矿物多采用表皮系数表征钻井流体的储层伤害程度,但无法定量表征工作流体具体性能指标与储层伤害程度的关系,不能有效指导现场钻井流体性能优化和调整。而大数据方法在多因素分析上优势明显,基于此,建立多参数钻井流体储层伤害模型,实现工作液优化以保护储层。为此,收集9口绒囊钻井流体完钻井及邻井6口其他钻井流体完钻井的现场数据,以绒囊钻井流体完钻井与邻井平均日产量差为目标函数表征储层伤害程度,以钻井流体密度、表观黏度、塑性黏度、漏斗黏度、动塑比、动切力、pH值7项参数为自变量,首先运用多元回归方法建立多参数模型,然后利用剥茧算法寻找储层伤害主控因素后建立钻井流体储层伤害数学模型,明确绒囊钻井流体性能与平均日产量差之间的定量关系。研究发现,表观黏度、密度、动塑比、pH值是决定储层伤害的主控因素,其回归系数分别为-1.561、0.428、-0.535、1.60,表明随着表观黏度、动塑比的增加绒囊钻井流体储层伤害程度加深,随着密度、pH值的增加绒囊钻井流体储层伤程度减小。利用钻井流体储层伤害数学模型指导延5-V1井水平段钻井流体性能调整,投产后的平均日产气量提高近800 m³。结论认为,相比试井等矿场评价方法,大数据方法既能准确诊断伤害程度又能为现场钻井流体性能优化提供理论依据,同时也为矿场评价储层伤害提供一种方法,现场应用效果明显。

The epidermal coefficient is mostly used to describe the reservoir damage degree of drilling fluid, but cannot quantitatively characterize the relation between the specific performance index and the reservoir damage degree, and effectively guide the optimization and adjustment of field drilling fluid performance. However, for the big data method, it has obvious advantages in multi-factor analysis. Based on this, a multi-parameter drilling fluid reservoir injury model is established to realize the working fluid optimization to protect the reservoir. To this end, the on-site data of nine completion wells withe fuzzy ball drilling fluid and six other adjacent completion wells are collected. The average daily output difference is took as the target function, with seven parameters of drilling fluid density, apparent viscosity, plastic viscosity, funnel viscosity, dynamic plastic ratio, dynamic shear force, and pH value as the independent variables. Firstly, the multiple regression method is used to establish a multi-parameter model. Then the mathematical model of drilling fluid on reservoir damage is established by the main controlling factors found by cocoon stripping algorithm. Finally, the quantitative relationship between the fluid performance and the average daily yield is defined. The study found that the regression coefficients of apparent viscosity, density, dynamic plastic ratio, and pH value are -1.561, 0.428, -0.535, 1.60, respectively, indicating that the apparent viscosity of fuzzy ball drilling fluid caused greater damage to the reservoir, density, dynamic plastic ratio, while the density and pH value have the protection effect of reservoir. In the case of the adjustment of drilling fluid performance of the horizontal section in Well-Yan5-V1 guided by the regression model, the average daily production after commissioning is increased by nearly 800 m³. In conclusion, compared with the on-site evaluation methods such as well testing, the big data method can only accurately diagnose the damage degree but also provide theoretical basis for the optimization of site drilling fluid performance. Meanwhile, it also provide a method for evaluating reservoir damage, and the application effect on site is obvious.