"Optimization of Regulatory Control Systems"

Testing a control system utilizing the Protuner methodology is to designed to gain knowledge about the system dynamics and then put that knowledge into the DCS to optimize the control of each unit operation and its associated loops. The Protuner tuning algorithms mathematically calculate the dynamic gain and phase shift of the process transfer function at all frequencies. The Table of Tuning in the Loop Analysis Report is optimum tuning for the loop based on gain margin and closed loop damping factor requirements. The objective is to test the system to identify data that accurately defines each loop transfer function. When this is done correctly, the tuning is as close to optimum as possible. As with any newly commissioned control system, it is often necessary to make modifications as more is learned about the specific requirements of each successive unit operation and the various individual control loops.


A process control system is installed to control what the plant produces. The process control system converts raw materials and energy into usable products normally through an intricate series of processing steps. In order to produce a uniform product, which consistently meets the changing production demands, systems must be in place to produce the final product at the lowest cost while minimizing the variance throughout the entire processing cycle. The regulatory controls are the foundation of any control system. They include the control of the flows, pressures, temperatures, and other variables, that are critical to the total operation of the system. Each successive unit operation, as a function of the quality of the regulatory control, introduces variations, which accumulate and reflect back throughout the process. In most cases we find the regulatory control loops de-tuned with typical startup type tuning. De-tuned feedback controller’s integral action will eventually eliminate any error and provide stable setpoint control at steady state. De-tuning of the system’s controllers avoids troublesome oscillations (thereby also avoiding troublesome control‖). Along with de-tuning, oscillating measurement signals were also found to be excessively damped‖ to mask control design problems, interaction, and equipment problems that should not be solved by de-tuning, but should be addressed to eliminate the sources of the problems. In other words, the control system are often installed and tuned with startup parameters‖ provided an excellent man machine interface, but typically provided poor control of the process under changing load and operating conditions.


This first step in the procedure is an analysis of the system drawings and the algorithm coding used in the control system. The knowledge gained is used to design a loop testing sequence.

  • Define the systems “Unit Operations”
  • Start at the beginning of the process
  • Test and optimize the operation of the fast responding non-interactive flows and pressure loops
  • Test and optimize the inner cascade loops
  • Optimize the outer cascade loops
  • Systematic analysis and optimization of the interactive loops that effectively de-couples them
  • When testing and tuning any loop, all the other loops in the defined “Unit Operation” are also trended.


Connect the Protuner to record the inputs (process variable measurement signals) the controller output signals and the setpoint signals of the loops in the system being tested. Our connection to the control system was made utilizing the Protuner 32 Data Acquisition Program and the OPC interface driver to connect to the Yokogawa OPC Server over the plant network. The following test procedure is recommended for loop analysis testing:

1.   Inspect the installed control equipment in the system, insure that the correct anti-aliasing filters are set, and record the “as found”‖    loop tuning and setup parameters. The anti-aliasing filters in all transmitters should be set between 1.3 and 2.0 seconds, based on       the 1-second update rate of the controllers. The test data indicated that there still was excess noise on many of the loopstested.            therefore, the filtering may not have been correctly set in the transmitters.
2.   Along with trending the variables of the loop being tested, all of variables of the other loops in the defined “Unit Operation”‖ are also trended during the loop testing. Record the closed loop control at both steady state, and the closed loop response to a small setpoint change up and down. During the testing on the Jubilee, we set up the Protuner Data Acquisition to trend all the loop variables each night. The overnight trend data on each loop was examined before open loop tuning testing was started.
3.   Place the loop in manual and then record a series of small step tests in the controller output. For self-regulating processes, the typical initial test is two bumps up, three down, and one up. For integrating processes, the typical test is to bump up-down and then down-up.
4.   The test data is analyzed to determine the measurement noise, the hysteresis plus deadband of the installed equipment, the process transfer function, and a table of optimum tuning parameters. If system design, measurement, or equipment problems are identified, an action plan is developed to address and correct the problem. Once the problem is fixed, the test would then be re-run. Along with the Protuner calculated tuning parameters on the Loop Analysis Report, the tuning procedure also determines the transfer function of the loop tested. The transfer function of each loop was examined for both open loop and closed loop response in the Bode or frequency domain. In most cases, the transfer function of the loop was entered into the Protuner Simulator. The loop response was then tested off-line with the various (slow, medium, and fast) tuning to changes in both setpoint and load upsets.
5.   The new tuning parameters are entered into the controller and the loop is tested in closed loop. If any problems such as stick-slip cycling in fast loops, or integral cycling in integrating loops, are found (from the closed loop testing with optimized tuning parameters in the controller) special non-linear controller algorithms are evaluated as one of the many options to fix the problem.
6.   The installed control strategy of each unit operation is analyzed. In many cases the installed control strategy will not provide the optimum desired control. When the installed control strategy cannot be optimized, the report documents the findings and recommends changes. In the Jubilee system, the separator vessels were found to have a form of split-ranged control of overhead pressures. Following the approval of the change request, the overhead pressure controls were reconfigured to better match the way the operators actually run the system.