From ILI Data to Repair Decisions : What PHMSA’s Proposed Repair Criteria Rule Asks of Your Program

By: Joe Pikas

When I got into corrosion in 1969, evaluating a corroded joint of pipe meant a book about six inches thick. Pages of diameters and wall thicknesses and what was allowed. You checked the maximum length and a maximum depth, and it handed back an answer that was very, very conservative. Which meant we dug up and repaired a lot of pipes that did not need repairing. 

PHMSA’s proposed repair criteria rule would change the basis of that decision. Here is what it says, and then I will tell you why it lands harder on your records than on your math. 

What the proposed rule would actually do 

On July 8, 2026, PHMSA published a notice of proposed rulemaking titled Repair Criteria for Hazardous Liquid and Gas Transmission Pipelines, under Docket PHMSA-2025-0019. The comment period closed September 8 with 37 comments on the docket, and a public meeting of the advisory committees is expected. So keep in mind this is still a proposal, not a final rule. 

The core move is toward predicted failure pressure ratio, which is simply the predicted failure pressure divided by MAOP for gas, or MOP for liquids. Here is where the immediate response triggers would land. 

Condition	Proposed immediate-response trigger	What changes
Cracks	Failure pressure ratio of 1.1 or below, or depth above 70 percent of wall	Depth threshold for immediate crack response rises to 70 percent
Preferential seam corrosion	Failure pressure ratio of 1.1 or below	Down from a ratio of 1.25
Interacting dents	Mechanical damage with gouging	Adds gouging, and excludes non-mechanical metal loss at 10 percent depth or less
Metal loss (hazardous liquid)	Failure pressure ratio of 1.1 or below	The 80 percent depth criterion is retained alongside it

 

The response schedule for gas transmission also moves out of 192.933 and consolidates into 192.714. The timelines themselves do not change: one year for near-term conditions in high- consequence areas, two years elsewhere. 

And the rule opens up which methods you are allowed to use. For metal loss, API 579 and Psqr would join ASME B31G, RSTRENG, and other models shown to produce comparable results. I mentioned Psqr during our October webinar and said I would not get into it that day. Well, it is now named in a proposed federal rule, so it is worth your time. For cracks, fracture mechanics methods would be acceptable, including API 579 Level II or III, Modified Ln-Sec, and Raju-Newman. There is an engineering critical assessment option for dents as well. 

Two more pieces matter more than they look. Section 192.712, and a new 195.415, would require a subject matter expert to perform the anomaly evaluation, and that evaluation must account for uncertainties including tool tolerance. And an anomaly requiring a response must be discovered within 180 days after the ILI tool is retrieved. 

Why this is a departure from how most of us learned it 

That six-inch book was B31G, original and later modified. A rule of thumb dressed in tables, and it was the best we had. 

RSTRENG came along in the late 90s and took multiple increments along the river-bottom profile instead of one length and one depth. That made a real difference in accuracy. It did not get accepted immediately either. It took the industry a while. 

I was fortunate enough to sit on the PRCI task groups for both RSTRENG and AC mitigation. On the second revision, Dr. John Kiefner made the call to keep the method slightly on the conservative side, so we would not land too close to the failure line. He was thinking about the unknowns out there. Pre-1970 ERW pipe. SCC. Cracking in the welds. He was right to do it, and the method is still slightly conservative today because of that decision. 

So be clear about what the proposed rule is doing. It is not lowering the bar. It is recognizing that an engineering calculation gives you a better answer than a table does and then asks you to stand behind the calculation. Those are two different things, and the second one is where the work is. 

The dig has to agree with the file 

Here is what catches people, and it has nothing to do with math. 

Too often, the data in the ditch does not match the records in the office. 

If the pipe is supposed to be 0.375 wall thickness and you find 0.25 in the ditch, you are in trouble. I have seen that more than once. One of those was a gas line where the wrong wall thickness had been installed, and the pipe had to be replaced. 

I have also seen crews take readings on the wrong joint of pipe entirely. Somebody back in the office noticed the numbers should have been deeper; they went back out, and the measurements had come off a different joint. That is the cost of excavating a pipeline twice, recoating it twice, and losing the time between. 

Now put the proposed rule on top of that. If your anomaly evaluation must account for tool tolerance, and a subject matter expert must sign off on it, then your validation dig is not just a repair trip. It is the check on the ILI call, and the quality of the data you bring back out of the ditch is part of the engineering record. 

Which raises the question I get asked more than any other: when do you quit on validation digs? There is no single answer, but you cannot even have the conversation unless the dig data, and the ILI data are in a form where you can compare them. 

And then there are the records 

I worked as a field engineer at a division office. When our file cabinets filled up, you know where the information went? The attic of the field office. I am sure a lot of it is still up there. I am sure there are people working in those offices today who do not know it is there. 

And properly archived is not the same as retrievable. I have seen records that went off to professional storage, categorized and kept dry, still fail to come back when an incident required them. 

Meanwhile, the regulators are asking for more, not less. Under MAOP reconfirmation, your corrosion history has to be traceable, verifiable and complete, and RSTRENG calculations are explicitly part of what supports the engineering basis. 

That is also the argument for running assessments across the full anomaly population rather than only the features above an immediate response threshold. The vendor hands you a filtered list so you can plan your digs, and that is useful. But the features below the threshold are where your corrosion growth rates come from. Run them all, keep them, and you can compare interval to interval and set your next reassessment on measurement instead of assumption. Skip them and you are guessing. 

No more shoving it in the attic and hoping it turns up. Those days are over. 

Where the workflow usually breaks

The path from an ILI run to a repair decision has four handoffs: the vendor’s anomaly list comes in, the assessment runs, the dig gets planned and executed, and the result goes back into the record. Most programs do all four. Fewer can produce the whole chain for one segment on short notice. 

At Technical Toolboxes, RSTRENG+ is built for the assessment end of that. Batch Run calculates across the full anomaly population instead of only the features above the dig threshold, the Zero-Out Method helps you pinpoint the sections that actually need further repair, and results carry GIS location so the field team and the office work from the same reference. 

On the capture end, the principle we build toward across our Asset Intelligence tools is the same: collect the data where the work happens, offline if you have to, and keep one record per asset holding the inspections, findings, photos and scan data together with an audit-ready history. Whatever system you use, that is the gap described above, and it is a workflow problem before it is a software problem. 

None of this does the engineering for you, and none of it replaces judgment. The proposed rule is explicit that a subject matter expert performs the evaluation. What good tooling does is make sure the expert’s work survives the handoff and can be produced on demand. 

What to do between now and a final rule 

Four things, none of which require you to wait. 

1. Read the NPRM. It is Docket PHMSA-2025-0019, published July 8, 2026. Comments are closed, but the proposal tells you where this is going. 
2. Check which assessment methods your procedures actually allow. If your documented process names only B31G, you will have work to do when the accepted method list expands. 
3. Pick a segment and try to produce its full assessment history today. Inputs, corrosion profile, predicted failure pressure, and scheduling determination. Time yourself. That number tells you more about your readiness than any gap analysis will. 
4. Look at how your validation of dig data gets captured and where it lands. If it goes into a notebook and gets typed up later, that is the weak link in the chain as this rule is tightening. 

The regulatory environment is going to keep changing. There is already a runway out to 2035. 2027 is a big year. 2028 is a big year. Get used to it. 

The operators who come through this well are not the ones with the most sophisticated analysis. They are the ones who can show their work.

Previously Recorded Webinar  

“PHMSA Changed the Rules: What Your Corrosion/Integrity Program Has to Prove Now”

Hosted by Joe Pikas 

Joe Pikus draws on decades of pipeline engineering, corrosion, and NACE standards experience to discuss the relationship between field measurements, engineering decisions, and accessible pipeline records.

Watch Video