Constant composition expansion (CCE) experiments provide critical relative-volume and density information describing the thermodynamic behavior of reservoir oils and gases under varying pressure. These properties are vital inputs for hydrocarbon reservoir engineering, as they impact how oil and gas move through the reservoir during production. However, the need for specialized personnel, high-end equipment and measures taken to ensure safety in handling high pressure fluids often render the CCE experiments expensive and slow. This work introduces a Local Interpolation Method (LIM), a proximity-informed, end-to-end CCE fluid properties prediction Artificial Intelligence (AI) model that leverages domain expertise and synthetic Pressure–Volume–Temperature (PVT) data archives that mimics the actual data. The AI model generates surrogate CCE behavior for new fluids, thereby reducing the need for completing laboratory CCE measurements when sufficiently similar fluids exist in the available archive and neighborhood support is strong. Each new fluid is embedded in a compositional–thermodynamic descriptor space, and its response is inferred from a small neighborhood of thermodynamically similar fluids. Within this locality, the LIM combines hybrid local interpolation for key scalar properties (such as saturation-point quantities and expansion endpoints) with shape-preserving reconstruction of monophasic and diphasic relative-volume curves, enforcing continuity at saturation and consistency between relative volume, density and compressibility. The workflow operates purely at inference time and does not require case-specific retraining. Application to a curated archive of CCE tests shows that LIM reproduces key CCE features with very good agreement to existing data across a range of fluid types, indicating that proximity-based AI modeling can substantially reduce reliance on new CCE experiments while maintaining engineering-useful agreement for compositional simulation workflows. Under leave-one-out evaluation on 488 CCE tests, mean curve-level Mean Absolute Percentage Error (MAPE) is 0.07% for monophasic relative volume and 0.07% for monophasic density. For well-supported neighborhoods (Tiers 1–3, n = 376), mean MAPE is 0.04% for both, with 2.65% for derived compressibility and 1.78% for diphasic relative volume. The workflow is automated in software to facilitate reproducible inference on operator-owned archives and can reduce turnaround time and laboratory burden in well-supported neighborhoods. The proposed AI model uses available experimental data owned by each operator and does not use others’ data while respecting the data privacy and data ownership.
Fotias et al. (Tue,) studied this question.