Abstract
•Uncertainty quantification was performed to measure the reliability of natural gas liquefaction process.•Surrogate model using polynomial chaos expansion approach was used for sensitivity analysis.•Sobol′ sensitivity indices can be obtained directly from the surrogated gPC model.•Study helps the robust design by evaluating the bounds and reliability based on confidence levels.
The practical quantification of a model's ability to describe information is extremely important for the practical estimation of model parameters. Hence, in this study, a complex sweet natural gas refrigeration chemical process was selected for uncertainty quantification (UQ) and sensitivity analysis (SA). A computer code was generated to create a hybrid digital simulation system (HDSS) to connect two commercially important software programs, namely Matlab and Aspen Hysys. Monte Carlo (MC) and Halton based quasi-MC (QMC) methods were used for uncertainty propagation (UP) and uncertainty quantification (UQ). A surrogate model based on the polynomial chaos expansions (PCE) approach was applied for SA. Sobol′ sensitivity indices were evaluated to identify influential input parameters. The proposed PCE methodology was compared with a traditional MC based approach to illustrate its advantages in terms of computational efficiency and acceptable accuracy. The results indicated that UQ and SA help in an in-depth understanding of the chemical process determining the feasibility and improving the operation based on reliability and consumer demands. This study used in the robust design by evaluating the bounds and reliability based on confidence levels and thereby increasing the reliance of the process at hand.
[Display omitted]