Are We Reasonably Certain that Reasonable Certainty Adequately Defines Uncertainty in Our Reserves Estimates?

Words can have many meanings, while numbers tend to have only one, so using the language of probability can better communicate the potential risks and uncertainties in developed oil and gas projects, which all employ uncertain forecasts of time, hydrocarbon price, cost, and reserves. Yet, rather than assign a probability value to the Proved or 1P deterministic estimate, SPE's Petroleum Resource Management System (PRMS) uses loosely defined expressions such as "reasonable certainty" to convey a high degree of confidence that the deterministically estimated hydrocarbon volumes will be recovered. A literature review is conducted to highlight what biases people are prone to, and to reveal what people consider to be reasonable certainty. The study recounts the provenance of the phrase "reasonable certainty" in the oil and gas industry. Using published work on cognitive bias by acknowledged oil and gas consultants, drawing on research by Harvard University statisticians, and carrying out our own tests on industry professionals, the authors demonstrate that there are consistent, and quite different perceptions of the language of quantitative probability from those suggested in the PRMS. Tests show repeatedly that industry professionals believe the probability value that corresponds to "reasonable certainty" is in the range 70-75%. This is far less than the implied 90% figure assigned to the probabilistic low estimate of recoverable volumes, generally taken as equivalent the 1P deterministic case. The divergence between deterministic language and associated and perceived probability is more pronounced when one considers the other reserves classes of 'Probable' and 'Possible'. Such perceptions lead to overly optimistic expectations and can result in poor investment decisions. There is a reluctance to assign probability values to deterministic reserve classes, but that should not prevent the assignment of more appropriate language within PRMS. Project team members, experts, and stakeholders must be aware that they are all prone to cognitive bias, which when coupled with overly optimistic perceptions of uncertainty, leads to wrong decisions being made, and to projects being mismanaged. The PRMS text could be changed to better describe the likelihood of the outcome of an event. The continued overuse of the loosely defined adjective 'reasonable' should be avoided. To convey probability greater than 90% or to describe the Low or Proved 1P case, the PRMS would better inform its users by using expressions like "very high probability", "almost certain", or "very likely"