In today’s dynamic business environment, manufacturers are under extreme economic pressures. Market globalization is resulting in intense pressures to reduce manufacturing costs to compete with lower wages and raw material costs of emerging countries. Competition exists between international companies to provide the highest quality products and to maximize plant throughputs with fewer resources, although meeting ever changing customer needs.
These marketing challenges must be met although fully complying with public and regulatory policies. Process Variability To deliver acceptable returns to their shareholders, international industry leaders are realizing they must reduce raw material and scrap costs while increasing productivity. Reducing process variability in the manufacturing processes through the application of process control technology is recognized as an effective method to improve financial returns and meet global competitive pressures.
The basic objective of a company is to make a profit through the production of a quality product. A quality product conforms to a set of specifications. Any deviation from the established specification means lost profit due to excessive material use, reprocessing costs, or wasted product. Thus, a large financial impact is obtained through improving process control. Reducing process variability through better process control allows optimization of the process and the production of products right the first time. The non-uniformity inherent in the raw materials and processes of production are common causes of variation that produce a variation of the process point. A process that is in control, with only the common causes of variation present, typically follows a bellshaped normal distribution (figure 2-1). A statistically derived band of values on this distribution, called the +/-2 sigma band, describes the spread of process variable deviations from the set point.
This band is the variability of the process. It is a measure of how tightly the process is being controlled. Process Variability (see definition in Chapter 1) is a precise measure of tightness of control and is expressed as a percentage of the set point. If a product must meet a certain lower- limit specification, for example, the set point needs to be established at a 2 sigma value above this lower limit. Doing so will ensure that all the product produced at values to the right of the lower limit will meet the quality specification. The problem, however, is that money and resources are being wasted by making a large percentage of the product to a level much greater than required by the specification (see upper distribution in figure 2-1).
The most desirable solution is to reduce the spread of the deviation about the set point by going to a control valve that can produce a smaller sigma (see lower distribution in figure 2-1). Reducing process variability is a key to achieving business goals. Most companies realize this, and it is not uncommon for them to spend hundreds of thousands of dollars on instrumentation to address the problem of process variability reduction. Unfortunately, the control valve is often overlooked in this effort because its impact on dynamic performance is not realized. Extensive studies of control loops indicate as many as 80% of the loops did not do an adequate job of reducing process variability. Furthermore, the control valve was found to be a major contributor to this problem for a variety of reasons. To verify performance, manufacturers must test their products under dynamic process conditions.
These are typically performed in a flow lab in actual closed-loop control (figure 2-2). Evaluating control valve assemblies under closed-loop conditions provides the only true measure of variability performance. Closed-loop performance data proves significant reductions in pro cess variability can be achieved by choosing the right control valve for the application.
The ability of control valves to reduce process variability depends upon many factors. More than one isolated parameter must be considered. Research within the industry has found the particular design features of the final control element, including the valve, actuator, and positioner, are very important in achieving good process control under dynamic conditions. Most importantly, the control valve assembly must be optimized or developed as a unit.
Valve components not designed as a complete assembly typically do not yield the best dynamic performance. Some of the most important design considerations include: Dead band Actuator/positioner design Valve response time Valve type and sizing Each of these design features will be considered in this chapter to provide insight into what constitutes a superior valve design. Dead Band Dead band is a major contributor to excess process variability, and control valve assemblies can be a primary source of dead band in an instrumentation loop due to a variety of causes such as friction, backlash, shaft windup, relay or spool valve dead zone, etc..
Dead band is a general phenomenon where a range or band of controller output (CO) values fails to produce a change in the measured process variable (PV) when the input signal reverses direction. (See definitions of these terms in Chapter 1.) When a load disturbance occurs, the process variable (PV) deviates from the set point. This deviation initiates a corrective action through the controller and an initial change in controller output can produce no corresponding corrective change in the process variable. Only when the controller output has changed enough to progress through the dead band does a corresponding change in the process variable occur. Any time the controller output reverses direction, the controller signal must pass through the dead band before any corrective change in the process variable will occur.
The presence of dead band in the process ensures the process variable deviation from the set point will have to increase until it is big enough to get through the dead band. Only then can a corrective action occur. Dead band has many causes, but friction and backlash in the control valve, along with shaft wind-up in rotary valves, and relay dead zone are some of the more common forms. Because most control actions for regulatory control consist of small changes (1% or less), a control valve with excessive dead band might not even respond to many of these small changes. A wellengineered valve should respond to signals of 1% or less to provide effective reduction in process variability. However, it is not uncommon for some valves to exhibit dead band as great as 5% or more. In a recent plant audit, 30% of the valves had dead bands in excess of 4%. Over 65% of the loops audited had dead bands greater than 2%.
Figure 2-3 shows just how dramatic the combined effects of dead band can be. This diagram represents an open-loop test of three different control valves under normal process conditions. The valves are subjected to a series of step inputs which range from 0.5% to 10%. Step tests under flowing conditions such as these are essential because they allow the performance of the entire valve assembly to be evaluated, rather than just the valve actuator assembly as would be the Figure 2–3. Effect of Dead Band on Valve Performance A7154 / IL case under most bench test conditions. Some performance tests on a valve assembly compare only the actuator stem travel versus the input signal. This is misleading because it ignores the performance of the valve itself. It is critical to measure dynamic performance of a valve under flowing conditions so the change in process variable can be compared to the change in valve assembly input signal. It matters little if only the valve stem changes in response to a change in valve input because if there is no corresponding change in the controlled variable, there will be no correction to the process variable.
In all three valve tests (figure 2-3), the actuator stem motion changes fairly faithfully in response to the input signal changes. On the other hand, there is a dramatic difference in each of these valve’s ability to change the flow in response to an input signal change. For Valve A the process variable (flow rate) responds well to input signals as low as 0.5. Valve B requires input sig begins responding faithfully to each of the input signal steps. Valve C is considerably worse, requiring signal changes as great as 10% before it begins to respond faithfully to each of the input signal steps. The ability of either Valve B or C to improve process variability is very poor. Friction is a major cause of dead band in control valves. Rotary valves are often very susceptible to friction caused by the high seat loads required to obtain shut-off with some seal designs. Because of the high seal friction and poor drive train stiffness, the valve shaft winds up and does not translate motion to the control element. As a result, an improperly designed rotary valve can exhibit significant dead band that clearly has a detrimental effect on process variability. Manufacturers usually lubricate rotary valve seals during manufacture, but after only a few hundred cycles this lubrication wears off. In addition, pressure- induced loads also cause seal wear. As a result, the valve friction can increase by 400% or more for some valve designs. This illustrates the misleading performance conclusions that can result from evaluating products using bench type data before the torque has stabilized.
Valves B and C (figure 2-3) show the devastating effect these higher friction torque factors can have on a valve’s performance. Packing friction is the primary source of friction in sliding stem valves. In these types of valves, the measured friction can vary significantly between valve styles and packing arrangements. Actuator style also has a profound impact on control valve assembly friction. Generally, spring-and-diaphragm actuators contribute less friction to the control valve assembly than piston actuators. An additional advantage of spring-and-diaphragm actuators is that their frictional characteristics are more uniform with age.
Piston actuator friction probably will increase significantly with use as guide surfaces and the O-rings wear, lubrication fails, and the elastomer degrades. Thus, to ensure continued good performance, maintenance is required more often for piston actuators than for springand- diaphragm actuators. If that maintenance is not performed, process variability can suffer dramatically without the operator’s knowledge. Backlash (see definition in Chapter 1) is the name given to slack, or looseness of a mechanical connection. This slack results in a discontinuity of motion when the device changes direction. Backlash commonly occurs in gear drives of various configurations. Rack-and-pinion actuators are particularly prone to dead band due to backlash. Some valve shaft connections also exhibit dead band effects.
Spline connections generally have much less dead band than keyed shafts or double-D designs. While friction can be reduced significantly through good valve design, it is a difficult phenomenon to eliminate entirely. A well-engineered control valve should be able to virtually eliminate dead band due to backlash and shaft wind-up. For best performance in reducing process variability, the total dead band for the entire valve assembly should be 1% or less. Ideally, it should be as low as 0.25%. Actuator-Positioner Design Actuator and positioner design must be considered together. The combination of these two pieces of equipment greatly affects the static performance (dead band), as well as the dynamic response of the control valve assembly and the overall air consumption of the valve instrumentation.
Positioners are used with the majority of control valve applications specified today. Positioners allow for precise positioning accuracy and faster response to process upsets when used with a conventional digital control system. With the increasing emphasis upon economic performance of process control, positioners should be considered for every valve application where process optimization is important. The most important characteristic of a good positioner for process variability reduction is that it be a high gain device. Positioner gain is composed of two parts: the static gain and the dynamic gain. Static gain is related to the sensitivity of the device to the detection of small (0.125% or less) changes of the input signal. Unless the device is sensitive to these small signal changes, it cannot respond to minor upsets in the process variable.
This high static gain of the positioner is obtained through a preamplifier, similar in function to the preamplifier contained in high fidelity sound systems. In many pneumatic positioners, a nozzle-flapper or similar device serves as this high static gain preamplifier. Once a change in the process variable has been detected by the high static gain positioner preamplifier, the positioner must then be capable of making the valve closure member move rapidly to provide a timely corrective action to the process variable. This requires much power to make the actuator and valve assembly move quickly to a new position. In other words, the positioner must rapidly supply a large volume of air to the actuator to make it respond promptly. The ability to do this comes from the high dynamic gain of the positioner. Although the positioner preamplifier can have high static gain, it typically has little ability to supply the power needed. Thus, the preamplifier function must be supplemented by a high dynamic gain power amplifier that supplies the required air flow as rapidly as needed. This power amplifier function is typically provided by a relay or a spool valve. Spool valve positioners are relatively popular because of their simplicity.
Unfortunately, many spool valve positioners achieve this simplicity by omitting the high gain preamplifier from the design. The input stage of these positioners is often a low static gain transducer module that changes the input signal (electric or pneumatic) into movement of the spool valve, but this type of device generally has low sensitivity to small signal changes. The result is increased dead time and overall response time of the control valve assembly. Some manufacturers attempt to compensate for the lower performance of these devices by using spool valves with enlarged ports and reduced overlap of the ports. This increases the dynamic power gain of the device, which helps performance to some extent if it is well matched to the actuator, but it also dramatically increases the air consumption of these high gain spool valves.
Many high gain spool valve positioners have static instrument air consumption five times greater than typical high performance two-stage positioners. Typical two-stage positioners use pneumatic relays at the power amplifier stage. Relays are preferred because they can provide high power gain that gives excellent dynamic performance with minimal steady-state air consumption. In addition, they are less subject to fluid contamination. Positioner designs are changing dramatically, with microprocessor devices becoming increasingly popular (see Chapter 4). These microprocessorbased positioners provide dynamic performance equal to the best conventional two-stage pneumatic positioners. They also provide valve monitoring and diagnostic capabilities to help ensure that initial good performance does not degrade with use. In summary, high-performance positioners with both high static and dynamic gain provide the best overall process variability performance for any given valve assembly. Valve Response Time For optimum control of many processes, it is important that the valve reach a specific position quickly. A quick response to small signal changes (1% or less) is one of the most important factors in providing optimum process control. In automatic, regulatory control, the bulk of the signal changes received from the controller are for small changes in position. If a control valve assembly can quickly respond to these small changes, process variability will be improved. Valve response time is measured by a parameter called T63 (Tee-63); (see definitions in Chapter 1). T63 is the time measured from initiation of the input signal change to when the output reaches 63% of the corresponding change. It includes both the valve assembly dead time, which is a static time, and the dynamic time of the valve assembly. The dynamic time is a measure of how long the actuator takes to get to the 63% point once it starts moving.
Dead band, whether it comes from friction in the valve body and actuator or from the positioner, can significantly affect the dead time of the valve assembly. It is important to keep the dead time as small as possible. Generally dead time should be no more than one-third of the overall valve response time. However, the relative relationship between the dead time and the process time constant is critical. If the valve assembly is in a fast loop where the process time constant approaches the dead time, the dead time can dramatically affect loop performance. On these fast loops, it is critical to select control equipment with dead time as small as possible. Also, from a loop tuning point of view, it is important that the dead time be relatively consistent in both stroking directions of the valve. Some valve assembly designs can have dead times that are three to five times longer in one stroking direction than the other. This type of behavior is typically induced by the asymmetric behavior of the positioner design, and it can severely limit the ability to tune the loop for best overall performance. Once the dead time has passed and the valve begins to respond, the remainder of the valve response time comes from the dynamic time of the valve assembly. This dynamic time will be determined primarily by the dynamic characteristics of the positioner and actuator combination. These two components must be carefully matched to minimize the total valve response time.
In a pneumatic valve assembly, for example, the positioner must have a high dynamic gain to minimize the dynamic time of the valve assembly. This dynamic gain comes mainly from the power amplifier stage in the positioner. In other words, the faster the positioner relay or spool valve can supply a large volume of air to the actuator, the faster the valve response time will be. However, this high dynamic gain power amplifier will have little effect on the dead time unless it has some intentional dead band designed into it to reduce static air consumption. Of course, the design of the actuator significantly affects the dynamic time. For example, the greater the volume of the actuator air chamber to be filled, the slower the valve response time. At first, it might appear that the solution would be to minimize the actuator volume and maximize the positioner dynamic power gain, but it is really not that easy.
This can be a dangerous combination of factors from a stability point of view. Recognizing that the positioner/ actuator combination is its own feedback loop, it is possible to make the positioner/actuator loop gain too high for the actuator design being go into an unstable oscillation. In addition, reducing the actuator volume has an adverse affect on the thrust-to-friction ratio, which increases the valve assembly dead band resulting in increased dead time. If the overall thrust-to-friction ratio is not adequate for a given application, one option is to increase the thrust capability of the actuator by using the next size actuator or by increasing the pressure to the actuator. This higher thrust-to-friction ratio reduces dead band, which should help to reduce the dead time of the assembly. However, both of these alternatives mean that a greater volume of air needs to be supplied to the actuator. The tradeoff is a possible detrimental effect on the valve response time through increased dynamic time. One way to reduce the actuator air chamber volume is to use a piston actuator rather than a spring-and-diaphragm actuator, but this is not a panacea. Piston actuators usually have higher thrust capability than springand- diaphragm actuators, but they also have higher friction, which can contribute to problems with valve response time. To obtain the required thrust with a piston actuator, it is usually necessary to use a higher air pressure than with a diaphragm actuator, because the piston typically has a smaller area. This means that a larger volume of air needs to be supplied with its attendant ill effects on the dynamic time. In addition, piston actuators, with their greater number of guide surfaces, tend to have higher friction due to inherent difficulties in alignment, as well as friction from the O-ring. These friction problems also tend to increase over time. Regardless of how good the O-rings are initially, these elastomeric materials will degrade with time due to wear and other environmental conditions. Likewise wear on the guide surfaces will increase the friction, and depletion of the lubrication will occur.
These friction problems result in a greater piston actuator dead band, which will increase the valve response time through increased dead time. Instrument supply pressure can also have a significant impact on dynamic performance of the valve assembly. For example, it can dramatically affect the positioner gain, as well as overall air consumption. Fixed-gain positioners have generally been optimized for a particular supply pressure. This gain, however, can vary by a factor of two or more over a small range of supply pressures. For example, a positioner that has been optimized for a supply pressure of 20 psig might find its gain cut in half when the supply pressure is boosted to 35 psig. Supply pressure also affects the volume of air delivered to the actuator, which in turn determines stroking speed. It is also directly linked to air consumption. Again, high-gain spool valve positioners can consume up to five times the amount of air required for more efficient high-performance, two-stage positioners that use relays for the power amplification stage. To minimize the valve assembly dead time, minimize the dead band of the valve assembly, whether it comes from friction in the valve seal design, packing friction, shaft wind-up, actuator, or positioner design. As indicated, friction is a major cause of dead band in control valves. On rotary valve styles, shaft wind-up (see definition in Chapter 1) can also contribute significantly to dead band. Actuator style also has a profound impact on control valve assembly friction.
Generally, spring-and-diaphragm actuators contribute less friction to the control valve assembly than piston actuators over an extended time. As mentioned, this is caused by the increasing friction from the piston O-ring, misalignment problems, and failed lubrication. Having a positioner design with a high static gain preamplifier can make a significant difference in reducing dead band. This can also make a significant improvement in the valve assembly resolution (see definition in Chapter 1). Valve assemblies with dead band and resolution of 1% or less are no longer adequate for many process variability reduction needs.
Many processes require the valve assembly to have dead band and resolution as low as 0.25%, especially where the valve assembly is installed in a fast process loop. One of the surprising things to come out of many industry studies on valve response time has been the change in thinking about spring-and-diaphragm actuators versus piston actuators. It has long been a misconception in the process industry that piston actuators are faster than spring-and-diaphragm actuators. Research has shown this to be untrue for small signal changes. This mistaken belief arose from many years of experience with testing valves for stroking time. A stroking time test is normally conducted by subjecting the valve assembly to a 100% step change in the input signal and measuring the time it takes the valve assembly to complete its full stroke in either direction. Although piston-actuated valves usually do have faster stroking times than most spring-and-diaphragm actuated valves, this test does not indicate valve performance in an actual process control situation. In normal process control applications, the valve is rarely required to stroke through its full operating range. Typically, the valve is only required to respond within a range of 0.25% to 2% change in valve position.
Extensive testing of valves has shown that spring-and-diaphragm valve assemblies consistently outperform piston actuated valves on small signal changes, which are more representative of regulatory process control applications. Higher friction in the piston actuator is one factor that plays a role in making them less responsive to small signals than springand- diaphragm actuators. Selecting the proper valve, actuator, positioner combination is not easy. It is not simply a matter of finding a combination that is physically compatible. Good engineering judgment must go into the practice of valve assembly sizing and selection to achieve the best dynamic performance from the loop. Figure 2-4 shows the dramatic differences in dead time and overall T63 response time caused by differences in valve assembly design
This is a good read. RE:”Extensive studies of control loops indicate as many as 80% of the loops did not do an adequate job of reducing process variability”. It is shame that many companies do not employ enough people (or the right people) to do this sort of checking and tuning of process equipment. They are too busy counting the money saved by getting rid of good people.. oh well, thats life. People with these skills are becoming rare indeed.