This article describes the most important methods and technologies that are developed at IFE in order to be able to build and use large-scale process simulators. You will find information on the application areas for process models in the related article on Process Simulation.
The Institute has a close co-operation with Kongsberg Maritime (www.km.kongsberg.com) on process modeling and simulation technology. The technology described below is used in ASSETT, which is Kongsberg Maritime’s commercial simulation tool.
What are the building blocks of a simulator
The most important computations in a simulator take place in mathematical models. Principles and laws from physics, chemistry and mechanics are used to formulate mathematical equations for the process behavior. The equations are translated into ”computer language” and supplied with mathematical solution algorithms. Then the model may be used to predict future process behavior.
A modern simulator also has an interactive, graphical user interface. In this GUI the user can assemble a total simulator by picking models from a model library. The user makes a ”drawing” of the total process where the models are connected into a process network. Each model can be given specific input parameters, and while the simulator ”runs”, it is possible to inspect the simulation results.
For the process to behave as needed, it must be controlled by a control system. This control system can be simulated just as the process, but it is also common to connect the simulator to a real control system and let it control the simulated process. This is especially the case for training simulators, and when the purpose of using a simulator is to verify the real control system.
An object oriented modeling approach
The principle used by IFE to make a total plant simulator may be denoted ”modified, modular sequential”. A module is a model of a piece of equipment or a process controller, for example a pump, and models exist for all necessary types of equipment. The same module is used repeatedly for all the pumps of the same type in a plant, but with individual parameters and data. I object oriented language the module may be called a class, and each time it is reused, we make a new class instance. The program executes all the instances in turn, in a determined sequence, and repeats this for as long as the simulator is ”running”. An important modification of this main principle is the use of a network solver, as explained later.
Thermodynamical methods for petroleum processing are mainly focused on computing the effects of changes in pressures and temperatures on material properties and phase transitions (for example the condensation of gas into liquids). IFE has developed a thermodynamic toolkit that are used by all the process models in the simulator. The toolkit has thermodynamic data as well as computational methods. Each process model may therefore be made quite independent of the thermodynamic properties of the fluids that flow in the plant.
A network of model instances
The model instances in the simulator constitute a network where fluids flow between instances. Changes in the pressure at one place in the process may quickly propagate and influence the flow at other locations in the network. In order to simulate without computational problems, the pressures and flow rates must be solved simultaneously for all the model instances. This is done by the network solver, nws, which collects information from all the instances into a total set of equations and solves them. Before the simulation is started, the network solver will do an analysis of the network and decide an optimal solution sequence, so that the network can be solved very efficiently.
A very important research area for the Institute is the development of knowledge and mathematical models of multiphase flow, a mixture of gas, oils, water, and chemicals in pipelines. See the related article on Multiphase Flow Technology. Often there is a need to simulate what happens when one or several such pipelines are connected to a processing facility, or when multiple processing plants are interconnected by such pipelines. IFE has developed specialized numerical methods and modeling techniques with a new ”multiphase network solver”, Muffin, for solving large networks with a mix of multiphase pipelines and process equipment.