Select and Size Control Valves Properly to Save Money
This is the second of a two-part article on control valves. Part one appeared in the September 1999issue of Control Engineering.There is a saying, "When momma ain't happy, nobody's happy." A parallel could be said about control valves, "When control valves aren't working, the whole loop's not working.
This is the second of a two-part article on control valves. Part one appeared in the September 1999issue of Control Engineering.
There is a saying, 'When momma ain't happy, nobody's happy.' A parallel could be said about control valves, 'When control valves aren't working, the whole loop's not working.'
An analysis of 31 control valves in a pulp and paper mill several years ago revealed 70% of the valves had significant operating problems (see Positioners cause most control valve problems diagram). Similar analysis in other industries has produced similar results. Even though control valves may be the single biggest contributor to control loop cycling, even though numerous articles have been written about the importance of sizing, selecting, and maintaining control valves, as a group control valves don't receive the respect they deserve.
Respect at last
While providing process improvement services to the pulp and paper industry Bill Bialkowski, president of EnTech Controls (Toronto, Ontario, Canada), recognized control valve dynamics depend on many factors including: valve design and size, actuator design and size, positioner performance, process conditions, and test methods. When all factors are considered control valves turn out to be complex systems that require careful engineering.
Mr. Bialkowski explains the problem with many control valves, 'Controllers expect the control valve to move and actually change the fluid flow right away for output changes as small as 0.1%. Unfortunately control valves have difficulty reversing direction and/or responding to small input changes as they tend to stick in their last position. Even under ideal conditions it takes time to convert a change in the milliamp signal to a pneumatic signal, build pressure on (or release pressure from) the actuator diaphragm, and obtain control valve movement. It turns out, control valves often respond in a fraction of a second to large input changes but, because of friction and mechanical clearances, the same valve takes 30 or 40 seconds to respond to a single 0.2% change. In the real world controllers are not 'patient enough' to wait that long. When the controller's initial small change was not satisfied additional control algorithm calculations produced additional milliamp output changes. By now the valve is moving but it's going to go too far. The controller views this as overshoot and begins new calculations to reverse the valve's direction. The result is a control valve induced limit cycle.' (See Avoid control valve induced limit cycle diagram).
Mr. Bialkowski adds, 'The problem becomes more complex when control valve sizing is considered. Many installed control valves are oversized to ensure more than adequate capacity. This results in a high process gain and thus small input changes to the control valve produce big flow changes.'
In 1992 EnTech released its Control Valve Dynamic Specification and almost immediately valve manufacturers began accessing their products' ability to comply with this new performance standard. Today, many control valves meet or exceed the performance standards defined in the original EnTech control valve spec. ISA (Research Triangle Park, N.C.) recognized the significance of the EnTech specification and convened the SP75.25 Control Valve Dynamic Testing committee to write a standard that promotes uniform specifying, testing, and reporting of control valve dynamic performance (see Online Extra box).
ISA's SP75.25 committee is being careful to define the standard on how uniform testing is to be conducted and reported. ISA is avoiding establishing performance criteria, that is being left for each manufacture to determine and report. What users will eventually see in manufacturer's literature is a consistent set of installed control valve performance characteristics.
Why users should be interested in the installed characteristics and control range is explained by Dennis Beckman, chemical industry performance consultant of Fisher Controls (Marshalltown, Ia.), 'The style and size of the valve has a significant affect on 'loop process gain' and thus control loop performance. Prior to EnTech's specification, valve selection typically considered maximum flow rates, rangeability (ratio of maximum to minimum controllable flow rates), and inherent valve characteristics. These methods fail to recognize the installed characteristics of the process and the control loop components (see CE , Feb. '99, p. 77). Since the valve provides a variable gain it is important to size and select a control valve that is sufficiently linear to stay within specified gain limits over the operating range of the system. If too much gain variation occurs in the control valve, the controller has less flexibility to control the process.'
An installed loop process gain (percent transmitter span divided by percent controller output) of 1.0 is the most desirable but good dynamic performance can be achieved if the installed loop gain is between 0.5 and 2.0 (see Installed flow characteristics influence rangeability diagram).
In terms of valve selection, butterfly valves have the smallest control range and globe valves have the largest. Eccentric plug and segmented ball valves fit in between.
Selection of the valve style is only part of assembling a well performing control valve 'system.' Attention to detail must be applied to each system component, especially positioners (see Positioners cause most control valve problems diagram).
When properly applied new positioners added to existing control valves can help reduce process variability.
Positioner rules-of-thumb guidelines evolved over the years, but recent improvements in positioner designs are changing many of these rules (see Online Extra box).
Modern positioners use a variety of internal control and feedback mechanisms to provide adjustable gain and make a positioner 'ready to jump' on small input changes.
Positioner manufacturers recognize the contribution of positioners and have on-going efforts to further improve the 'variable time to respond' problem described earlier. A promising technique is to use a positioner's digital intelligence to manage an adaptive gain algorithm and provide a consistent gain response regardless of the size of change to the input signal.
One form of insanity believes different results will be achieved by continuing to do things the same way.
Efforts by ISA's SP75.25 committee and the control valve manufacturers are useless if users don't take an active interest and insist that new control valves comply with the standard. Users should also take steps to make existing control valves come as close to compliance as possible.
Users serious about augmenting control valve performance to improve product quality need to do four things:
Acquire knowledge on how to benchmark existing control valve performance;
Ensure persons responsible for purchasing, engineering, and maintaining control valves understand the importance of understanding control valves as systems;
Ensure control loops are properly tuned; and
Regularly revisit the above three steps and ensure all five control loop elements are behaving as expected (see CE , Feb. '99, p. 77).
Users who give existing control valves some attention and respect, can expect to see their efforts pay big rewards.
Control Valve Characteristics Summary
Here are the relative differences in three commonly used control valve designs.
High-performance butterfly valve
One, usually equal percentage
One, usually linear
High flowing pressure drops
Usable control range
Source: Control Engineering with data from Fisher Controls
Understanding actuator and control valve bench settings
Control valves are often stroked for operational verification before and/or after they are installed, but before operational process conditions are present. (Similar to jogging a motor to verify wiring operation and rotational direction.) When persons conducting the stroke verification are unfamiliar with the term and purpose of 'bench settings,' undesired and unnecessary field adjustments often result.
When sizing and selecting a spring and diaphragm actuator for a control valve considerations include:
Operating pressures at the inlet and outlet of the valve;
Operating and shutoff pressure drops across the valve;
Friction introduced by the packing material;
Loading pressure changes on the actuator diaphragm;
Effective area of the actuator diaphragm;
Stroke and travel of the valve plug;
Push down to open or close characteristics;
Secondary forces acting on the valve plug; and
Characteristics of the spring.
In the absence of actual upward, downward, and side forces exerted by the process, calculations are made that permit assembling and adjusting the control valve on an assembly bench so it works properly when installed and operating.
Therefore, under conditions of zero pressure in the valve body and zero pressure differential across the valve's trim parts, a control valve may stroke from 0-100% with much different pressures applied to the diaphragm than the normal 3-15 psig (0.2-1.0 bar). For example, when tested on an assembly bench, a control valve may not begin to move until 4.7 psig is applied to the diaphragm. A different control valve may reach 100% travel with just 12 psig pressure applied to the diaphragm. Neither example necessarily indicates an improperly adjusted control valve and actuator, in fact the opposite is more likely to be true. A control valve that strokes from 0-100% with exactly 3-15 psig when setting on a test bench may not perform as required when installed and subjected to actual operating forces.
Mr. (or Ms.) Process Engineer, do yourself a favor. Be sure your installation and maintenance contractors understand bench setting, and before they make field adjustments to 'fix' a control valve so it stokes correctly, have them talk to you.
Control Valve Dynamic Specification
EnTech Control's Control Valve Dynamic Specification is a 20-page specification with three sections: nonlinear, dynamic step response, and valve sizing. Each section includes lengthy explanations, supporting graphics, recommendations, default values, and space for user data. A glossary of terms is located at the end.
The nonlinear section establishes the maximum allowed valve tracking nonlinearities allowed for deadband, step resolution, and total hysteresis.
Control valve tracking nonlinearities represent the inability of the valve system (valve, actuator, and positioner) to faithfully track changes in the input signal and ensure that changes in flow coefficients (Cv) actually occur.
Ideally, the valve should track input changes with a travel gain of one, however mechanical clearances, and dynamic and static friction make it difficult for large valves to smoothly respond to small changes in input. Failing to recognize and address valve tracking nonlinearities as part of a control valve specification can result in a control valve that jerks and yanks its way from position to position.
Dynamic step response section establishes the minimum and maximum step range that produces consistent dynamics.
Minimum step size is the lower range limit and depends on total hysteresis and magnitude of inconsistent movements. Minimum step size is valve design dependent and is frequently double the total hysteresis. Minimum step size values are stated for nominal, fine, and very fine step changes. Valves capable of changing flow rates with very fine step changes are usually more expensive.
Maximum step size values of nominal, wide, and very wide are the upper range limits over which a valve is nearly linear and depends on velocity limited movements. The wider the upper range limit, the more capable the valve design.
Step response specification establishes the amount of time, after a change in input signal, the output reaches 86.5% of the final steady state value.
The valve sizing section defines the maximum allowed flow limit cycle amplitude as a percent of nominal flow. Nominal, fine, and very fine values are expressed for the percent of: minimum step size, flow gain, and flow limit cycle.
Limit cycle amplitude can be predicted as one half the minimum step size times the flow gain.
Flow gain is calculated by dividing flow units (i.e., gpm) by percent of valve travel and dividing the results by the percent flow rate at the expected operating point.
A variation in the process gain portion of the valve sizing section helps establish the degree of difficulty expected in tuning the control loop; a useful bit of information when selecting an appropriate control algorithm.
For more information about EnTech's Control Valve Dynamic Specification, visit
ISA's SP75.25 Control Valve Dynamic Testing Subcommittee Update
Feb. '99 'Turn Problem Loops Into Performing Loops' 'Adaptive controller tracks process changes'
Mar. '99 'Optimize Existing Processes to Achieve Six Sigma Capability'
May '99 'Assessing Control Loop Performance' 'Open-Loop Response Testing Improves Process'
Sep. '99 'Control Valves: Match Size with Application'