Control loops have to deal with disturbances and setpoint changes, yet operator displays give a static picture. The simple identification of the loop deadtime and calculation of the process variable (PV) rate of change can be used to provide intelligent trends and future trajectories that not only help the operator but also offer equipment and environmental protection, optimization of setpoint response, and process analysis.
Trend charts are next to useless if they don’t have the right time scale. The minimum chart time scale should be 10 loop deadtimes because control loop cannot completely reject a disturbance or achieve a new setpoint in < 5 deadtimes. The time frame of interest increases as loops are tuned slower for robustness (ability to ensure a non-oscillatory response for deteriorations in process dynamics, such as an increase in deadtime). Deadtime is the easiest dynamic parameter to identify. The time to the start of the change in the PV rate of change after a major change in output, is the deadtime. If the initial PV rate of change is close to zero, the identification is even easier.
The PV rate of change (dPV/dt) multiplied by the deadtime can provide a PV value one deadtime into the future on the same time scale. Slide 1 in Test1A-Loop-1-Present-Future shows how the future PV predicts what is going to happen to the PV in the future. This can be useful for operators to understand the future impact of manual and automatic actions. Human impatience and lack of understanding of deadtime typically leads to overreaction in manual and too much reset action (too small of a reset time) in automatic as noted in the March 2011 modeling and control blog “Are we Misleading our Operators?”.
The calculation of dPV/dt in slide 2 uses a deadtime block to create an old PV that when subtracted from the present PV creates a delta PV. The delta PV added to the present PV gives a predicted PV one deadtime into the future. The delta PV divided by the deadtime is dPV/dt. The use of the deadtime block is essential because the method:
- provides a continuous train of dPV/dt values (no aliasing or calculation delay as experienced in traditional techniques that use sample interval)
- improves the signal to noise ratio
- is inherently consistent with the first principle that any change made in the loop is not seen until one deadtime into the future.
For me the most impressive improvement in setpoint response is from Bang-Bang logic that holds the PID output at an output limit for fastest approach to setpoint. When the future PV is close to the setpoint, the output is set and held at the final resting value for one deadtime to minimize overshoot before being released for feedback control. This opportunity to optimize setpoint response is described in the Control, May, 2005 article “Full throttle batch and startup response”.
The dPV/dt has many uses. The March 2011 modeling and control blog “A Calculation so Simple yet so Powerful” lists 7 opportunities. Other recognized prospects afforded by the calculation of dPV/dt are:
- Smart extension of the trend chart time scale for slow approaches to setpoint
- Smart logic to optimize reset time to prevent overshoot from too small of a reset time and faltering from too large of a reset time
- Cooling rate profile analysis for reaction rate and crystallization rate analysis
- Tradeoff between yield and production rate in batch cycle time optimization
- Surge prediction and recovery (precipitous drop in flow)
- Pressure relief prevention (rapid increase in pressure)
- Environmental violation prevention (rapid approach to RCRA pH limit)
This whole deal reminds me of the song “THE FUTURE AIN’T WHAT IT USE TO BE” I heard last Saturday night at a Meat Loaf concert in the Austin City Limit Moody Theatre.