Dynamic simulation is moving in leaps and bounds to cut out unnecessary process prototypes, and also to improve system performance and reliability, writes Jim Jacoby and Ian Willetts.
The integration of turbomachinery dynamic simulation control functions and hardware into a single high reliability platform (such as Tricon with robust control algorithms) has led to significant improvements in system performance and operability.
In the past, a compressor train might have required separate controllers for suction pressure, recycle and turbine or motor speed, as well as separate relay sequencing and alarm panels, a separate vibration monitoring package and a separate over speed trip protection package.
Now, only the over speed protection system must remain separate to meet the requirements of industry standards. While continual improvements are made to the system hardware and software, the success of a project is often determined during commissioning and startup.
To minimise commissioning and startup risk, we recommend a high fidelity simulation to plant operators and control system developers whenever the systems are less than straightforward. In addition to running the Tricon control program, the simulation includes part of the process plant associated with the rotating equipment train. Under the simulation environment, both steady state conditions and dynamic control responses can be observed.
Four areas of benefit can be realised: Cost savings; Time savings; Safety; and availability.
Dynamic simulation
Dynamic simulation was first used in the turbomachinery business more than 25 years ago to validate the sizing of the recycle valves on large LNG compressor trains. These studies required detailed data on the entire piping system, taking months to complete. While these studies were called ‘dynamic’, they could not be viewed in real time. They calculated the dynamic response of the system, but only provided the results as data tables, which were very costly to optimise.
As digital systems became available for turbine and compressor control, the desire to test and demonstrate these controls dynamically emerged. The first simulations generally were not much more than a tieback of the outputs to the inputs with some first-order time lags added to make the actions more authentic. These tieback simulators did not have much process reality, but did provide a reasonable technique for testing the control software and provide real-time results.
Types of simulation
In the last five years, the sophistication and capabilities of dynamic simulators has improved dramatically. These improvements are the results of innovative algorithms for combining the rigorous steady-state calculation methods with transient dynamic modelling. Also, the improved performance of personal computers now allows quicker iteration of results. Most of the new first principles dynamic simulation packages have incorporated object-based modelling, dramatically simplifying the configuration process.
Since some of the current generation dynamic simulators are based on the same first-principles calculations of the steady-state process simulators, they are able to closely match the state calculations of the steady-state simulators. Currently there are no industry standards for the definition of the fidelity of a simulator, but the following two categories are widely accepted.
Low fidelity — The tieback simulators that were developed for closed-loop testing of digital control programs are considered low fidelity. This level of simulation can be accomplished with additional hardware (I/O targets for the control system or in software with the simulation running in another controller and I/O data exchange directly to memory locations). Even though a low-fidelity dynamic simulation is useful for validating the control program and training the operators, it is not suitable for making accurate predictions on the performance of the physical equipment (such as the compressors and ancillary equipment, recycle valves or the effects of the process environment on the control program).
High fidelity — A high-fidelity simulator includes both realistic fluid and equipment dynamic calculations. The inertia of the machinery train is also included in this type of simulation to improve the reality of transient performance predictions.
High-fidelity dynamic simulators are also characterised by the use of rigorous calculations of fluid physical properties including a rich library of components and thermodynamic methods.
Sophisticated thermodynamics calculations allow realistic prediction of the system performance throughout the process model. Flash calculations provide the ability to realistically model the behavior of a two-phase process such as knockout between stages of a wet gas compressor or quench evaporation on the suction side of a refrigeration compressor.
At IPS we use the DynSim modelling package together with TRISIM Plus control stimulation package. TRISIM provides a software interface between the Dynsim model of the plant and the Tristation emulator. DynSim also executes actual Tricon code without going through a conversion or translation. This allows the engineer to test the actual control configuration file rather than an approximated emulation of the logic.
Dynamic simulation uses
In addition to using a dynamic simulator to validate a control program within a process environment, there are numerous other uses, depending on the features and fidelity of the simulation package.
Equipment sizing and selection — If the simulator is used during the plant design phase of a project, the information obtained from a dynamic simulation study can be used to determine if the size of equipment is adequate for transient conditions. Obviously it is less costly to make a design change than to find out during the startup phase that a valve, pipe or vessel requires changing.
Equipment to think about here includes: suction drums; recycle valves; valve actuators and accessories; suction throttle valves; and control program design.
Control selection & verification — Control logic validation is the first of these. Open loop testing of a closed-loop controller can demonstrate that a controller moves in the correct direction based on the direction of the error. Unfortunately very little quantitative information is available with this type of test. A closed loop simulation allows a thorough evaluation of the behaviour of the system and the associated controls in a safe and efficient manner.
Well designed simulators such as DynSim include the ability to define and store initial conditions. These initial conditions allow returning to a saved state quickly and easily to perform repeated testing with different control configurations or control settings. For example, starting the turbine may take additional time due to warm up timers or ramp rates. If the function that is being tested requires reaching normal operating speeds, an initial condition stored for the normal operating speed allows the test engineer to return to this starting condition within seconds and also execute repeated runs quickly and easily.
Other dynamic simulation uses include: control program Factory Acceptance Test (FAT); validation of control program changes; system pre-tuning; training and safety; operator training; control system engineer training; process engineer training; troubleshooting; and also all those ‘what-ifs’, which allows a process or control system engineer and operator to test out their own ideas about how to better operate the process using the dynamic simulation tool, without risk to the plant, personnel or profits.
Choosing a simulator
The fidelity of the modeling package has a direct bearing on simulator usage. With today’s software, high-fidelity simulations can be built and tested in just a matter of weeks providing an incredible return on investment considering the uses and benefits described.
A serious problem with simulators that are fully simulated is shelf life. Since they simulate the control program, any time the control program changes, the simulator must be updated. Eventually, the simulator will become out of sync with the plant controls, rendering the simulation useless. IPS’ solution with TRISIM Plus uses virtual stimulation to connect the DynSim plant model to the Tristation emulator, eliminating the need to simulate the control program.
The actual control configuration file is used together with the dynamic model of the process. This reduces the work during implementation, eliminates the source of errors in the simulation and assures that the simulator stays in sync with the actual plant and control system. This ensures an extended return on investment over time.
Conclusion
Dynamic simulation brings together controls, process and operations. The vastly improved performance of modern PCs and dynamic simulation packages over systems available only a few years ago have provided control systems designers and process engineers with an excellent set of capabilities for the evaluation, testing, implementation and modification of complex turbomachinery control programs in a process environment. Costly mistakes are avoided and commissioning and startup times are reduced. Significant cost savings are realised. In addition, these same tools can form the basis of a realistic operator training simulator.
[Jim Jacoby is the director of the turbomachinery controls portfolio and Ian Willetts is the global consulting director of Invensys Process Systems (IPS).]