After a fuel gas explosion the building is gone but the pressure history is not. Damage distribution, debris trajectories and glass breakage together behave as a distributed, if crude, pressure gauge. Used carefully, they constrain what kind of event occurred. Used loosely, they support almost any narrative someone wants to advance.
The physical distinction
A deflagration propagates by heat and mass transfer, with a flame front moving subsonically relative to the unburned gas ahead of it. A detonation propagates by a shock wave, supersonically, with the reaction zone coupled to the shock. The two produce markedly different pressure histories: deflagrations generate comparatively slow pressure rise and can vent effectively through openings, while detonations produce near-instantaneous pressure jumps that venting cannot relieve.
Accidental fuel gas events in buildings are overwhelmingly deflagrations. Deflagration-to-detonation transition requires conditions — sustained confinement, repeated obstacles, sufficient run-up length — that ordinary residential geometry rarely supplies.
Why the distinction is worth establishing
The two mechanisms imply different antecedent conditions. Establishing which occurred narrows the plausible fuel volume, concentration distribution and degree of confinement, which in turn bears on how long gas had been accumulating and where it entered.
It also bears on what could reasonably have been designed against. Explosion venting sized for a deflagration is not protection against a detonation, and the standards that govern such provisions treat them as distinct problems.
Damage as a pressure record
Window glass fails at low overpressures and is a sensitive early indicator. Unreinforced masonry, stud walls and roof structures each have characteristic failure thresholds spanning roughly an order of magnitude, so the pattern of what failed and what survived brackets the peak pressure across the structure.
Direction matters as much as magnitude. Wall panels displaced outward, debris throw distances and the orientation of failed connections together indicate where the pressure was highest and how it moved. A systematic blast-effects survey records these as vectors and works back toward a source region.
What the survey cannot resolve alone
Overpressure mapping constrains the location of the gas cloud, which is not the same as the location of the leak or the location of the ignition source. A cloud that accumulated in a crawlspace and ignited at an appliance three rooms away produces damage centred on neither the leak nor the appliance.
Nor does the survey by itself establish concentration. The same peak pressure can be produced by a small volume near stoichiometric or a larger volume well off it, and distinguishing those requires the accumulation analysis rather than the damage analysis.
Modelling and its assumptions
Computational fluid dynamics reconstruction of gas dispersion and combustion is an established technique and can be genuinely informative. It is also assumption-heavy in ways that are easy to underweight. A CFD result depends on an assumed leak rate, an assumed duration, an assumed ventilation condition, an assumed ignition location and an assumed geometry — several of which are the very things in dispute.
The useful discipline is sensitivity analysis: demonstrating which conclusions hold across the plausible range of each assumption and which depend on a particular choice. A model presented as a single deterministic answer invites, and usually receives, exactly that challenge.
Confinement and congestion
Flame acceleration in a deflagration is driven principally by turbulence generated as the flame front passes obstacles. Congested geometry — joists, furniture, piping runs, partition walls — accelerates the front and raises peak pressure, sometimes substantially, for the same fuel and concentration.
This is why nominally similar events in nominally similar buildings produce very different damage, and why importing a peak pressure figure from another incident is unsound without matching the geometry.
Locating the ignition source
Overpressure analysis and ignition analysis are separate exercises that constrain each other. Any device or event capable of delivering sufficient energy within the flammable cloud is a candidate: pilot lights and burner ignitions, thermostats and relays, refrigeration compressors and their controls, light switches, static discharge, and electrical faults.
Elimination here is usually partial rather than complete, because the candidate that ignited the cloud is often consumed in the event. What the physical evidence more reliably supports is a constraint on ignition location, which the damage distribution and the accumulation analysis together bound. Where an appliance is asserted as the ignition source, its as-found control positions and gas valve condition are evidence and should be documented before the unit is moved.
Where these analyses are challenged
Challenges cluster predictably: that the damage survey sampled selectively, that failure thresholds were taken from tabulated values without accounting for the actual construction and its condition, that a detonation was asserted where the evidence supports only a fast deflagration, and that the CFD reconstruction assumed its conclusion in its inputs.
Work that documents the survey completely, states its thresholds and their sources, and shows the range of model inputs tested tends to withstand those challenges. Work that reports only a peak pressure and a conclusion generally does not.
The framing that holds up best treats the overpressure evidence as a constraint rather than an answer — it rules out scenarios that cannot have produced the observed damage, and the surviving set is narrowed further by the accumulation, migration and ignition analyses. Conclusions that depend on the damage survey alone are doing more work than that evidence can carry.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.