The primary objective of the present work is to propose a new methodology that combines topological data analysis (TDA) and physics-informed artificial intelligence (PhysAI) models to enable automated reserve estimation and reliable field optimal recommendations as new data becomes available. Due to the resilience and efficient nature of the approach, we can deliver a massive number of forecasts on a regular basis. In this way, engineers and decision-makers can rely on a practical approach to explore multiple unconventional field development strategies on a timely basis.
Identifying relevant correlations and performance production drivers from large volumes of disparate and noisy data is a daunting task that geoscientists and engineers frequently encounter in their daily work. This process is necessary for building coherent and representative datasets that are eventually consumed by repetitive iterations with existing modeling and forecasting tools. Failing in meeting certain levels of data quality, proper attribute and model selection criteria can severely compromise the reliability and efficiency of desired outcomes and decisions. This problem poses various challenges as the confidence in field decisions vary wildly with respect to the geological uncertainty, number of wells and complexity of operations envisioned in an exploitation plan. The general practice is that the larger the field project the coarser the data resolution becomes in both space and time. Correspondingly, modeling and forecasting have to be replaced by simpler and more practical workflows that could allow assessing a large number of wells and scenarios. In any case, it is always desirable to achieve high levels of both efficiency and accuracy regardless of the complexity and size of the field project.
During the last years, there has been a significant interest in finding insightful, accurate and fast ways to model shale reservoirs. Shale modeling is a multiphysics and multiscale problem that involves multiphase flow on complex fracture systems subject to various geomechanical forces induced by different completion and operational configurations. There is a high level of uncertainty associated with properties, fracture geometry and the transient nature of the reservoir behavior. As a result, ambiguity manifests in disparate models yielding nearly identical history matching with radically different forecasts and estimated ultimate recovery (EUR). This is a problem that essentially depends on three tightly connected components: (a) data quality and representativeness on the phenomena under study, (b) reliable extraction of key performance drivers from data, and (c) model accuracy from performance drivers.
