«

»

Dec
08

Exceptional Opportunities in Process Control – Control Valve Rangeability

There are a lot of ways of looking at rangeability. Nearly all of them lead to the wrong conclusion as to what type of valve is best for process control. Some of the absolute worse valves for control (e.g. on-off piping valves) have the highest stated rangeability.

Valve rangeability is particularly important for pH control, batch control, startup, and plant turndown (see Control Talk column Downturn Turndownin Control July 2009 issue)

From, a piping view point, a full bore ball valve might be thought to have the highest rangeability because when the valve wide open, the flow path is nearly an open unobstructed section of pipe. A conventional butterfly would not be far behind because the only obstruction is a disc that could be almost horizontal when wide open.

Another definition of valve rangeability I have heard is the maximum flow divided by the minimum flow where the actual flow characteristic deviates by some specified margin from the specified inherent flow characteristic. Based on this definition, a linear trim (linear inherent characteristic) is stated to have the best rangeability. This approach is bogus in that the installed characteristic will be different and the controller can compensate for a deviation in characteristic through reset action.

The largest controllable flow divided by the smallest controllable flow is the definition of valve rangeability from a control viewpoint. Just being able to pass a high flow for a given valve size or adherence to an inherent valve characteristic does not mean the valve has high rangeability for control. You need to look at the installed valve characteristic where the percent flow is plotted versus stroke. Note the plot uses percent flow so the magnitude of how much flow the valve passes is not the issue.

For liquid service, the ratio of the pressure drop of the valve wide open to the valve fully closed can be used to show the effect of pump and piping design on the installed characteristic. This valve drop ratio varies from 1.0 where the frictional loss from the piping is negligible (entire difference between pump discharge and destination pressure is available as a pressure drop across the valve) to a minimum of about 0.05 where the valve drop at wide open is about 5% of the system pressure drop for energy conservation (decreased pump head and hence size). Figures 7-47a through 7-47c in the attached ControlValveRangeability.pdf excerpt from the ISA book The Essentials of Modern Measurements and Final Elements – A Guide to Design, Configuration, Installation, and Maintenance show the effect of valve drop ratio on the installed characteristic for linear, equal percentage, and modified percentage inherent characteristic. These figures show that a linear trim distorts to an undesirable type of quick opening characteristic where there is a burst of flow near the closed position followed by a noticeably decreasing valve gain (valve sensitivity) above 30% open. Conversely, the equal percentage trim becomes more linear as the valve drop ratio decreases. The curves for the equal percentage trim shown in Figure 7- 47b are for a conservative rangeability parameter equal to 100 (R=100). Many valves designed for superior throttling service have a larger R that would lower all of the curves in Figure 7-47b near the closed position.

Some progress has been made in a more realistic assessment of valve rangeability based on changes in slope of the installed valve characteristic and hence changes in the valve gain (more commonly referred to as the process gain). The lowest controllable and highest controllable flow depends upon where the slope decreases to less than 1/4 of its maximum thereby putting a limit on the change in process gain of 4:1. Based on this criterion, a sliding stem valve has a better rangeability than a ball valve or the conventional disc butterfly as seen in Figures 7-48a through 7-48c in the excerpt.

Heat exchanger temperature and inline composition control loops often benefit from the increase in gain with stroke offered by an equal percentage characteristic because it helps compensate for the decrease in temperature or composition process gain as the flow through the valve increases. In fact there is theoretically an exact linearization possible for a valve drop ratio of 1.0, because the slope (valve gain) of the inherent equal percentage characteristic being proportional to flow exactly cancels out the process gain inversely proportional to flow.

For vessel level, pressure, and temperature control loops, the process gain is so small that the allowable controller gain is way above the controller gain used. Consequently, changes in valve gain have a negligible effect.

Flow loops have a linear process gain so the valve gain linearity affects tuning. The effect of this is minimized by the use of reset rather than gain action.

I have suggested for more than 20 years that a more absolute accounting of valve rangeability from a control perspective would be to take the stick-slip near the closed position and use this as the X coordinate and use the corresponding Y coordinate on the installed valve characteristic as the minimum controllable flow. You cannot control tighter than the limit cycle from the resolution limit near the closed position. Based on this criterion, valves with a minimum sticktion near the seat and a percentage type of characteristic would offer the best rangeability. Sliding stem valves with a percentage trim, a valve drop ratio of 0.25 or higher, low friction correctly tightened packing, diaphragm actuators, and digital positioners would have the best rangeability and the best dynamics. If the pressure drop allocated to the control valve is less than 10% of the system drop to save energy, the nonlinearity of the installed characteristic of most trims becomes potentially detrimental to loop tuning and performance.

I got on a roll listening to Bob Seger’s “Roll.”