The “bathtub curve” is an engineering term that refers to the reliability of a product or system. In short, it is the idea that when a product is new, there is a brief “burn-in” period, where the rate of failure is relatively high. What follows is a period where failures are relatively constant and random before a third period where failures begin to occur more due to the natural lifecycle of the product nearing its end. In pipeline systems, we can look at the incident rate by the decade of installation to gauge whether this engineering concept applies to pipeline systems.
Gas Transmission
PHMSA’s Gas Transmission mileage by decade data lets us examine how pipeline rates of failure by decade installed might reflect the lessons of the bathtub curve. First, we can only examine significant incident rates by the installation decade for failures due to Pipeline and Valve Sites, precisely the performance measure PHMSA shares in their Gas Transmission Performance Measures. However, PST also wanted to consider compressor stations to understand if there is a relationship between the age of those stations and significant pipeline failures. While the relationship between compressor stations and mileage is not linear, we can safely assume greater mileage requires more compressor stations. Plot 1 below shows that, with or without compressor stations included in the analysis, there appears to be some element of “burn-in” time with pipeline failure rates. Without compressor stations, the incident rate nearly tripled from pipelines installed in the 2000s to 2010s (.6 incidents per 1,000 miles to 1.6). However, when including compressor stations, that relationship becomes even more striking from the 2000s to the 2010s. While the 2020s haven’t shown quite the same leap, we will look at how this decade’s incident rate develops compared to the 2010s.
Many miles of pipeline continue to operate despite operators lacking knowledge of their installation year. PHMSA’s mileage by decade data on gas transmission lines separates these unknown miles, which allows us to take a closer look at the incident rate among “unknown”-aged pipelines. Plot 2 looks at these failure rates on pipeline and valve sites and compares them to the original analysis in Plot 1. The result is striking: unknown pipelines fail at rates nearly three times higher than the next highest incident rate. While these pipelines are currently phasing out of operation, their impact is considerable compared to all other pipeline ages.
Hazardous Liquids
Similarly, we can examine these same onshore significant failure rates in Hazardous Liquids lines. While in this case, we continue to ignore any failures involving platforms and storage, we can compare incident rates on Pipeline and Valve sites alone or including Pump/Meter Station Equipment and Piping. Either way, we see a similar story in Plot 3, where there is some evidence that early failures (or burn-in) occur in pipeline systems but are not quite as significant as the incident rates we see in pipelines at the end of their life cycle.
