How to avoid control design pitfalls with dynamic simulation
Adopting multi-purpose dynamic simulation (MPDS) solutions as a design standard can help with visualization of control interactions and control pre-tuning and ease challenges in later stages of the project, including startup. Simulation is part of the Industrial Internet of Things (IIoT).
Using multi-purpose dynamic simulation (MPDS) has helped several large companies visualize control interactions, helps with controller pre-tuning, and eases startup challenges and later updates, leading to adoption of MPDS as a design standard; integrating simulation and design benefits loop tuning for better level control, valve positioning, and pump speed, among other areas. Simulation is part of the Industrial Internet of Things (IIoT). See related examples and figures. Proven in more than 800 projects, dynamic simulation models can be developed along with plant designs and can evolve through the project execution, aiding in every stage from design through operation:
1. Design: A model is initially built to test feasibility and design operability. Static simulators traditionally used during this phase provide the basic information but miss some important operational and control details.
2. Distributed control system (DCS) and programmable logic controller (PLC) configuration: Implementation of the control logic into the platform that is chosen by the client is tested with the aid of the dynamic simulator. At this point, the model has been developed to provide the controls with all of the required inputs and outputs.
3. Start-up: Once the controls configuration is validated, start-up scenarios can be tested in the dynamic simulator, without risk of lost production or increased downtime, to ensure the plant is operable. The operators can start receiving training well before the plant's construction is completed.
4. Operations: At this point the model is a mature, accurate representation of the plant and can be used as a training tool and test bed for plant modifications. The concept of MPDS can be summarized in Figure 1.
The value of dynamic simulation for controls design
Traditionally, control logic is tested through simple loopback testing, which involves linking the control loop output back to its input, usually adding some expressions to account for process lag and direction of change. The control loops are tested individually, and it becomes very difficult to develop expressions that describe the physical process on which two control loops may operate.
Dynamic simulation is often added to a project to address the limitations of simple loop testing. The control logic, already created in a PLC or DCS, is connected to a dynamic simulation model, which generates inputs for the control system and evaluates the control response. This has proven to be a cost-effective way to test controls before commissioning, usually resulting in shorter commissioning and start-up times.
Usually, three approaches are available when applying dynamic simulation to a controls checkout-tie-back, medium fidelity, and rigorous simulation. The benefits and drawbacks of each approach are shown in the table.
Controls checkout approach
Even when a rigorous model is built for checkout, adding simulation after the controls are built may be too late for some aspects of the design. Using dynamic simulation during the controls design phase can work as a proof-of-concept method to validate the control strategy, while its implementation in the control code is tested during the subsequent controls checkout phase. This way, expensive rewriting of the control code (along with correcting logic diagrams and documentation) is avoided.
Using dynamic simulation
Because of the nature of dynamic simulation, the control loops interact with each other through the model. These interactions may cause unexpected behavior, which becomes apparent to the designer while working on alternative solutions for these unforeseen problems. This concept is shown in Figure 2.
Figure 2 shows part of the process of an oilfield pumping station. Oil, water, and gas extracted from a well are separated in a vessel with a weir. Production of oil and water are measured and then combined to be pumped to a processing plant several miles away. The objective is to control levels of oil and water on either side of the weir. The oil box volume, on the right of Figure 2, is much smaller than the left side of the vessel of Figure 2, making the oil level very sensitive to overflow rate.
In the original control concept shown in Figure 2, the interface level controls the pump speed, while the oil level is controlled by the ratio of the valves. When oil level increases, valve 1 opens, and valve 2 closes. The water level would increase this way, making the interface controller increase the pump speed, finding a new stable operating point. The strategy seems adequate, but when run on a dynamic simulator, it was found that the system could not reach a steady state (see Figure 3).
The vessel starts empty, and when the interface level reaches 5 ft, the pump is started. When the oil level meets its set-point, the system becomes unstable (see Figure 4). As the valves move, the interface level changes-this changes the rate of overflow of oil, in turn affecting the oil level. This creates a continuous cycle that can't be eliminated through tuning. Individual loop testing would not have identified this issue, since the decoupled loops would work as designed. The proposed solution, tested with the dynamic simulator, is shown in Figure 5.
The output signals of both level controllers are sent to the pump. Each level controller acts on a dedicated valve. The system characteristics have not changed. The pump and valves are the same size, and the controller tuning is the same as from the previous example.
Figures 6 and 7 show that as the oil level increases, the oil valve opens and the pump is commanded to increase speed, while the interface level is controlled by its own valve and its contribution to the total pump speed. This way disturbances to the interface and overflow rate are prevented, resulting in a system that is easier to control.
Using a dynamic simulator
The aforementioned example could have been solved with many different control strategies. The key point is that the problem only surfaced when the dynamic simulator was used. With this in mind, any process that can be accurately modeled in a dynamic simulator is a candidate for such analysis. The only requirement is that the simulator has complete libraries that can model both the process and the controls and that it can solve with sufficient precision to produce an appropriate set of data. Dynamic simulation for control validation has been successfully applied to many industries including:
- Upstream oil and gas: There are many instances in which a dynamic simulator has been employed to verify a control strategy.
- Liquid natural gas (LNG): The simulator has been used to validate control strategies on compression and gasification plants and in studying control and protection methods for LNG compressors.
- Steam and utility systems: Dynamic simulation has been successfully used in testing and designing feed-forward control loops. One particularly important application is to prevent pressure sags during steam generator trips. n Mining operations: Dynamic simulation has been applied to many mining processes, from ore transportation to chemical processing.
- Refineries: Anti-surge control studies have become an industry standard for dynamic simulation. The scope of these studies has been extended to include interactions with upstream and downstream equipment; i.e., to evaluate the effect of compressor-recycle on column pressure and adapt the controls to address issues.
- Flare systems: Dynamic simulation is recognized as an adequate design method according to the American Petroleum Institute (API)'s standard 521, pressure-relieving and depressuring systems. As well, a dynamic simulator can be used to evaluate the effectiveness of high integrity protection systems (HIPS).
The added value of dynamic simulation
Control strategies may look good on paper and may pass simple loopback testing that is done during coding, but sometimes complex behavior and control interactions are only seen once the controls are tied to the process they control. A dynamic simulator can show these issues earlier in the project, and once the issues are recognized, the controls design team can work with the simulation team to develop and test a new strategy. This type of collaboration also prepares the teams to work together on the next phase-the controls checkout. Dynamic simulation should be included early in the design phase of the project in order to maximize the added value and cost-cutting opportunities it provides.
Andres Lipinski is a senior consultant of Dynamic simulation studies at Schneider Electric. Dr. Ian Willetts is the simulation and training global practice director at Schneider Electric Simsci. Edited by Emily Guenther, associate content manager, Control Engineering, CFE Media, firstname.lastname@example.org.
- How to avoid control design issues with dynamic simulation
- Understanding dynamic simulation
- The multi-purpose dynamic simulation (MPDS) workflow process.
What are the requirements to use dynamic simulation?