The Puzzle
Today, we have an unprecedented amount of data available to solve this pipeline risk puzzle. Let’s say we want to understand internal corrosion potential on a natural gas pipeline. We examine some recent ILI results, looking for internal corrosion metal loss indications. We find some. Are they occurring at bottom o’clock positions of the pipe circumference? If so, that is a clue. We plot the ILI anomalies in GIS, add aerial photography, add topography, and look for more clues. Do we see clusters of metal loss at possible low spots—where the pipe is crossing creeks, valleys, etc? Let’s overlay elevation data—are there steep inclinations here where liquids/solids could accumulate and persist? Are we close to gas inputs, where historical liquid excursions (carryovers) might have accumulated and might first impact piping?
Next, we examine gas quality records and the performance record of the input gas streams that might have put contaminants into the gas stream. Given this, we need to understand the chemistry—what combinations of chemicals and environmental factors could be generating corrosion and at what rates? Then we can study fluid flows, thermodynamics, and hydraulics to understand how contaminants might behave inside the product stream. For those who like engineering detective work—isn’t such sleuthing compelling?
This is essentially what good risk assessment is doing. But it is far more efficient than what we would-be detectives can do individually. The risk assessment can broadcast our detective work over tens of thousands of miles of pipelines almost instantly. This effectively replaces thousands of man-hours of investigation and instantly puts key information into the hands of decision-makers.
It really is exciting to see large quantities of data drawn into a model and immediately see meaningful, actionable information come out. Turning data into information ensures that the right decisions can be made.
The risk assessment should add clarity. Some risk assessments add complexity. The real world is sufficiently complex that no unnecessary complexity should be tolerated. In a good risk assessment, if complexity appears, it should only be because the underlying science is complex.
Assessment is of course, just the beginning of risk management. Even with complete understanding of risk—via the risk assessment—we still have the challenges of how to manage this risk. Again, a host of factors comes into play: how much risk reduction is warranted? How quickly should risk reduction occur? Which is better—much risk reduction at a specific location or more modest risk reduction but over many miles of pipeline? All strive to answer the key underlying question: how safe is ‘safe enough’?
How Risk Assessment Helps
Achieving safety while undertaking a potentially dangerous activity means identifying and managing risks. Although they seem simple in concept, pipelines are actually complex, dynamic systems, operating in often-challenging environments and subject to a vast and varying array of integrity threats.
While risk has always been an interesting topic to many, it is also often clouded by misconceptions. Many equate risk analyses with requirements of huge databases, complex statistical analyses, and obscure probabilistic techniques. In reality, good risk assessments can be done with only moderate effort and even in a data-scarce environment. Risk assessment should be a useful tool that improves all aspects of pipelining.
Beyond the desire for a straightforward approach, there also seems to be an increasing desire for more sophistication in risk modeling. This is no doubt the result of an unprecedented number of practitioners pushing the boundaries as well as more widespread availability of data and more powerful computing environments. Today, it is easy and cost-effective to consider many more details in a risk model. Initiatives are currently under way to generate more widespread, complete, and useful databases to further our knowledge and to better support the detailed risk modeling efforts.
The desire for ‘more’—more accuracy, more knowledge, more decision-support—is also fueled by the knowledge that potential consequences of incorrect risk management are higher now than in the past and will likely continue to increase. Aging infrastructure, system expansions, and encroaching populations are primary drivers of this change. Regulatory initiatives reflect this concern in many parts of the world.
Robustness Through Reductionism
The best practice in risk assessment is to assess major risk variables by evaluating and combining many lesser variables, generally available from the operator’s records or public domain databases. This is sometimes called a reductionist approach, reducing the problem to its subparts for examination. This allows assessments to benefit from direct use of measurements or evaluations of multiple smaller variables, rather than a single, high-level variable, thereby reducing subjectivity. If the subparts—the details—are not yet available, then higher level inputs must suffice.
The reductionist approach also applies to the physical dimensions of the system. The risk for a pipeline is assessed as the sum of the risk of its components, where the components are the pipe, fittings, valves, tanks, pumps, compressors, meters, etc.
A critical belief underlying this book is that all pertinent information should be used in a risk assessment. There are very few pieces of collected pipeline information that are not useful to the risk assessment. The risk evaluator should expect any piece of information to be useful until he absolutely cannot see any way that it can be relevant to risk or decides its inclusion is not cost-effective.
Any and all experts’ opinions and thought processes can and should be codified, thereby demystifying the experts’ personal assessment processes. The experts’ analysis steps and logic processes can be replicated to a large extent in a risk assessment model.
A detailed risk assessment should ultimately be ‘smarter’ than any single individual or group of individuals operating or maintaining the pipeline—including that retired guy who ‘knew everything’.
It is often useful to think of the assessment process as ‘teaching the model’. We ‘tell’ the model what we know and what it means to know various things. We are training the model to ‘think’ like the best experts and giving it the benefit of the collective knowledge of the entire organization and all the years of record-keeping.
At the same time, a balance between completeness and complexity much be struck–Intelligent Simplification.
Changes from previous approaches
Previous risk assessment approaches served us well in the past. They helped support decision making by crystallizing thinking, removing subjectivity, and helping to ensure consistency. But the era of many older approaches has passed, due to increased expectations as well as the now superior analyses techniques and availability of powerful and inexpensive computer tools.
Our regulators, attorneys, neighbors, and other stake holders are no longer satisfied that we can successfully manage risk using tools that are not modern and robust. We now have strong, reliable, and easily applied methods to estimate actual risks, and no longer must accept the compromises generated by intermediate scoring schemes or statistics-centric approaches. This is a part of avoiding missteps and dispelling myths.
There is no longer any valid reason to use a relative, scoring type risk assessment approach. There is also no reason to adopt the statistics-centric ‘classical’ QRA approaches. We now have updated techniques and a powerful, but simple framework to capture and more efficiently use all available information. When much more useful results are available with no additional cost or effort, why use lesser solutions?
The modern approach to pipeline risk assessment is presented here. It is superior—in accuracy, defensibility, and cost of analyses—to all alternative approaches since it incorporates the best and eliminates the weaknesses from others.
A substantial improvement in risk assessment methodology should not be a surprise. Changes to risk algorithms have always been anticipated, and every risk model—even the most advanced—should be regularly reviewed in light of its ability to incorporate new knowledge and the latest information. As part of such reviews, it is fairly easy to spot when a risk assessment is not meeting modern expectations.
This site presents the newer risk assessment methodologies for evaluating all aspects of pipeline risk. This approach reflects the advances in risk assessment technology from research & development efforts as well as years of input of pipeline operators, pipeline experts, and risk assessors.
Key Improvements
Sections of this site offer foundational and background information. The experienced, practicing risk manager may wish to move directly to the how-to sections. It is advisable to quickly become familiar with the most essential elements of the newer (begun in 1999) methodology presented. Central to this much-improved (and ever-improving) methodology are several key features:
- The abandonment of all scoring (point assignment systems) which is now replaced by measurements.
- The PoF triad—exposure, mitigation, and resistance—the essential ingredients to understand PoF.
- The use of OR and AND gate math.
- The use of both measurements and estimates to replicate an SME’s decision processes.
- The calculation of hazard zones to drive CoF estimates.
- Identification of simple list of essential elements of good risk assessment practice.
Many other aspects of risk assessment discussed here remain similar to previous approaches. Pipeline risk factors are generally well understood. It is only the better capturing of their role in risk that changes. The estimation of consequences has generally been more grounded in physics and engineering principles already. Fewer changes in those methodologies are warranted. The probability side of the risk equation has benefitted more from recent developments.
Armed with these key changes in methodology, the more experienced reader can find basic definitions and application nuances and then move to ‘hands-on’ portions to efficiently begin assessing risks. Otherwise, here’s another path to getting started.