Texas Oil & Gas Magazine Vol 4 Issue 1 | Page 25

CLICK TO JOIN OUR COMMUNITY: March 25-26, 2015 | Houston | Texas > reservoir pressure. Other data is possible to collect, but cost-prohibitive - such as bottom hole pressure data and lab analysis of fluid samples. There is still more data that we can collect but has a high measure of imperfectness - such as core samples. Geologic heterogeneity & natural fractures violate the assumptions in some of our most sophisticated models used for production data analysis. Pressure/stress dependent properties and multi-phase flow should also get an honorable mention when discussing complexities in modeling. We often don’t know whether wells can be treated as if they are isolated within the reservoir until months after we’ve had to make a decision regarding well spacing. An understanding of first principles can elucidate some of the possible interpretations involved in analysis, but there is no single answer - or in some cases, not enough data to provide any answer at all. And then we must base a business decision based upon it. What are the greatest advances in developing reservoir modeling technologies specific to unconventional oil & gas? I would have to say computational power, because it allows for an acceleration of the scientific method. The ability to test multiple hypothesis in simulation and determine which do and do not lead to plausible explanations of the behavior we empirically observe in production data accelerates our ability to learn and gain knowledge of unconventional well performance. What will you be presenting at the upcoming Reservoir Engineering for Unconventional Oil & Gas congress? The Transient Hyperbolic Model (THM) is a decline curve model that is equivalent in form to the analytic solution for a closed linear reservoir. It has three primary advantages: 1) The parameters of the model have physical significance and are well correlated to behavior of the reservoir - meaning they can be estimated a priority, not simply through empirical curve fitting; 2) Because the model is equivalent in form to an analytic solution, it is possible to compute the THM analytically as well - from any analytic or numerical history match, the parameters of the model can be calculated, and rapidly applied to other wells that do not have data sets that make analytic or numerical modeling possible; and 3) The THM represents the behavior of transient linear, transitionary, and boundarydominated flow, allowing modeling of full-life behavior of unconventional wells. If it can be fit to production data that is still within a transient or transitionary regime, given the evaluator can accurately express a belief about longterm behavior of the reservoir (i.e. not just the model). What are the reservoir modeling analysis methodologies that will be key in estimating EUR accurately and predicting production performance? One common feature of numeric, analytic, and empirical models, has been a recent trend of increasing complexity. For example, the industry observed complex geo-mechanical behavior in the Haynesville during the development boom circa 2007. This spurred the development of simulators that could model how reservoir properties change during the depletion process. Analytic models led to semianalytic models that combine some of the power of numerical simulators. There are numerous empirical models proposed to describe transient flow. This can be overwhelming to a reservoir engineer deciding which combination to use under what conditions. This circles back to having computational power available to analyze these different models within a reasonable period of time. One step further, which would improve the abilit y to make inferences about future performance from all models, is the addition of statistical uncertainty evaluation to any model. If we know our models are not accounting for every complexity (because we do not know every complexity), then it is important we generate a range of possible interpretations for any given model. Discrete model fits do not tell us the probability of achieving the model’s predictions, nor assist in tuning the model over the long term - how is one to judge what was a correct or incorrect, or good or bad, forecast, in an objective manner? If we focus on the outcomes of various possible model fits, and how likely those outcomes are, we would gain much confidence in our ability to estimate future performance in all cases, by our new found ability to judge and learn from how well we forecast in the past. REGISTER HERE www.unconventional-reservoir-engineering.com www.american-business-conferences.com + (1) 800 721 3915 [email protected]