Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

CONTROL ENGINEERING 2018 1

Characterization of Open-Loop Dynamic


Process
D. Efren, R. Pedro

All the investigations and projects should begin assigning a II. THEORETICAL FUNDAMENTS
process model. This paper will analyze and explain three
different types of methods to identify models that have certain A. Tangent Method
characteristics. These methods are the tangent method, tangent The tangent method consists initially with an open-loop
and point method, and finally the two point method, that are test, that is done by putting the controller on the manual mode
based on the graphical approach. and then, make the load change, the result of the response
curve is documented until the new steady state level is
Keywords: Tangent, Line, Method, Point, Step, Inflection. reached.
After that process the curve is analyzed, so then the dead
time and the response rate is found by drawing a tangent line
I. INTRODUCTION at the inflection point of the response curve. The dead time is
calculated at the cross section at the middle of the tangent line

T HE identification of a model of the dynamic characteristics


of a system is a fundamental part found on any model-
based controller design. There will be a big influence on
and the base of the old steady-state level.

the values of the parameters that are going to interact with the
control system.
The analysis will be made by the following model:

𝑘𝑒 '()*
𝐺 𝑠 =
𝜏𝑠 + 1

To find the transfer function, we need to make an open loop


test, that will relate the output and the input variables.
Control tuning methods used to characterize the dynamic
response of a process from an Open-Loop step test require to
first determine the dynamic parameters of the model, such as
process gain (k), the time constant (𝑡/ ), and the dead time (τ). Gain KM=K time delay Ө = t0, and time constant = ፒ

Here the gain is showing how the controlled variable is 𝑘𝑒 '()*


𝐺 𝑠 =
changing according to the chance in the output, the dead time 𝜏𝑠 + 1
shows the time that the controller takes to notice the change in
the transmitter output, the time constant indicates the speed in
which the controlled variable is changing.
B. Tangent and Point Method
Afterwards the gain is calculated, we are going to use three
methods to estimate the dead time and the time constant for As Suggested by Murrill in 1967, the tangent and point
method is acknowledged as the straightforward method of the
each response
modeling process. The model consists by certain parameters
estimated from the open loop step response.
This method changes compared to the tangent method in the
estimate of the time constant, but it also has things in
common, like the way the dead time is calculated, which is the
same way as the tangent method.
CONTROL ENGINEERING 2018 2

First, we need to find at which point the step response reaches


the 63.2% of the total steady state change, this point is defined
as t1.
The time constant then is calculated by
𝜏 = 𝑡1 − 𝑡/

𝜏 is the process time constant and the 𝑡/ is the dead time


This method has a shorter estimate of the time constant,
therefore this results as a more conservative controller.

C. Two Point Method


Figure 1. Time step response
In the two point method we use a 63.2 point as we did on the
tangent and point method, also we need to use another one that In order to solve the problem, we need to find the 3 constant
is located when the step response reaches the 28.3% of the we have in the transfer function using the 3 graphic methods we
steady-state change, and we define it as t2. saw in part II.
A. Tangent Method

Seeing the graph of the gain power we found that k will be


the difference between the first steady state and the last
𝜏 = 1.5(𝑡1 − 𝑡2) steady state. So, k will be:
𝑡/ = 𝑡1 − 𝜏

On this method we find shorter values of the time constants


𝑘≈7
and longer values of the dead time compared to the tangent
and point method, but this method we assumed that is more Now, to find 𝑡8 and 𝜏 we draw the tangent on the inflexion
precise because we don’t include our own perception to point in the step response graph. 𝑡/ is given by the
analyze the points difference between 0 and the point where the tangent
crosses the first steady state in the time axis (P1). We find 𝜏
as the difference between the point where the tangent
crosses the second steady state (P2) and P1. This is shown
III. CASE STUDY RESULTS
in figure 2.
The process model we are given has the following transfer
function:

𝑘𝑒 '()*
𝐺 𝑠 =
𝜏𝑠 + 1

Where 𝑘 is the gain power of the system, 𝑡/ is the dead time


and 𝜏 is the time constant.
The step response of the process is given by:

Figure 2. tangent method representation

Graphically, we found that


CONTROL ENGINEERING 2018 3

𝑡/ = 41.6 𝑠𝑒𝑐
And,
𝜏 ≈ 89 𝑠𝑒𝑐

So, the transfer function of the problem is:

7𝑒 '(>?.@)*
𝐺 𝑠 =
(89)𝑠 + 1

B. Tangent and Point Method

In this method we know that the gain power will be the same
than the last one, so: Figure 3. Representation of tangent and point method
A. Two-Point Method
𝑘 ≈ 7
To solve the problem using this method, we must find the
And the death time will be the same also: value where the gain reaches the 28.3% (t2) and the same
point than in the tangent and point method where the gain
𝑡/ = 41.6 𝑠𝑒𝑐 reaches the 63.2% (t1). These points are shown in figure 4.

To find 𝜏, we obtain from figure 3 that the point where the


gain reaches the 63.2 percent of the gain is:

𝑃 63.2% ≈ 4.4
So,
𝑡1 ≈ 100

Now the time constant is obtained by:

𝜏 = 𝑡1 − 𝑡/
𝜏 = 100 − 41.6
𝜏 = 58.4 𝑠𝑒𝑐

The transfer function now is

7𝑒 '(>?.@)*
𝐺 𝑠 =
(58.4)𝑠 + 1 Figure 4. Two-point method representation

Then the point where the gain reaches the 28.3% corresponds
to:

𝑃(28.3%) = 1.98

And,

𝑡2 ≈ 70 𝑠𝑒𝑐

With t1 and t2 we obtain the time constant as:

𝜏 = 1.5(𝑡1 − 𝑡2)
𝜏 = 1.5(100 − 70)
𝜏 = 45 sec
CONTROL ENGINEERING 2018 4

And 𝑡/ will be given by:


𝑡/ = 𝑡1 − 𝜏
𝑡/ = 100 − 45
𝑡/ = 55 𝑠𝑒𝑐

The transfer function of the system results as:

7𝑒 '(>E)*
𝐺 𝑠 =
(55)𝑠 + 1

IV. CONCLUSION
In conclusion for this project we saw that the three methods
provide almost the same result for the constant of the transfer
function except for the tangent method which gave us different
values of the constants.

We can resume that in our experience the less effective


method was the tangent method, this is because this method is
completely graphic and our constant values would depend on
our individual perception about getting the inflexion point in
order to get the tangent of the graphic function. On the other
hand, the most effective method we got was the two-point
method which gave a precise value of the constants because in
this method we obtained the points of interest with certain
values that are already given to us by the method itself and not
getting the values visually as in the other two methods.

This project was critical for our understanding about the


differences of the three methods we saw to get the constants
values given the plot of the step response of the transfer
function of the system. As we saw earlier because of our
understanding we could determine which one of the method
worked out better for our particular case.

REFERENCES
[1] K. J. Åström and T. Hägglund, Benchmark systems for PID control,
Preprints IFAC workshop in digital control. Post, present and future of
PID control, Terorsa, Spain. April 2000, 181-182.
[2] P.W. Murrill, Automatic control of processes, International Textbook
Co., 1967.
[3] T.E. Marlin. Process control, designing processes and control systems
for dynamic performance, McGraw-Hill International Editions, 1995.
[4] Lewis M. Gordon, “Feedback Control Modes”, Process Automation
Series, Foxboro-McGraw-Hill, Inc., 1985, p17
[5] Armando B. Carripio, Tuning of Industrial Control Systems, Instrument
Society of America, 1990, pp. 43-44
[6] J.G. Zieglar & N.B. Nichols, “Optimum Settings for Automatic
Controllers”, Trans. ASME, Nov. 1942, pp. 759-768
[7] Thomas B. Kinney, “Tuning Process Controllers”, Process Automation
Series, Foxboro-McGraw-Hill, Inc., 1985, pp. 19-24
[8] T. Senbon & F. Hanabuchi, Instrumentation Systems: Fundamentals and
Applications, Springer-Verlag, 1991, pp. 49-50

You might also like