Simulation of Domestic Boiler Control

AS WHITE: SIMULATION OF DOMESTIC BOILER CONTROL
Simulation of Domestic Boiler Control

A. S. White
School of Engineering and Information Sciences,
Middlesex University,
Hendon,
London NW4 4BT
United Kingdom
Email a.white@mdx.ac.uk
Abstract A simple simulation using SIMULINK is presented to model the thermal/control behaviour of a domestic gas boiler.
The simulation was the first step in a research programme to investigate the efficacy of control using an oxygen sensor. A brief
review of boiler control is used to outline current practice and research developments. Ratio control in several forms has been
investigated. Performance improvements using control incorporating an Oxygen sensor are examined. These indicate that a
modest fuel saving of the order of 5% is possible from the addition of the sensor to the control loop. The performance of the sensor
is insensitive to small changes in position and to a 10% variation in the time constant of the sensor. The fuel air ratio varies
throughout the heating cycle with only a small period when either system achieves accurate control
Keywords-Boiler contr; ratio control; oxygen sensor; SIMULINK; PID control
I.
described by partial non-linear differential equations. The

heat exchange to the water and hence to produce steam is
also a process described by equations that can only be
solved completely with extensive use of computer aided
design packages. Fortunately control systems can be
designed with much simpler information based on a series
of "lumped" parameters.
The first major analysis of the problem was given by
[2]. They outlined simpler equations that describe the
whole process well, including the heat exchange to the
water, the steam generation. The effects of combustion are
allowed for by assuming that the combustion temperature is
proportional to the fuel flow rate. Subsequent authors have
followed this analysis. Prior to the work of Chien the
design of control for boilers was largely ad hoc.
The combustion temperatures depends on; burner
design, combustion chamber design, fuel flow rate, airflow
rate and the heat exchanger characteristics.
INTRODUCTION
One of the first applications of control practice to

industrial technology was the use by James Watt to control
the generation of steam in his famous steam engine. This
was paralleled in the 1970's by the first use of Fuzzy control
by [1] also applied to the control of a steam engine.
However the use of automatic controls to alter process
applications, of which combustion is one, was not advanced
until the 1940's when the Proportional plus Integral plus
Derivative feedback controller was devised almost
independently in the USA and the UK by engineers in the
process industries. The theoretical principles, which were
established largely by the beginning of the 1950's, were
those termed now "classical" control. The advent of fast
digital computers in the 1960's allowed the development of
"state-space" control with the optimal controllers now used
widely in servomechanisms and process control as well as
the power industry using Direct Digital control.
III.
II.
BOILER CONTROL
BOILER PROCESSES
Techniques used in boilers are similar to other methods
developed for process control and used terminology
somewhat different to the main control community wedded
to servomechanisms.
Reference [3] describes these
techniques in a seminal work. Many industrial procedures
in use today are included in this reference with the
exception that most are now implemented in digital form
using microprocessors. These procedures were classical in
The processes involved in boiling water to produce

steam in power plants are very complex and non-linear. The
analysis of the combustion process depends strongly on the
burner configuration and the shape and size of the
combustion chamber. It would involve an extensive use of
Computer Fluid Dynamics (CFD) analysis to gain a
thorough understanding of the flow regimes that are
IJSSST, Vol.11, No. 4
22
ISSN: 1473-804x online, 1473-8031 print

manufacturers' data, to be almost always Proportional plus
Integral (PI) devices. Since the basic component controllers
are PI the tuning procedures are well known for use by
installation technicians. Modern controllers would require a
different order of competence.
formulation. Boilers as with other process control suffer

delay times in the process and special methods have to be
used to control those [4] tuning PID controllers with neural
networks. Many of the techniques being researched are
quite sophisticated because of the non-minimum phase
response of heat exchange systems. Current research is
typified by [5], who designed optimization for load
following, [6] who applied modern H control, [7] who used
Lyapunov control. Fuzzy control has also been investigated
but little evidence is available in order to judge its' merits.
A substantial amount of current research is either Eastern
European or Chinese in origin and translations are not
readily available. Possible controllers that have merit are
Neural Networks combined with Fuzzy systems. These
controllers would be trained how to operate and hold some
advantages over other devices. It would be easier to set up
than some more complex controllers and could deal with
very complex situations.
Robust optimum PID control is described by [8] who
obtained asymptotic solutions to the eigenstructure
assignment comparing favorably with loop recovery
methods.
Reference [9] also examined robust control using loop
shaping to achieve good PID performance.
IV. OTHER SIMULATION

Apart from the very sophisticated simulators for
training power station staff the use of simulation in control
of boilers is scarce. Two examples are described below.
5
4
7
6
A. oiler Control Practices
9
9
Commercial boilers in practice are governed by

ABMA-Boiler 301 and other standards. The current state of
the art is described by [10]. Reference [11] described a
simulator used to assist with the implementation of PLC
controllers.
In a Domestic boiler there are only three quantities that
are varied; the water flow rate, the airflow rate and the fuel
flow rate. Control of the steam generation, for power, is
normally performed by controlling the ratio of air to fuel to
achieve satisfactory combustion, with the ratio being held at
a value to guarantee adequate combustion without
dangerous conditions resulting. Central heating "boilers" do
not normally produce steam. However there is little
difference from a control point of view.
Very simple central heating systems are operated in an
"on-off" or "bang-bang" mode. This type of operation
would not allow the implementation of oxygen sensors in
any meaning full sense and will not be considered further
here.
Basic procedures currently used in power generation
and in central heating called ratio control [12], are shown in
figure 1. Here the temperature is measured and fed to the
master controller, which compares the measured value to the
set point and it sends a request for action to the fuel valve
controller. The air flow controller acts in a set ratio to the
fuel flow controller. There are two variants, cross limiting
control and cross limiting with O2 trim (fig 2& 3). The
controllers used in operating the valves appear, from
Figure 1 Ratio Control
Reference [13] describes a simulation of several marine

applications including a boiler with PI control of drum
pressure and air turbine flow rate using a PACE TR10
analogue computer, however no results were shown.
More recently, [14] describe a Simulink simulation of a
coal-fired steam boiler. Most of the details of the simulation
are hidden but some results of optimum
5
4
LS
HS
9
9
F
Figure 2 Cross-limiting control
23
The maximum gas flow rate is 0.7 l/s and the central
heating water flow rate and hence the flow rate for the
primary heat exchanger is constant at is 0.37 l/s. With the
domestic heating enabled, the take off is 12 l/min using a
mains supply at ambient temperature Te.
The gas valve appears to be a propriety servo gas valve
with PID control of pressure. Data from a typical
commercial Honeywell valve VK4105G has been used to
model this behavior. This has a time constant of 1 second.
The valve controls pressure and this relationship is roughly
proportional to the square of the Pulse Width Modulation
(PWM) signal in percentage. Since the flow rate in any
valve is proportional to the square root of pressure
difference, we can justifiable say that the control of flow
rate is approximately linear with PWM.
The heat exchanger characteristics were not available
and modeling this part is left to a more advanced and later
model. The heat exchange process is much lower than the
control actions.
From discussions with the engineers at Ideal it appears
that the fan is controlled via PWM speed control. The flow
rate is NOT measured and a rotational speed is selected and
controlled via a PID controller rather than a phase-locked
loop. They did not reveal how long it took the fan to ramp
up to speed, nor did they supply any details of the fan inertia
etc. in order to calculate it. From measurements on other
boilers the time constant of the fan is between 12 and 20
seconds.
Since the control sequencing of the microprocessor is
not available, discrete representation of the control system
has not been attempted. Due to the long time constants this
is not likely to cause serious errors. No data is yet available
to indicate any connection between the air/fuel ratio and the
burner performance and this has therefore been omitted
from this simple first
%O2
5
4
7
6
9
Figure 3 Cross-limiting with O2 trim control
combustion control are given and details of investigating a

faulty oxygen sensor are shown. The control used is a statespace optimal control scheme. Other work of these
important researchers is unfortunately in Czech.
Reference [15] examines a discrete model of a boiler
identified using a least squares weighting. They illustrate
various methods of PID controller design. Reference [16]
analyses model development for large scale boiler plant
giving a full nonlinear thermodynamic model with control
inputs using a MATLAB representation. This problem was
solved using constrained optimization. The use of an
Oxygen sensor in control of a domestic boiler is described
by [17]. They used control of two loops, one for
temperature, and the other controlling the percentage of
Oxygen. This is not the same as the ratio control used here
as this work has three loops to control the airflow separately
as well as the temperature and air/fuel ratio. However the
results of the controller used by Conte are within 10% of the
optimum ratio.
VI. EQUATIONS OF THE SYSTEM
V. DESCRIPTION OF SYSTEM
In order to produce a SIMULINK model we need to

describe the physical and control systems. These equations
are simplified due to the lack of experimental and
commercial data. The simplified model described is the
minimum to describe the physical boiler operation.
The system that has been modeled is a central heating

boiler based on an Ideal modulating model m30100
combination boiler. This machine uses natural gas as fuel.
It is shown schematically in figure 4.
%CO
qa
A. Physical Equations
We assume that no losses in heat occur except via the
heat exchangers. The energy balance for the Primary Heat
exchanger HEX1 is given by:
Ts 1
PWM
FanGain
6600/100
42/6600
mC
1
Figure 5 Fan control schematic
dT
m C T T Q
dt
10
10
For the secondary plate heat exchanger we obtain two

equations.
24

C. Fuel control
The fuel flow is always positive hence we need to limit
the fuel output to a maximum of 0.7 l/s and there will
possibly be a minimum flow rate too for stable operation of
the burner (this is not specified at this time). The fuel valve
has a time constant of 1 second. A schematic of the system
is shown in figure 7 below:
Ignoring the change of internal energy in the primary

circuit we have:
m C
12
dT
m C T T Q 0
dt
10
The domestic hot water has however a change in

internal energy to account for:
VII. SIMULINK MODEL
dT
mC
m C T T Q
dt
20
20
Simulink is a computer simulation package that was

devised by Mathworks and runs with MATLAB. The
package is an icon driven simulator using standard control
system functions to represent integrators, delays, scopes and
signal sources. The model is provided in two versions, one
figure 8, just includes the boiler with simple ratio control
and the second, figure 9 that includes the O2 sensor control
system.
We can substitute equation 2 into 1 and also take

Laplace transforms, hence:
Q Q
T
s
1
10
A. Fuel control
The boiler equations are set up at the top of the model
diagram as two variables T10 is obtained by integrating the
inputs from Qb and Qw. This explains why the response
curves increase linearly with constant fuel input. The other
variable T20 is obtained using a simple transfer function with
a time constant. In the real boiler a thermistor measures the
temperature T10 at a location down-stream of the heat
exchanger hence a time delay is incurred and the thermistor
has its' own time constant.
The time delay is constant
because the water flow rate is fixed. The error in T10 from
the required value, using a step function to ask for a change
in set point after a given time, is fed to a PID controller
which then requires a fuel rate via a maximum limit of 0.7
l/s. The lower limit has not been incorporated at this time.
This completes the boiler loop. The model incorporates
scopes at various points so that several variables can be
monitored. The second loop in this model is that for the
airflow again using a PID controller. A set-point in fuel flow
rate is devised from knowledge of the required operational
fuel/air ratio and a measure of the airflow and the real fuel
flow rate is subtracted to obtain an error which is then acted
upon by the PID controller. Since for the boiler represented
the airflow is not measured a signal from the controller is
used instead as would be in practice. An estimate of the real
fuel/ air ratio is computed in the model.
And from 3 we obtain:
m C
m C
T
s
T Q
mC
mC
20
This can be arranged:
m C
T Q
m C
T
m C
s
m C
20
Thus if we know Qw ,Cp, Cv and the relevant masses and

mass flows we can simulate the response of the heat
exchange system.
B. Physical Equations
The fan characteristic supplied showed contradictory
data. However it is clear that the maximum flow rate at
6600 rev/min was 42 l/s. Fan speed is controlled by PID
and Pulse Width Modulation (PWM). This control is much
faster than the thermal processes described here and can be
approximated by a constant gain. This model is shown in
Figure 5. In ratio control as outlined earlier the input to the
airflow controller is the difference in fuel flow rate demand
from a set-point value. The rest of the airflow controller is
therefore shown in figure 6:
B. Sensor control
The prime difference in the two models (figure 9) is the
third loop introduced for the sensor to control the airflow.
The representation of the sensor includes a variable time
delay as the flow rate now varies and this is included. The
sensor is modeled as a simple time constant and gain. It is
the assumed that an output can be computed from the sensor
25

can variation of the airflow due to disturbances. This is an
important reason for having the sensor control as the normal
ratio control method does not allow disturbances to be
eliminated.
output to give fuel/air ratio at that time instant. The setpoint is fixed and a PID controller sends a signal via fixed
limits to the airflow controller as before. A block is include
so that a real airflow measurement can be included late as
OXYGEN
SENSOR
Flue
Fan
m a
Primaryheat
T10
exchanger(HEX1)
Ti
Burner
Water
pump
m 1
Waterfrom
Te
2
mains m
T20
Central
heating
water
To
radiators
Fuelflow
f
rate m
From
radiators
Figure 4 Schematic of boiler installation
26
+u(t)
+qfset
indistinguishable. However the large effect is a much better

control of air fuel ratio as shown in figure 15. This diagram
shows that neither system adheres to desired value of
Fuel/Air Ratio, but the system with the O2 sensor is closer to
the desired FAR for more of the cycle. The overall fuel
used by the ratio controller since the air flow rate is fixed is
greater and this for an average leaner mixture. This implies
a fuel reduction of between 2 and 3% for continuous
operation after startup. The ratio controlled system is
unlikely to be able to meet new standards on combustion.
The output is within 5% of the correct fuel/air ratio when
the oxygen sensor is used. This broadly agrees with the
work of Conte et.al.
%CO
PID
qf
Limiter
0100%
PWM
Figure 6 Airflow control
+qfreq
IX.
PID
The theoretical aspects of boiler control and central

heating control are well understood and analysis for control
purposes is a reasonable expectation.
Current control practice in power plants and central
heating is similar but less complex in central heating
applications. The best practice in power plants already
includes the use of oxygen sensors to allow closed loop
control of combustion.
More modern controllers have been designed for power
applications but have not been applied to central heating
systems.
Their use would have to be practically
implementable and yield substantially better performance.
Since the controllers are already based on microprocessors
the cost of implementation is not likely to be more but the
development cost would be substantially greater to prevent
safety problems.
A SIMULINK model has been developed for ratio
control and for O2 sensor control. These models indicate
that the O2 sensor control is more effective in controlling
air/fuel ratio and achieving closer combustion control
reducing CO2 emissions.
T s 1
PWM
Figure 7 Fuel control
VIII.
CONCLUSIONS
PERFORMANCE
The performance of the model has been verified in the

sense that the numbers produced agree with numerical and
calculus evaluations of the equations. However it has not
been validated in the sense that it has been compared to real
data from the given boiler. It seems unlikely that this data
exists or that we can obtain data from another similar
machine.
However it would be simple to do so
experimentally and it could be the basis for further work or
grant application.
ACKNOWLEDGMENT
This research was
Development Agency.
A. Detailed outputs
The cyclic heat input to the domestic hot water is
shown in figure 10. The response of the primary hot water,
the secondary hot water temperature and fuel flow rate are
given in figures 11, 12 & 13. Apart from the gains of the
sensor PID controller every other constant is held the same.
Here we show the comparison of the two systems. There is
little difference between them (Figs 11, 12 & 13) in this
case and the O2 sensor system is clearly able to
When the sensor position and time constant are altered there
is only a second order effect on the overall response for a
variation of about 30% in values of position and time
constant.
The curves shown in figure 14 are
supported
by
the
London
REFERENCES
[1]
[2]
[3]
[4]
[5]
27
T.J. Procyk and E. H. Mamdani, A linguistic Self-organising

Process Controller, Automatica, 15, 1, pp.15-30, 1979.
K. L. Chien, E I Ergin, C Ling and A. Lee. A Dynamic Analysis of
a Boiler, Transactions of ASME, November, pp.1809-1819, 1958.
S.G. Dukelow, The Control of Boilers, ISA press, 1986.
D.H. Kim, PID Tuning of a Boiler Control System using Immune
Algorithm Typed Neural Network, Computational Science ICCS
2004, 3037, pp. 695-698.
J. H. Mortensen, T. Moelbak, P. Andersen and T. S. Pedersen,
Optimization of boiler control to improve the load-following
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
capability of power-plant units, Control Engineering Practice, 6,

12, pp. 1531-1539, 1998.
W. Tan, H.J. Marquez and T. Chen, Multivariable Robust
Controller Design for a Boiler System, IEEE Trans. on Control
Systems Technology, 10, 5, Sept., pp.735-742, 2002.
H.W. Rder, and G.S.Rsel, Strategies for adjustment of the
complete boiler control system in power units, Annual Review in
Automatic Programming, 12, 2, pp. 430-433, 1985.
H. Zargarzadeh;M. Motlagh and M. Arefi, Multivariable robust
optimal PID controller design for a non-minimum phase boiler
system using loop transfer recovery technique, Control and
Automation, 16th Mediterranean Conference, Digital Object
Identifier: 10.1109/MED.2008.4602170, pp. 1520 - 1525
Zhiqi Hu; Hanbai Fan; Yujun He, Robust Controller Design for
Boiler System; Intelligent Computation Technology and Automation
(ICICTA), International Conference, 1, Digital Object Identifier:
10.1109/ICICTA.2008.430, pp. 445-449.
J. Gilman, Boiler Control Systems Engineering, ISA Press, 2004.
I. Yoshiharu and N. Eiichi, An Auxiliary Boiler Control Simulator
for Training Marine Engineers on the PLC systems, J. Japan
Institution of Marine Engineering, 38, 3, pp. 159-163, 2003.
C.A. Smith and A.B. Corripio, Principles and Practice of Automatic
Process Control, Wiley, 2nd Ed., 1997.
C. Read, Marine Engineering Applications, Analogue Computer
Applications, Symposium held at Paisley College of Technology,
1967, Ed A McKenzie, pp.82-94.
P. Neuman, B. Sule, P. Zitek and T. Dlouhy, Non-linear engineering
simulator of a coal fired steam boiler applied to fault detection of
optimum combustion control, IFAC Conf. on Fault detection,
supervision and safety for technical processes, 2000, Budapest,
Hungary, pp.905-910.
S. Simani and S. Beghelli, PID controller design application based
on boiler process model identification, Decision and Control, 2007
46th IEEE Conference, Digital Object Identifier:
10.1109/CDC.2007.4434092 , pp. 1064-1069
E. Aranda, M. Frye, Chunjiang, Qian, Model development, state
estimation, and controller design of a nonlinear utility boiler system;
Industrial Technology, ICIT. IEEE International Conference on
Digital Object Identifier: 10.1109/ICIT.2008.4608412, pp. 1- 6
G. Conte, M. Cesaretti and D. Scaradozzi, Combustion control in
domestic boilers using an oxygen sensor, Control and Automation,
MED '06. 14th Mediterranean Conference, 2006, pp. 1-4.
m2
Mass of water in domestic hot water side of plate
heat exchanger
Mass flow rate of water in primary heat exchanger
m 1
Qb
Heat energy input from fuel
Qw
Heat energy input from primary circuit to
Mass flow rate of gas in the boiler
Greek
1
2
Effectiveness of primary heat exchanger

Effectiveness of secondary heat exchanger
GLOSSARY
HL or LL-High or low limit.

HS or LS -High or low switch.
Set Point (SP)-The desired value of a variable.
Temperature sensors-these are usually thermocouples or
platinum resistance thermometers.
KEY TO FIGURES
1
2
3
4
5
6
7
8
9
Roman
Specific Heat at constant pressure
Specific Heat at constant volume
Mass of water in primary heat exchanger
Mass of water in primary side of plate heat exchanger
Mass flow rate of water of domestic hot water
domestic water
s
Laplace variable
T1o
Temperature at exit of primary heat exchanger
T20
Temperature at exit of secondary heat exchanger
for domestic water
Environmental temperature at entry to secondary
Te
heat exchanger
Ti
Temperature at entry of primary heat exchanger
LIST OF SYMBOLS
Cp
Cv
m1
m12
m 2
m f
28
water flow
temperature sensor
exhaust
set point
master temperature controller
ratio computation
flow controllers
to burner
flow sensors
This model employs ratio control of fuel/airflow for a domestic boiler. Hotwater supply is triggered by a heat input
0.4
s+0.4
qf/qa
1
kc
cyclic demand
Qw
burner f/a ratio

recip
kd
0.4
Qw
20
Display
-0.0001686
environmental
tempTe
20
Display1
real airflow
T10
0.0714
T20
heat exchanger
time constant
airflow
measurement
0.0
f/a setpt
s
Integrator
kb
fuelenergy ka
computed fuel
-Kqf
Product2
qdf
1
air Controller PID
T ransport
Delay
gas valve
s+1
airdisturbance
30
66
s+1
tempcorr
gas flow limits1
PWM
N
temp sensor
zeropwm
airflowsignal
air Controller1
PID
fan gain
measured temp CHW
0.0
2
temperror
setpointtemp
fan dynamics
5s+1
Temperature control loop
airflow
Airflow Control Loop
Figure 8 Ratio Boiler Control

This model employs O2 sensor ratio control of fuel/airflow for a domestic boiler. Hotwater supply is triggered by a heat input
burner f /a ratio
qf /qa
burner f /a ratio 1
1
-K s+0.4
1
To
recip
u
-K-
cyclic demand
Qw
s+1
Variable
Transport Delay
axd
0.0446
O2 sensor
kc
kd
Qw
T20
heat exchanger
time constant
0.4
f/a setpt
1
nI
1
u
1/qd
Display
environmental
tempTe
fuel /air ratio controller
1t
u
O
20
T10
rfldel
1
s
-K -
limits to f /a ratio
kb
real airflow
Integrator
fuelenergy ka
-K-
computed fuel
qf
Product 2
1
airflow
measurement
air Controller
qdf
Transport
Delay
PID
gas valve
s+1
airdisturbance
Display 1
30
temp sensor
zeropwm
gas flow limits 1
PWM
N
66
5
tempcorr
1
s+1
airflowsignal
air Controller 1 PID
fan gain
measured temp CHW
-K2
temperror
fan dynamics
setpointtemp
1
5s+1
Temperature control loop

airflow
Airflow Control Loop
Figure 9 Boiler Control with O2 sensor
29
Heat Transfer to Domestic Hot Water

30
25
20
W
k/
w
Q
15
10
20
40
60
80
100
Time/s
120
140
160
180
200
Figure 10 Simulated demands of Heat transfer to domestic water
Boiler Water temperature

80
70
60
g
e
d/
1
T
er
ut
ar
e
p
m
et
r
et
a
w
r
eil
o
B
50
40
30
O2 sensor
Ratio control
20
10
20
40
60
80
100
Time/s
120
140
160
180
200
Figure 11 Temperature in primary water circuit
30
Domestic Heating Water Temperature

60
50
40
g
e
d/
2
T
er
ut
ar
e
p
m
e
T
30
20
10
O2 sensor
ratio control
20
40
60
80
100
Time/s
120
140
160
180
200
160
180
200
Figure 12 Response of domestic hot water to demand
Fuel flow Rate for O2 and ratio Control

0.8
0.7
0.6
0.5
1s
/l
et
ar
w
ol
F
l
e
u
F
0.4
O2 sensor
ratio control
0.3
0.2
0.1
0
-0.1
20
40
60
80
100
Time/s
120
140
Figure 13 Fuel flow rate
31
Variation in Air/Fuel Ratio for varying Sensor time constant and position
0.09
0.08
0.07
0.06
oi
t
a
R
ri
A
l
e
u
F
0.05
0.04
0.03
0.02
0.01
0
10
15
20
25
Time/s
30
35
40
45
50
Figure 14 Effect of varying sensor position and time constant
Fuel /Air Ratio for Ratio and Oxygen Sensor Control

0.12
0.1
Ratio Control
FARD = 0.0714
0.08
oi
t
ar
ri
A/
l
e
u
F
Oxygen sensor
FAR=0.068
0.06
FAR= 0.0654
FAR=0.7162, time=20.302s
0.04
0.02
10
20
30
40
Time/sec
50
60
70
80
Figure 15 Comparison of sensor and ratio control
32

Simulation of Domestic Boiler Control

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Simulation of Domestic Boiler Control

Загружено:

Авторское право:

Доступные форматы

AS WHITE: SIMULATION OF DOMESTIC BOILER CONTROL