The steady-state and dynamic behaviour of multicomponent packed distillation columns has been simulated using two film plug flow and back-mixing models. The model solutions have been obtained employing orthogonal collocation on finite elements. The use of Jacobi, Legendre or Hermite polynomials within the finite element procedure are compared with an approach employing the Galerkin criterion. Results from the three polynomial procedures are very similar. However, in cases where more than three collocation points are required the incorporation of cubic Hermite polynomials gave the most efficient routine in terms of CPU time on a VAX 8820 computer. The Galerkin procedure proved to be the least efficient. A number of comparisons are made with results obtained both theoretically and experimentally by previous workers. In the majority of cases the predictions of column behaviour using the finite element procedures of this work give better agreement with experiment than analytical and finite difference solutions reported by other researchers. Orthogonal collocation is seen also to give stable solutions with similar CPU times to those of the finite element procedures but it is not possible to place the grid points at desired locations. Single and two variable control of overheads and bottoms compositions is examined both for a simulated industrial scale column and for a column situated at B.P. Chemicals, Baglan Bay, South Wales. Perturbations in feed composition are utilised as the disturbance with reflux and vapour flowrates as the manipulated variables. Conventional PID feedback control is applied using previously published tuning algorithms as well as a new criterion (the 'decrease gain' procedure) based upon a method due to Yuwana and Seborg (1982).
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:637219 |
| Date | January 1993 |
| Creators | Hapoglu, H. |
| Publisher | Swansea University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.002 seconds