Key points are not available for this paper at this time.
Abstract This study benchmarks the predictive performance of commercial multiphase flow simulators for flow assurance in Brazilian pre-salt offshore production facilities (OPFs). A dataset of 350 field measurements from seven OPFs was used to compare ALFAsim and OLGA against the in-house mechanistic model Parque das Baleias (PB), under identical geometry, fluid property, and boundary condition inputs. Steady-state simulations evaluated pressure, temperature, and liquid holdup, while transient cases examined well shut-in and restart operations, including hydrate formation risk. In steady-state analysis, all simulators achieved engineering-level accuracy, with pressure predictions within ±20% of field data in over 94% of cases for ALFAsim, 99% for PB, and comparable results for OLGA. The in-house simulator (PB) exhibited the lowest mean absolute percent error (4.78%), reflecting optimized closure relationships for pre-salt conditions. Temperature predictions were consistent overall, though ALFAsim tended to underestimate values relative to OLGA and PB slightly. Liquid holdup predictions showed similar trends, with some divergence in flowline and riser sections. Transient simulations for two OPFs with BSW (basic sediment and water) greater than 20% showed strong agreement between ALFAsim and OLGA in predicting pressure, temperature, holdup, and flow rate trends. Both models reproduced phase segregation, seabed-induced cooling, and parameter stabilization within two hours after restart. Hydrate risk analysis indicated no hydrate formation during the four-hour shut-in, although longer durations could lead to crystallization. Results confirm that, when configured consistently, commercial tools can reliably simulate pre-salt flow conditions. The study underscores the value of robust PVT characterization, consistent boundary conditions, and systematic validation against field measurements for both steady-state and transient operations.
Almeida et al. (Sat,) studied this question.