The best wireless settings provide an effective compromise between the need to maximize battery life and meet loop performance objectives. Since power consumption is greatest in transmitting an update, battery life can be significantly extended by reducing unnecessary updates.
The improvement in battery life can be achieved by increasing the time interval for periodic reporting (default update rate) and the threshold sensitivity for exception reporting (trigger level). However, an increase in the update time interval will increase the integrated error for unmeasured disturbances. The effect is similar to what was discussed in the first Control Talk Blog "What is the Best PID Execution Time?"
The update time interval is the default update rate for large trigger level settings. For small trigger level settings, the update time interval is shorter and variable.
For slow loops, such as level and temperature, the effect of update time interval can be estimated by the same equations for the effect of PID execution time. For slow loops, the update time interval is often much less than the integral time setting making the increase in integrated error negligible for unmeasured disturbances.
Since the update time from the minimum default update rate is presently much larger than the minimum PID execution time, the increase in deadtime is a concern for fast loops. Since the PID execution time can be set as low as 0.1 sec, the additional deadtime was significant mostly for liquid pressure and furnace pressure loops that manipulate variable speed drives. The scope of fast loops affected by wireless is larger. The additional deadtime for unmeasured disturbances is ½ the update time interval since the latency in the measurement result transmitted is negligible.
It would seem that that wireless would not be suitable for flow loops until you consider tuning practices, essential functionality for common applications, and the advantage of an enhanced PID developed for wireless.
Flow loops are detuned to deal with the nonlinearity of the control valve. Consequently, update time intervals of 4 seconds or less do not normally require retuning from the additional deadtime. For longer update time intervals, the use of the enhanced PID provides stable control with existing tuning settings. In fact, the controller gain can be increased to the inverse of the largest open loop gain which occurs at the PID output where the slope of the installed valve characteristic is greatest. Adaptive tuning can reduce the effect of the valve nonlinearity on tuning by scheduling the tuning settings as a function of PID output. Slides 28 and 29 in ISA-AW-2011-Wireless-Measurement-and-Control-Opportunities-Presentation.pdf show that the use of enhanced PID with a 16 second default update rate could be tuned to provide a single correction that achieved a complete setpoint change and full disturbance rejection. In comparison, the traditional PID gain had to be decreased by a factor of 4 to prevent excessive oscillations.
Many flow loops have their setpoint driven by a primary process loop. The cascade action will correct for prolonged upsets that affect the primary process variable. The main purpose of the secondary flow loop is to isolate the nonlinearity of the valve from the primary loop and to immediately respond to the demands of the primary loop. Often the short term deviations in the secondary loop due to secondary disturbances, such as pressure, are not seen in the primary loop due to the effect of process volumes. Thus, pressure disturbances are corrected by a stable secondary loop before they are seen in column or vessel level, pH, or temperature control. Even for inline control, the use of an enhanced PID for the primary loop besides the secondary loop enable tight setpoint and load control for a default update rate of 16 seconds for the flow loop and 60 seconds for the static mixer pH loop as seen in slides 31 and 32. The key here is to insure the primary loop is at least 4 times slower than the secondary loop per the common cascade rule by selecting the pH primary default update rate to be 4 times the flow default update rate since most the loop deadtime is from the update rate.
The trigger level can reduce the effect of a large default update rate by reporting a change in the measurement that is large enough to be a concern. If an unmeasured disturbance must start to be rejected in less than the default update rate, the trigger level should be less than ½ the allowable short term error in the process variable. The key words here are “short term” error because most process engineers and operators will state unreasonably tight control requirements from getting obsessed with numbers after the decimal point in digital displays or thinking of persistent long term errors. In reality, short term errors from higher trigger levels are rarely seen in important primary process variables due to the attenuating effects of process volumes.
The trigger level should be set to screen out noise if the noise does not need to be visible for process analysis. The judicious use of trigger level can reduce valve dither and the need for a signal filter besides improving battery life.
Some guidelines:
- For unmeasured disturbance rejection, the increase in deadtime is ½ the update time
- For unmeasured disturbance rejection in primary process loops, the update time should be less than the 10% of the integral (reset time)
- For flow loops, the traditional PID must be detuned for update times more than 4 seconds
- For compressor surge control, liquid or polymer pressure control, and furnace pressure control, default update rates are presently not fast enough
- For cascade control, the default update rate of the secondary loop should be less than ¼ the primary loop default update rate
- The enhanced PID enables the use of larger update times and eliminates the need for detuning when the additional deadtime causes oscillations
- The trigger level can be decreased to provide faster updates
- The trigger level should be less than ½ the real allowable short term error
- The trigger level should be large enough to prevent updates from ignorable noise
Loops must be carefully monitored, especially for ability to handle abnormal conditions.