Basement Gas Deflagration: Meter Data Versus Damage
A 10 m x 7 m x 2.4 m basement (168 m-cubed) has two 0.6 m x 0.6 m windows high on the walls. A natural gas explosion at 21:40 blew one window out and left the other intact. The gas water heater pilot, 0.45 m above the floor, is the suspected ignition source. The utility meter shows 0.6 cubic feet per hour standing pilot use and a 24-hour total of 62 cubic feet, of which 14 cubic feet is accounted for by the range and dryer; the meter tests as functioning normally. The homeowner smelled gas 'for a while', then opened the door at the top of the basement stairs two minutes before the event to air the space out. Damage includes first-floor joists lifted about 40 mm, interior partitions leaning outward, and the steel basement door found 6 m out in the yard.
1. Methane LFL is 5.0 percent and stoichiometric is about 9.5 percent by volume. Compute what the unaccounted gas could produce in this basement and evaluate whether the meter data is consistent with a flammable cloud.
Read the worked answer
Thresholds: at 5.0 percent, the basement would need 0.05 x 168 = 8.4 m-cubed (about 297 cubic feet) to be uniformly flammable; at 9.5 percent, about 16.0 m-cubed (about 564 cubic feet).
Available gas: unaccounted is 62 minus 14 = 48 cubic feet, or about 1.36 m-cubed, which mixed uniformly would give only 0.81 percent across the basement, far below the LFL. The entire 62 cubic feet gives only about 1.05 percent.
The key insight: uniform mixing is the wrong model. Methane has a specific gravity near 0.55 and stratifies at the ceiling. That 1.36 m-cubed can form a cloud of 1.36 divided by 0.05, or about 27 m-cubed, at exactly the LFL. Spread over the 70 m-squared floor, that is a ceiling layer roughly 0.4 m thick. So the recorded release is at the low end but not physically impossible, provided the gas stratified thinly and was subsequently mixed downward.
Consequences for the investigation: do not declare the meter data exculpatory. Instead, test duration and instrumentation: diaphragm meters under-register flows near their low-flow threshold, so a slow leak running for days or weeks may never appear in a 24-hour total. Obtain multi-week consumption history, pressure tests, and the leak survey records, and look for an unmetered source such as a second service or a propane cylinder.
2. Why does the pilot at 0.45 m matter, and what role did the door being opened two minutes earlier play in producing an ignition?
Read the worked answer
Stratification problem: a 0.4 m ceiling layer sits entirely above a pilot at 0.45 m, so a stably stratified cloud would never reach the ignition source. The event therefore requires a mechanism that moved gas downward or mixed the layer into the breathing zone.
The opened door supplies that mechanism: opening the stair door creates a strong inflow of cool air at the bottom of the stairwell and a displacement and mixing front that pushes the buoyant layer down and dilutes it toward the flammable range. It also supplies oxygen. The two-minute interval is exactly the kind of mixing transient that turns a static, non-ignitable layer into a flammable one, and it explains why the ignition occurred when it did rather than hours earlier.
Additional tests this points to: check the second window and the stair door for inward-then-outward damage, look for a smoke or dust 'clean-out' band at the layer height, and interview the homeowner about the interval between opening the door and the event. Reconstruct the sequence rather than asserting a single cause.
Characterization: this is a deflagration of a premixed fuel-air cloud, not a backdraft. A backdraft requires an oxygen-starved, fuel-rich compartment fire; here the fuel was unburned gas and the compartment had no significant fire load. Mislabeling it changes the ventilation and ignition analysis entirely.
3. Can the steel door found 6 m away be used to quantify overpressure? Reason it through and state what the damage collectively supports.
Read the worked answer
Order-of-magnitude check: a 0.9 m x 2.0 m steel door of roughly 40 kg thrown 6 m implies a launch velocity near the square root of (g x R), about 7.7 m/s at a favorable angle, and kinetic energy near 1.2 kJ. If the deflagration accelerated it over about 0.05-0.1 s, the average force is on the order of 3-6 kN over 1.8 m-squared, or roughly 2-3 kPa (a fraction of a psi).
Why that number is soft: the door may have been unlatched, may have been driven partly by the deflagration wind and gas jetting, and may have been carried by debris. Thrown distance is a weak, non-conservative gauge and should not be presented as a measured overpressure.
What the damage collectively supports: joists lifted 40 mm and partitions leaning outward indicate a genuine structural overpressure in the low psi range, which is consistent with a deflagration of a few kilograms of fuel-air mixture in a partially vented basement, and inconsistent with a small localized flash. Test it against the blast-damage gradient: expect the heaviest damage at the ignition end and directional bending of partitions away from the cloud center.
Report language: describe the door displacement as corroborative, quantify with explicit assumptions and uncertainty, and never present a single damage feature as a calibrated pressure measurement.