Speaking at the SAS day at Texas A&M university, SAS’ Keith Holdaway offered three case studies on the use of analytics—specifically, SAS’ Semma methodology. Semma, ‘sample, explore, modify, model, assess’ has been applied to complex upstream problems by Shell, Saudi Aramco and an unnamed ‘major Barnett Shale operator.’ The commonality between these case histories are that they involve multi-variate optimization and are therefore amenable to a statistical approach.
Shell used Semma to identify the most significant variables in a set of ‘fragmented, unreliable, sparse data’ from 211 wells in the Pinedale anticline field. Some 2400 frac stages were studied to provide Shell with an understanding of what distinguished poor from exceptional wells. The program also allowed the company to develop new models for pressure depletion and to eliminate unnecessary completion stages. Overall, optimization opportunities were identified in 25% of the wells studied. The results of this study have been presented as SPE 135523.
The unnamed Barnett Shale operator was confronted by a large number of interacting parameters including the impact of proppant volume on production, the difficulty of isolating significant variables that impacted the fracking process and how multiple facets of the region’s complex geology interacted with its engineering approach. Semma was applied to an 11,000 well data set to show which variables made the most important contribution to key performance metrics such as drilling cycle times and well lifetime production. Semma allowed the operator to develop ‘consistent and repeatable’ workflows and to identify opportunities for future cost reduction. The study brought a 30% reduction in proppant costs and a new modeling methodology for de-risking new plays.
Operating a mature field, Saudi Aramco realized the need for a change in its business process to combat production deferments. Previously, multiple domain silos created process inefficiencies, slowed analytical efforts and led to inconsistent data quality. The tools available to analyze the large numbers of wells were limited and results from ‘deterministic’ well studies were inconsistent. Again, SAS’ technology was leveraged to achieve a 25% reduction in deferred production and improved forecast accuracy. Aramco now has speeded data collection with standardized and automated production surveillance. This presentation was extracted from SPE 141110.
Josh Wills described how ‘data science’ is performed using SAS and Cloudera’s Hadoop implementation. Data science falls somewhere between statistics and software engineering. For big data, whether from click streams or oilfield sensors, the value is only realized when data scientists can access all the data at once. Here the tools of the trade are SAS’ Lasr and Cloudera-ML. Read the SAS day presentations here.
© Oil IT Journal - all rights reserved.