International Journal on Mechanical Engineering and Robotics (IJMER)

__________________________________________________________________________

Modeling and Simulation of prototype of boiler drum level control

1
Keyur Solanki, 2Jalpa Shah, 3Nishith Bhatt
Institute of Technology, Nirma University, Essar Steel Ltd. Hazira, Surat
Email: [email protected], [email protected], [email protected]

Abstract - This paper represents an approach for An intense decrease in this level may expose boiler
controlling a very crucial parameter of boiler i.e. level of tubes, allowing them to become overheated and
the boiler drum using PID controller. IMC based PID damaged. An increase in this level may cause
tuning method is used with feed forward and feedback interference with the process of separating moisture
strategy is used to control two element drum level. Besides
this paper is also describes the modeling of the process for
from steam within the drum, thus the efficiency of the
level control and implemented it in simulink. Hardware boiler reduces and carrying moisture into the turbine [2].
model has also been developed and proved open loop Typically, there are three strategies used to control drum
validation for theoretically derived model & practical level. With each successive strategy, a refinement of the
model, further practical and simulation responses are previous control strategy has been taken place. For
compared with respect to rise time, settling time and extent of the load change requirements, the control
maximum peak overshoot. strategy depends on the measurement and control
Keywords – Drum level, IMC based PID technique, Feed equipment.
forward – feedback control strategy, Modeling.
The three main options available for drum level control
I. INTRODUCTION are discussed below:

Boiler is defined as a closed vessel in which steam is A. Single Element Drum Level Control
produced from water by the combustion of fuel. In The single element control is the simplest method for
boilers, steam is produced by the interaction of hot flue boiler drum level control system. It is least effective
gases with water pipes which is coming out from the form of drum level control which requires a
fuel mainly coal or coke. Also, chemical energy of measurement of drum water level and feed water control
stored fuel is converted into the heat energy and heat valve. It is mainly recommended for boilers with modest
energy is absorbed by the water which converted in to a change requirement and relatively constant feed water
steam. condition. The process variable coming from the drum
Drum Level Control Systems are used extensively level transmitter is compared to a set point and the
throughout the process industries. Control system is difference is a deviation value. This signal is given to
used to control the level of boiling water contained in the controller which generates corrective action output.
boiler drums and provide a constant supply of steam. If The output is then passed to the boiler feed water valve,
the level is too high, flooding of steam purification which adjusts the level of feed water flow into the boiler
equipment can occur. If the level is too low, reduction in drum.
efficiency of the treatment and recirculation function.
Pressure can also build to dangerous levels. A drum
level control system tightly controls the level whatever
the disturbances, level change, increase/decrease of
steam demand, feed water flow variations appears.
This work represents an approach for controlling a very
crucial parameter of boiler i.e. level of the boiler drum
using PID controller. Besides, this paper is also
describes the modeling of the process for level control.
II. BOILER DRUM LEVEL CONTROL
Boiler drum level control is critical for the protection of Fig. 1. Single element drum level control
plant and safety of equipment. The purpose of the drum B. Two Element Drum Level Control
level controller is to bring the drum level up to the given
set point and maintain the level at constant steam load.

ISSN (Print) : 2321-5747, Volume-2, Issue-2,2014

1
 International Journal on Mechanical Engineering and Robotics (IJMER)
__________________________________________________________________________
A two-element system can do good job under most The feed forward strategy is applied in this work is
operating conditions. Two-element control involves described below:
adding the steam flow as a feed forward signal to the
Consider the generalized process shown in fig 4. It has
feed-water valve . Two-element control is primarily
an output y, a potential disturbance d, and an available
used on intermediate-size boilers, in which volumes and
manipulated variable m.
capacities of the steam and water system would make
the simple total level control inadequate because of
“swell.” Total level control is undesirable when it is
detected by sensors that are insensitive to density
variations, such as the conductivity type. Displacement
and Differential pressure type transmitter sensors are
preferred from this perspective because they respond to
hydrostatic pressure. Smaller boilers, in which load
changes may be rapid, frequent, or of large magnitude,
will also require the two-element system

Fig. 4. Block diagram of feed-forward controller

The disturbance d (also known as load and process load)
changes in an unpredictable manner and our control
objective are to keep the value of the output y at desired
levels. A feedback control action takes the following
steps:
 Measures the value of the output (flow, pressure,
Fig. 2. Two element drum level control liquid level, temperature, composition) using the
C. Three Element Drum Level Control appropriate measuring device. Let ym be the value
indicated by the measuring sensor.
This control system is ideally suited where a boiler plant
consists of multiple boilers and multiple feed water  Compares the indicated value ym to the desired
pumps or feed water valve has variation in pressure or value ysp (set point) of the output. Let the deviation
flow. It requires the measurement of drum level, steam (error) be e = ysp – ym.
flow rate, feed water flow rate and feed water control  The value of the deviation e is supplied to the main
valve. By using cascade control mechanism level controller. The controller in turn changes the value
element act as a primary loop and flow element act as a of the manipulated variable m in such a way as to
secondary loop and steam flow element act as a feed reduce the magnitude of the deviation e. usually,
forward controller. Level element and steam flow the controller does not affect the manipulated
element mainly correct for unmeasured disturbances variable directly but through another device
within the system such as boiler blow down. Feed water (usually a control valve), known as the final control
flow element responds rapidly to variations in feed element.
water demand either from the feed water pressure and
steam flow rate of feed forward signal.  The feedback controlled system of fig 4 which is
called closed loop. Also, when the value of d or m
changes, the response of the first is called open
loop response while that of the second is the closed
loop response.
Feedback controller takes action as:
By reducing the block diagram of fig 4, we have

If set point does not change output must not change in

ideal case.

Fig.3. Three element drum level control

III. CONTROL STRATEGY

ISSN (Print) : 2321-5747, Volume-2, Issue-2,2014

2
 International Journal on Mechanical Engineering and Robotics (IJMER)
__________________________________________________________________________
So, from above calculation forward controller is ∂ ρv ∂ P ∂ ρ1 ∂ P ∂Vv ∂V 1
classical lead lag type compensator. W s h−W fe=V v +V 1 +ρv + ρ1
∂ P ∂t ∂ P ∂t ∂t ∂t
[3]

IV. MODELING ∂P ∂P ∂V 1 ∂V1

The mathematical model of the boiler system is
W s h−W fe=−V 1 K 1 +V 1 K 2 −ρ v + ρ1
∂t ∂t ∂t ∂t
described in this section where two main equations has [4]
been obtained i.e. the drum level and pressure equations. ∂P ∂P ∂D ∂D
Both equations consider the level and pressure as state W s h−W fe=−V 1 K 1 +V 1 K 2 −ρ v π r 2 + ρ1 π r 2
variables, and are obtained using mass and energy ∂t ∂t ∂t ∂t
balances of the boiler system considering both liquid --------- [5]
and steam phases. ∂P ∂D
W s h−W fe= [ V 1 k 2−V 1 K 1 ]+ [ ρ π r 2−ρv π r 2 ]
The following assumptions are made for this model: ∂t ∂t 1
-------------- [6]
 The drum is a perfect cylinder. Energy balance:
 The heat exchange surface between vapor and W s h h v −W fe heo +Q ∂[ ρ 1 h1 V 1+ ρ v hv V v ]
sww=¿ ¿
∂t
liquid is planar.
 The water in both phases (liquid and vapor) at the V v =−V 1; Because, steam volume decrease or increase
drum is at the saturated conditions. as water level increase or decrease.
Mass flow rate balance [3]
W s h h v −W fe heo +Q ∂[ ρ 1 h1 V 1− ρ v h v V 1]
Based on mass flow rate balance, the equations are as sww=¿ ¿
∂t
follows:
∂ ρ1 ∂V1 ∂ h1
D=height of water ∈theboiler drum . W s h h v −W fe heo +Q sww =V 1 h1 + ρ 1 h1 + ρ1 V 1 −h v
∂t ∂t ∂t
W sh =mass steam flow . ------- [7]
W fe=mass water flow . ∂ ρ1 ∂ P ∂V1 ∂ h1
Qsww =heat flow rate between the furnace metal∧liquidW. s h h v −W fe heo +Q sww =V 1 h1 ∂ P ∂ t + ρ1 h1 ∂t + ρ1 V 1 ∂ P
ρ1=density of saturated water . --------- [8]
ρ v =density of saturated stead . Putting the value of K 1 , K 2 , K 3 , K 4in equation 8
d=height of the boiler drum . ∂P
h1 =enthalpy of saturated water .
W s h h v −W fe heo +Q sww =
∂t
[ π r 2 d h1 k 2+ ρ1 π r 2 d k 4−hv π r 2 d
h v =enthalpy of saturated steam --------- [9]
From equation 6
∂D
∂[ ρv V v + ρ1 V 1] W s h−W fe− [ρ π r 2−ρv π r 2 ]
W s h−W fe= ----------------------- ∂P ∂t 1
∂t =
[1] ∂t [ π r 2 d k 2−π r 2 d k 1 ]
∂ ∂ ∂ ∂ ∂P
Wsf- Wfwf = vv ρv + ρv vv+ v1 ρ1 + ρ1 v1 ----- Putting the value of in to equation no. 10
∂t ∂t ∂t ∂t ∂t
[2] A=π r 2 d h1 k 2+ ρ1 π r 2 d k 4−hv π r 2 d k 1−ρ v π r 2 d k 3
ρ v =a0 +a 1 P+a 2 P2 2 2
∂ D [W ¿ ¿ s h h v −W fe heo +Q sww ][ π r d k 2−π r d k 1 ] −AW s h
ρ1=b0 +b1 P+b2 P2 =
∂t [ ρ1 h1 π r 2−ρ v h v π r 2 ][ π r 2 d k 2−π r 2 d k 1 ]−A
∂ ρv ∂ ρ1
=k 1=a1 +2 a2 P =k 2=b 1+ 2b 2 P On substituting the appropriate values, we have
∂P ∂P ∂D
=1.87× 10−3
V 1=π r 2 D ∂t
Converting equation in to Laplace transform
∂V 1 2∂D SD(S) =1.87 ×10−3
=π r
∂t ∂t 1.87 ×10−3
D ( S )=
S
V. PID TUNING METHOD

ISSN (Print) : 2321-5747, Volume-2, Issue-2,2014

3
 International Journal on Mechanical Engineering and Robotics (IJMER)
__________________________________________________________________________
IMC based PID tuning procedure is used in this work Where, lem is the filter tuning parameter to vary the
whose description is as follows: [4][5] speed of the response of closed loop system. Now the
low pass filter can be of three types:
Consider a process model Gp*(s) for an actual process
or plant Gp(s). The controller Qc(s) is used to control If we focus on setpoint changes, the form of filter used
the process in which the disturbances d(s) enter into the is f(s) = 1/( lem* s+1) ^ n. Here, n is the order of the
system. The various steps in the Internal Model Control process.
(IMC) system design procedure are
If we focus on good tracking of ramp set point changes
Factorization: It means factoring a transfer function into the filter of the form used is
invertible (good stuff) and non invertible (bad stuff)
f(s) = (n. lem. s + 1)/ (lem* s+1) ^ n
portions. The factor containing right hand plane (RHP)
or zeros or time delays become the poles in the inverts If we focus on good rejection of step input load
of the process model when designing the controller. So disturbances the filter of the form use is f =
this is non invertible portion which has to be removed ( gamma.s+1)/( lem* s+1) ^ n where gamma is any
from the system. constant.
Mathematically it is given as Equivalent standard feedback controller:[6]
G¿p (s)=G¿p ¿ Now we compare with PID Controller transfer function
Where, For first order : Gc(s) = [Kc . (Ti .s + 1)]/ (Ti . s)
G¿p ¿
And find Kc and Ti ( PI tuning parameters).
G ¿p ¿
Similarly for 2nd order we compare with the standard
Usually we use all pass factorization.
PID controller transfer function given by :
Ideal IMC controller:
Gc(s) = Kc . [(Ti .Td .s^2+Ti . s+1)/Ti . s]. [ 1/ Tf . s+1]
The ideal IMC controller is the inverse of the invertible
Where
portion of the process model.
T = Tau (any constant)
It is given as Ti = integral time constant
Td = derivative time constant
Qc*(s) = inv[ Gp*(-)(s)]
Tf = filter tuning factor
Adding Filter: Now we add a filter to make our Kc = controller gain
controller proper.A transfer function is said to be proper
Now we perform closed loop simulations for above
if the order of the denominator is at least as great as the
procedure and adjust lem (lemda) considering a trade off
order of the numerator. If they are exactly of the same
between performance and robustness (sensitivity to
order the transfer function is said to be semi-proper.If
model error).
the order of the denominator is greater than the order of
the numerator the transfer functions is strictly proper. Drum level control Transfer function:
Thus a controller can be physically implemented if it is
proper. So to make the controller proper mathematically 1.78 ×10−3
G p=
it is given as s
Qc(s) = Qc*(s) f(s) = inv [ Gp*(-)(s)] f(s) 1
f s=
Where f(s) is a low pass filter.
λs+ 1
q s=G−1p fs
Low pass filter [f(s)]: In order to improve the robustness
of the system the effect of model mismatch should be qs
q c=
minimized. Since mismatch between the actual process 1−q s G p
and the model usually occur at high frequency end of the s × 1000
systems frequency response, a low pass filter f(s) is q c=
usually added to attenuate the effects of process model 1.78 × ( λs+1 )−1000 × 1.78× 10−3
mismatch. 1
q c=
Thus the internal model controller is usually designed as 1.78 × 10−3 × λ
the inverse of the process model in series with the low If, λ=1,then k c =561.80
pass filter i.e If, λ=2,thenk c =280.90
Qc(s) = Qc*(s) f(s) = inv[ Gp*(-)(s)] f(s). VI. SIMULATION
Where f(s) = 1/( lem* s+1) ^ n

ISSN (Print) : 2321-5747, Volume-2, Issue-2,2014

4
 International Journal on Mechanical Engineering and Robotics (IJMER)
__________________________________________________________________________

76

75.5

75

% height of water
74.5
Process value
74

73.5

73

72.5

72

71.5

71

70.5

70
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
Fig. 5. Simulink model of two element drum level time(sec.)
control Fig.8. Response of simulink model
Open loop validation: Practical swelling & shrinking response by applying
disturbance 10 second on & off by solenoid valve in
In our process we have derived theoretically boiler drum steam flow, below graph is shown
level control process is pure integrator process. if we
give small step change to integrator in open loop 67.7
Process value

% he ight o f w ater
strategy it will go to the infinity mode so we have 67.6
implemented in closed loop mode to control the process.
open loop mode prove that our theoretically derived 67.5

process validate to the practical system. open loop

practical response is shown below. 67.4

67.3
85
Process value
% H eight of w ater

67.2
80

67.1
0 5 10 15 20 25 30 35 40 45 50 55 60
75
time(sec)
Fig.9. Swelling and shrinking response
70
Delay Rise Max Settling
time(td) time(tr) peak time(ts)
65
overshoot
60
Ideal 14 sec. 6 sec. 0.4 25 sec
0 10 20 30 40 50 60 70 Practical 15 sec. 13sec. 0.5 50 sec
time(sec)

Fig. 6. Open loop validation VII. CONCLUSION

With different lemda value imc based pid tuning
IMC based pid tuning for lemda = 2 is implemented
response is shown below, lemda is tuning parameter that
because it’s give less overshoot. In above comparison
will vary the speed of response.
table of delay time, rise time, settling time is shown.
6
Difference between ideal & practical is due to transfer
function of control valve & i/p converter, which is not
% H e ig h t o f w a te r

5.5
Process value
Step input
5
kept in ideal simulation. Difference for Delay time &
4.5

4
rise time is due to pump pressure which is injecting
3.5 water inside the drum is not matching ideally &
3
practically
2.5

1.5
VIII. REFERENCES
1

0.5
[1] Boiler Control and Optimization S. G. Duklow
0
0 10 20 30 40 50 60 70 80 90 100 110 120
(1970) B. G. Liptek (1985, 1995) X. Cheng, R.
time(sec.) H. Meeker, JR. (2005) Inputs by G. Liu (2005).
Fig. 7.IMC lambda=2 response. [2] Thesis by Roopal Agrawal, “Internet based Data
Ideal response of imc based PID tuning by lemda=2 Logging and Supervisory Control of Boiler Drum
chosen because it is giving minimum overshoot. Level Using Labview”, a thesis submitted for
Practical imc based pid tune in PLC for boiler drum partial fulfillment of the requirement for the
level control that was give below practical response.

ISSN (Print) : 2321-5747, Volume-2, Issue-2,2014

5
 International Journal on Mechanical Engineering and Robotics (IJMER)
__________________________________________________________________________
award of the degree of M.Tech in Electronics and regulating (integrating) processes” University of
Instrumentation Engineering-May 2012. Connecticut, Department of Chemical, Materials
and Biomolecular Engineering U-3222, 191
[3] Enrique Arriaga-de-Valle,Graciano Dieck-
Auditorium Road, 06269-3222 Storrs CT, United
Assad” Modeling and Simulation of a Fuzzy
States Received 3 August 2006; accepted 18
[4] Supervisory Controller for an Industrial Boiler” January 2007 Available online 10 April 2007.
Electrical Engineering Department ITESM,
[8] Book: chemical process control, An introduction
Monterrey Campus 2501 E. Garza Sada
to theory and practice George Stephanopolous.
Monterrey, NL Mexico, CP 64849.
[9] Andris Sniders, Toms Komass” Simulation of
[5] Antonio Visioli university of Brescia, Via Branze
Multi link Invarient Control System for Steam
38, I-25123 Brescia, Italy” Research Trends for
boiler” Engineering for rural Development
PID Controllers” Acta Polytechnica Vol. 52 No.
Jelgava, 23.-24.05.2013.
5/2012.
[10] T. Rajkumar,V. M. Ramaa Priya and K.Gobi”
[6] Linkan Priyadarshini, J.S Lather” Design of
boiler drum level control by Using Wide Open
IMC-PID Controller For Higher Order System
Control With Three Element Control System”
and Its Comparison with Conventional PID
AbhinavInternational Monthly Refereed Journal
Controller” International Journal of Innovative
of Research In Management & Technology
Research in Electrical, Electronics,
Volume II, April-2013.
Instrumentation and control engineereing Vol. 1,
Issue 3, June 2013. [11] Juan J. Gude and Evaristo Kahoraho” Control
Considerations in a Drum Level Control
[7] Jeffrey E. Arbogast, Douglas J. Cooper”
Prototype” IEEE ETFA’2011.
Extension of IMC tuning correlations for non-self


ISSN (Print) : 2321-5747, Volume-2, Issue-2,2014

6