(599g) Dynamic Simulation of an Ethylene Recycle Compressor System

The Linde Engineering Division of The Linde Group plans and delivers process plants for air separation, liquefaction of natural gas, separation of gases, and for the production of olefins and synthesis gas. Compressors and turbines are key components in such plants. Therefore, a safe operation of those machines over extended periods of time has to be ensured. Additionally, transient operation associated with performance control and startup or shutdown has to be considered.

Linde Engineering uses dynamic simulations for both design and analysis of machine protection systems (MPS). The main purpose of a machine protection system is to ensure safe, i.e. non-surging operation during transient conditions such as normal stop or emergency shutdown events. Moreover, the controller design needs to be verified with respect to stability and controllability.

This contribution outlines a dynamic simulation project for the analysis of an Ethylene Recycle Compressor system and for the MPS design. For both purposes reliable equipment models are required. For steady state simulation, the machine vendor provides performance maps that describe the correlation of rotational speed, suction flow, compressor head and efficiency. However, such a performance map is valid for distinct operating conditions only. Under different operating conditions, ideally the machine vendor must provide a new set of performance maps. For steady state simulation, such an approach could be considered. However when performing dynamic simulations with suction conditions varying from integration step to integration step, such a procedure becomes rather unhandy. Hence, an algorithmic approach must be used to describe variations in suction conditions with respect to the normal operating conditions that are described in the performance map. The authors' approach to flexibly model compressor performance maps and the implementation in a proprietary process simulator is presented.

OPTISIM^® is the proprietary equation oriented process simulator used in Linde Engineering for steady-state process design and rating simulation, steady-state optimization as well as dynamic simulation and dynamic optimization. For the calculation of physical properties, the same proprietary multiphase property package is used as for all simulation applications in order to assure the highest possible level of consistency. In OPTISIM^®, a plant model is generated using unit operations from a model library. These unit operations are connected via material, energy and information streams and need to be parameterized in order to reflect a specific process unit.

Using the transformation of the performance map within the process simulator, it is shown, that dynamic simulation is the key tool to verify the topological concept for a compressor MPS. In particular, an Ethylene Recycle Compressor system providing cooling to an olefin plant is analyzed in detail. First it is shown, that the dynamic model reflects the process design reasonably well, even though different software tools have been used for process design calculation and dynamic simulation. Then, the process topology, together with the bypass valves designed by the machine vendor, are used to carry out an emergency shutdown (ESD) simulation, as such an event is both, sufficient and necessary for machine protection. During an ESD, the drive of the compressor is switched off and immediately after the detection of such an event the bypass valves are forcibly opened. If the machine starts to surge thereafter, then the MPS is inadequate. This analysis will show that a narrow view of the compressor system alone without considering the interactions with the overall process it is operating in may result in incorrect sizing of the bypass valves. A solution will be presented which offers sufficient machine protection and which is robust against slight modifications in the process topology.

Moreover, a dynamic model offers full flexibility to evaluate the control concept of the system. Several pressure control strategies will be analyzed and their expected performance under various disturbance scenarios will be compared.