Physics Inspired Machine Learning for Solving Fluid Flow in Porous Media: A Novel Computational Algorithm for Reservoir Simulation

The Physics Inspired Machine Learning (PIML) is emerging as a viable numerical method to solve partial differential equations (PDEs). Recently, the method has been successfully tested and validated to find solutions to both linear and non-linear PDEs. To our knowledge, no prior studies have examined the PIML method in terms of their reliability and capability to handle reservoir engineering boundary conditions, fractures, source and sink terms. Here we explored the potential of PIML for modelling 2D single phase, incompressible, and steady state fluid flow in porous media. The main idea of PIML approaches is to encode the underlying physical law (governing equations, boundary, source and sink constraints) into the deep neural network as prior information. The capability of the PIML method in handling reservoir engineering boundary including no-flow, constant pressure, and mixed reservoir boundary conditions is investigated. The results show that the PIML performs well, giving good results comparable to analytical solution. Further, we examined the potential of PIML approach in handling fluxes (sink and source terms). Our results demonstrate that the PIML fail to provide acceptable prediction for no-flow boundary conditions. However, it provides acceptable predictions for constant pressure boundary conditions. We also assessed the capability of the PIML method in handling fractures. The results indicate that the PIML can provide accurate predictions for parallel fractures subjected to no-flow boundary. However, in complex fractures scenario its accuracy is limited to constant pressure boundary conditions. We also found that mixed and adaptive activation functions improve the performance of PIML for modeling complex fractures and fluxes.

CFD Analysis of Refrigeration Cycle Ejector

The use of ejector cycles for increased performance and efficiency is becoming more prevalent in industry. The goal of this study is to evaluate an ejector using Computational Fluid Dynamics (CFD) to evaluate flow patterns, perform trade studies varying the type of refrigerant, and determine the entrainment ratio for each working fluid, over a range of boundary condition pressures, set at points along the ejector’s flow path. The 2012 Toyota Prius V is one of the first automobiles using an ejector cycle in their internal cabin refrigeration system. The DENSO Corporation ejector hardware was used as the basis for the creation of geometry for the CFD mode of the ejector. Three working fluids were simulated, R-134a, R-245fa, and R-1235yf. The primary findings of this study were as follows. The CFD study here indicates that R-245fa performs the best out of the three working fluids, when examining their entrainment ratios (ratio of secondary to primary flow rates in the ejector). For all three working fluids, the entrainment ratio was seen to peak performance at an ejector inlet pressure of 1.75 × 105 Pa. The ejector mixing chamber pressure and ejector outlet pressure boundary conditions also witnessed a rise in entrainment ratios, during an increase of their respective pressure values.

Buoyant forces promote tidewater glacier iceberg calving through large basal stress concentrations

Iceberg calving parameterisations currently implemented in ice sheet models do not reproduce the full observed range of calving behaviours. For example, though buoyant forces at the ice front are known to trigger full-depth calving events on major Greenland outlet glaciers, a multi-stage iceberg calving event at Jakobshavn Isbræ is unexplained by existing models. To explain this and similar events, we propose a notch-triggered rotation mechanism, whereby a relatively small subaerial calving event triggers a larger full-depth calving event due to the abrupt increase in buoyant load and the associated stresses generated at the ice–bed interface. We investigate the notch-triggered rotation mechanism by applying a geometric perturbation to the subaerial section of the calving front in a diagnostic flow-line model of an idealised glacier snout, using the full-Stokes, finite element method code Elmer/Ice. Different sliding laws and water pressure boundary conditions are applied at the ice–bed interface. Water pressure has a big influence on the likelihood of calving, and stress concentrations large enough to open crevasses were generated in basal ice. Significantly, the location of stress concentrations produced calving events of approximately the size observed, providing support for future application of the notch-triggered rotation mechanism in ice-sheet models.

Buoyant forces promote tidewater glacier iceberg calving through large basal stress concentrations

Iceberg calving parameterisations currently implemented in ice sheet models do not reproduce the full observed range of calving behaviours. For example, though buoyant forces at the ice front are known to trigger full-depth calving events on major Greenland outlet glaciers, a multi-stage iceberg calving event at Jakobshavn Isbræ is unexplained by existing models. To explain this and similar events, we propose a notch-triggered rotation mechanism whereby a relatively small subaerial calving event triggers a larger full-depth calving event due to the abrupt increase in buoyant load and the associated stresses generated at the ice-bed interface. We investigate the notch-triggered rotation mechanism by applying a geometric perturbation to the subaerial section of the calving front in a diagnostic flowline model of an idealised glacier snout, using the full-Stokes, finite element method code Elmer/Ice. Different sliding laws and water pressure boundary conditions are applied at the ice-bed interface. Water pressure has a big influence on the likelihood of calving, and stress concentrations large enough to open crevasses were generated in basal ice. Significantly, the location of stress concentrations produced calving events of approximately the size observed, providing support for future application of the notch-triggered rotation mechanism in ice-sheet models.

Discrete element method approach to modelling VPP dampers

In this paper, we present a novel approach to modeling and analysis of Vacuum Packed Particle dampers (VPP dampers) with the use of Discrete Element Method (DEM). VPP dampers are composed of loose granular medium encapsulated in a hermetic envelope, with controlled pressure inside the envelope. By changing the level of underpressure inside the envelope, one can control mechanical properties of the system. The main novelty of the DEM model proposed in this paper is the method to treat special (pressure) boundary conditions at the envelope. The model has been implemented within the open-source Yade DEM software. Preliminary results are presented and discussed in the paper. The qualitative agreement with experimental results has been achieved.