(195c) Advanced Model-Based Dynamics Optimization of a Commercial Natural Gas Recovery Unit

Natural gas liquids (NGLs) are the main contributor to the profitability of natural gas processing plants. During the recent period of relatively low gas prices NGLs spot prices have experienced a slower drop remaining relatively steady. Therefore, driving operating plants towards higher NGL production can lead to a more profitable position. In this work, we look to use model-based optimization on a selection of NGL plants to optimize them towards higher profitability and/or lowest energy consumption.

Such studies are meaningful only when based on sound, high-fidelity models of the plant, ensuring that “optimal” solutions obtained satisfy all important safety and operability constraints. In the typical workflow a model is built in commonly used flowsheeting tools based on the Sequential Modular approach. Limitations in this approach result in engineers having to resort to manual trial-and-error simulations with no guarantee of obtaining an optimal solution or even one that satisfies all the important constraints. In this work models are being built in an Equation Oriented environment which allows the use of formal mathematical optimization methods to guarantee an optimal solution satisfying all constraints is found.

In this paper we use the example of an on-going study of a NGL Plant in Abu Dhabi region to explain the practical benefits of Advanced Model-based Optimization. We will describe and show how the key first step in such a study is validation: ensuring that model predictions of key results are matching the actual plant behavior. We will describe how the model is then used to understand the best operational strategies for the facility through model-based optimization. This demonstrates how the approach used overcomes the limitations of the typical workflow and models used. In particular, the use of tools based on the Sequential Modular approach is limited in optimization capabilities due to its problematic handling of non-standard specifications, closing of recycle loops and the unavailability of gradients. The trial-and-error simulations used in this approach are not only time consuming and error prone but often limited in the operating decisions and constraints that can be considered. We will show how using an Equation Oriented environment overcomes these limitations and allows an optimal solution satisfying all operational and safety constraints to be found using mathematical optimization methods.

Finally we will show how this leads to the set-up of operating guidelines to allow operators to drive the plant towards optimal operation; giving them the ability to respond to changes in the plant or economic conditions quicker and with more confidence than previously possible. In addition to moving the plant towards optimal operation, the models will also provide the ability to address a range of operational situations.