Forty Rules, One Blind Spot: How PHMSA's April Sprint Exposed the Mixed-Asset EAM Problem

On April 24, 2026, PHMSA issued 40 notices, final rules, and direct final rules in a single rulemaking package — covering gas transmission pipelines, hazardous liquid lines, underground natural gas storage, and LNG terminals. Comments close June 23. For the pipeline safety community, this is an intense sprint. For operators who manage pipelines alongside processing facilities, compression stations, or storage depots, it is something more disorienting: a compliance exercise that cuts across two fundamentally different categories of physical infrastructure at the same time.

And most of them are running two separate EAM systems.

A Rulemaking That Respects No Silos

The April package did not arrive alone. It followed PHMSA's January 2026 final rule on integrity management alternatives for class location changes (effective March 16), and the January 1 effective date of updated API RP 1171/1170 v2 requirements for underground gas storage. In practical terms, gas utilities and midstream operators are managing overlapping compliance timelines that span both their linear networks and their discrete facilities — with documentation requirements that assume a unified view of asset condition across both.

That unified view is exactly what most organisations do not have.

The median energy utility operates with a GIS-linked system for its network assets and a traditional CMMS or EAM for its facilities. These systems were built for different purposes, carry different asset taxonomies, and rarely share a common asset identifier. Reconciling them for a compliance submission is a manual task: export from one, normalise with the other, resolve discrepancies, and hope the timestamps agree. One utility integration analysis published this year found that valve records updated in GIS regularly take days or weeks to propagate to EAM systems, with field crews manually re-entering work order data in the gap.

That latency is a nuisance in steady state. In a regulatory sprint with a June 23 comment deadline, it is a liability.

The Cascade Problem No Silo Can See

There is an operational dimension that the compliance discussion tends to obscure. The assets covered by PHMSA's April package are not independent objects. Underground gas storage facilities sit at the intersection of two asset worlds that physically interact. The wellbore and reservoir are network assets: nodes in a pressure graph where injection capacity at one well affects draw rates and reservoir pressure across the formation. The surface infrastructure — compressors, separators, metering stations — is a hierarchy of discrete equipment governed by parent-child maintenance relationships.

A compressor degradation event does not stay in the compressor maintenance record. It reduces injection capacity, which changes reservoir pressure, which affects the pressure balance on the pipeline network feeding the storage facility. If your linear asset management system and your discrete asset system are not sharing data in real time, that failure chain is invisible at the network level until something downstream registers an anomaly.

This is what network-aware reasoning is designed to prevent: understanding assets not as isolated objects but as nodes in a dependency graph, where condition changes propagate through topology in ways that a site-level maintenance system cannot model.

For a deeper look at how network modeling differs from equipment hierarchy management — and why the distinction matters for AI-driven maintenance — see our linear asset management guide.

One Platform, Both Topologies

UniEAM's architecture starts from a different premise: that a mixed-portfolio operator should not have to choose between a system that understands networks and one that understands hierarchies. Both need to be native to the same platform, reasoning over the same asset graph.

For linear assets — gas pipelines, water mains, electric distribution lines, rail corridors — the platform uses GIS-native graph representations. Cascade failure models propagate through network topology. An anomaly at one node generates risk scores for connected segments, not just a work order at the anomaly location. Inspection priority is set at the network level, weighted by downstream exposure.

For discrete assets — compressor trains, pump stations, LNG vaporisers, water treatment plants — the platform uses ISO 14224-aligned equipment hierarchies. Failure modes roll up from components to functional systems to functional locations. OEE-grade telemetry feeds condition-based maintenance triggers. Parent-child failure relationships are explicit, not inferred from correlation.

UniEAM's agentic platform connects both. AI agents run continuously across the unified asset graph, correlating telemetry from the pipeline network with equipment health data from the facility layer. When a compressor shows degradation signatures, the agent models the downstream network effect before proposing a maintenance action — not after an incident makes the connection obvious. Every recommendation carries a complete decision log: input data, model version, threshold that triggered the action, and the human approval that executed it.

Audit-Ready Is Not a Feature

PHMSA's integrity management frameworks share a common thread: you must not only manage assets well, you must demonstrate that you are doing so, with traceable rationale. Why was this pipeline segment prioritised for inline inspection? What data supported the risk assessment that deferred a hydrostatic test? How did a compressor overhaul deferral factor into network-level risk scoring at the connected storage facility?

In a siloed architecture, assembling those answers means pulling records from multiple systems and hoping the timestamps agree. In a unified platform, the decision log is built in — available for regulators, not reconstructed for them.

Machine learning models can now forecast asset failures six to twelve months in advance with accuracy rates above 85%, according to recent industry benchmarks. But a model trained on pipeline condition data cannot incorporate the maintenance state of the compression stations at either end if that data lives in a separate system with no shared asset identifier. The AI is only as useful as the graph it can reason over. Fragment the graph, and you fragment the intelligence.

With 40 rulemakings in play and the comment window closing in days, the operators who will feel this most acutely are those whose compliance data still lives in two systems that do not talk to each other. The deeper issue is that audit-readiness and operational intelligence are the same architectural requirement: asset data in one place, described consistently, reasoned over as a unified graph.

Curious how UniEAM handles both? Request a demo.