Simulators use math to create virtual processes
Complex sets of equations can help designers of processes, objects, or control systems experiment and test solutions without prototypes.
Imagine these situations:
- A designer creates a model of a part for an automotive enclosure on a computer. As part of that process, the computer generates instructions that can be uploaded to a machining center to make the item without any additional programming.
- An engineer creates a new chemical processing unit for a lager plant that will make a new product from an existing feedstock. When the new unit is actually built, it performs exactly as expected.
- To analyze a piping failure, an engineer uses a combination of finite element analysis (FEA) and computational fluid dynamics (CFD) to identify material stresses and effects caused by flowing materials that caused weakening of a welded joint.
All these are possible because of simulation programs that use complex mathematical processes to create virtual representations of actual objects or processes. In the simulator, real life is reduced to mathematical relationships that can be expressed in different types of equations.
There are dozens of different kinds of simulation platforms because there are so many kinds of applications and requirements. Some are highly specialized with a high degree of fidelity and others more general. Typically, they are designed to use one particular type of mathematical function, which makes a given platform suited to a specific type of analysis.
Tony Lennon, industry marketing manager for industrial automation at MathWorks, describes Simulink as a broadly applicable platform for many types of systems. “We can model any kind of dynamical system, anything that’s described by a differential equation,” he says. “The mathematics underlying Simulink deals with differential equations. This creates a simulation environment where you can mathematically express different types of systems, so it opens many possibilities. You can do thermodynamic systems, mechanical systems, hydraulic systems, electrical systems, and you can approach a problem from a multi-domain point of view. Seldom today do you only have one kind of system.”
Other platforms are more specifically suited to particular applications, such as chemical processing. Most companies that provide large-scale DCSs also offer a simulation platform that works with the control system. For example, Honeywell Process Solutions has its UniSim Design suite that is designed to help users design and optimize process units before they’re built. “Engineers can rapidly evaluate the most profitable, reliable, and safest design,” says Rafael Coronel, global business manager, engineering effectiveness for Honeywell Process Management. “Estimates suggest that on-site design changes made during commissioning constitute 7% of the capital cost of a project. Simulation enables companies to evaluate the impact of their design decisions earlier in the project.”
Using simulators does not stop when the plant is built. A user may have to move to another type of platform to do different things, but all aspects of design and operation can enter in. “Once you design the process unit with one platform, you can start up and operate it with another, and optimize it with a third,” says Joseph McMullen, SimSci-Esscor simulation software product marketing manager at Invensys Operations Management. “We characterize it as plant lifecycle simulation. It starts in the design phase with conceptual engineering. In the operating phase we look at start-up and shut down, DCS logic, etc. Then we optimize, but since things change in a process, you eventually come back to the start again.”
Other kinds of math
While the systems discussed so far depend on ordinary differential equations, FEA and CFD platforms are built on partial differential equations. The result is an ability to make different and often more complex calculations of fluid behavior in a pipe or where stresses appear in a structural component.
However, this capability comes at a cost. The more complex the calculations, the more computing power and number crunching time is required to support the system. The same applies to model fidelity. Your ability to model a process may depend on determining how much product is going to flow through a specific valve under various conditions. Perhaps it’s enough for your purpose to know simply the flow at a given pressure. That should be a relatively simple matter. On the other hand, you may need to know a detailed picture of turbulences and a cross-sectional flow profile as the liquid moves into the next segment of pipe. These are much different pictures and will require different methods to generate each one.
Ultimately, simulation platforms all have the common objective of designing or testing something without actually building it. As Lennon summarizes it, “An ideal would be that I do everything in simulation, all the steps along the way. I then produce one test prototype, and it works, first time, exactly as we designed it. That’s the whole idea.”
Peter Welander is a content manager for Control Engineering. firstname.lastname@example.org
For more information, visit: