Lightning, holdover, and telling a natural ignition from a human one
Lightning ignitions can smoulder for days before they are detected, which puts the strike and the fire on different dates and makes the distinction from human causes less obvious than it sounds.
Utility ignition: what the hardware and the fault data show
Where a wildfire is attributed to electrical infrastructure, the conductors, the hardware and the operator's own recorded data usually contain the answer.
Finding a square metre in a hundred thousand acres
A fire that burns a landscape still started at one point. Origin area determination in wildland fire runs backward along the fire's own directional record.
The two ordinary vehicle fires: fluid on a hot surface, and a chafed harness
Most non-collision vehicle fires trace to one of two mechanisms. Distinguishing them is usually possible, and it determines whether the answer is maintenance, repair work or a manufacturing defect.
Stranded energy: handling evidence in electric vehicle fires
High-voltage battery fires reignite hours or days after suppression. That behaviour is a safety problem, a storage problem and an evidence-preservation problem at the same time.
Did the fire start before the collision or after it?
Sequencing a vehicle fire against a collision changes who is responsible for what. The physical evidence and the recorded data usually constrain the order more tightly than either does alone.
Canine alerts and laboratory confirmation in ignitable liquid analysis
A canine alert is a screening tool that directs sampling. It is not a laboratory result, and the difference between the two is a recurring point of failure in fire investigations.
Negative corpus, and what NFPA 921 replaced it with
Concluding a cause by eliminating every alternative is not the same as proving it. The distinction between legitimate elimination and reasoning from ignorance is where many cause opinions are lost.
Origin determination after flashover, when the patterns stop cooperating
Flashover degrades the fire patterns that origin determination depends on. What survives, what does not, and how a defensible origin opinion is built when the compartment burned to completion.
Tracing gas from the leak to the building when the piping is gone
Gas rarely travels from a failed main to a structure in a straight line. Reconstructing the path is often possible even after emergency repairs have removed the evidence.
Deflagration or detonation: what overpressure evidence can and cannot settle
Structural damage after a gas explosion functions as a distributed pressure gauge. Reading it well constrains the event; reading it loosely produces confident conclusions the evidence does not support.
Odorant fade and the explosions nobody smelled coming
The most common statement after a gas explosion is that nobody smelled gas. That is often true and rarely exculpatory — several mechanisms strip odorant before it reaches a nose.
Arcing, overheating, and the limits of what fire patterns can tell you
Fire patterns show where a fire burned longest — not necessarily where it started. Why electrical origin determinations need physical corroboration beyond pattern analysis.
What a defensible combustible dust hazard analysis has to establish
A CDHA is written to prevent an incident and then read backwards by people looking for what it failed to say. What the standard requires, and where these disputes actually land.
When a receptacle is the origin and when it is a victim
Receptacles and terminations are both a common alleged origin and a common casualty. The physical differences between a connection that failed and one that was destroyed.
Why the second explosion does the damage
Primary dust deflagrations are usually survivable. The secondary event, fed by accumulation on surfaces no process drawing shows, is what levels facilities — and it is largely a housekeeping question.
Documenting an electrical fire scene before the excavation begins
Most electrical fire evidence is destroyed during debris removal rather than by the fire itself. What has to be recorded, and in what order, for the analysis to remain possible.
Testing dust as handled, not as supplied
Kst, MEC and MIE govern how a facility is protected. All of them shift with particle size and moisture — which is why the sample the values came from is often the weakest link.