved to 0.1 g, logs mass loss in real time. An orifice plate with a differential-pressure transducer yields exhaust mass flow:

$\dot{m}_e = C \sqrt{\Delta P / T_e}$

Here $C$ folds in duct geometry and the calibration constant. Then $\Delta P$ gives the pressure drop across the orifice, and $T_e$ the exhaust temperature. Finally, a 632.8 nm helium–neon laser photometer measures smoke obscuration in the duct. Babrauskas and Mulholland added that photometer in 1987. Its arrival explains why the current ISO 5660-1 title now spells out “smoke production rate (dynamic measurement)”.

What is heat release rate, and why does it dominate?

Heat release rate measures the power a burning sample throws off per unit area, in kW/m². It predicts fire growth better than any other single property.

Babrauskas and Peacock (1992) called it “the single most important variable in characterizing the flammability of products and their consequent fire hazard.” At first that claim jars. After all, smoke and toxic gases kill far more people than flames do.

However, HRR predicts how fast a room reaches those lethal conditions. Once a fire really takes hold, it floods the space with smoke and carbon monoxide anyway. The material’s own toxic potential barely shifts that. So the metric that tracks growth rate wins on hazard.

Peak HRR simply takes the maximum of that curve. It governs whether a room tips into flashover. It also drives the radiative feedback that ignites nearby fuel. Time-to-flashover in the ISO 9705 room-corner test tracks cone pHRR closely. Östman and Tsantaridis (1994) first published that link.

How do you read the shape of an HRR curve?

The curve shape reveals the burning mechanism, and often matters more than the peak itself. Schartel and Hull laid out a four-part taxonomy in 2007 that labs still follow.

Thermally thin materials — textiles, films, collapsed foams — give one sharp spike. The whole sample pyrolyses at once. As a result, available fuel sets the peak, not steady burning physics.

Thick non-charring polymers burn on a broad plateau instead. Thick cast PMMA shows this best. Once its pyrolysis zone balances against the imposed flux, burning settles into a quasi-steady state.

Thick charring solids — wood and many composites — produce a two-peak signature. First, surface pyrolysis drives an early peak. Then an insulating char forms and the curve dips. Finally the thermal wave reaches the back face. A second peak then rises, often above the first.

Intumescent systems flip the pattern. A small early peak appears, then the swelling char blocks both heat and volatiles. Suppression can run dramatic. Biomass-based intumescent coatings cut pHRR by 83–90% and THR by 49–87% against a bare steel substrate.

Typical peaks at 50 kW/m² span nearly two orders of magnitude. Neat polypropylene rages at roughly 1,275 kW/m². Polyethylene lands between 900 and 1,400. Cast PMMA holds its quasi-steady 650–900 kW/m². Rigid polyurethane foam runs 370–740 kW/m². Flexible PUR foam runs 200–450, and solid pine peaks modestly at 150–250. Glass-fibre/phenolic aircraft laminates stay under 200 kW/m². Yet a carbon-fibre/iPP composite once hit 1,738 kW/m².

Quoting a cone calorimeter peak alone rewards transient spikes, though. A 1.6 mm HIPS coupon can touch 1,200 kW/m² and burn out inside 90 seconds. Yet a 25 mm charring board may never clear 150 kW/m². It can still release five times the total energy. In short, the peak tells half the story.

What does total heat release add?

THR integrates the HRR curve over time, in MJ/m². It answers the question pHRR cannot: how much fuel does this material hold?

$\text{THR}(t) = \int_{0}^{t} \text{HRR}(\tau)\, d\tau$

Physically, THR gives the area under the curve — the total fire load per unit exposed area. That makes it directly useful downstream. Eurocode EN 1991-1-2 builds the fire-load densities of its parametric fire curves from data of exactly this kind.

THR also unlocks a deeper property, the effective heat of combustion or EHC. Divide total heat output by the mass the specimen actually lost.

$\text{EHC} = \frac{\text{THR} \cdot A}{\Delta m} \quad [\text{MJ/kg}]$

Here $A$ gives the exposed area. Then $\Delta m$ gives the mass loss. EHC always falls below the gross heat of combustion from a bomb calorimeter (ISO 1716). The ratio between them defines combustion efficiency $\chi$, typically 0.7–0.95 for well-ventilated flaming.

That gap makes flame-retardant (FR) science tractable. A drop in EHC after adding an FR signals gas-phase action — halogens or phosphorus disrupting radical chemistry inside the flame. But a THR cut with no EHC change points to condensed-phase action. Think char formation, or heat-sink fillers such as aluminium hydroxide (ATH) and magnesium hydroxide (MDH). Schartel and Hull lean on this split, and so does FAA report AR-05/14.

THR also covers pHRR where pHRR misleads. A low-peak, high-THR material smoulders along for a long time and still delivers a big fuel load. But a high-peak, low-THR material flashes and dies. So most codes clamp THR inside fixed windows. EN 13823 SBI uses THR at 600 seconds. Yet IMO FTP Code Part 5 caps Q_t at 0.7 MJ for bulkhead linings and 2.0 MJ for deck coverings.

What is MARHE, and why did EN 45545-2 pick it?

MARHE, the maximum average rate of heat emission, tracks the highest running average of HRR across a test. It matches full-scale fire outcomes better than any other cone calorimeter number.

Two metrics still leave a gap. pHRR spikes too readily. And a plain average HRR drowns in long low-emission tails. Neither one forecasts full-scale behaviour well. So fire scientists wanted one scalar that fused peak, timing, and duration.

The math takes two steps. First, average the HRR curve from test start:

$\text{ARHE}(t) = \frac{1}{t} \int_{0}^{t} \text{HRR}(\tau)\, d\tau$

Then take the maximum of that running average:

$\text{MARHE} = \max_{t}\, \text{ARHE}(t) \quad [\text{kW/m}^2]$

Because ARHE accumulates heat before it maximises, one number ends up sensitive to four things at once: time to ignition, peak intensity, total heat released, and when the peak lands. An early sharp peak lifts MARHE far more than a late one.

That reach explains the fit with full scale. Luo and colleagues wrote in 2022 that MARHE “objectively and comprehensively reflects the heat release performance of the samples in the whole combustion process.” A Nordtest screening study ranked MARHE the second-best cone predictor of SBI and room-corner performance, behind only pHRR.

So MARHE anchors EN 45545-2, the European railway fire-safety standard. Measured per ISO 5660-1 at 50 kW/m², it serves as the principal pass/fail heat-release parameter across most of the 26 requirement sets (R1–R26). Each set then grades by hazard level: HL1 for low risk, HL2 for standard passenger service, HL3 for long tunnels and underground running.

The thresholds run tight. Below sits a representative slice.

Requirement set	Application	HL2	HL3
R1 (interior vertical/horizontal surfaces)	Side walls, partitions, interior doors	MARHE < 90 kW/m²	MARHE < 60 kW/m²
R6 (seat shells F1C/F1D)	Passenger seat shell base and back	MARHE < 90 kW/m²	MARHE < 60 kW/m²
R7 (exterior surfaces, gangway interiors)	Body-shell walls, gangways	MARHE < 90 kW/m²	MARHE < 60 kW/m²

The US FRA report DOT/FRA/ORD-19/39 reproduces the full tables. For complete seats, EN 16989 reports MARHE in kilowatts alongside total smoke production. Maritime practice picks it up too: Part 10 of the IMO 2010 FTP Code applies MARHE to fire-restricting materials on high-speed craft. Part 5 of that same code, however, keeps a separate radiant-panel method with Q_p and Q_t. Different scenarios, different metrics.

Certified rail composites now hit these numbers routinely. Intumescent veils on aramid honeycomb, for instance, clear the 90 kW/m² HL2 bar with margin.

Which other cone calorimeter numbers matter?

Five more numbers ride along with HRR, THR, and MARHE. Together they turn a single cone calorimeter run into a material fingerprint.

Time to ignition (TTI), in seconds, scales with thermal inertia $k\rho c$ and the square of the ignition-temperature difference. Below a critical flux of roughly 10–20 kW/m², most polymers and cellulosics never ignite at all.

Mass loss rate, in g/m²·s, pairs with EHC to diagnose FR mechanism — the same gas-phase versus condensed-phase split from earlier.

Smoke production rate (SPR), its integral TSP, and specific extinction area (SEA) together describe visibility loss. SEA hovers near 100 m²/kg for clean-burning PMMA. Yet it climbs past 1,000 m²/kg for aromatic polymers like PVC and polystyrene. Those same materials dump the most soot into the upper layer.

CO and CO₂ yields, in kg per kg of mass lost, quantify how completely the fuel burns. Under vitiated air — the subject of ISO/TS 5660-5:2020, the controlled-atmosphere cone — CO yield from PMMA more than doubles as oxygen falls from 21% to 14%. An underventilated compartment therefore does not merely burn slower. It also turns far more toxic.

Finally, the Fire Growth Rate Index (FIGRA) deserves attention. EN 13823 defines it formally as the maximum of HRR(t) divided by t. Labs routinely compute a cone version as pHRR over time-to-peak, or over TTI. Morgan and Bundy (2007) measured cone FIGRA across UL-94-rated plastics in a NIST study. Their values ran from roughly 0.5 kW/(m²·s) for a brominated V-0 polycarbonate up to 16 kW/(m²·s) for non-FR polypropylene. That thirty-fold spread opens up even between materials of similar pHRR.

Typical cone calorimeter values at 50 kW/m²

For a sense of scale, here follow typical numbers. Every row comes from a horizontal cone calorimeter test at 50 kW/m².

Plastics and foams:

Material	Thickness	TTI (s)	pHRR (kW/m²)	THR (MJ/m²)	EHC (MJ/kg)
PMMA (cast, black)	25 mm	50–90	650–900 (quasi-steady)	800–1,200	~24
HDPE	5 mm	60–80	900–1,400	130–160	~40
Polypropylene (neat)	3.2 mm	25–50	1,275–2,200	120–150	~40–46
HIPS (non-FR)	3.2 mm	~50	~1,265	116	34
PC + brominated FR	3.2 mm	78	343	70	21
PVC (non-FR)	3.2 mm	23	243	51	14
Rigid PUR foam	25 mm	2–5	370–740	30–80	15–20
Flexible PUR foam	20 mm	2–10	200–450	20–40	20–25

Wood products and composites:

Material	Thickness	TTI (s)	pHRR (kW/m²)	THR (MJ/m²)	EHC (MJ/kg)
Pine / spruce (solid)	10–18 mm	20–40	150–250 (two peaks)	40–100	12–15
OSB / plywood (untreated)	11–18 mm	30–60	170–250 initial peak	50–90	12–15
FR-treated wood composite	11–18 mm	>50	300-s avg 58–100	30–40% lower	lower
GFRP / phenolic (aircraft)	2–4 mm	50–100	100–200	10–25	8–15
Carbon-fibre / iPP	2–4 mm	30–60	up to 1,738	60–90	25–35

The FR-treated wood row deserves a note. USDA Forest Products Laboratory work cut the 300-second average HRR of wood composites from 118–218 kW/m² down to 58–100 kW/m² through treatment alone.

Treat every figure above as a typical range, never a material constant. Grade, thickness, pigment, and backing all shift them. Within one lab, standard uncertainty on pHRR sits near 5–10%. Round-robins between labs widen that to 15–25% for TTI and THR. So a sound report quotes the mean of three replicates. It also states heat flux, thickness, frame status, and backing.

What corrupts a cone calorimeter result?

Edge effects and sample prep cause most of the damage in a cone calorimeter run. Both leave fingerprints that a careless reader will miss.

The retainer frame covers only a few millimetres of specimen edge. For composites and sandwich panels, pyrolysis gas escaping sideways adds spurious early HRR. Tsantaridis and Sandinge and colleagues both measured swings of 10–15% between framed and unframed runs. So every report must state its frame set-up.

Moisture content governs TTI. That makes the 48-hour conditioning at 23 °C and 50% RH non-negotiable. Backing choice matters nearly as much. Swapping standard ceramic fibre for an aluminium plate visibly reshapes the second peak of a thick charring sample.

Intumescent coatings raise a subtler problem. As they swell toward or past the 25 mm gap, the sample surface sees a higher flux. Some labs answer with a 50 mm gap. Others add a wire grid above the sample. Thermoplastics that melt and drip need the 5 mm aluminium foil edge lip. That lip keeps molten polymer inside the holder.

Thickness deserves the loudest warning. In the Schartel and Hull data, 1.5 mm PMMA gives a fuel-limited spike. At 6 mm the same polymer yields a broader single peak. At 25 mm it settles into a near-steady plateau. One material, three wildly different curves. So thickness belongs beside every number you quote.

How far does bench-scale data carry?

Not as far as a building. A cone calorimeter runs a bench-scale, well-ventilated, one-dimensional forced-flaming test. Real compartments break all four assumptions.

The bench cannot reproduce wall re-radiation, enclosure feedback, oxygen vitiation, or flame spread in three dimensions. Flame heat-flux feedback onto a cone sample reaches only about 19–37 kW/m², depending on polymer. Flashover rooms, by contrast, routinely see 40–100 kW/m². That gap shapes how engineers use the data.

Mid-scale tests bridge it. The EN 13823 SBI test burns a 1.5 m corner specimen against a 30 kW propane sandbox. It produces the FIGRA, THR₆₀₀ₛ, and SMOGRA values that drive Euroclass A2/B/C/D ratings for building products. Moreover, Efectis sells software that predicts SBI behaviour from a single cone HRR history. Twelve 0.01 m² cone coupons can stand in for three 2.25 m² SBI panels. That saves a lot of material and money.

At full scale the ISO 9705 room-corner test remains the reference. Yet bench and room tests still disagree more often than anyone would like. Making small-scale data carry that far, in fact, gave MARHE its reason to exist.

Key takeaways

A competent cone calorimeter report never collapses a material into one number. Each parameter answers its own question. First, pHRR shows how fiercely the material peaks. Second, THR shows how much total fuel it holds. Third, MARHE shows the worst sustained burn across the run. Next, TTI shows how quickly the material gets there. Then EHC reveals whether a flame retardant works in the gas phase or the condensed phase. Finally, SEA and the CO yields describe smoke and toxic hazard.

Codes pick their metric to match the scenario. EN 45545-2 anchors on MARHE, because rail fires burn for minutes inside confined tunnels. IMO FTP Part 5 limits Q_t and Q_p, because ship linings must never push a compartment toward flashover. Euroclass rests on SBI-derived FIGRA and THR₆₀₀ₛ, because building products matter most during the first ten minutes of growth.

Grasping why each metric exists — and reading curve shapes rather than peaks — separates a fire safety engineer from a data consumer. More than forty years on from that first bench at NBS, the cone calorimeter still earns its floor space for one reason. It yields every one of these numbers from a single 100 mm square, at bench-scale cost. And its core principle holds to 5% no matter what the material contains.

Cite this article

Dinh, D. C. (2026, April 18). Cone Calorimeter 101: HRR, THR, and MARHE Explained (Updated July 25, 2026). PyroRisk. https://pyrorisk.net/blog/cone-calorimeter-101/

Get citation →

---