Optimizing Re-Crosslinkable Preformed Particle Gel Injection Parameters Under Reservoir Uncertainty in Strongly Heterogeneous Carbonate Reservoirs

Preferential water channeling in strongly heterogeneous carbonate reservoirs causes ineffective water circulation and uneven interlayer sweep, limiting remaining-oil recovery and complicating re-crosslinkable preformed particle gel (RPPG) treatment design. A representative CMG-STARS sector model, informed by field-derived properties and the reported Mishrif Formation architecture at Halfaya, coupled size-dependent RPPG transport, retention, swelling, staged re-crosslinking, and permeability reduction with flow redistribution. Single-factor and orthogonal simulations established injection-parameter ranges. Response-surface modeling, Morris screening, and Sobol variance decomposition quantified sensitivities and interactions among eight formulation and treatment inputs. Injection concentration, rate, timing, and volume were jointly optimized using designed exploration and controlled evolution (DECE) in CMOST, followed by full-physics verification. With optimized injection settings fixed, uncertainty in 18 reservoir and RPPG-retention parameters was propagated using surrogate-assisted Monte Carlo analysis. Results showed the following. (1) All four injection parameters affected recovery factor nonlinearly. Injection volume and timing dominated, with Sobol total-effect indices of 46.1% and 32.6%, respectively. Interactions accounted for approximately 64% of the total effect of injection volume. (2) DECE optimization achieved a final recovery factor of 43.8%, exceeding the orthogonally selected scheme and baseline waterflooding by 1.6 and approximately 6.1 percentage points, respectively. (3) Under the prescribed uncertainty distributions, the probabilities of the final recovery factor equaling or exceeding 40.30%, 43.43%, and 45.58% were 90%, 50%, and 10%, respectively. The deterministic result (43.8%) was close to the median (43.43%) and lay within the main uncertainty-response range, supporting robustness under the assessed conditions. The integrated workflow supports quantitative RPPG treatment design and performance assessment under reservoir uncertainty.