15 Feb, 2013
Earning Versus Learning
Drilling and development decisions often involve high levels of uncertainty about various geologic parameters. The industry has become adept at addressing these uncertainties in prioritizing drilling decisions based on expected monetary values (the “Earning” part). Early wells are typically those with high expected monetary values, ideally some combination of lower cost, higher success value, and higher chance of success. What is often overlooked, however, is the value of the information learned for properly sequencing the wells (the “Learning” part).
Using Early Results to Optimize Later Outcomes
Results of early wells can change assessments of subsequent wells – you may change the order of drilling, change how wells are drilled, or completely eliminate some previously planned wells. Getting the sequence right, especially for the first few wells, can have a significant impact on overall value achieved. That is, it’s important to balance “Earning versus Learning” for individual wells in order to maximize earning over the entire program.
Analysis of these sequential decisions with dependencies can get quite complicated if there are more than a few wells involved. The challenge is twofold: assessing the multiple geologic dependencies across a spatial network of opportunities, and then determining the optimal sequence given different outcomes as drilling proceeds.
The past few years have seen significant research in this area, with new methodologies employing complex mathematical approaches to handling the complexity. These include information-theoretic approaches with dynamic programming (Bickel and Smith, 2008), stochastic dynamic programming and “multi-armed bandits” (Brown and Smith, 2011), and Bayesian Networks and Markov random fields (Martinelli, et al, 2012).
Practical Approaches that Break the Analytical Logjam
If these sound complicated and hard to implement, they usually are. They require specialized expertise within organizations. While it’s a worthy goal to develop this expertise, what do the rest of us do in the meantime? Fortunately, as with most decision analytic techniques, there are simpler and easier-to-understand methods that, while not as rigorous, often provide enough insight into the key issues to allow us to proceed.
For example, the complicated methods described above provide some basic characteristics of what typically makes a well a good candidate for early drilling, a good candidate for waiting on additional information, or one that is likely robust enough to drill at any time in the sequence. We can use these characteristics to guide discussions about how to trade off “Earning versus Learning”.
Alternatively, we can break the problem into smaller, logical clusters and use basic value of information approaches to establish the optimal sequence within clusters. In our Advanced Decision Analysis (ADA) course, we perform one of these smaller sequencing analyses, and discuss how this could be expanded to gain insight into the larger problems.
All of these methods, whether simple or complex, require a thorough understanding of probability theory and how to assess and model dependencies. Our Petroleum Risk and Decision Analysis (PRD) course provides the fundamentals of probability theory and value of information methods, including numerous exercises. Meanwhile, our Advanced Decision Analysis (ADA) course provides more hands-on and advanced applications of these concepts.
TIM NIEMAN is President of Decision Applications, Inc., a San Francisco area based decision analysis consulting firm. His firm performs decision and risk analysis for various organizations facing complex decision problems. His recent oil and gas consulting work includes risk analysis of pipeline routing; risk analysis for deepwater flow assurance; portfolio analysis for budgeting E&P R&D portfolios; and development of methods for assessing new basin entry opportunities. Other recent work includes development of remediation and reuse strategies for impaired properties, including former refineries and pipelines; numerous projects for the Yucca Mountain proposed nuclear waste repository; work on mountain top coal mining, unconventional oil and gas drilling, basin-wide water management and climate change issues; and cancer causation modeling for national health organizations.
Mr. Nieman was formerly Senior Decision Analyst for Geomatrix Consultants, an Oakland based geological and environmental consulting firm. Prior to that, he was Director of Operations for Lumina Decision Systems, a decision analysis consulting and software firm. And prior to that, he spent 14 years with Amoco as a geophysicist, economist, and risk and portfolio analyst. He has a B.S. in geology and an M.S. in geophysics from Michigan State University, and an MBA from Rice University.
Courses now available: