ONLINE OPTIMIZATION OF AN ETHYLENE PLANT

A simulation model has been developed for the petrochemical complex of OMV Deutschland GmbH, Burghausen (Germany) as part of the Advanced Plant Management System (APMS(R)). The scope of the work was to make a model that matched plant data with sufficient accuracy to benefit from reconciled plant data and optimized setpoints. This paper describes the plant model, the development phase, the tuning of the model and the on-line and off-line performance. The model of the petrochemical complex consists of the refinery unit, the ethylene plant and downstream treatment units including their interactions. The emphasis in modelling was put on the ethylene plant, composed of ten furnaces with five different geometries and a complex feedstock system. The furnaces are modelled rigorously with the use of the programs SPYRO(R), FIREBOX(R), CONVEC and TES(R). Models related to other units, e.g. coker, HDS and Pyrotol are based upon empirical relations. The equation oriented flowsheeting program TISFLO(R) is used to ensure fast solutions of the large and complex problem. In total, the plant model consists of more than 5000 linear and non-linear equations. Final model tuning was done on-site in periods of stable plant operation. The predictive power of the model has been confirmed by the reconciliation of over 450 plant data. In optimization a total of 106 constraints and 37 decision variables are present. The results of optimization show that it is beneficial to modify prevailing operating conditions. The improvements on profit registered in the past are between 1 and 3%.