Every incident begins with an origin. Every failure has a root. Asritas helps investigators reconstruct what happened, uncover why it happened, and build conclusions that can withstand technical, operational, and regulatory scrutiny. The reasoning is strengthened. The judgment remains human.
From the DCS historian, field logs and the emergency sequence, Asritas reconstructs the sixteen seconds exactly as they unfolded — the first step in tracing the failure back to its origin.
Board operator confirms feed rate stable at 142 m³/h. Reactor E-401 outlet temperature reading 318 °C, within the 305–340 °C operating band. No open alarms.
T-99 (E-401 outlet skin TC) holds at 318 °C while redundant thermowell TW-99B climbs. Deviation not flagged — reading remained inside configured range clamp.
PT-402 records 41.2 barg climbing at ~0.4 bar/s. Design pressure of V-402 is 46 barg. Temperature controller continues firing charge heater H-401.
Ductile overpressure failure initiates at the heat-affected zone of the longitudinal weld. Relief valve PSV-402 does not lift. Loss of primary containment.
Board operator opens BDV-410 to flare. Unit isolated. No personnel in the immediate area. Incident classified LOPC-2 (Loss of Primary Containment).
Each exhibit is examined against the design basis, then reasoned against every other — surfacing the causal relationships that rule some failure modes out and point toward the root.
Fracture surface shows ductile dimpling with beach marks consistent with thermal-mechanical fatigue initiating at the weld HAZ. No pre-existing inclusions or manufacturing voids detected.
Manual isolation valve on the PSV-402 inlet found ~40% open. Car-seal broken and absent. Diverts a portion of relief capacity away from the flare header.
March turnaround work order for a flush of V-402A has no closing car-seal check recorded. Management-of-Change MOC-2231 left open. Line-up not verified at restart.
Testimony is read against the instrument record, so investigators can see where a person's account confirms the data — and where someone noticed what the instruments could not.
“Around 03:13 I noticed the T-99 trend went flat while the downstream temperatures kept climbing. It didn't add up, but the DCS never raised an alarm so I had nothing to action on.
“There was a rising high-pitch whine near the E-401 skid maybe a minute before it let go. I called it in on the radio. By the time I reached the stairwell I heard the release.
“V-402A was opened during the March turnaround for a flush. I can't find a record of it being car-sealed shut afterward. That step lives on the turnaround punch list, not the daily rounds.
Asritas replays the raw historian for the team. A single miscalibrated thermocouple held its reading inside the alarm band — so the controller kept firing while the vessel climbed toward failure.
Candidate failure modes are not ranked by a score. Each is held against every exhibit in turn, and the explanations that cannot survive contradiction fall away — until one remains that accounts for the whole record.
Reasoning is qualitative and traceable. Every conclusion links back to the evidence and testimony that supports it — nothing is asserted without a source.
Fractography indicates fatigue initiation at the weld HAZ, not a pre-existing inclusion or fabrication flaw. Vessel passed last internal inspection with no relevant findings.
Manual actions during the event were correct given the information displayed. The board operator was reacting to instrument readings that were themselves incorrect. Not a root cause.
A −43 °C offset on skin TC T-99 caused TIC-401 to over-fire H-401, driving V-402 above design temperature and pressure. Relief valve protection was diverted by the partially-open V-402A left after the March turnaround.
A −43 °C calibration drift on thermocouple T-99 caused the controller to over-fire heater H-401, driving V-402 beyond design temperature and pressure. The relief path was bypassed by valve V-402A, left 40% open after the March turnaround — preventing PSV-402 from venting the overpressure.
Calibration procedure gap — no field verification of T-99 vs TW-99B
Car-seal management — V-402A not restored on turnaround punch list
MOC-2231 closed without relief-path line-up confirmation