Reservoir simulators model the highly nonlinear partial differential equations that represent flows in heterogeneous porous media. The system is made up of conservation equations for each thermodynamic species, flash equilibrium equations and some constraints. With advances in Field Development Planning (FDP) strategies, clients need to model highly complex Improved Oil Recovery processes such as gas re-injection and CO2 injection, which requires multi-component simulation models. The operating range of these simulation models is usually around the mixture critical point and this can be very difficult to simulate due to phase mislabeling and poor nonlinear convergence. We present a Machine Learning (ML) based approach that significantly accelerates such simulation models. One of the most important physical parameters required in order to simulate complex fluids in the subsurface is the critical temperature (Tcrit). There are advanced iterative methods to compute the critical point such as the algorithm proposed by Heidemann and Khalil (AIChE J 26,769–799, 1980) but, because these methods are too expensive, they are usually replaced by cheaper and less accurate methods such as the Li-correlation (Reid and Sherwood 1966). In this work we use a ML workflow that is based on two interacting fully connected neural networks, one a classifier and the other a regressor, that are used to replace physical algorithms for single phase labelling and improve the convergence of the simulator. We generate real time compositional training data using a linear mixing rule between the injected and the in-situ fluid compositions that can exhibit temporal evolution. In many complicated scenarios, a physical critical temperature does not exist and the iterative sequence fails to converge. We train the classifier to identify, a-priori, if a sequence of iterations will diverge. The regressor is then trained to predict an accurate value of Tcrit. A framework is developed inside the simulator based on TensorFlow that aids real time machine learning applications. The training data is generated within the simulator at the beginning of the simulation run and the ML models are trained on this data while the simulator is running. All the run-times presented in this paper include the time taken to generate the training data and train the models. Applying this ML workflow to real field gas re-injection cases suffering from severe convergence issues has resulted in a 10-fold reduction of the nonlinear iterations in the examples shown in this paper, with the overall run time reduced 2- to 10-fold, thus making complex FDP workflows several times faster. Such models are usually run many times in history matching and optimization workflows, which results in compounded computational savings. The workflow also results in more accurate prediction of the oil in place due to better single phase labelling.
相似文献 |
|||||||
|
|||||||
|
排序方式： 共有36条查询结果，搜索用时 78 毫秒
31.
Paul F. Scott Pedro Carreira Kieran Cleary Rod D. Davies Richard J. Davis Clive Dickinson Keith Grainge Carlos M. Gutiérrez Michael P. Hobson Michael E. Jones Rüdiger Kneissl Anthony Lasenby Klaus Maisinger Guy G. Pooley Rafael Rebolo José Alberto Rubiño-Martin † Pedro J. Sosa Molina Ben Rusholme ‡ Richard D. E. Saunders Richard Savage An&#;e Slosar Angela C. Taylor David Titterington Elizabeth Waldram Robert A. Watson § Althea Wilkinson 《Monthly notices of the Royal Astronomical Society》2003,341(4):1076-1083
32.
Source subtraction for the extended Very Small Array and 33-GHz source count estimates 总被引：1，自引：0，他引：1
Kieran A. Cleary † Angela C. Taylor ‡ Elizabeth Waldram Richard A. Battye Clive Dickinson § Rod D. Davies Richard J. Davis Ricardo Genova-Santos Keith Grainge Michael E. Jones ¶ Rüdiger Kneissl G. G. Pooley Rafael Rebolo José Alberto Rubiño-Martín Richard D. E. Saunders Paul F. Scott An&#;e Slosar &#; David Titterington Robert A. Watson 《Monthly notices of the Royal Astronomical Society》2005,360(1):340-353
We describe the source subtraction strategy and observations for the extended Very Small Array (VSA), a cosmic microwave background interferometer operating at 33 GHz. A total of 453 sources were monitored at 33 GHz using a dedicated source subtraction baseline. 131 sources brighter than 20 mJy were directly subtracted from the VSA visibility data. Some characteristics of the subtracted sources, such as spectra and variability, are discussed. The 33-GHz source counts are estimated from a sample selected at 15 GHz. The selection of VSA fields in order to avoid bright sources introduces a bias into the observed counts. This bias is corrected and the resulting source count is estimated to be complete in the flux-density range 20–114 mJy. The 33-GHz source counts are used to calculate a correction to the VSA power spectrum for sources below the subtraction limit. 相似文献
33.
Muhammad Asif Robert Alexander Tahira Fazeelat Kieran Pierce 《Organic Geochemistry》2009,40(8):895-901
Sediment extracts and crude oils have been shown to contain methyl substituted biphenyls and dibenzothiophenes, with isomer distributions suggesting a geochemical relationship between the two compound classes. Laboratory simulation experiments have shown that carbon catalyses the reaction between surface adsorbed sulfur and biphenyl to form dibenzothiophene. Similarly, the methyl substituted biphenyls reacted to yield corresponding methyl dibenzothiophenes. We suggest that the widespread distribution of dibenzothiophene and alkylated dibenzothiophenes in sediments and crude oils is the result of a catalytic reaction of biphenyl ring systems and surface-adsorbed sulfur on the surface of carbonaceous material. 相似文献
34.
35.
36.
Sheth Soham Heidari Mohammad Reza Neylon Kieran Bennett James McKee François 《Computational Geosciences》2022,26(1):1-11
|
| 设为首页 | 免责声明 | 关于勤云 | 加入收藏 |
|
Copyright©北京勤云科技发展有限公司 京ICP备09084417号 |