February 2025
Special Focus: Digital Technologies
Strategies to resolve process variability via regulatory process control
This article discusses strategies to address process instability for effective process control.
Process optimization without process stabilization often yields limited financial results. This is because the true optimum typically lies closer to process limits, and instabilities prevent operations near these boundaries, leading to lost utilization of asset potential. Therefore, stabilizing the process using process control is essential for complete process optimization. However, many plants face challenges resolving these instabilities even after investing in best-in-class process control software and technologies. This article discusses strategies to address process instability for effective process control.
The role of process control in stability. The primary purpose of process control is to keep the process at the desired setpoints by minimizing variability, implementing effective disturbance rejection and stabilizing the process back to setpoints after disruptions. Here, common reasons for process instabilities are explored and actionable strategies and suggested to resolve them via regulatory control.
Proportional integral derivative (PID) control in manual mode. PID controllers are critical to maintain stable setpoints. When a PID controller is in manual mode, it cannot actively reject disturbances, leading to process variability.
FIG. 1 shows the difference in process variability before and after taking a controller onstream in automatic mode. It shows how the PID controller minimized process variability after being placed in auto mode. As can be seen, the PID controller rejected disturbances seen in manual mode and thereby kept the process at a constant setpoint.
FIG. 1. The PID controller keeps the process stable when placed in automatic mode.
It is therefore necessary for a process plant to note the number of PID controllers in auto mode as a key performance indicator (KPI). If the PID controllers are not in auto mode, it may be due to malfunctions. The following steps can be used to resolve this situation:
- Stoke check the actuators to ensure they follow the given output target.
- Check the controller’s configuration in the distributed control system (DCS) and ensure that the correct control algorithm is used (Type B is common), as suitable for the application.
- Input the initial tuning PID parameters. To estimate the initial tuning parameters, use either a closed-loop tuning tool (that uses past data to identify process dynamics and) to obtain PID parameters or a traditional step test method.
- Ensure setpoint tracking in case cascade control is enabled for bumpless transfer in auto mode.
Instability originating from a PID controller. If a PID controller exists but its PID values are input randomly or without proper understanding (i.e., poorly tuned), it will become the source of process variability. However, if a specific controller is showing instability, then it does not necessarily mean that its tuning parameters need adjustment. Quite often, a PID controller may be interacting with a continuous disturbance from another controller that requires tuning: this is also known as interaction. In such a case, the controller in this example is a victim and not the source. FIG. 2 shows multiple PID controllers that were re-tuned and stabilized. To resolve this situation:
- To identify if the PID controller is the root cause of instability, switch it to manual mode. If fluctuations subside, the PID needs re-tuning.
- If switching to manual mode does not resolve the process instability, then it is either interacting or it may have a sensor or actuator problem. Tuning this single controller often stabilizes multiple interacting controllers.
FIG. 2. PID controllers stabilized after re-tuning.
Instability originating from sensors. High transmitter noise can mimic disturbances, causing the PID controller to react unnecessarily—i.e., the sensor noise causes the PID controller to interpret the variation as a disturbance, impeling the PID controller to make continuous attempts to reject a disturbance that does not exist. FIG. 3 shows the reduction in the process variable via a sensor filter.
FIG. 3. Using a signal filter reduced transmitter noise, which reduced the process variability.
To resolve this situation:
- As a rule of thumb, keep the sensor noise below 2% of the transmitter’s range
- If the transmitter is noisy, use the signal filter in the DCS to reduce it. Noise signal filters can be applied in 0.1-sec increments
- To distinguish between sensor noise and real disturbances, either apply the signal filter or switch the PID controller to manual mode to see if the instability subsides. Instability due to noise does not subside in manual mode.
Instability from actuators. Actuators are final control elements that impact processes. The output from the PID controller is sent to the control value as a target valve position. If these control elements do not follow the PID controller’s output or its defined valve position every second, this will lead to process instability. The control value position is governed by its positioner. Modern valves have inbuilt PID controllers to control the valve position. If the tuning of the valve positioner PID control is not done correctly, it will lead to a continuous cycling of valve positions. This can be resolved by the following:
- Modern valves often include internal PID positioners that require separate tuning.
- Proper tuning of the valve positioner can stabilize the valve and reduce instability in the associated PID controllers. This cascades stability across downstream interacting controllers.
FIG. 4 shows a modern valve that has a PID controller within it to control position. The valve positioner controller was unable to meet its target position prior to tuning the positioner. However, after tuning the valve position, it stabilized at its setpoint. FIG. 5 shows the resolution in the flow stability by tuning the control valve positioner.
FIG. 4. Tuning the valve position control inside a valve.
FIG. 5. Stabilization of process parameters by actuator positioner control tuning.
Instability originating from Interactions. When two PID controllers in the same layout are aggressively tuned, then both try to meet their setpoint, leading to cycle disturbances and interaction between them. For example, a disturbance in one PID controller can trigger a rejection response that impacts other controllers, leading to plant-wide oscillations. This can be resolved by the following:
- One strategy used is to detune some controllers to reduce interactions—sacrifice control precision where business performance allows [e.g., an aggressively tuned level controller on a feed tank, destabilized downstream distillation column PID controllers (FIG. 6)]. Detuning the level controller can stabilize the entire column.
- Identify the root cause and break the interaction cycle by briefly switching the interacting controllers to manual mode one at a time. Typically, controllers managing the largest flow impact others the most. If switching to the PID controller to manual mode results in stability in all the controllers, then the root cause has been found. Next, tune this controller, then move downstream to each controller in the sequence and switch to auto mode, thus resolving any instability. For example, the level controller on a feed tank was aggressively tuned to maintain its target level so product would not flow out of the tank. However, the flow feeds to the distillation tower. This led to a fluctuation in the column, causing various process problems related to reliability issues. Simply detuning the tank level controller stabilized the entire column. No matter how much tuning was done on the distillation column, it would not have resolved the instability until the source tuning was resolved.
FIG. 6. Detuning the feed tank level’s control led to stabilization in the downstream process controllers.
Instability originating from the APC controller. APC is supervisory control that sits on top of PID controllers and regulates their setpoints in a multi-variable process. If an APC is not tuned, designed or its controller variables (which are PID controllers) are not tuned well, this would also lead to fluctuations. This can be resolved by the following:
- Ensure accurate APC modeling, including proper inferential variable selection and step testing.
- Address model mismatches and tune APC parameters, such as setting realistic upper and lower limits for controlled variables. Note: An APC is free to move the controlled variables within upper and lower bounds. If those cases are set broadly, it will seem to be causing large variations since it is allowed to do so.
- Always verify that PID controllers under APC supervision are well-tuned. A poorly tuned PID controller will propagate fluctuations through the APC.
Takeaway. Resolving process instabilities require a systematic approach that addresses each control layer, from supervisory APC systems to the final actuator. Each layer must be correctly tuned and configured to achieve overall stability, process optimization and profitability. There are no shortcuts; every problem must be addressed at its source. Additional references on how APC can resolve process instabilities are available in the articles “The process control journey: Primary process control—Part 1 and Part 2,” published in previous issues of Hydrocarbon Processing.
Comments