Fundamentals of Multi-phase Flow in Porous Media

Multi-phase flow dynamics are central to the performance of any enhanced oil recovery (EOR) operation, but they become especially critical in thermal processes where injected heat fundamentally alters fluid properties and rock-fluid interactions. In a typical reservoir, multiple phases—oil, water, and gas—occupy the pore space simultaneously. The way these phases distribute and move under varying pressure and temperature conditions determines how effectively hydrocarbons can be displaced toward production wells. Key concepts such as wettability, capillary pressure, and relative permeability govern the equilibrium and flow of each phase. When heat is introduced, viscosity decreases dramatically, interfacial tensions shift, and phase behavior can change (e.g., vaporization or compositional changes). Understanding these dynamic responses is the foundation for designing efficient thermal recovery strategies.

Wettability, the tendency of a fluid to spread on the rock surface, can be altered by temperature. For instance, many reservoirs are naturally water-wet, but heating can shift wettability toward oil-wet or mixed-wet states, affecting oil mobility. Capillary pressure curves, which describe the relationship between phase saturation and pressure difference, also change with temperature and must be incorporated into reservoir models. Relative permeability, a measure of how the presence of one phase impedes the flow of another, is perhaps the most critical input. Thermal recovery processes often cause relative permeability curves to shift, especially at the leading edge of a steam front or combustion zone. Without accurate relative permeability data at reservoir conditions, predictions of oil recovery become unreliable.

Furthermore, the injection of heated fluids introduces a new phase—steam or hot combustion gases—that interacts with the existing multiphase system. The movement of these phases is governed not only by viscous forces but also by gravity, capillary forces, and compressibility. In many thermal projects, the injected steam flows preferentially through high-permeability layers, leading to early breakthrough and poor vertical sweep. This phenomenon, known as gravity override or steam channeling, is a direct consequence of multi-phase flow dynamics and is a major challenge that must be mitigated. A thorough grasp of these fundamentals allows engineers to design injection schemes that achieve more uniform displacement and higher ultimate recovery.

Thermal Recovery Methods and Their Multi-phase Flow Challenges

Thermal recovery techniques have been deployed for decades, particularly in heavy oil and oil sands reservoirs where oil viscosity is too high for conventional recovery. Each method introduces heat in a different manner, and the resulting multi-phase flow patterns vary substantially. Below we examine the most common techniques and the flow dynamics associated with each.

Steam Flooding (Steam Drive)

In steam flooding, high-quality steam is injected continuously into the reservoir through dedicated injector wells. Steam is much less dense than the resident oil and water, so it tends to rise to the top of the formation due to gravity segregation. As steam flows through the reservoir, it transfers heat to the rock and fluids, reducing oil viscosity and causing oil to move toward the producers. The multi-phase flow behavior in a steam flood is characterized by a three‑zone structure: a steam zone near the injector, a hot water bank, and an oil bank ahead of the thermal front. Understanding the relative permeability of steam, hot water, and oil at elevated temperatures is critical for predicting steam breakthrough timing. Early breakthrough can lead to large volumes of steam channeling to producers without sweeping the intervening oil, wasting energy and reducing thermal efficiency. Engineers often use horizontal wells or injection strategies such as pattern modifications to mitigate this.

Steam-Assisted Gravity Drainage (SAGD)

Developed for oil sands and extra‑heavy oil, SAGD relies on gravity to drain heated oil downward. Two horizontal wells are drilled, one above the other. Steam is injected into the upper well, creating a steam chamber that grows upward and outward. Heated oil and condensate drain by gravity into the lower producer. The multi-phase flow within the steam chamber involves co‑current and counter‑current flow of steam, water, and oil. Steam rises and flows laterally at the chamber boundary, while mobilized oil and water flow downward. The dynamics of condensate drainage and steam trap are essential: if the producer draws down too much, steam can break through and reduce the efficiency of heat transfer. Accurate modeling requires two‑phase (or three‑phase) relative permeability data that account for temperature and pressure variations within the chamber. SAGD has proven highly successful in the Athabasca oil sands, but its performance depends strongly on reservoir heterogeneity and the presence of lean zones or shale barriers that disturb the vertical gravity drainage pattern.

In-Situ Combustion (ISC)

In‑situ combustion involves igniting a portion of the reservoir oil and injecting air or oxygen to sustain a combustion front. The heat generated reduces oil viscosity, vaporizes lighter fractions, and cracks heavy components. Multi-phase flow complexity here is extreme: a moving combustion front creates a burning zone, a coke zone, a steam bank (from vaporized water), a hot water bank, and an oil bank, all preceding the gas front. In addition to three flowing phases (oil, water, gas), the combustion process introduces chemical reactions that alter fluid compositions in real time. The morphology of the combustion front is highly sensitive to the oxygen flux, reservoir pressure, and the permeability distribution. Flow instabilities can cause the front to finger or channel, bypassing large oil volumes. Operators must carefully balance air injection rates and manage water injection to improve combustion propagation and sweep efficiency. Understanding the multi-phase flow of combustion gases (carbon dioxide, nitrogen, steam, and light hydrocarbons) is crucial for predicting front stability and oxygen utilization.

Hot Water Injection

Hot water injection is conceptually simpler but less thermally efficient than steam injection. It avoids the phase change of steam but still reduces oil viscosity through heat transfer. The multi-phase flow in hot water injection is predominantly two‑phase (oil and water) at elevated temperatures. The viscosity ratio between oil and hot water is more favorable than at cold conditions, improving displacement efficiency. However, the absence of a gas phase means that gravity override is less severe, and the overall thermal front moves slower than in steam flooding. Hot water injection is often used as a follow‑up process after steam flooding to recover residual oil or to control steam channeling. The relative permeability curves at elevated temperatures are key inputs for forecasting performance.

Electrical Heating

Electrical heating methods, such as ohmic heating or induction heating, are applied in situations where steam injection is impractical—for example, in deep thin zones or permafrost regions. Heat is generated directly in the reservoir by passing an electrical current through the formation or by using heating elements. The multi-phase flow behavior is similar to hot water injection, but the temperature distribution can be more uniform because heating occurs in situ. Electrical methods may also change the salinity and pH of formation water, affecting interfacial tensions and relative permeability. The main challenge is ensuring even heat distribution and avoiding hot spots that could vaporize water locally, introducing a steam phase. The interaction between electrically induced thermal gradients and existing fluid saturations requires sophisticated reservoir simulation that couples thermal, electrical, and fluid flow physics.

Key Mechanisms Influencing Recovery Efficiency

The success of any thermal recovery project hinges on a handful of multi-phase flow mechanisms that directly control oil displacement and sweep. Among the most important are viscous fingering, gravity segregation, steam override, and phase trapping.

Viscous Fingering

When a less viscous fluid (e.g., steam or hot water) displaces a more viscous oil, the interface becomes unstable, and fingers of the invading fluid penetrate the oil bank. This instability reduces volumetric sweep and leaves bypassed oil behind. In thermal recovery, the viscosity contrast can be enormous—steam viscosity is about 1/100th that of heavy oil at reservoir temperature. Even after heating, the mobilized oil may still be orders of magnitude more viscous than the injected phase. Understanding the conditions that trigger viscous fingering and the impact of reservoir heterogeneity is essential for designing injection rates and well patterns that minimize fingering.

Gravity Segregation and Steam Override

Density differences between fluid phases cause gravity segregation: lighter fluids rise, heavier fluids sink. In steam flooding, the low‑density steam preferentially travels along the top of the formation, overriding the oil column. This is the most common cause of poor vertical sweep in thick reservoirs. The effect is exacerbated by high vertical permeability and low injection rates. To combat override, operators may use horizontal wells, inject steam at high rates to create a more stable front, or alternate steam injection with water injection (cyclic steam stimulation). The key to predicting override is the balance between viscous forces (which tend to push steam horizontally) and gravitational forces (which drive it upward). Dimensionless numbers such as the gravity number and the steam override number are used to assess the severity.

Capillary Trapping and Relative Permeability Hysteresis

During thermal processes, oil and water can become trapped in small pore throats due to capillary forces. As saturation changes—especially during temperature cycling in cyclic steam stimulation—relative permeability can exhibit hysteresis, where the flow behavior differs depending on whether saturations are increasing or decreasing. This hysteresis can reduce the effectiveness of subsequent injection cycles. Accurate modeling of trapped saturations and hysteresis loops is necessary for reliable recovery predictions.

Phase Behavior and Component Transfers

Heating can vaporize light ends from the oil, creating a gas phase that moves separately. This gas phase can strip additional hydrocarbon components and condense in cooler zones, altering oil composition and viscosity. The interaction between thermal effects and compositional changes adds another layer of complexity to multi-phase flow. Equation‑of‑state models that account for temperature‑dependent phase equilibria are needed to simulate such behavior.

Modeling and Simulation of Multi-phase Thermal Flows

Given the complexity of the multi-physics processes involved, reservoir simulation is indispensable for evaluating thermal recovery projects. Modern numerical simulators (e.g., CMG STARS, Schlumberger’s ECLIPSE Thermal, and open‑source packages like MRST) incorporate coupled solutions for fluid flow, heat transfer, and chemical reactions. They solve conservation equations for mass, energy, and momentum in a porous medium with multiple phases, accounting for temperature‑dependent rock and fluid properties, relative permeability, capillary pressure, and phase transitions.

The main challenge in thermal simulation is the high computational demand. Because temperature fronts propagate slowly compared to pressure transients, fine grids and small timesteps are required to capture the sharp thermal gradients. Additionally, the strong nonlinearity of the system, especially near phase boundaries, demands robust numerical schemes. Recent advances in adaptive mesh refinement and parallel computing have made large‑scale thermal simulation feasible. Still, history matching thermal recovery processes remains difficult due to the scarcity of high‑temperature relative permeability and capillary pressure data. Laboratories must perform corefloods at realistic stress and temperature conditions to provide the required inputs. Errors in relative permeability can lead to order‑of‑magnitude errors in predicted oil recovery.

An ongoing area of research is the development of reduced‑order models and proxy models that can accelerate simulation without losing key physics. These models can be used in optimization workflows to find optimal injection rates, steam quality, and well placement. Additionally, data‑driven approaches are emerging: machine learning algorithms trained on simulation outputs can predict production responses for new scenarios, though they still rely on high‑fidelity physics models to generate training data.

For further reading on simulation techniques, the Society of Petroleum Engineers (SPE) provides many technical papers. One recommended resource is the OnePetro database where one can search for terms like "thermal reservoir simulation" or "multi-phase flow in steam flooding." Another authoritative source is the U.S. Department of Energy’s Office of Fossil Energy, which publishes reports on enhanced oil recovery and thermal process modeling.

Advances in Monitoring and Control of Multi-phase Flow

Real‑time monitoring of multi-phase flows in thermal recovery is essential for making operational adjustments that improve sweep efficiency and reduce costs. Traditional surveillance tools—temperature logs, pressure sensors, and production fluid sampling—provide valuable but localized data. More recent technologies offer full‑field insights. For instance, distributed temperature sensing (DTS) using fibre‑optic cables installed along wells can generate continuous temperature profiles that reveal steam breakthrough zones and hot spots. In SAGD operations, DTS data are used to infer the shape of the steam chamber and to detect subcool conditions near the producer.

Cross‑well seismic tomography can image changes in saturation and temperature between wells, providing a 3D picture of the thermal front. Tracer tests, where a chemical or radioactive compound is injected with steam and detected at producers, help quantify fluid pathways and identify channeling. With the rise of the Internet of Things (IoT), operators can now integrate these diverse data streams into real‑time decision‑making platforms. Machine learning algorithms can assimilate monitoring data and automatically adjust injection parameters to maintain optimal conditions.

An emerging field is the use of intelligent wells with inflow control valves (ICVs) that can regulate the flow from different zones. In a thermal project, ICVs can shut off layers where steam breakthrough has occurred, forcing steam to sweep less‑swept intervals. This kind of active control is a direct application of understanding multi‑phase flow dynamics: we know that steam tends to flow where resistance is lowest, so we use mechanical intervention to redistribute it. For more details on intelligent well applications in thermal recovery, see a technical paper from SPE’s Thermal Recovery topic page.

Future Directions and Research Opportunities

The drive to improve thermal recovery economics and reduce environmental footprint is pushing research into several promising areas. One is the use of additives such as foam or nanoparticles to control mobility of the injected phase. For example, steam foams can dramatically reduce steam mobility, suppressing override and improving vertical sweep. The complex multi‑phase flow of foam in porous media is still being studied, but field tests have shown encouraging results.

Another avenue is hybrid processes that combine thermal recovery with solvent injection. Processes like Expanding‑Solvent SAGD (ES‑SAGD) inject light hydrocarbons along with steam to enhance oil viscosity reduction and reduce steam‑to‑oil ratios. The multi‑phase flow in these processes involves simultaneous mass transfer of solvents between phases and complex condensation behavior. Understanding the interplay between thermal and compositional effects requires advanced simulation and experimental work.

Machine learning and artificial intelligence are being applied to predict relative permeability at high temperatures from limited experimental data, to generate complex reservoir models from seismic attributes, and to optimize injection schedules in real time. However, these data‑driven models still need to be grounded in physics to avoid extrapolating into unphysical regimes. Combining physics‑based simulation with machine learning (sometimes called “physics‑informed machine learning”) is a rapidly growing field that holds great promise for thermal recovery.

Environmental considerations are also driving innovation. Reducing greenhouse gas emissions from steam generation is a major concern. Alternative energy sources such as solar‑generated steam or downhole electrical heating can lower carbon intensity. The multi‑phase flow dynamics in these non‑conventional heating methods may differ from conventional steam injection (e.g., lower temperature steam or distributed heat sources), requiring new modeling approaches. Additionally, there is growing interest in in‑situ combustion with oxygen rather than air, which eliminates nitrogen injection and produces a nearly pure CO₂ stream that can be sequestered. The multi‑phase flow of oxygen‑rich combustion gases introduces new safety and chemical interaction aspects.

Finally, advances in high‑performance computing and experimental techniques (such as micro‑CT imaging of flow at reservoir conditions) are providing unprecedented insights into pore‑scale physics. These findings can be upscaled into reservoir‑scale models, improving their predictive power. As computational resources continue to grow, full‑field thermally reactive compositional simulations with billions of grid cells may become routine, enabling even more accurate optimization of thermal recovery projects.

Conclusion

Multi‑phase flow dynamics underpin every aspect of thermal recovery, from the initial displacement of oil by steam or hot water to the complex reaction fronts in in‑situ combustion. A robust understanding of how fluid phases interact—governed by relative permeability, capillary forces, gravity, and phase behavior—is essential for designing recovery processes that maximize hydrocarbon extraction while minimizing energy use and environmental impact. As the industry moves toward more challenging resources and stricter environmental targets, continued research into the fundamental physics of multi‑phase thermal flow, coupled with advances in monitoring, simulation, and machine learning, will be key to unlocking the full potential of thermal enhanced oil recovery. The field remains rich with opportunities for innovation, and the engineers and scientists who master these dynamics will lead the way in sustainable energy production.

For additional authoritative information on this topic, readers may consult the "Thermal Methods of Enhanced Oil Recovery" chapter in the book Principles of Enhanced Oil Recovery (ScienceDirect), or explore the latest papers in the Journal of Petroleum Science and Engineering.