Professional Documents
Culture Documents
Part 20: Performance Measurement - The Rest of the Story (see below)
Part 19: How to Exploit Process Variability
Part 18: Are Your Valves Killing You?
Part 17: How to Tune for Both Setpoint Changes and Upsets
Part 16: How to Control a Process With Long Dead Time
Part 15: How to Manage Your Plant's Performance in Real-Time
Part 14: Which Tuning Method Should You Use?
Part 13: How to Automatically Find a Model Of Your Process
Part 12: Safety Factor: The Most Important Tuning Parameter
Part 11: Eliminate Cycling In Your Plant
Part 10: Quickly Tune Slow Loops
Part 9: How to Linearize Your Process
Part 8: When Should You Use Derivative Action?
Part 7: Plant Data for PID Tuning & Modeling
Part 6: Choosing the Best Filter
Part 5: When to use Honeywell A,B,C,D, Real or Ideal Algorithms
Part 4: The Best Sample Interval for Process Control
Part 3: Loop Stability - The Other Half of the PID Tuning Story
Part 2: Ziegler-Nichols Tuning Rules & Limitations
Part 1: PID Algorithms and Units
Harris Index
A performance measure typically discussed by academics is the Harris Index.
The Harris Index looks at the error signal which is the process variable or
measurement minus the setpoint. The Harris measures the ratio between the
error variance and the variance achievable by a minimum variance controller.
The larger the value, the poorer the performance of the loop. The Harris Index
1 = perfect
larger = poorer performance
0 = perfect control
1 = poorest control
Performance Categories
In general, there are three broad categories of control loop performance in
plants.
1. The first and typically rarest are loops that are performing well.
2. The second is loops that are oscillating because of valve hardware or
over-aggressive PID tuning. The Harris index used as a performance
metric will catch these.
3. An operators typical response to a loop that is oscillating is to de-tune it.
The third category are loops that were oscillating and have been de-tuned.
The Harris index will not "catch" these loops.
0 = perfect control
larger = % performance improvement possible
Conclusions
PlantTriage provides everything you need to manage your plants performance in
real-time. Browser-based reports make pinpointing and diagnosing problems
easy. They also make tracking performance easy. PlantTriage includes a host of
analysis tools for optimizing control systems including valve wear analysis, noise
and filter analysis, spectral analysis, response time analysis, simulation,
linearization, and characterization and PID Tuning.
2004 ExperTune Inc.