Real Time Implementation of Model Reference

Adaptive Controller for a Conical Tank

Anna Joseph & J Samson Isaac

Department of Electronics & Instrumentation Engineering,
Karunya University, Coimbatore, Tamilnadu - 641114
E-mail : [email protected], [email protected]

Abstract Nonlinear process control is a difficult task in

process industries. Conical tank level control is one among
them. Real processes often exhibits nonlinear behavior, time
variance and delays between inputs and outputs. This paper
aims at implementing an adaptive control algorithm for a
conical tank system. System identification of the non-linear
process is done using black box modelling and found to be
first order plus dead time (FOPDT) model. Here
Proportional integral (PI) controller based on Ziegler
Nichols (ZN) method is designed initially and the results are
compared with model reference adaptive controller
(MRAC).The real time implementation of process is done in
LabVIEW using NI DAQ (BNC 2120). Better controller
performance and error can be minimized by using MRAC
than that of the ZN tuned PI controller.

The majority of the control theory deals with the

design of linear controllers with linear systems. PID
controllers proved to be a perfect controller for simple
and linear processes. When it comes to the control of
nonlinear and multivariable processes, the controller
parameters have to be continuously adjusted.
Conventional controllers are widely used in industries
since they are simple, robust and familiar to the field
operator. Practical systems are not precisely linear but
may be represented as linearized models around a
nominal operating point, the controller parameters tuned
at that point may not reflect the real time system
characterestics due to variations in process parameters.
So an adaptive control mechanism is designed for
controlling the nonlinear tank system.

Keywords Conical Tank, Model Reference Adaptive

Controller (MRAC), PI Controller, LabVIEW

I.

Here the system used is a conical tank and is highly

nonlinear due to the variation in area of cross section of
the level system with height. Conical tank finds wide
application in petrochemical, papermaking, water
treatment industries etc.

INTRODUCTION

Control of industrial processes is a challenging task

for several reasons due to their nonlinear dynamic
behavior, uncertain and time varying parameters,
constraints on manipulated variable, interaction between
manipulated and controlled variables, unmeasured and
frequent disturbances, dead time on input and
measurements. The control of liquid level in tanks and
flow between the tanks is a basic problem in process
industries. In many processes such as distillation
columns, evaporators, reboliers and mixing tanks, the
particular level of liquid in the vessel is of great
importance in process operation. A level that is too high
may upset reaction equilibria, cause damage to
equipment or result in spillage of valuable or hazardous
material. If the level is too low it may have bad
consequences for the sequential operations. So control
of liquid level is an important and common task in
process industries.

Tuning a controller is the adjustment of process

parameters. Since conical tank is highly nonlinear we
make use of model reference adaptive controller to
control the water level. The proposed method can adjust
the controller parameters in response to changes and
disturbances in plant by referring to the reference model
that specifies properties of the desired control system.
In this work the process model is experimentally
determined by using system identification technique.
The method adopted here for system identification is
step test and is done in real time with LabVIEW using
NI DAQ. The conventional controller tuning is
accomplished using Zeigler Nichols based PI controller
settings and the performances are compared with
MRAC based on settling time and Integral Squared
Error (ISE).
ISSN : 2319 3182, Volume-2, Issue-1, 2013

57

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

II. EXPERIMENTAL SETUP

conical tank. The actual storage tank water level sensed

by the level transmitter is feedback to the level
controller & compared with a desired level to produce
the required control action that will position the level
control as needed to maintain the desired level. Now the
controller decides the control action & it is given to the
V/I converter and then to I/P converter. The final control
element (pneumatic control valve) is now controlled by
the resulting air pressure. This in turn control the inflow
to the conical tank & the level is maintained.

A real time experimental setup for highly nonlinear

conical tank is constructed. The process control system
is interfaced with LabVIEW using USB-based DAQ
module to the Personal Computer (PC). The laboratory
set up for this system is shown in Figure 1,it consists of
a conical tank ,a water reservoir, pump, rotameter, a
differential pressure transmitter, an electro pneumatic
converter (I/P converter), a pneumatic control valve, an
interfacing USB based DAQ module and a Personal
Computer (PC).

The specifications of the tank are as follows :

Height
Volume
Bottom Diameter
Top Diameter
Angle
Material

:
:
:
:
:
:

80 cm
33.5 litres
7.62cm
36.62 cm
10deg
Stainless Steel

III. SYSTEM IDENTIFICATION

3.1 Mathematical Modelling
The process considered here is a conical tank
system shown in Figure 3 in which the level of the
liquid is desired to maintain a constant value. This can
be achieved by controlling the input flow rate into the
tank. Here q is the inlet flow and qo is the outlet flow.
= density of the liquid in the system Kg/cm3

Fig. 1: Conical Tank Setup

= density of the liquid in the inlet stream Kg/cm3

2.1 Process
The control parameter chosen here is the level.
Capacitance sensor and level transmitter arrangement
senses the level from the process and converts into
electrical signal. Then electrical signal is fed to the I/V
converter which in turn produces propotional voltage
signal to the computer.

= density of the liquid in the outlet stream Kg/ cm3

V = total volume of the conical tank
= volumetric flow rate of the inlet stream LPH
= volumetric flow rate of the outlet stream LPH.
R = Maximum radius of the conical tank
r = Radius of the conical tank at steady state
H = Maximum height of the conical tank
h = Height of the conical tank at steady state

Fig. 2: Block Diagram of Closed Loop system

Figure 2 shows the block diagram of a closed loop

system. The control system maintains water level in a

Fig. 3 : Conical Tank

ISSN : 2319 3182, Volume-2, Issue-1, 2013

58

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

Using the law of conservation of mass:

Assuming the room temperature as constant, the density

of water is same throughout.

6.

Compute

7.

Note down t2

8.

Then time constant

9.

Process gain Kp is PV/V where v is change in

the input in volts.

Therefore

Applying the steady state values, and solving the

equations (1) and ( 2) , for linearzing the non - linearity
in the conical tank,

Where

constant,

is the process gain,

Rt hs

is the time
2

Fig. 4: Process Reaction Curve

3.2 Black Box Modelling

At a fixed inlet flow rate, outlet flow rate, the

system is allowed to reach the steady state. After that a
step increment in the input flow rate is given, and
various readings are noted till the process becomes
stable in the conical tank. The experimental data are
approximated to be a FOPDT model. The model
parameters chosen here at 25-35cm operating range is:

System identification is a methodology for building

mathematical models of dynamic systems using
measurements of the systems input and output signals,
ie, system identification utilizes input-output
experimental data to determine a system's model. In real
time implementation, system identification of nonlinear
process is done using step test fig 4.
Steps to find transfer function are:
1.

Note down initial steady state value of process

variable (ISS).

IV. PI CONTROLLER

2.

Give noticeable change in input at time (t1).

3.

Note the time delay.

The PI control scheme consists of propotional term

& integral term. The PI controller equation is given as:

4.

Observe the change in the process variable and note

down new sready state (NSS).

u ( t ) K P e p K i e ( t ) dt

5.

Find total change in process variable.

ISSN : 2319 3182, Volume-2, Issue-1, 2013

59

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

u ( t ) controlleroutput
K p propotional g ain
K i int egral g ain
The propotional term makes a change in the
controller output propotional to the current error value.
If the propotional gain value is high it results in a large
change in the controller output for given error. Very
large values for propotional term make the system
unstable. The contribution of integral term will be
propotional to the error and also the time. Since the
integral term takes into account the past values of the
error to give the current controller output. So the error
accumulated over time will be multiplied by the integral
gain and added to the controller output.

Fig. 6: Response of process with PI controllerfor

setpoint 30cm

There are different methods for tuning a PI

controller of which the simplest method is the Zeigler
Nichols method. But the Zeigler Nichols method can be
used to tune a controller for a process if the ratio of
dead time (td ) to time constant () must be within the
range .1< <.6 .So here ZN method cannot be
employed, thus we use trial and error method. The
criteria for selecting the PI gains was integral squared
error (ISE) ie; the controller setting which gives less ISE
was selected as the gains of the controller. The real time
implementation results are shown in fig 5 and 6.

V. MODEL REFERENCE ADAPTIVE CONTROLLER

The MRAC scheme is presented in fig 7.It consists
of four blocks such as process, controller, reference
model and adaptor. The reference model is an ideal
model and its output ym(t) directly denotes the required
dynamic response. The adaptive regulation process of
the controller parameters is described as follows: when
the input value r(t) is set to the controller, it is also
simultaneously added to the reference model input; at
the initial stage, since the origin parameters of
controlled object are unknown, the controlled
parameters are not determined causing the output
response y(t) not in accordance with ym(t) and e(t) is
produced. When e(t) is introduced into the adaptive
regulation loop, through the calculation by adaptive
laws and then proper dynamic signal of changing the
controller parameters is derived to make the y(t) get
approaching to ym(t), i.e.e(t) 0 with adaptive process
ceased.

Fig. 5: Block diagram of PI controller

Fig. 7: Block diagram of MRAC

The adjustment mechanism of MRAC system

constructed by adaptive control rule called MIT rule
which performs the algorithms as following:
ISSN : 2319 3182, Volume-2, Issue-1, 2013

60

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

Tracking error:

Cost function:
MIT Rule says that the time rate of change of is
proportional to negative gradient of J.
That is

Controller law:
Fig. 8: Response of Plant with MRAC for setpoint 30cm.

Update rule:

VI. COMPARATIVE RESULTS

Tuning parameter is
Adjustable parameter is

The comparative results of the controller for setpoint 30

cm are shown in the below table.

Control signal is u and command signal is

TABLE 6.1 Comparative results of process performance

parameters
PROCESS
PERFORMANCE
PARAMETERS
Rise time(sec)
Settling time(sec)
Steady state error (cm)

uc,
Where e denotes the model error and is the controller

parameter vector. The components of

are the

sensitivity derivatives of the error with respect to .The

parameter is known as adaptation gain. The MIT rule
is a gradient scheme that aims to minimize the squared
model cost function .The real time implementation of a
processs is shown in fig 7 and 8.

PI

MRAC

189
115
.95

111
128
.41

Table 6.1 Comparative results of performance indices

CONTROLLER
PI

ISE
11397.2

IAE
7967.8

MRAC

2463.7

1729.8

VII. CONCLUSION & FUTURE WORK

The nonlinearity of the conical tank is analyzed.
Conventional PI controller and MRAC is implemented
in LabVIEW and is tested in real time for the nonlinear
tank. The result shows that better controller performance
and error is minimized in model reference adaptive
controller. In future we can optimize MRAC using soft
computing techniques such as Genetic Algorithm.
VIII. REFERENCES
[1]

Anandarajan.R,
M.Chidambaram,
T.Jayasingh,
Limitations Of A PI controller for a first order
nonlinear
process
with
dead
time,ISA
Trancastions,Vol.45.No.2,pp 185-199,2006

[2]

Pankaj Swarnkar, Shailendra Jain,R.K.Nema, Effect

of Adaptation Gain on System Performance for Model
Reference Adaptive Control Scheme Using MIT

Fig. 7: Block Diagram of the plant with MRAC in LabVIEW

ISSN : 2319 3182, Volume-2, Issue-1, 2013

61

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

Rule, WorldAcademy Of Science, Engineering And
Technology,pp 621-626,2010
[3]

Rathikarani.D and D.Sivakumar, Adaptive PI

Controller for a nonlinear system, Dept.Of
InstrumentationEngineering,
Annamalai
University,Sensors
&
Transducers
Journal,Vol.109,Issue 10,pp 43-58,2009.

[4]

J
Satheesh
Kumar
,
P
Poongodi
and
K.Rajasekaran:Modelling and Implementation of
Labview Based Non-linear PI Controller for a Conical
Tank. Journal of Control & Instrumentation Volume 1,
Issue 1, November, 2010, Page 1-9.

ISSN : 2319 3182, Volume-2, Issue-1, 2013

62