西南石油大学学报(自然科学版)
西南石油大學學報(自然科學版)
서남석유대학학보(자연과학판)
JOURNAL OF SOUTHWEST PETROLEUM UNIVERSITY(SEIENCE & TECHNOLOGY EDITION)
2014年
1期
65-73
,共9页
孙立春%郭丽娜%何娟%吴梦阳%倪军娥
孫立春%郭麗娜%何娟%吳夢暘%倪軍娥
손립춘%곽려나%하연%오몽양%예군아
随机地质建模%响应面%代理模型%不确定性%地质模型%概率
隨機地質建模%響應麵%代理模型%不確定性%地質模型%概率
수궤지질건모%향응면%대리모형%불학정성%지질모형%개솔
stochastically geomodeling%response surface%proxy%uncertainty%geomodel%probability
针对应用随机地质建模技术进行不确定性分析中的难点问题,提出了使用响应面代理模型进行不确定分析的方法及其研究流程;分析了此方法应用过程中的关键点,即根据地质特征,充分考虑各建模环节的不确定性,开展敏感性参数分析,应用响应面方法建立代理模型,生成训练/验证模型及应用代理模型进行结果预测时采样算法的要求。基于代理模型的预测结果,可方便地得到定量评价参数相对真实的概率分布。在此基础之上,筛选对应于某一概率下的随机地质模型,有助于进一步定量评价油田开发风险与潜力,为开发方案的编制及优化服务。
針對應用隨機地質建模技術進行不確定性分析中的難點問題,提齣瞭使用響應麵代理模型進行不確定分析的方法及其研究流程;分析瞭此方法應用過程中的關鍵點,即根據地質特徵,充分攷慮各建模環節的不確定性,開展敏感性參數分析,應用響應麵方法建立代理模型,生成訓練/驗證模型及應用代理模型進行結果預測時採樣算法的要求。基于代理模型的預測結果,可方便地得到定量評價參數相對真實的概率分佈。在此基礎之上,篩選對應于某一概率下的隨機地質模型,有助于進一步定量評價油田開髮風險與潛力,為開髮方案的編製及優化服務。
침대응용수궤지질건모기술진행불학정성분석중적난점문제,제출료사용향응면대리모형진행불학정분석적방법급기연구류정;분석료차방법응용과정중적관건점,즉근거지질특정,충분고필각건모배절적불학정성,개전민감성삼수분석,응용향응면방법건립대리모형,생성훈련/험증모형급응용대리모형진행결과예측시채양산법적요구。기우대리모형적예측결과,가방편지득도정량평개삼수상대진실적개솔분포。재차기출지상,사선대응우모일개솔하적수궤지질모형,유조우진일보정량평개유전개발풍험여잠력,위개발방안적편제급우화복무。
To solve the difficulty problems,which the stochastic geological model is uncertain,the paper presented the response-surface proxy model to carried out the uncertainties method and its research process. The uncertainties with the geomodel can’t be easily and correctly disclosed with respect to the process of the stochastically modeling that is limited by computer’s capacity and time consumption. In the whole process,some following crucial things should be considered serious-ly. All uncertainties with the geomodeling should be fully taken into account in compliance with the geological characteristics of reservoirs. Sensitivity analysis would be carried out prior to the generating of proxy model. The sampling algorithms for generating both the training and the validation models could be different from the predicating of the target through proxy model. The probabilistic distribution of the target variables(oil in place as an example)can be understood through proxy model to a great extent. Therefore,these stochastic geomodels can be picked out with respect to the probabilities in terms of the target. In this sense,risks and potentials for the development of a reservoir can be appraised further based on the uncertainty studies.