Speaking at the 2010 Palisade Risk Conference in London last month, Statoil’s John Zhao advocated ‘putting more science in cost risk analyses.’ Quantitative analysis allows the measurement of risk probability and the forecast of consequences. According to Zhao, such analysis performed by Wall Street’s ‘quants’ forecast the recent economic crisis but ‘they were ignored by the management generals.’ While Monte Carlo analysis using tools such as Palisade’s @RISK are popular in oil and gas, the simplistic line-item ranging exercise fails to capture large capital project contingencies. Empirical data has shown that many disastrous cost overruns were due to poorly evaluated contingent risks. To show management a complete risk, both systemic risks that history shows to be likely and specific risks with ‘discrete’ probabilities need to be con-sidered. The technique is to blend continuous probability distribution functions (PDF) for project cost estimates with discrete PDFs from a project risk register.
Zhao describes current approaches as ‘delinquent’ because they ‘intuitively respond to risks, derive statistics from mathematical models and lack empirical validation and business knowledge.’ This leads to non-credible analyses that are not trusted by management.
Zhao works through a cost risk analysis example to illustrate the effects of various inputs such as the type of probability distribution, correlations and dependencies and historical data. Zhao recommends ‘double triangle’ distributions and a ‘risk register table augmented with risk discrete MC functions and contingencies. Statoil’s approach also handles correlated risks such as the way the value of field piping hours correlates with scaffold labor hours.
Historical data can hold valuable clues to model building and for reality checks. But ‘the sad fact is that companies are poor at keeping historical cost data’. One analysis suggested project cost in the range $170-230 million at 90% confidence. But calibrating with historical data brought the confidence rating down to 66%.
Risk analysis is a practice that can be very confusing, even for the experts. There is insufficient academic research and very little mature, pragmatic empiricism. Putting more science into risk analyses is a start but there are more unexplored facets to the overall goal of integrating the entire value chain of oil and gas risk assessment.
Zhao’s novel approach also aligns with the AACEi’s1 Recommended Practices (RP 2009) for cost risk analyses. More from www.oilit.com/links/1005_7.
1 Association for the Advancement of Cost Engineering— www.oilit.com/links/1005_8.
© Oil IT Journal - all rights reserved.