Poor controller tuning contributes to wear and tear, often forcing the valve to move five or 10 times more than is actually required for good control. This article assesses the cost of poor tuning and provides some direction for the person looking to reduce valve maintenance costs.
For a PID control loop, the valve is usually the most costly component, typically representing 50% to 75% of the purchase cost of the control loop. However, it often represents closer to 90% of the ongoing maintenance costs.
These costs are well understood. Valve seals wear, packing leaks, components corrode, and linkages and springs suffer the stresses of thousands, even millions of cycles. In most industrial plants, a planned maintenance program combats the effects of this wear. It is typical for plants to spend on average between $400 and $2,000 per valve per year.
A good maintenance program will evaluate valve performance and routinely rebuild or replace valves that are out of specification. Work orders are written and the jobs are scheduled during plant shutdowns. Plant maintenance personnel are usually stretched quite thin during these shutdowns, so much of the work is completed with overtime pay or through third-party contractors, further increasing the cost.1
Wear and tear is caused by excessive operation of the valve. If the valve is constantly in motion, then the springs and linkages are being constantly stressed and fatigued. Positioner arms and other devices with moving components are also being worn in the process.2,3
As Free as the Air We Breathe?
A well-hidden cost of control valves is compressed air. Large, expensive compressors operate with electricity 24 hours a day. If you know your plant's cost per kilowatt-hour, you can estimate the cost per year for 1 standard cubic foot per minute (SCFM) of air: 0.25 hp/SCFM x 0.745 kW/hp x 24 hr/day x 365 day/year x cost/kWh.
At a typical industrial rate of $0.06/kWh, this is roughly $98/year for each SCFM. And this doesn't count the capital, depreciation, and maintenance costs for the compressor, dryer, and distribution system. To determine the cost of air for control valve actuation, we need to know a little about the size of the valve actuator. Table I provides some typical air uses, based on actuator size and type.
Since it costs only $100 to $200 per year to tune and optimize a control loop, the savings on air alone justifies the effort with more than 200% payback.
The amount of valve movement plays a significant role in the amount of air required for the valve. For good control, of course, the valve must move. But just how much the valve must move depends on many factors, including the size of the valve, the process gain, and the size and frequency of disturbances and setpoint changes.
Top Control's experience is that most controllers, when properly tuned, require the valve to move between 1,500% and 20,000% per day; half of the time responding to noise, and half of the time to disturbances. A poorly tuned controller may move the valve two to 50 times as much.
Said another way, a properly tuned valve makes the equivalent of between 15 and 200 full strokes each day. This works out to between 1% and 14% per minute. Obviously, if we can reduce the valve movement, we can reduce the air usage. This in turn reduces valve maintenance needs, air costs, and process variability.
While we're on the topic of air losses: During a shutdown, assign someone to go hunting for air leaks. When the plant is quiet, they can quickly pinpoint air leaks (a spray bottle filled with soapy water helps). They can fix these leaks as they go, and the payback will be tremendous.
One other way to save on compressed air costs is to use an inlet guide vane to pre-spin the air entering the compressor. This reduces compressor-operating costs by approximately 10%.
Figure 1: Remove Derivative
Compared to PID (blue) with a small, 1-sec. derivative, the same PI without derivative (red) reduces valve motion seven-fold.
How Tuning Affects Valve Costs
Of course, air costs are not the ultimate goal of the control system. A properly tuned controller will react when the process requires it, but will not over-react to process noise. In addition to tuning, a filter can be used to further reduce the noise that is seen by the controller.
Based on Top Control's experience, most control loops will see a two or three times reduction in valve movement when tuned properly. This happens through the proper use of P, I, and D parameters. Using these techniques, we have seen valve movement reduced by as much as 200 times.
Derivative (D) setting has the greatest impact on valve movement. Derivative action responds to the rate of change of the process variable (PV), or the error. Since noise represents quick changes to the PV, the derivative action responds quickly, moving the valve. In effect, derivative action will amplify noise.
Derivative is used primarily when we need aggressive tuning on relatively slow loops. For fast loops, such as flow and most pressure loops, removing the derivative action will result in an immediate improvement in performance for the process, and a significant reduction in movement of the valve. Figure 1 shows the impact of the removal of derivative action from a fast control loop. In this case, the elimination of the derivative action has resulted in a seven-times reduction in valve movement. And the PV still arrives at setpoint in about the same time. But clearly, there is more work to do on the tuning of this loop.
Figure 2: Tune the Loop
The red line is the same as the PI control shown in Figure 1. Optimal tuning (green) results in another two-times reduction in valve movement while improving control.
If we choose more optimal tuning, we will get even more reduction of valve movement. Figure 2 shows the improvement when choosing more optimal tuning. This results in another two-times reduction in valve movement, while improving control. So the net effect of tuning and filtering is a five to six times reduction in valve movement.
There is one other thing we can do to reduce valve movement without seriously affecting control performance. We can add a filter. A properly sized filter will reduce the noise, with only a minor impact on the controller performance. Figure 3 illustrates the impact on both valve movement and PV. This results in a further 60% reduction of valve movement, and there is little adverse consequence to the control.
Figure 3: Filter Properly
Filtering yields a further 60% reduction of valve movement.
If we compare the initial state, as found, with the final state, using optimal tuning and a filter, we can see that the control has improved slightly, and the valve movement has been reduced dramatically (Figure 4). In this case, there is a 20-times reduction in valve movement.
In our experience, proper configuration, tuning, and filtering result in a five times reduction in valve movement. This optimization requires roughly one hour per loop and costs $100 to $200 per loop. The control performance will be better and the valve maintenance will be reduced by a factor of two.
Figure 4: Add It Up
Comparing the initial as-found with the final result, there is a 20-times reduction in valve movement with slightly improved control.
Production Costs
In addition to the impact on valve maintenance costs, the impact of poor loop performance on overall production is quite remarkable. It often totals between 0.5% and 3% of production. This may be found in waste or scrap product, recycle, or in increased raw material and energy costs. While it is hard to predict this specific impact in advance, it is often the largest benefit of proper controller tuning.
Excess valve movements cost thousands of dollars per year for each valve. Through the proper use of tuning and filtering, valve movement in most loops can be reduced by a factor of two to 10 times. This can reduce valve maintenance and air use costs dramatically, while maintaining or improving control loop performance.
Using modern tools and methods to optimize loops takes only one or two hours per loop per year. The time and effort are very quickly paid back in reduced valve maintenance costs, reduced air usage, and reduced production costs (Table II).4
George Buckbee is a control engineer for Top Control, Hubertus, Wis., where he both teaches process control and solves plant control system problems. He holds a B.S. in chemical engineering from Washington University in St. Louis, and an M.S. in chemical engineering from the University of California at Santa Barbara. He is a registered professional engineer and a senior member of ISA. George has been actively applying process control for more than 15 years, including roles in plant maintenance, engineering, project management, teaching, and consulting. Over the past 10 years, he has been focusing on improving process operating costs through the use of control systems. Contact him at george.buckbee@topcontrol.com.
All figures and simulations were completed using ExperTune software.
References
1. Ruel, Michel, "Valve Health Certificate," Chemical Engineering, November 2001, pp. 62-65.
2. Liptak, Bela, Instrument Engineer's Handbook: Process Control, Chilton, Radnor, Pa., 1995.
3. Ruel, Michel, "Loop Optimization: Before You Tune," Control Magazine, Vol. XII, No. 03 (March 1999), pp. 63-67.
4. Buckbee, George, "Achieving Huge ROI through Controller Tuning & Optimization," Proceedings, ISA 2001, Houston, 2001.