Modelling and Optimization of an Industrial Steam Network

Abstract

Steam utility systems supply process units with the required heat, rotational power, feed for steam methane reforming et cetera. This work is based on the steam network located at Shell Raffinaderij Nederland in Pernis.
The refinery has grown over the years which has resulted in an extensive and complex steam network which interconnects old and new process units. Operating the network is difficult, especially predicting the effect of changes in the network or finding the optimal operating mode where costs are minimized. The aim of this study is to create a model of the steam network which can predict these effects and able to find the optimal operating point.

The steam network is simulated in two models: one focused on hydraulic calculations and one on operational cost optimization. Both models are created in the software i-Steam from PIL, which is specialized in simulating and optimizing steam utility systems. The hydraulic model incorporates the pipelines and all steam producers and consumers connected to the network. The model is validated for four scenarios which arechosen to include varying ambient conditions and operating modes. The model is applied for testing the
effects of a new unit which must be connected to the existing steam network. The optimization model is focused on steam and electricity production and uses the current market conditions, such as the prices of natural gas, electricity, water and CO2 emission. Another factor in optimizing the system is by comparing the
steam production price with the steam price which can be taken in from the power generation plant 2 (PGP2).
The software uses GAMS as an optimization tool which applies multi integer nonlinear programming to find the operating mode with the minimal operating cost.

The pressure and temperature validation results for the hydraulic model are mostly within the 2.83% accepted uncertainty window. For certain areas the model indicates large steam surpluses over 10 t/h, which are probably due to bad flowmeters on the steam consuming side. The simulations of the new project resulted in the most promising connection to the existing steam network. For 18 barg steam consumption, the new unit should be connected to LS4413 with a backup connection for redundancy to LS980019. For 3 barg steam production,
the economic feasibility of laying 250 meters of pipeline in order to be able to export 4 t/h has to be further researched. The optimization model is validated using monthly timestamps over a two-year period. Optimization of these timestamps resulted in an average operating cost reduction of 1-3 %, with one peak at 6%.

The hydraulic model can simulate the effects of operating scenarios, retrofits and locates the bottlenecks in the steam network. The optimization model reduces the operating costs while providing each process unit with the required steam. Both models can help the employees working at the refinery to study new scenarios and advice in operating the network at its optimal point.