Imagine you have a big outdoor celebration planned for Saturday. You check the weather, and there’s a 90% chance of rain. Will you change your plans?
While the practice of weather forecasting is not a novel concept, its accuracy and reliability depend heavily on the availability of historical and current data. Predictive models require a substantial amount of information to forecast future outcomes with precision. Many fields, including the water industry, have historically struggled with limited access to comprehensive data. However, the world is witnessing a transformation due to the proliferation of digital technologies.
Most pipelines have been in service for a very long time – before modern record keeping. Often, utilities have limited information on their condition. Once installed, pipelines are continuously deteriorating, and some are now reaching the end of their design life. Managing this tide of aging assets with limited information and financial resources is a top challenge for utilities around the world.
Today, technology and data analytics are delivering new insights into pipeline performance, and utility management strategies are evolving as a result – shifting from reactive to proactive. Utilities can use assessment data to prioritize their most urgent needs, plan timely interventions, and avoid running assets to failure. However, utilities can unlock even more value from pipeline condition data by analyzing historical deterioration trends and evaluating risk of failure over the long term.
The ability to be proactive and adapt to changing circumstances using data reflects the evolving synergy between technology, decision-making, and strategic planning.
Let's not forget the first conundrum: planning an outdoor celebration amidst a high likelihood of rain! Want to outsmart the weather and keep the good times rolling? How about considering a versatile event setup that seamlessly transitions between indoor and outdoor spaces? This way, you can confidently celebrate rain or shine, all while appreciating the power of both foresight and flexibility!
Now imagine you are planning to replace a critical transmission main. You assess the condition of the pipeline and discover only two pipes are expected to fail over the next 30 years. Will this information change your plans for the asset?
Utilities can predict the future condition of their pipelines and estimate when specific pipes should be reinspected or considered for repair or replacement. The ability to forecast helps utilities better plan and budget for pipeline maintenance in the coming years. Having this foresight supports justification of renewal needs and leads to more efficient allocation of limited financial resources to the most urgent priorities.
Ultimately, leveraging insights from a prediction model gives utilities more flexibility in their pipeline management approach. The utility in the example above may defensibly decide to extend the life of their transmission main with targeted interventions.
Note that as the prediction extends further into the future, uncertainty associated with the predictions increases. To facilitate longer-term planning, utilities should consider conducting periodic reassessments to recalibrate the baseline data. Additionally, we advise updating the predictive model as new condition data becomes available.
Statistical models predict how long a pipe can operate until it is likely to fail or exceed certain structural thresholds.
How can I leverage remaining useful life analysis to stay a step ahead?
STEP 1 ‒ Collect Condition Data
Determine the baseline or last known condition of the pipeline. Collect high-resolution condition assessment data to effectively predict the remaining useful life (RUL) of prestressed concrete cylinder pipes (PCCP) and metallic pipes. Broken wire wraps are the best indicator of degradation in PCCP, while metallic degradation analyses are based on wall thickness measurements.
STEP 2 – Define an Intervention Limit
An intervention limit is the point at which you should consider taking action, which could range from a reinspection to a repair or even replacement of the pipe. The asset owner can define this limit based on their risk tolerance or by using information from a structural analysis. Finite element analysis produces a performance curve that helps determine the structural yield or strength limit of the pipe. The structural yield limit is used most often as the intervention limit. It represents the number of broken wire wraps or amount of wall loss that will result in the pipe yielding, depending on the pipe’s operational conditions and internal pressure.
STEP 3 – Analyze Remaining Useful Life
Estimating RUL requires an understanding of a pipe’s historical deterioration patterns – the change in pipe condition over time. The output of the predictive model is the range of expected broken wire wraps (PCCP) or corrosion growth (metallic) for each pipe, in each year over the prediction horizon. By applying the intervention limit, we can predict which pipes will require action over the next 20+ years.
How can I improve the accuracy of the forecast?
Just as weather forecasts are not always right, pipe degradation forecasts also come with degrees of uncertainty. However, there are ways to increase confidence in the RUL predictions, thereby improving long-term planning.
Tool Resolution and AccuracyCollecting accurate, high-resolution data is the first step to boosting performance of the predictive model.
Augment the DataCalculating a pipe’s RUL with first-time inspection data carries more uncertainty. This is because we must make assumptions about when the pipe began degrading. We typically assume degradation started at the time of installation, which can lead to an overestimation of the pipe’s remaining life.
However, Xylem has over two decades of pipeline inspection and monitoring data. This means we can augment our model with degradation rates from pipes similar to the subject pipeline in size, age, material, and class. Augmenting the data leads to better predictive outcomes.
Include Reinspection DataIncorporating periodic reinspection data into the model also decreases uncertainty. We can estimate a more accurate degradation pattern with two or more known condition datasets. With more data, the actual and predicted pipe condition become more closely aligned.
These predictive analytics derive more actionable information from utilities’ hard-earned inspection and monitoring data. Layering these results with additional datasets can reveal even more valuable information. As utilities continue their digital journey, we look forward to exploring new ways to breakdown data silos, connect the dots, and deliver insights that help optimize infrastructure reliability.