Clairton coke oven explosion: draw the temporary energy path

The short answer: the U.S. Chemical Safety Board (CSB) found that pressurized water entered the enclosed space between the nearly closed gates of a 50-psig, cast-iron isolation valve at U.S. Steel's Clairton Coke Works. The water overpressurized and fractured the valve, opening a path for toxic, flammable coke oven gas to escape and explode. A normal-operation P&ID could show the gas topology, but the job also needed a temporary-work drawing, equipment limits, a stepwise procedure, and a facility-siting view.

The CSB released its final report on August 10, 2026. It identifies the valve overpressure as the cause, inadequate procedure and hazard analysis as contributors, and occupied-building siting as a contributor to the severity. This article explains the drawing lesson from those findings; it does not independently investigate the event or replace the report.

Evidence status: final CSB report, last checked August 25, 2026. The CSB calls the internal valve sequence “likely” because there was no basement surveillance video, pump process data, valve pressure measurement, or remote gate-position indication. The explanatory diagram below preserves that boundary and does not assign a more precise pressure, gate position, ignition source, or worker action than the report supports.

What happened at Battery 13

The processed coke oven gas system supplied Batteries 13 and 14 through a common line that split at a tee. Each branch had a manually operated isolation valve. The Battery 13 valve was an 18-inch double-disc gate valve made in 1953, refurbished in 2013, and rated for 50 psig. The gas system normally operated at about 1 psig.

Workers sometimes found that coke oven residue prevented an isolation valve from closing fully. U.S. Steel had a valve-exercising procedure that mentioned steam at no more than 10 psig, but it did not describe washing the valve seats with water. The CSB found that high-pressure water washing had nevertheless been used on an ad hoc basis for at least three years.

On August 11, 2025, workers connected a contractor's positive-displacement pump truck to a cleanout port located between the valve's two gates. They applied water while lowering the gates. The exact supplied pressure is unknown: the pump had no recorded process data, and the valve had no pressure sensor. The CSB established only that the pump could produce pressure significantly above the valve's 50-psig rating, especially against a blocked discharge.

The CSB's likely sequence is:

1. The double gates started open, so water could leave through the piping.
2. Workers lowered the gates and applied water through the cleanout port.
3. With the gates fully or nearly closed, water filled the enclosed cavity between them.
4. Continued pumping raised the pressure in the valve body and cap.
5. The cast-iron valve fractured, and the damaged gates no longer sealed the gas line.
6. Coke oven gas escaped into the basement, rose toward the bench level, ignited, and exploded 24 seconds after a radio evacuation order.

The explosion killed two people, injured 11 others—five seriously—and caused an estimated $52.5 million in property damage. Three routinely occupied rooms were less than 20 feet above the source piping and were not designed to withstand an explosion.

The drawing lesson: temporary energy changes the system boundary

A P&ID is good at controlled connectivity: the gas source, branch, isolation valve, cleanout connection, downstream battery, normal pressure, and references to line and valve specifications. It is not automatically a safe-work procedure, valve cutaway, pump curve, energy-control plan, or facility-siting study.

The job introduced an external energy source that did not exist in normal operation. Once the pump truck was connected, the review boundary had to expand. The relevant system was no longer just “coke oven gas piping at about 1 psig.” It was gas piping plus a positive-displacement water pump, a cleanout port between two moving gates, a possible blocked cavity, the valve's pressure capability, the order of operations, people in the affected area, and buildings above the release point.

That is the durable rule: when temporary equipment can add pressure, flow, heat, electricity, motion, or chemical inventory, redraw the job state before approving the work. Do not assume the normal P&ID still represents the highest-energy case.

Use three controlled views, not one overloaded sheet

For work like this, the review packet should answer three different questions.

These views may be separate controlled documents. The important point is the cross-check: every temporary connection on the work sketch must land on a known system point; every step must stay inside verified equipment limits; every credible release or overpressure case must have an identified control; and the affected-area decision must match the facility layout.

Worked review: connect an external pump to a valve cleanout

Start with a proposed task: “Use a pump truck to wash residue from the isolation-valve seats.” Before choosing a pump setting or writing steps, walk the energy path.

1. Locate the injection point. Confirm where the cleanout port enters the valve and whether it is upstream, downstream, or between closure elements. A symbol on a P&ID may need a vendor cutaway or equipment drawing to answer this.
2. Enumerate every gate state. Draw fully open, partially closed, and fully closed. For each state, trace where injected liquid can exit. Treat “no visible outlet” as a blocked-flow case, not a drafting omission.
3. Compare source capability with every component. Use maximum credible delivered pressure—not normal process pressure or an engine-speed nickname. Include hoses, fittings, ports, valve body, cap, and any trapped section.
4. Define the engineered pressure-control path. Show the approved pressure limitation, relief or return route, measurement point, and safe destination. Do not invent these features from the diagram; a qualified engineer and the controlling procedure must establish them.
5. Control hazardous energy. OSHA's lockout/tagout rule requires documented energy-control procedures where unexpected energization or release of stored energy can injure people, including specific isolation steps and verification. Contractor and site procedures also have to be coordinated.
6. Map affected people and buildings. A line diagram cannot establish blast resistance or safe siting. Use the facility plan and the hazard analysis to decide who must be notified, relocated, or excluded.
7. Make the stop conditions visible. Unexpected valve resistance, leakage, gas alarms, missing pressure indication, or a lost discharge path should route to a defined safe state—not an improvised next step.

The CSB's recommendations follow the same structure: written valve-washing procedures with overpressure controls, a hazard analysis or safety validation, worker involvement and training, facility-siting evaluation and mitigation, and a broader process-safety management system.

Common failure modes in maintenance drawings

• Drawing only the permanent plant: the temporary pump, hose, jumper, blind, bypass, or portable power source never appears in the review.
• Using normal pressure as the limit: a 1-psig process is treated as low risk even though connected equipment can supply far more pressure.
• Showing a valve as one symbol: internal cavities, cleanout ports, gate position, and trapped-volume behavior remain invisible.
• Writing an activity instead of a state sequence: “wash valve” does not say which paths are open, blocked, isolated, relieved, measured, or verified at each step.
• Treating the P&ID as a facility layout: schematic proximity is mistaken for physical separation, occupied-building exposure, or evacuation direction.
• Adding plausible safeguards after the fact: a relief device, gauge, or interlock is drawn because it would make the diagram look safe, without evidence that it existed.
• Erasing uncertainty: unknown pump pressure or exact gate position is replaced with a confident number or animation. In an incident reconstruction, “not established” is a required label.

What this reconstruction cannot prove

The figure does not reproduce the plant's controlled P&ID, valve geometry, building layout, or CSB causal diagram. It is an original explanatory plate based on the final report. The CSB could not fully determine the precise worker-action sequence or delivered water pressure, and it located ignition at or near the Battery 14 oven closest to the transfer area without identifying a more specific ignition source. Those unknowns remain unknown here.

The report also discusses standards applicability, process-safety coverage, cast iron in hazardous service, and post-incident actions in much greater detail. Read the full report before using this event in a hazard review or training program. A diagram can structure the evidence; it cannot turn a public report into project-specific engineering approval.

A concrete next step

Take one upcoming temporary connection—pump truck, nitrogen purge, steam hose, electrical jumper, or bypass—and draw both the normal state and every work state. List the temporary source's maximum capability, component limits, discharge or relief path, isolation and verification points, stop conditions, and affected area. If any arrow ends at “unknown,” stop the approval and resolve that input.

You can use ChatDiagram's P&ID maker to draft the permanent and temporary topologies, then attach the controlled procedure, equipment data, and facility plan for qualified review. Keep inferred or missing facts visibly marked; never let the renderer complete a hazardous-energy path by guesswork.

Cite this article

Whitfield, Ray. “Clairton coke oven explosion: draw the temporary energy path.” ChatDiagram, August 25, 2026. https://www.chatdiagram.com/blog/clairton-coke-oven-explosion-diagram.