Fire Origin Engineering and Heat Release

A burning chair releases energy at a rate that can range from a few kilowatts in the smouldering early phase to several megawatts at peak involvement. That number drives virtually everything else. Flame height correlates directly with HRR via the Heskestad correlation, so a calculated flame height can be compared against scorch or ignition marks on overhead elements. The time to flashover in a compartment can be estimated from the HRR growth rate and the room geometry using Babrauskas and Peacock's equations.

In casework, HRR enters the analysis in two ways. First, an investigator may want to know whether the proposed ignition source (a candle, a smouldering cigarette, a small electrical arc) could have had a high enough HRR to ignite an adjacent item within the observed time window. Second, they may compare the peak HRR estimated from the damage pattern against published calorimeter data for the fuel load involved to check consistency.

Compartment fire behaviour is usually described in four phases. In the growth phase, fire spreads from the ignition source through its immediate fuel load and generates a rising hot-gas plume. The plume fills the upper volume of the room with a buoyant layer of hot gas and smoke. As more fuel burns, the layer deepens and its temperature climbs.

1. Ignition and incipient growth
   The fire is small and localised. HRR is low, typically under 50 kW, and conditions are survivable. Origin indicators are most reliable in this phase because the fire has not yet redistributed char.
2. Free-burning growth
   The fire is spreading through its primary fuel load. The hot layer descends, radiative feedback from the ceiling to floor items increases, and HRR rises, often following a t-squared growth model. At around 600 degrees Celsius in the upper layer, radiation to floor level reaches roughly 20 kW/m2, enough to ignite loose paper.
3. Flashover
   All exposed combustibles in the compartment ignite near-simultaneously. HRR spikes by one to two orders of magnitude. Post-flashover conditions destroy many pre-flashover origin indicators and create the majority of the structural damage investigators examine.
4. Fully developed (ventilation-limited)
   The fire is now constrained by oxygen supply through openings. Burning rate is controlled by the ventilation factor (opening area times the square root of opening height). Incomplete combustion products, especially CO, peak in this phase.
5. Decay
   Fuel is consumed. HRR falls. The compartment begins to cool. Post-fire, the pattern of deep charring or calcination in the area of lowest residual fuel load is sometimes misread as an accelerant pattern; it is often simply the area that burned longest.

A fire inside a building will shift between fuel-limited and ventilation-limited regimes as it grows, as windows break, and as doors are opened or closed. Understanding which regime was active at any given moment matters because the two produce different damage signatures.

The ventilation factor, written as Av times the square root of Hv (where Av is the area of openings and Hv is their height), directly sets the air mass flow rate into a ventilation-limited compartment. Investigators can calculate it from measured opening dimensions at the time of burning and compare it against the burning rate implied by the damage to test whether a ventilation-limited scenario is physically consistent.

Fire Dynamics Simulator (FDS) is a large-eddy-simulation CFD code developed by NIST, with substantial contributions to key sub-models made by VTT Technical Research Centre of Finland. Given a geometry defined in SmokeView or PyroSim, an HRR input curve, and surface thermal properties, it produces time-step-by-time-step predictions of temperature, velocity, species concentrations, and visible-light extinction throughout the volume.

Zone models such as CFAST (also from NIST) divide a compartment into a two-zone approximation: a hot upper layer and a cool lower layer, each treated as well-mixed. Zone models run in seconds, which makes them useful for rapid sensitivity testing, but they cannot resolve spatial detail within a zone. An FDS run on a finely resolved grid may take hours or days on standard hardware but provides the spatial fidelity needed to predict where a specific overhead element ignited.

• Input requirements: accurate room geometry, HRR curve for the ignition scenario, material thermal properties (conductivity, density, specific heat, ignition temperature), and vent conditions. Garbage-in, garbage-out applies with full force here.
• Validation obligation: NFPA 921 requires that any model used in fire investigation be validated for the specific application, that its uncertainty be disclosed, and that inputs be defensible. The NIST FDS Validation Guide documents the model's accuracy against full-scale experiments.
• Investigative use: models are used to test hypotheses (could this ignition source have grown to flashover in the time reported?), not to determine origin by themselves. Physical evidence drives the hypothesis; the model tests its physical plausibility.
• Courtroom hazard: colourful FDS animations carry high persuasive weight with juries. Expert witnesses must clearly distinguish between what the model predicts and what actually happened, and must explain the assumptions behind every input.

NFPA 921 (Guide for Fire and Explosion Investigations) is the primary reference standard in the United States and is widely referenced in courts globally. Its chapter on engineering analysis (Chapter 21, Failure Analysis and Analytical Tools, in the 2021 edition) sets out what engineering tools investigators may use, how they should be applied, and how findings should be integrated into the overall investigation.

The chapter distinguishes two categories of engineering use. Calculation methods, meaning hand calculations using fire-dynamics equations for flame height, plume temperature, flashover threshold, and HRR from ventilation, are well-established and relatively straightforward to document. Computer fire models are more powerful but carry greater disclosure and validation obligations.

A critical limitation NFPA 921 articulates is that engineering analysis cannot determine ignition cause. It can say whether a proposed scenario is physically consistent with observed outcomes. It cannot, on its own, say that an accelerant was or was not present, or that a specific person set a fire. Those conclusions require scene evidence. The engineering analysis supports the physical investigation; it does not replace it.

Several HRR and fire-dynamics calculations appear repeatedly in fire investigations and civil litigation. Understanding the equations, their inputs, and their limits is what distinguishes defensible engineering testimony from unsupported assertion.

• Flame height (Heskestad correlation): estimates visible flame length from HRR. Used to test whether a candle or small burner could have directly impinged on an overhead fuel at the reported height.
• Time to flashover (Thomas, Babrauskas): estimates the minimum HRR needed to flash a compartment and the time required given a t-squared growth rate. Used to check whether a reported ignition-to-rescue timeline is consistent with the observed damage.
• Ventilation-limited HRR: for a sealed or nearly sealed compartment, the maximum burning rate is set by oxygen mass-flow through vents. If inferred HRR from damage exceeds the ventilation limit, an alternative scenario is needed.
• Plume temperature (McCaffrey correlation): estimates gas temperature at a given height above a fire of known HRR. Useful for asking whether a smoke detector at a given ceiling height would have activated before reported discovery.
• Heat flux to a target (point-source model): estimates radiant heat flux from a fire to a nearby item to test whether piloted or unpiloted ignition was possible within the geometry of the scene.

Fire engineering analysis can move a finding from a subjective reading of char patterns to a quantified, testable claim. The corresponding risk is that false precision in modelling can make a plausible but untested story appear authoritative. Several notorious wrongful arson convictions, reviewed by NFPA 921 working groups and the Innocence Project, were supported by physical evidence misread in the light of discredited arson indicators and reinforced by inadequately scrutinised engineering-style testimony.

The safeguards are explicit in NFPA 921. Every engineering analysis must document its inputs and the source of those inputs. Every model output must be compared against physical evidence; a model that predicts flashover in two minutes when the damage pattern suggests ten minutes should prompt a re-examination of the inputs, not confidence in the model. Uncertainty must be stated. A conclusion framed as definitively proved when the inputs carry large ranges is misleading. The investigator's job is to reduce uncertainty to the minimum possible and then honestly report what remains.