Thermal traditional enhanced oil recovery (EOR) screening has long relied on static lookup tables that filter candidate reservoirs using fixed, isolated property cutoffs such as depth, net pay, permeability, and viscosity. While useful for preliminary sorting, these static matrices fail to capture non-isothermal thermodynamics, transient geomechanical stress fields, dynamic fluid-phase behaviors, and the stringent economic and environmental constraints governing modern heavy oil assets. This manuscript introduces a dynamic, four-stage integrated decision architecture that transforms thermal EOR (TEOR) screening from a passive collection of property thresholds into an actionable, physics-governed engineering workflow. The primary novelty of this work lies in its holistic synthesis across five previously siloed operational domains: steam-solvent hybrids, late-stage SAGD with non-condensable gas (NCG) co-injection, electromagnetic/radio-frequency heating, machine learning applications, and CCUS-coupled recovery. Unlike previous reviews that analyze standalone recovery techniques or unconstrained “black-box” predictive models, this framework explicitly integrates operational boundary exceptions (for example, Vacuum Insulated Tubing, Downhole Steam Generation, and solvent prepads) that extend traditional depth, thin-pay, and tight-matrix screening limits. Furthermore, it embeds concrete, physics-informed machine learning applications—ranging from facies proxy models and spatial–temporal steam chamber predictors to dynamic Steam-Oil Ratio (SOR) optimizers and real-time failure detectors—directly into sequential screening gates. Every parameter cutoff across Cyclic Steam Stimulation, Steam-Assisted Gravity Drainage, In Situ Combustion, and Electromagnetic Heating is thoroughly justified through fluid dynamics, heat transfer thermodynamics, geomechanics, and combustion kinetics. By systematically evaluating candidate reservoirs through four sequential decision gates, Thermodynamic Viability, Container Integrity, Lifecycle Carbon and Economic Balancing, and Modular Field Pilots with Real-Time Sentinel Monitoring, this architecture provides petroleum engineers and asset managers with an end-to-end roadmap to de-risk capital deployment, optimize thermal efficiency, and align heavy oil operations with global decarbonization imperatives.
The results of a numerical modeling of the laminar water flow and mixed conductive–convective heat transfer in highly elongated heatsink with continuous rectangular microchannels, and a width-to-length ratio of 625 (i.e., 0.2 mm to 125 mm), are presented, along with two methods of fluid entry (along and perpendicular to the base of the heatsink) for a wide range of surface heat flux density q from 30 to 2000 kW/m2 and water volume flow rate from 10 to 50 L/h (ReD=26−128). It has been demonstrated that: (1) The elongated microchannel maintains its effectiveness in heat removal by water, even under conditions of high q-value; however, in the terminal sections the temperature can approach the maximum permissible operating level for high-loaded silicon electronic components (~95–100 °C). (2) For microchannel heatsinks, the direction of the incoming flow is not of significant influence, since the microchannel structure, even with a height of 2.5 mm, exhibits substantial diffusivity and the flow in the inlet area rapidly turns along the heat distribution base; this flow further possesses the same characteristics as in the case of the input flow parallel to the axis of the microchannels. The developed model allows for the optimization of geometric, kinematic and thermal parameters of highly efficient microchannel devices, depending on the expected operating conditions.
Pin-fins are widely used to enhance heat transfer in compact heat exchangers, turbine cooling passages, and electronic devices, but their complex geometries make accurate thermal–fluid prediction computationally expensive. This paper presents a geometry-aware multiscale (GAMS) graph neural network (GNN) for predicting steady turbulent flow and heat transfer in a two-dimensional channel containing arbitrarily shaped pin-fin geometries. An automated framework integrating geometry generation, meshing, and ANSYS Fluent simulations was developed to construct the training dataset. Pin-fin geometries were parameterized using piecewise cubic splines, generating 1000 unique configurations through Latin Hypercube Sampling. Each simulation was converted into a graph representation, where nodes contained spatial coordinates, normalized streamwise position, one-hot boundary indicators, and signed distance to the nearest wall. These graph-based features were used to train the GNN to predict the temperature, velocity magnitude, and pressure fields directly from geometry. The network achieved excellent predictive accuracy, successfully capturing boundary layers, recirculation zones, and upstream stagnation regions while reducing computational wall time by 2–3 orders of magnitude compared to conventional CFD simulations. Overall, the proposed GNN provides a fast, reliable surrogate modeling framework for complex thermal–fluid flow configurations.
This study presents a comprehensive analysis of two advanced cooling methods immersion and bottom plate cooling applied to a cylindrical Lithium-ion battery cell (21700 format) for electric vehicles (EVs). Using 3D Computational Fluid Dynamics (CFD) simulations, the thermal performance of both methods was evaluated under various discharge rates (1C, 2C, and 3C) and coolant flow rates. The results demonstrate that immersion cooling offers significantly better thermal management than bottom plate cooling. Specifically, immersion cooling reduced the maximum cell temperature by 34%, maintaining a peak of 50.6 °C at a 3C discharge rate, whereas bottom plate cooling resulted in a peak temperature of 77.7 °C. Additionally, immersion cooling achieved a superior heat transfer coefficient of 225.08 W/m²K, compared to 193.49 W/m²K for bottom plate cooling. Immersion cooling also exhibited a more uniform temperature distribution, reducing temperature gradients and enhancing battery lifespan, especially at high discharge rates. At a lower discharge rate of 1C, the temperature difference between the two methods was minimized to 3 °C, indicating that bottom plate cooling remains effective under less demanding conditions. Furthermore, immersion cooling significantly reduced pressure drop by 59%, enabling more efficient coolant flow and lower energy consumption. The findings underscore the advantage of immersion cooling in high-performance applications, offering improved thermal control, energy efficiency, and operational safety
The application of forced flow and suction has a significant impact on enhancing the efficiency of thermal management systems. This study explores the influence of forced flow and suction on improving the performance of thermal management systems, particularly focusing on their effects on laminar flow and heat transfer in Cu/water nanofluid over a rotating disk. Entropy analysis is also conducted under these conditions to assess the thermodynamic implications. The governing nonlinear equations describing the flow and heat transfer processes are transformed using Kármán transformations and solved numerically using MATLAB. Key parameters are analyzed to evaluate their impact on flow dynamics, heat transfer, average entropy generation, and the Bejan number, with graphical. The findings reveal that radial forced flow and suction significantly enhance momentum and heat transfer. However, while the addition of nanoparticles increases the skin-friction coefficient, it reduces the local Nusselt number. The average entropy generation number increases with higher forced flow, suction, and rotational Brinkman numbers, whereas the average Bejan number exhibits a declining trend. Conversely, nanoparticles decrease the average entropy generation number and increase the average Bejan number. The tabular data reflect that the skin-friction coefficient and Nusselt number increase up to 122.6% and 9.1% at forced flow parameter 𝜆 = 1 when compared to the flow situation with no forced flow (𝜆 = 0). These quantities also show an enhancement of up to 200.3% and 966.6% at suction parameter 𝑊𝑠 = −2 when no suction is applied (𝑊𝑠 = 0). For the nanoparticle volume fraction 𝜙 = 0.2 , the skinfriction coefficient augments up to 138.3% while the Nusselt number depreciates up to -1.5% when nanoparticle volume fraction is absent (𝜙 = 0).
