论文部分内容阅读
介绍一种用来预测有圆形相连水体的油气藏的水体参数和原始油气地质储量的新方法。油气藏的水侵量通过van-Ever-dingen和Hurst(VEH)的非稳态水侵方法表示。应用该方法除了可以预测原始油气地质储量外,还可以预测水体的参数。这些参数包括水侵系数B、无因次水体半径ReD、把真实时间t转换为无因次时间tD的时间转换因子c。还可以应用参数B和c预测水体的储水系数hCt和流动系数kh/μ。为预测水体参数,运用最小二乘法求解物质平衡方程,却引出了非线性回归问题。因为在拉普拉斯空间的解法要比在真实空间内的解法简单,所以可以应用拉普拉斯变换的数值反演,结合非线性回归分析求解最小二乘法所需的VEH的解对水体参数的偏导数。还要应用Levenberg方法预测参数值,以确保当初始假设值和真实值差距较大时,参数可以收敛于真实值。模型对于原始油气地质储量N、Gi以及水侵系数B是线性的;对于无因次水体半径ReD、时间转换因子c是非线性的。若假设一组c和ReD的值,便可以算出N、Gi以及B的值,也就可以进一步算出误差平方和。
A new method for predicting the water parameters and original oil and gas geologic reserves of reservoirs with circular connected water bodies is presented. Water intrusion volume is expressed by the unsteady intrusion method of van-Ever-dingen and Hurst (VEH). In addition to predicting the original oil and gas geological reserves, this method can also predict the parameters of the water body. These parameters include water intrusion coefficient B, dimensionless water body radius ReD, time-to-convert factor c that transforms real time t into dimensionless time tD. Parameters B and C can also be used to predict the water storage coefficient h Ct and the flow coefficient kh / μ. In order to predict the water body parameters, solving the equation of material balance by the least square method leads to the nonlinear regression problem. Because the solution in Laplace space is simpler than the solution in real space, numerical inversion of Laplace transform can be applied, and the nonlinear regression analysis can be used to solve the VEH solution required by least square method. Partial derivative. Levenberg’s method is also used to predict parameter values to ensure that the parameter converges to the true value when the initial assumed value is significantly different from the true value. The model is linear with respect to the original oil and gas geologic reserves N, Gi and water intrusion coefficient B; for the dimensionless ReD, the time conversion factor c is nonlinear. Assuming a set of c and ReD values, the values of N, Gi, and B can be calculated, and the square sum of errors can be further calculated.