Speaking at last month’s meet of the Houston chapter of the Society of Petroleum Evaluation Engineers, Ryder Scott’s Miles Palke observed that main use of reservoir simulation is to test various field development scenarios and evaluate investment opportunities. But companies may pressure their consultants to use such ‘scoping’ models for reserves estimation. Elsewhere, models developed to forecast vertical well performance may be used to project undeveloped horizontal well performance.
Palke observes that in such cases, ‘A disconnect exists between the original purpose and proposed use of the model.’ In such circumstances, an in-depth review of the model’s internals is required to ascertain its applicability to the new use case. Palke recommends a ‘reality check,’ comparing model output with traditional analytical techniques. He also provides a check list1 of tests that can be applied even when the engineers who originally set up the model are no longer around.
When reviewing the history match process, reviewers need to be aware that some simulation software allows unreasonable changes to be made to inputs. Pore volumes may exceed gross volumes, a modeled aquifer may not exist or residual saturations may tend to zero. A serious problem can arise when minor tweaks to parameters impact history in a minor way but significantly affect predictions. ‘Don’t be tricked by very good matches of single phases or cumulative volumes at the end of history.’ These can simply indicate that the modeler has set the dominant phase’s rates so that the simulator hits the volumes. Reviewers should home in on the transition from simulation to prediction and also checkout the reasonableness of the base case with no changes in operating conditions or well count. Building on Palke’s analysis, Ryder Scott is developing a set of metrics for reservoir model reviews. More from www.oilit.com/links/1106_5.
1 This article was abstracted from Ryder Scott’s informative June-August newsletter www.oilit.com/links/1106_4.
© Oil IT Journal - all rights reserved.