A Physicochemical Property-Based Predictive Model for Enhanced Oil Recovery Efficiency

Technical Summary: A Physicochemical Property-Based Predictive Model for Enhanced Oil Recovery Efficiency

Problem Statement
Enhanced Oil Recovery (EOR) is critical for extracting remaining oil from mature reservoirs, yet predicting the efficiency of various EOR techniques remains a significant challenge. Existing predictive models frequently rely on empirical multipliers, compensatory factors, or "black-box" machine learning approaches that lack generalizability beyond their specific calibration datasets. These methods often obscure the underlying physical mechanisms, making it difficult for engineers to screen methods or optimize injection parameters without costly laboratory experiments. There is a distinct gap in the literature for a model that predicts EOR efficiency across multiple methods using only measurable physicochemical properties, without arbitrary compensatory factors.

Methodology
The paper proposes a novel mathematical model that predicts the oil recovery factor ($R$) as a multiplicative combination of dimensionless ratios, each representing a distinct physical mechanism. The model is governed by Equation (1), which integrates the following measurable inputs:

• Fluid and Rock Properties: Oil viscosity ($\mu_o$), absolute permeability ($k$), injection velocity ($v_{inj}$), and formation water salinity ($S_{sal}$).
• Thermal and Pressure Conditions: Injection temperature ($T_{inj}$), injection pressure ($P_{inj}$), and Minimum Miscibility Pressure ($P_{MM}$).
• Chemical Properties: Surfactant/polymer concentration ($C_{inj}$), Interfacial Tension (IFT, $\sigma$), latent heat of vaporization ($h_{fg}$), CO2 diffusion coefficient ($D_{CO2}$), and mass adsorption on rock ($\Gamma$).
• Chemical Affinity: A novel term based on the Hildebrand solubility parameter ($\delta$), defined as $exp[-k_{int} \cdot |\delta_{oil} - \delta_{compound}|]$, which quantifies the fundamental interaction between the injected compound and the formation oil.

The model incorporates specific physical interpretations for each term:

• Mobility Ratio: Viscosity reduction is modeled via a fractional flow theory exponent (0.25).
• Thermal Effects: Temperature dependence follows an Arrhenius-type relationship, with a reference temperature of 150°C.
• Miscibility: The pressure term captures the sharp increase in recovery when injection pressure exceeds the MMP.
• Synergy: An ASP synergy factor ($\beta_{ASP}$) accounts for the combined effects of alkali, surfactant, and polymer.
• Adsorption Loss: A term accounts for the reduction in active compound concentration due to rock adsorption.

Crucially, the model contains no arbitrary compensatory factors. The only calibrated constant is the chemical interaction constant $k_{int} = 0.15$, which represents the sensitivity of recovery to chemical affinity.

Validation and Results
The model was validated against 20 real experimental cases extracted from peer-reviewed literature, covering nine distinct EOR categories: Steam flooding, CO2 flooding, Polymer flooding, Surfactant flooding, Foam flooding, and Alkaline-Surfactant-Polymer (ASP) flooding.

• Performance Metrics: The model achieved a Mean Absolute Error (MAE) of 0.39% and a Maximum Absolute Error of 2.0%.
• Accuracy Distribution: 90% of the cases (18 out of 20) had an error of less than 1%, and 100% of the cases had an error of less than 2%.
• Key Observations:
  • The model accurately captured the strong temperature dependence in steam flooding (e.g., Liaohe field experiments).
  • It correctly predicted the monotonic increase in recovery with polymer concentration.
  • The chemical affinity term significantly improved predictions for surfactant and ASP flooding, where chemical compatibility is essential.
  • The pressure term accurately reflected the recovery jump when CO2 became miscible with oil.

Significance and Claims
The paper claims that this work provides a scientifically robust and genuinely predictive tool for EOR efficiency. Its primary significance lies in:

1. Physical Rigor: By relying exclusively on measurable physicochemical properties and avoiding arbitrary fitting parameters, the model offers a transparent view of the underlying physics governing EOR.
2. Generalizability: The model successfully predicts efficiency across diverse methods (thermal, gas, and chemical) and reservoir conditions without method-specific recalibration.
3. Chemical Affinity Integration: The introduction of the Hildebrand solubility parameter term addresses a critical gap in modeling the interaction between injected compounds and reservoir oil.
4. Practical Application: The model was applied to the Potiguar Basin in Brazil, projecting recovery factors for various methods (with ASP showing the highest potential) and demonstrating its utility for screening EOR strategies in mature fields.

Alignment with Sustainable Development Goals
The paper discusses the model's relevance to UN Sustainable Development Goals (SDGs):

• SDG 7 (Affordable and Clean Energy): By maximizing recovery from existing fields, the model helps extend the productive life of mature reservoirs, reducing the need for new exploration and drilling, thereby enhancing energy security and reducing the environmental footprint of new infrastructure.
• SDG 13 (Climate Action): The model supports the optimization of CO2-EOR operations, which simultaneously enhance oil recovery and sequester CO2 underground. The paper estimates that field-scale application could store approximately 1,000 tonnes of CO2 per km², contributing to climate change mitigation.

Limitations and Future Work
The authors acknowledge that the model assumes homogeneous reservoir properties and does not currently account for heterogeneity, fracture networks, or time-dependent effects such as polymer degradation. Future work aims to extend validation to carbonate and fractured reservoirs, incorporate time-dependent kinetics, and integrate the model with machine learning for automatic property estimation from well logs.