Statistical modeling of geopressured geothermal reservoirs

Abstract

Identifying attractive candidate reservoirs for producing geothermal energy requires predictive models. In this work, inspectional analysis and statistical modeling are used to create simple predictive models for a line drive design. Inspectional analysis on the partial differential equations governing this design yields a minimum number of fifteen dimensionless groups required to describe the physics of the system. These dimensionless groups are explained and confirmed using models with similar dimensionless groups but different dimensional parameters. This study models dimensionless production temperature and thermal recovery factor as the responses of a numerical model. These responses are obtained by a Box-Behnken experimental design. An uncertainty plot is used to segment the dimensionless time and develop a model for each segment. The important dimensionless numbers for each segment of the dimensionless time are identified using the Boosting method. These selected numbers are used in the regression models. The developed models are reduced to have a minimum number of predictors and interactions. The reduced final models are then presented and assessed using testing runs. Finally, applications of these models are offered. The presented workflow is generic and can be used to translate the output of a numerical simulator into simple predictive models in other research areas involving numerical simulation.

Response surface modeling of CO2dynamic storage efficiency factor in high permeability thick sandstones

“…showed it to be a reliable and accurate space‐filling technique. Afonja and Ansari among other researchers have applied this technique to design numerical experiments for EOR and geothermal projects.…”

Optimization and uncertainty quantification model for time-continuous geothermal energy extraction undergoing re-injection

“…Other technologies proposed improving efficiency and economics by including the circulation of carbon dioxide or combination with natural gas production [9,10,11]. Subsurface properties are often poorly characterized and therefore associated with significant uncertainties, leading to inaccurate predictions and optimization [12,13,14,15,16]. Therefore, optimizing field recovery performance under various time-dependent uncertainties is crucial to improve efficiency.…”

“…With rigorous training and validation, the low-fidelity model is expected to capture the governing physics as in the high-fidelity model but with significantly less computational effort. The training and validation process usually involves the use of Design of Experiments (DoE) for sampling [36,37,38,14,39,40,41]. When the output is time-dependent, the response is often built by several models constructed at discrete time intervals [42].…”

“…This approach is frequently utilized in petroleum engineering applications to perform resource assessment and decision-making in field development [36,38]. In geothermal applications, [14] used it for uncertainty quantification for thermal recovery factor of the injection scheme in a tilted geothermal reservoir. [43] conducted uncertainty quantification for power capacity in the Ngatamariki geothermal field in New Zealand.…”

Response surface method for assessing energy production from geopressured geothermal reservoirs

“…These algorithms increase the efficiency of the Newton loop by creating the Jacobian matrices around previously simulated points instead of traditionally solving the flow equations (Ansari 2014;He and Durlofsky 2014;Cardoso and Durlofsky 2010). The third approach, which is used in this work, is to run the detailed model using specific combinations of factors sampled by experimental design and then fit a proxy response surface to the factor space (Ansari 2016). Experimental design and response models are popular and used widely (Fisher and Genetiker 1960;Mishra et al 2015).…”