- 期刊
利用基因演算法建立滲透率計算模式
Development of Permeability Calculation models by Genetic Algorithms
摘要
利用井下電測資料及滲透率計算模式，可以估算地層滲透率。然而，在這些計算模式中的係數，會隨著所計算的地層不同而異。而本研究的目的是由井下電測資料比對實際測得的岩心資料，並利用基因演算法(Genetic Algorithms Method)，求得適合於台灣海域深部地層之滲透率模式。這些模式可用作估算該區域之地層滲透率。 本研究分析台灣南部海域的四口井(X-9號井、X-10號井及X-11號井，井深介於3244呎至3450呎間)之井測資料，可得到頁岩質含量、孔隙率、地層電阻、含水飽和度及殘餘含水飽和度等地層參數。根據這些分析結果，以及岩樣的滲透率資料比對，採用基因演算法(Genetic Algorithms Method)，求得適合於台灣南部海域地層滲透率的計算模式。 本研究所採用的滲透率計算模式為Wylie-Rose，Coates-Dumanoir及Porosity三種模式，利用每單口井的井下電測資料以及岩樣之滲透率資料，經由基因演算法而求得各種模式之相關係數值。利用每單一口井所採用的同種滲透率模式(例如：Wylie-Rose模式)所得之結果(有三口井而得三個方程式)非常相近，而將三個方程式的參數平均後得到的三種滲透率計算模式分別為： κ=3287.147*(ψ4.761/Swir4.616) (Wylie-Rose model)； κ=((7383703*ψ2*0.513)/(0.5134*0.608))2 (Coates-Dumanoir model)； κ=10((-3.136)+22.224*ψ) (Porosity model)
並列摘要
We can estimate the permeability of formation by using well log data and permeability calculation models. However the coefficients of calculation models will change along with the computer formation. The purpose of this study is to derive the permeability calculation models for the deep formations in southern Taiwan sea area by using well log and genetic algorithms to compare the actual core samples. These calculation models can estimate the permeability of the deep sea area. This study analysis well log data from four wells (X-9、X-10and X-11) in southern Taiwan sea area (the depth of wells are between in 3244 ft to 3450 ft). After analysis, we can obtain the shale content、porosity、formation resistivity、water saturation and water irreducible etc. The permeability calculation models have been developed by using Genetic Algorithms optimization based on the results from well log data and permeabilities data from the core samples. This study choice three permeability calculation models are Wylie-Rose、Coates-Dumanoir and Porosity models. To obtain the coefficients of model by using Genetic Algorithms optimization based on the results from well log data and permeabilities data from the core samples. For each permeability calculation model (for example：Wylie-Rose model), the model derived individually from each well (three are three different equations from three wells) shows very similar results. For averaging the parameter of the four equations, the results of the three models derived are：κ=3287.147*(ψ4.761/Swir4.616) (Wylie-Rose model)；κ=((738.3703*ψ2*0.513)/(0.5134*0.608))2 (Coates-Dumanoir model)；κ=10 ((-3.136)+22.224*ψ) (Porosity model).