Uncertainty and sensitivity analysis by Monte Carlo simulation: Recovery of trans-resveratrol from grape cane by pressurised low polarity water system


TURGUT S. S. , Feyissa A. H. , KÜÇÜKÖNER E., KARACABEY E.

JOURNAL OF FOOD ENGINEERING, vol.292, 2021 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 292
  • Publication Date: 2021
  • Doi Number: 10.1016/j.jfoodeng.2020.110366
  • Journal Name: JOURNAL OF FOOD ENGINEERING
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aquatic Science & Fisheries Abstracts (ASFA), BIOSIS, Biotechnology Research Abstracts, CAB Abstracts, Food Science & Technology Abstracts, INSPEC, Veterinary Science Database
  • Keywords: Sensitivity analysis, CFD-Monte Carlo combination, Pressurised water extraction, Grape cane, SUPERCRITICAL-FLUID EXTRACTION, SOLID-LIQUID EXTRACTION, MASS-TRANSFER, PHENOLIC-COMPOUNDS, OPTIMIZATION, MODEL, CO2, ANTIOXIDANTS, PROPAGATION, SOLVENT
  • Süleyman Demirel University Affiliated: Yes

Abstract

Mechanistic models used to describe heat, mass and momentum transfers involve output uncertainties, which arise from the assumptions made during the model development, and from the uncertainties in model input parameters. The objective of this study was to address the uncertainties and global sensitivities of the input parameters of a model which was developed for the pressurised low polarity water extraction (PLPW) system. Monte Carlo analysis with 1000 simulation was applied with a defined noise for each input parameters to visualise the uncertainty in the model predictions (total extraction time and extract concentration). Six sensitivity methods (scatter plot, standardised regression coefficient, correlation coefficient, Kruskal-Wallis test, differential analysis, and semi-variogram) were evaluated and compared to obtain the input parameters that were responsible for the output uncertainty. It was found that two (extraction solvent flow rate and particle porosity) out of the nine parameters were mainly responsible for the uncertainty. Results of the uncertainty and sensitivity analyses can be used to build reliable mechanistic models, interpret the model outputs, and prioritise future experimental efforts for PLPW system.