Emerson Process Management 53eA instruction manual Process Reaction Curve

MODEL 54eA	SECTION 15.0
	CALIBRATION - CONTROL

FIGURE 15-1. The Process Reaction Curve

6.When sufficient data have been collected, return the output signal to its original value using the sim- ulate test function. Maintain the controller in this manual mode until you are ready to initiate auto- matic PID control, after you have calculated the tuning constants.

Once these steps are completed, the resulting process reaction curve is used to obtain information about the overall dynamics of the system. It will be used to cal- culate the needed tuning parameters of the Model 54eA controller.

NOTE

The tuning procedure outlined below is adapted from "Instrumentation and Process Measurement and Control", by Norman A. Anderson, Chilton Co., Radnor, Pennsylvania, ©1980.

Information derived from the process reaction curve will be used with the following empirical formulas to predict the optimum settings for proportional and inte- gral tuning parameters.

Four quantities are determined from the process reac- tion curve for use in the formulas: time delay (D), time period (L), a ratio of these two (R), and plant gain (C).

A line is drawn on the process reaction curve tangent to the curve at point of maximum rise (slope) as shown in Figure 15-1. The Time Delay (D), or lag time, extends from "zero time" on the horizontal axis to the point where the tangent line intersects the time axis. The Response Time period (L), extends from the end of delay period to the time at which the tangent line intersects the 100% reaction completion line repre- senting the process stabilization value. The ratio (R) of the Response Time period to the Time Delay describes the dynamic behavior of the system.

In the example, the process Delay Time (D) was four seconds and the Response Time period (L) was 12 seconds, so:

R = L 12 seconds = 3

D4 seconds

The last parameter used in the equations is a plant gain (C). The plant gain is defined as a percent change in the controlled variable divided by the per- cent change in manipulated variable; in other words, the change in the measured variable (pH, conductivi- ty, temperature) divided by the percent change in the analog output signal.

The percent change in the controlled variable is defined as the change in the measured variable (pH,

83