Process control in the process industry

A. Sharma, Contributing Author, Jaipur, India

In this age of automation, process control in the process industries is essential. Every piece of process equipment is outfitted with automatic control to operate without human intervention and a safe shutdown in the event of an upset. The basic process control system (BPCS) and the safety instrument system (SIS) are used to configure this automation. As a result, every process engineer should be familiar enough with the process control to detect and troubleshoot any control issues. Consequently, process control engineers who understand both the process and the control system are now required.

This article will detail a fundamental process control system that is not based on the Laplace transform. Here, some features of an ideal controller are provided, as well as how process engineers should check these control systems during piping and instrumentation diagram (P&ID) review and execution.

Power and motive fluid failure. Control valves are the last control element in all industries, particularly the process industry. These control valves have a fail-safe mode that is determined by the safety of the equipment as well as the plant—this can be a fail to open, close or lock. However, the fail-safe position differs depending on whether the power or motive fluid fails. In the event of a motive fluid failure such as instrument air or control oil, the control valve will remain in the fail-safe position while power is still available. The control valve positioner will then get output based on the controller output; once the motive fluid is available, the control valve will receive the same position based on output.

When motive fluid is available but there is only a power failure—while backup power is generally available, a total blackout situation is considered here—all controllers enter into manual mode by default and the output (mill ampere) that goes from controller to positioner becomes zero (not 4 milliamps), and the control valve sees the zero milliamps and goes to a fail-safe position as a result.

The two cases appear to be the same. In both cases, the control valve enters fail-safe mode. However, the fail-safe design is exclusively dependent on motive fluid failure, and the valves operate owing to 0 milliamps upon power loss. This does not imply that the two cases are identical. As previously indicated, 95% of control valves operate similarly in both circumstances, but the remainder behave differently (fail-safe in motive fluid failure and in power failure, valves hold the same position or just opposite to motive fluid failure).

Synchronous proportional, integral, derivative (PID). The scan time of any controller is the amount of time it takes to calculate the output based on the input error. In general, the controller calculates the error (set point–process valve) and then modifies the output based on that error. This output is sent as current (milli amps) to the final control element (in the process industry, a control valve) (4 mA–20 mA). The frequency of the controller is referred to as the “scan time of the controller.” Conversely, the scan frequency of any measurement element of a measuring sensor varies depending on the process.¹

Although pressure and flow sensors are slightly faster and temperature and composition (analyzer) measuring sensors, depending on the process design, have a long reaction time and dead time. Analyzers are frequently multiplexed with a large number of streams to detect with a single gas chromatography (GC) instrument and to save money on capital. The measurement period of a single stream is extended by minutes to half an hour. Each measuring sensor has its own transfer function, as well as its own reaction and scan time.

The measured value is sent into the controller as a process value (PV), and the setpoint is determined by an operator or another controller. The controller's scan time and the response time of measuring sensors should be identical and synchronized to avoid process disturbance. If the controller is faster than the measuring element's response time, the controller calculated output can travel a long distance using old values. Some plants utilize asynchronous controllers, which compute the output based on the old PV value rather than the new PV value. The process is disrupted by this type of asynchronous controller, and the controller is run manually by the operator.

Slew rate. This is the rate at which the controller's output changes over time. Although a controller's output (MV) can be altered in milliseconds to seconds (depending on scan time), the ultimate control element (control valve) is slow and can take minutes to open. When the controller is faster than the final control valve response, the output is repeatedly transmitted to the control valve that is not opening owing to the delayed reaction.

The controller output may exceed the required opening, causing the process parameters to become unstable. In some circumstances, the controller may also reach apex value. Some control valves in plants operate in a separate fashion. When the controller's output is between 1% and 5%, it does not act; when it exceeds 6%–10%, it quickly opens. This causes the process parameters to be disrupted, forcing the controller to act in the opposite direction. To avoid such problems, the slew rate of essential controllers should be aligned with the control valve.

If the control valve is fast, the controller output should be quick as well, and vice versa. These two characteristics prevent the controller from turning on during normal operation. If the control valves are at their limits (open or completely closed) and the controller still has an error, the controller can compute output and increase until the error is zero. The controller's output can exceed this limit, causing the controller to overrange. To prevent wind-up, controllers have an anti-reset windup (ARW) feature that locks the output computation when the final element is near the limit.

The following recommendations and best practices to achieve smooth controlling can be used as check list:

1. Because the control valve fail-safe is determined by safety, the controller action should be determined by process needs [i.e., if the PV is increased, the control valve (MV) should also be increased, then use a direct controller, and vice versa.]²
2. PV-SV tracking should be enabled in every controller to avoid the bump during the switch from manual to auto.
3. The primary controller output (MV) should track the secondary controller setpoint in cascade control (when the secondary is not cascaded). The primary controller loop will be open while the secondary is not in cascade mode, displaying the initialed manual (IMAN), and PV-SV tracking should be enabled in the IMAN mode.
4. All override controller selectors should be checked based on process demand and external feedback tracking to avoid bump as well as windup of the controller that is not selected.
5. In split-range control, dead band or overlapping should be checked and one direction control (ODC) should be checked when only one valve is in operation.³
6. The interaction of two or more controllers should be checked using a decoupler, or tuned with the BLT tuning method.⁴
7. The range of the process valve (PV) and setpoint (SP) should be checked and rechecked with the instrument range.
8. Engineering units of PV, SV and MV should be checked by a process engineer.
9. The slew rate and scan time of the controller to be checked and synchronized.
10. In open loop (input open, output open or a power issue), the controller should be put in manual mode by default and should hold last good manipulated valve.
11. All alarm setpoints and hysteresis of alarm should be checked by a process engineer.
12. Generally, a PI controller should be used for easy tuning and fast action, except in some critical and slow loops.
13. A PV derivative controller should be used where proportional kick is not acceptable, such as a reactor where runaway can occur.
14. The anti-reset function limit should be checked.
15. All temperature and analyzer controllers should be PID to determine speed response and robustness.
16. To tune a controller, always do a bump test to determine the trend, then implement suitable parameters.
17. For critical controllers where process conditions change in every cycle, adaptive control should be used. This means tuning parameters automatically change based on process mode. HP

ACKNOWLEDGEMENT

This work is based on the author’s experience and learning and is not affiliated with any company. The author is grateful to Shree Krishna for support and advice during the writing of this article.

LITERATURE CITED

1. Sharma, A., “The two features of a PID controller,” International Society of Automation (ISA) blog, May 2022, online: https://blog.isa.org/the-two-features-of-a-pid-controller?_gl=1*if8ehk*_gcl_au*ODg2ODgxMzczLjE2ODYxNjk1OTY
2. Stephanopoulos, G., Chemical process control: An introduction to theory and practice, 1st Ed., PTR Prentice hall, Englewood Cliffs, New Jersey, January 1984.
3. McMillan, G. K. and M. C. Douglas, Process/industrial instruments and controls handbook, 5th Ed., McGraw-Hill, 1999.
4. Sharma, A., “Interaction and decouple control,” International Society of Automation (ISA) blog, June 2022, online: https://blog.isa.org/interaction-and-decouple-control?_gl=1*1dumdjh*_gcl_au*ODg2ODgxMzczLjE2ODYxNjk1OTY

ABHISHEK SHARMA works at an ethylene plant as a Process Engineer and is actively involved in the plant's smooth running and troubleshooting. He has more than 4 yr of experience working in the steam cracker unit and earned his B.Tech degree in chemical engineering with honors from the National Institute of Technology Raipur in India. Sharma has also finished a process control course from IIT Mumbai and trained under the American Institute of Automation and Process Control (AIAPC). He is an associate member of the IChemE and an active professional member of the American Institute of Chemical Engineers. Sharma is the author of more than seven articles, case studies and papers.