Simulation and data validation in multistage flash desalination plants

This paper describes the main concepts of a specific Fortran program tackling steady-state simulation and data validation (gross error detection and data reconciliation) for multistage flash desalination plants. The process simulation was carried out using an equation-oriented approach in which the decomposition of the system leads to a sensitivity matrix which is diagonal by blocks with extra diagonal blocks. The topology of this matrix is conserved according to a procedure which can be extended to an unsteady-state simulation of the process, justifying the interest of the adopted methodology. The validation of the raw process data was achieved by the implementation of a simultaneous strategy of gross error detection and data reconciliation, based on the bivariate method. In this approach we minimize an objective function using the maximum likelihood principle, which takes into account both contributions from random and gross errors, The complete mathematical model describing the process behaviour at steady state was considered as constraints in the process optimization. Results corresponding to the structural and operating conditions of the distiller N degrees 5 of Umm Al Nar East plant which was put into service in 1987/1988 in the Emirate of Abu Dhabi are presented. This package was developed at the School of Chemical Engineers in Toulouse (France), and the project was sponsored by the International Centre for Water and Energy Systems (ICWES) located in Abu Dhabi (United Arabs Emirates).