关键词: 页岩气, 产能评价, 流动方程解析法, 流动方程数值模拟法, 人工智能方法, 动态评价

Abstract: Shale gas production has exhibited high initial yields followed by a rapid decline. This rapid decline suggests that early-stage production rates were likely excessive, accelerating the depletion of reservoir productivity and adversely affecting the Estimated Ultimate Recovery (EUR) of shale gas wells. Therefore, accurate and reasonable productivity evaluation plays a key role in ensuring stable reservoir development. To identify the challenges of current shale gas productivity evaluation approaches and explore feasible solutions, this study analyzed the unique connotation of shale gas productivity and reviewed recent progress in three approaches: (1) analytical solution methods of flow equations; (2) numerical simulation methods of flow equations; and (3) artificial intelligence (AI)-based methods. The results revealed that shale gas productivity was highly stage-dependent, with substantial variations in dominant controlling factors, flow mechanisms, and flow regimes across different production stages. Early and late production stages exhibit distinct controlling factors, leading to differentiated perspectives across the various evaluation methods. The analytical solution method relied heavily on a deep understanding of flow mechanisms. Numerical simulation methods require extensive, high-quality datasets and strong reservoir engineering expertise for validation. AI-based methods faced challenges such as high opacity, limited interpretability, and poor generalization. Based on these findings, future research should focus on integrating shale gas flow mechanisms at both micro and macro scales. Emphasis should be placed on the multidimensional integration of geological modeling, stress-petrophysical evolution, fracture propagation, multiphase flow numerical simulation, and decline analysis, enabling more comprehensive productivity characterization. In addition, further work is needed to incorporate mechanism-informed constraints into machine learning algorithms, enhance model transparency through causal inference, and improve interpretability. These advancements aim to avoid the limitations in existing productivity evolution methods and support the development of robust and rational shale gas productivity evaluation models and methods, providing theoretical guidance for accurate well productivity prediction, production stabilization, and efficient resource development.

Key words: shale gas, productivity evaluation, analytical solution method of flow equations, numerical simulation method of flow equations, artificial intelligence methods, dynamic evaluation