基于吉洪诺夫正则化的人造天然气水合物地层骨架数字化制备技术

Digital preparation technology of artificial gas hydrate formation skeleton by Tikhonov regularization

以地质资料丰富、成藏条件良好的美国阿拉斯加北部斜坡Mount Elbert井Unit C-GH1井段水合物地层为参照对象,将岩心制作工艺、宏观物性和微观孔隙3个方面有机结合,采用控制变量、正交分析以及吉洪诺夫正则化等方法,建立岩心工艺参数、渗透率与孔隙度、孔喉直径之间的拟合函数关系及其相应的数学计算模型,从而开发出一套高效快速的人造天然气水合物地层骨架的数字化制备技术。研究结果表明:通过控制变量和正交分析分析得出的最优人造岩心配方与原位水合物目标地层骨架十分相近,宏观物性参数渗透率和孔隙度相对误差分别为2.00%和0.40%,微观孔隙形态主要特征与天然地层骨架趋于一致;岩心制作工艺参数、膨润土添加质量分数、黏结剂添加质量分数、压力和时间与宏观物性渗透率和孔隙度的相关性极高,通过一元回归得出其相关系数均在0.9以上;基于吉洪诺夫正则化进行的多元回归分析得出了高精度的二次渗透率和一次孔隙度预测函数模型,平均测试相对误差分别为2.65%和0.87%;利用所提出的预测模型思路和建立的数学模型可以在已知一方参数的情况下快速计算出岩心的渗透率、孔隙度以及孔喉直径等另外两方参数,从而提高制作工艺参数优选效率与提高人造岩心贴近程度。

Unit C-GH1 hydrate formation of the Mount Elbert Well in the North Slope of Alaska,which had abundant geological data and good reservoir conditions,was taken as the reference object.Three aspects parameters of prepare process,macroscopic physical properties and microscopic pore structure were organically combined.Using the methods of control variables,orthogonal analysis and Tikhonov regularization,the unary and multivariate functions among prepared process parameters,physical properties and pore structure parameters were established.A set of efficient and rapid digital preparation technology of artificial gas hydrate formation skeleton was developed.The results show that the optimal formula obtained by control variables and orthogonal analysis is very close to that in situ hydrate formation skeleton,with the relative error of 2.00%and 0.40%in permeability and porosity,respectively.The main characteristics of the microscopic pore morphology are consistent with those in situ gas hydrate formations.The preparation process parameters of bentonite,binders,pressure and pressuring time have extremely high correlations with macroscopic physical parameters of permeability and porosity,and the correlation coefficients are all above 0.9 through unitary regression.High-precision secondary permeability and linear porosity functional equations are obtained based on Tikhonov's regularization of multiple regression analysis,and the average test relative errors are 2.65%and 0.87%,respectively.The proposed prediction model and the established mathematical models can be used to quickly calculate the permeability,porosity,pore and throat diameters when one aspect parameters are known,so as to improve the efficiency of preparation process parameters selection and similarity of artificial cores.