Control Valve Dynamic Performance Specification
Published on : Friday 02-10-2020
Saurabh Bhardwaj explains how to get most out of Control Valves in the quest to design safe and profitable plants.
Control valves are the most important piece of instrumentation equipment in a control loop and act as final control elements to help achieve an optimum yield of desired product by tightly regulating process variable around the set point. Tighter regulation implies that process variable stays at its intended set point for majority of the plant operation. This translates to maximum productivity and less off-spec product and hence a profitable plant in operation with a good RoI (Return on Investment).
There is a common mistake made in sizing of most of the control valves. The intentions that lead to such a mistake may be good, but the results are gradually bad. In a typical control loop representation, it is comfortably assumed that a control valve will behave perfectly in following controller output signals and provided that the controller is tightly tuned, we should be able to achieve a perfect control. But in reality, we observe that such a perfect control is hardly realizable and what we get in return are the limit cycles. Limit cycles are constant amplitude oscillations of process variable around the set point and represent the inability of the controller to tightly regulate process variable around the set point. Limit cycles (Figure 1) usually happen in a control loop when any of the elements of the control loop (transmitter, controller or control valve) exhibit non-linear behaviour which make loop gain equal to 1. A control valve is found to exhibit such non-linearity in majority of control loops which is mainly contributed by dead band, stiction and the resulting total hysteresis. It is this non-linear behaviour of the control loop which is a single biggest contributor to poor control loop performance and destabilisation of process operation. The usual approach of sizing of control valve does not address the non-linearity expected from a control valve for a given application. It also does not address the variation expected in valve gain which plays a bigger role in determining tuning settings of PID controller residing inside the process control system and can be a big deciding factor on how the control loop performs.
Traditional valve sizing approach focusses mainly on selecting a valve with maximum capacity, minimal valve to system pressure drop and minimal leakage in addition to the mechanical selection in which valve style, materials and pressure rating are picked to ensure safe and reliable performance. It is usually thought that a valve with maximum capacity would deliver much more flow than required and hence bigger the valve size, better it is. What is not realised is that the process engineer has already built in a factor to make sure that there is more than enough flow in
the given maximum (e.g., 25% more than needed). Effect of non-linearities (stiction and dead band) increase when the valve size increases and with it the limit cycles, which cannot be eliminated by any amount of controller tuning unless PID controller has external reset feedback feature and control valve is provided with real time, fast and precise position feedback. A bigger valve size implies that the valve would usually throttle with a very narrow throttling range of its travel and more towards its closed position and hence reduced flow rangeability through the valve leading to poor valve and control loop performance.
The second mistake, which is usually made in control valve selection, is to choose a control valve with minimal seat leakage. Minimal seat leakage requires a tight seal between the valve closure member and valve seat and a tight packing around the valve stem (Leakage Class IV or greater). Both of these imply a dramatic increase in the non-linearities (stiction and dead band) and hence the limit cycles, which ultimately degrade the performance of the control loop, and put plant safety seriously at risk. Hence, it is very important to realise here is that a control valve can either do throttling or act as an on-off valve, but not both at the same time. Hence, if you want control valve to do its intended job of throttling, then have separate shut off valves upstream and downstream of the control valve.
The third mistake, which is often made when selecting a control valve, is minimising valve to system pressure drop. It is usually thought that a valve with minimal valve to system pressure drop ratio is good for saving energy of pumping fluid. But this turns out to be a disaster for control valve’s inherent flow characteristic. Choice of control valve inherent flow characteristic depends on whether the valve is installed in a very large piping length where majority of system pressure drop occurs across the piping and only a fraction of it occurs across the valve; or the valve is installed in a very small piping length where majority of the system pressure drop occurs across the valve and only a fraction of it occurs across the piping AND whether the process to be controlled is linear (flow, pressure, level) or non-linear (temperature, pH, composition, etc). So, the valve has to be selected in a way that it cancels out the non-linearity of the process or piping friction pressure drop (which varies with the square of the flow) and a linear installed flow characteristic is obtained with a constant valve gain (ideally!) to ensure a good valve performance with a tightly tuned controller. So, the valve drop less than 25% of the system pressure drop distorts the inherent flow characteristic of the valve where a linear flow characteristic starts behaving like a quick opening and equal percentage flow characteristic excessively flattens out near the closed position (Ref 2). Quick opening flow characteristic (Figure 2) has steep slope near the valve’s closed position, which indicates higher valve gain and this in turn is an indicator of greater non-linearity due to stiction and backlash.
Excessive flattening of the equal percentage flow characteristic near closed position imply that the slope stays zero in this region and valve response doesn’t change with the travel member movement. This translates to no flow response in this region where the flow characteristic flattens out and hence loss in sensitivity (Figure 3). (Figures 2 and 3 credit: Greg McMillan’s ISA presentation titled How to Get Most Out of Valves).
ISA TR 75.25.02 and EnTech Specification for control valve dynamic performance specification recommends valve gain between 0.5 to 2.0 %flow/%stroke (Ref 1&4) and a variation in valve gain from minimum to maximum throttling range of 10% to 90% of the valve travel for sliding
stem valve and 20 to 50 degrees for rotary valves (effect of non-linearities are greater for conventional rotary valve designs but improved designs with splined shaft to actuator connections and triple offset designs for butterfly valves reduce these effects). Valve gain outside of these limits produce undesirable performance.
Another misunderstood aspect of control valve specification is valve rangeability. Valve rangeability as depicted in most of the vendor catalogues is the rangeability of the valve inherent flow characteristics and is often thought to be ratio of maximum to minimum flow whose Cv is within specified inherent flow characteristic. What is not realised is that real valve rangeability is a function of valve stick-slip (stiction) and valve to system pressure drop ratio (Ref 3).
Selection of control valve is incomplete without appropriate type of actuator and positioner. Actuator and positioner should be selected to provide the best threshold sensitivity (signal change required before actuator/positioner starts responding). Best recommendations for actuator involve use of spring and diaphragm actuators sized for 150% of maximum thrust requirement. For case of large valve size where usual tendency is to specify piston actuators which have poor threshold sensitivity, high pressure spring and diaphragm actuators should be deployed. For positioners, always select smart, sensitive, digital positioner with high fill and exhaust rate and fast, precise, real time position feedback should be specified. With use of external reset feedback feature of PID with real time position feedback, limit cycles can be stopped (Ref 2).
With above factors in consideration, control valve specification sheet should mention dynamic performance specifications that I have addressed above (stiction, dead band and Valve gain). Additionally, it also includes Dynamic Step response specification sets the minimum step size, maximum step size, T86 response time and minimum position (Ref 1&4). Due to space limitation, I won’t be able to cover these additional aspects.
Let us realise the value of the profession and do our best to design safe and profitable plants.
References:
1. ANSI/ISA TR 75.25.02-2000 (R2010)- Control Valve Response Measurement from Step Inputs
2. Greg McMillan’s whitepaper titled Valve response: Truth or consequences How to Specify valves and positioners that don’t compromise control https://info.controlglobal.com/ebook- valve-response-truth-or-consequences
3. Greg McMillan’s Book titled Essentials of Modern Measurements and Final Elements in the Process Industry: A Guide to Design, Configuration, Installation, and Maintenance
4. EnTech Control Valve Dynamic Specification (Version 3.0 11/98) https://www.emerson.com/documents/automation/manuals-guides-control-valve-dynamic-specification-pss-en-67756.pdf
Saurabh Bhardwaj is an Instrumentation Engineer with close to 12 years of work experience in EPC companies. He has been involved in all the aspects of project execution from concept till commissioning. He is immensely influenced by work of Bela G Liptak, Greg McMillan, Greg Shinskey and Harold Wade on Instrumentation and Control and considers them as his role models. He has immense interests in developing proficiency and advancing his career as a Process Control Engineer.
