Physics-informed neural networks (PINNs) represent an emerging computational paradigm that incorporates observed data patterns and the fundamental physical laws of a given problem domain. This approach provides significant advantages in addressing diverse difficulties in the field of complex fluid dynamics. We thoroughly investigated the design of the model architecture, the optimization of the convergence rate, and the development of computational modules for PINNs. However, efficiently and accurately utilizing PINNs to resolve complex fluid dynamics problems remain an enormous barrier. For instance, rapidly deriving surrogate models for turbulence from known data and accurately characterizing flow details in multiphase flow fields present substantial difficulties. Additionally, the prediction of parameters in multi-physics coupled models, achieving balance across all scales in multiscale modeling, and developing standardized test sets encompassing complex fluid dynamic problems are urgent technical breakthroughs needed. This paper discusses the latest advancements in PINNs and their potential applications in complex fluid dynamics, including turbulence, multiphase flows, multi-field coupled flows, and multiscale flows. Furthermore, we analyze the challenges that PINNs face in addressing these fluid dynamics problems and outline future trends in their growth. Our objective is to enhance the integration of deep learning and complex fluid dynamics, facilitating the resolution of more realistic and complex flow problems.

Summary: In high-temperature, high-pressure (HTHP) wells with a limited safety margin during tripping operations, modifications in the properties of drilling fluid within the well are substantial. Accurate prediction of surge and swab pressures is crucial for secure drilling operations. With this research, we introduce a method for improving surge pressure calculations through a predictive model for drilling fluid rheological parameters. The research investigates the actual condition of the opened-ended pipe (OEP) drillstring configuration beneath a 5-in. drillpipe, considering different yield stress, annular ratios, and tripping velocities. Findings indicate that when the annular ratio is below 0.55 or above 0.65, the drillstring configuration consistently results in either internal pipe blockage or annular blockage, regardless of the yield stress or annular ratio. This method can be extended to various drilling tool combinations through extensive data analysis, facilitating field assessment of drillstring conditions. Additionally, a systematic investigation was simultaneously done on the impact of sensitive parameters, including tripping speed, pump displacement, and annular size, on the surge pressure at the drill bit. The results suggest that accounting for temperature and pressure effects on drilling fluid characteristics may lead to an underestimation of computed surge pressure. In an OEP configuration, higher tripping speeds are feasible; in contrast, in a closed-ended pipe (CEP) configuration, activating the pump during tripping may reduce swab pressure. This study provides both theoretical and operational guidance for managing wellbore surge pressure across different drillstring configurations in HTHP wells, along with a novel approach for accurately predicting wellbore surge pressure.

In the well shut-in test, the multiphase phenomena, including gas–liquid separation, liquid loading, phase transition, and gas–liquid substitution, affect the pressure recorder near the reservoir, where the pressure fluctuations induced by the multiphase flow phenomena are also recorded by the pressure recorders. The pressure buildup curve would exhibit a classical hump, and its derivative curve would display a downward plunge, which cannot well characterize the reservoir when it contains the influence of the multiphase flow phenomena. After well shut-in, the wellbore experiences four stages, which are no gas–liquid separation, gas–liquid separation, gas–liquid substitution, and wellbore-formation equilibrium. A wellbore pressure regression model is developed to eliminate the influence of multiphase flow phenomena on the pressure buildup curve. A phase transition model and a transient gas–liquid two-phase flow model are also developed and integrated into the regression model. Compared with commercial software CMG, the discrepancy of the phase transition model is less than 2.00%. Compared with field data, the maximum prediction discrepancy of temperature and pressure by the two-phase flow model is 1.45% and 1.40%, respectively. The accuracies of both the models satisfy the demand of solving the wellbore pressure regression model. Case studies are conducted to prove the validity of the wellbore pressure regression model. The results show that the developed model can effectively eliminate the hump on the pressure buildup curve. After regression, the downward plunge on the derived pressure buildup curve is removed, which proves that the developed model can effectively eliminate the influence of multiphase flow phenomena on the pressure buildup curve and allow the measured pressure data to reflect the reservoir characteristics without any distortions.

Drilling platforms need to be evacuated whenever drilling operations are carried out in typhoon-prone areas. During the evacuation period, the wellbore is in a static state for a long time. After gas intrusion, it brings a high risk to the unloading and well opening operations. Aiming at this problem, this paper considers the law of gas migration in the wellbore, and proposes a