Artificial Well Engineering Intelligence (AweI): Is It Drilling Engineer's Dream or Driller's Nightmare?

Abstract Applying artificial intelligence (AI) is exceedingly difficult for drilling operations as the system is overly complex and dynamic. As a result, more comprehensive domain-general engineering mapping, also known as "artificial well engineering intelligence," is required to predict operating parameters and problems with reasonable accuracy. This paper presents a detailed overview of engineering models that are interconnected in the form of microservices to provide a more logical solution as the well is drilled. It draws out some important findings and discusses ways that results can be infused with the work on explainable artificial engineering intelligence in realtime. The results argue the logical reasoning and mathematical proof. Drilling Engineer–Driller–Rig system interaction through AweI with interconnected subdomains requires tighter integration between various engineering models. To some extent, tractable abstract knowledge at the human level is derived from analytical reasoning through engineering models. Various engineering models are connected in the form of microservices, which can be called any number of times when the optimization is carried out. The results are transferred for physical actions either to the driller or control as set points. The method presented does not claim to address all the issues as a whole. This methodology attempts, however, to present a coherent adaptive model that provides more transparency to the algorithms that can be used as operational parameters for the driller. The analysis results have shown that the convergence was very quick in obtaining an optimal solution and the predictability in the test wells has shown the best solution results under uncertainty. It has also been found that the results provide reasonable threshold values when increased data is used as the well is drilled. As long as the driller stays within the operational region, the results have shown that the operating parameters are satisfying and good enough for the desirable outcome. In other words, a near-normal engineering solution is achieved. The two major interacting bottlenecks observed in the study are (1) the absence of domain-expertise and mapping the conceptual space and (2) the valuation of the results, which can be translated into practical operational parameters. The engineering microservices to derive engineering intelligence include the following: Torsional and lateral instabilities ROP coupled bit wear Hole cleaning Casing wear BHA Drill ahead Mechanical specific energy Hydro mechanical specific energy Motor stall weight (if motor present)