Professional Documents
Culture Documents
N - Modeling and Simulation of Some Parts of A Fluidized Bed Combustor
N - Modeling and Simulation of Some Parts of A Fluidized Bed Combustor
N - Modeling and Simulation of Some Parts of A Fluidized Bed Combustor
73
Modeling and Simulation of some Parts of a Fluidized Bed Combustor
Comparative statistical analysis has (Tb) wall temperature (Tw) and tube
become a ready tool for prediction in surface temperature (Tt). The definitions of
several engineering practices. Regression these variables are:
analysis was used to validate the clinker • Bed temperature (Tb), is the value
and cement mill in the production line of of temperature taken inside
Ashaka cement PLC. The resulting models bubbling bed of combustor using
were to predict effects of failure and thermocouple and temperature
production rates on electrical energy index meter.
in Ashaka cement production [2]. Models • Steam temperature (Ts), is the
were validated for implicit and explicit average temperature of steam
Runge-Kutta methods to solve combustion produced which is collected
problems such as heat losses to the walls, periodically via steam exit pipe.
influence of complex chemistry of flames • Tube surface temperature (Tt), is
and turbulence interactions [3]. Dynamic the average temperature of tube
models for a bubbling fluidized bed bioler surface obtained.
were validated in an operational 260 MW • Combustor wall temperature (Tw),
thermal power plant. The boiler process is the value of temperature taken at
data were used as input to the model and wall surface.
then compared the computer signal with The hypothesized relationship
measured output [4]. A group of between steam temperature as dependent
researchers solved a set of three models variable while bed, wall and tube surface
(simplified model, simulator and improved temperature as independent variables may
model) for some parameters of bubbling be written as:
fluidized bed combustor [5]. The Ts = bo + b1Tb + b2Tt + b3Tw + α …1
simplified model had air flow rate, fuel
flow rate, ambient air temperature and wall where;
pipe temperature while char mass, solid b0 is coefficient intercept of model that
temperature and solid mass were the gives steam temperature at ambient
steady state variables [5]. In order to condition .
improve the estimation of steady state b1 is coefficient that shows effect in
variables and overcome the problem of steam temperature due to increase in bed
running the model according to operating temperature
conditions, an Extended Kalman Filter b2 is coefficient of Ts associated with Tt
(EKF) was employed to obtain an
while holding Tb and Tw constant.
improved steady state estimation of
nonlinear systems [5]. The applicable b3 is coefficient of Ts associated with Tw
theory is discussed in the next section. while holding Tb and Tt constant.
2. THEORY OF COMPARATIVE α is other factors that influences steam
STATISTICAL VALIDATION temperature.
Multiple regression analysis is a statistical The assumptions used in our
tool that allows several factors to enter the application of regression analysis are:
analysis separately so that the effect of i. There exist a linear relationship
each can be estimated and to investigate between steam temperature on one
the relationship between the variables in a hand and the variables such as bed,
system [6,7]. Independent variables are tube surface and combustor wall
variables that change at start of a process. temperatures on the other.
The dependent variable is a variable that is ii. Multicolinearity not present
observed [8]. Some of these factors that between the independent variables
affect steam temperature are water, air and (bed, tube surface and combustor
fuel flow rates as well as bed temperature wall temperatures).
74
I. I.Ozigis, H. Dandakouta and G. Egbo
75
Modeling and Simulation of some Parts of a Fluidized Bed Combustor
From eqn. (2), the degree of freedom in addresses errors observed in the
denominator n2= 3. experiments. The new theoretical model
The critical values of T in statistical table, serves as reference data and a set of guess
T0.025 = 3.182 [9]. values for experiments were made for
Then compare 3.182 to obtained values for computation until a satisfactory
T-statistic in each analysis of variance. convergence is achieved for the modified
3.3 Computation Technique model. The modified models will improve
The data as shown in Appendices (I-X) for the estimation of the variables in both
simulated and experimental results of the simulation and experimental
fluidized bed combustor were entered into measurements of bed, steam, tube surface
computer when PASW [10] window was and wall temperatures. The comparative
opened. The variables were entered based analysis for bed, steam, tube surface and
on simulated and experimental data, and wall are shown in Appendices XI-XIV.
the program run for the comparative The new models equations were built by
regression output using the raw data. The summations of coefficients as shown in
experimental work extends the theoretical Tables (1-8) and equations (4 to 11) which
model to obtain an improved model that are similar to equation 1.
76
I. I.Ozigis, H. Dandakouta and G. Egbo
77
Modeling and Simulation of some Parts of a Fluidized Bed Combustor
1400
1200
Temperature(K)
1000
800
600 Experimental Bed(K)
Simulated Bed(K)
400
Improved Bed(K)
200
0
0 10 20 30 40 50 60 70
Time(mins)
Fig.1: comparative bed temperature against time for burning single fuel (lignite)
78
I. I.Ozigis, H. Dandakouta and G. Egbo
and steam flow rates of 0.014 kg/s and Appendix XII. The improved model for
0.012 kg/s, respectively when burning fuel the steam temperature gave a maximum
mix ratio of 90:10 (lignite: poultry value of 559 K. The correction factor
droppings) in the combustor. Comparative obtained through arithematic procedures
analysis carried out for the validation of results in an improved model as:
the model indicated a very good fitting of Ts = −2056 .1 + 0.4t − 0.023Tt + 7.81Tw
the data (R2 = 0.988) and standard error of … 13
the estimate of 2.4 % as shown in
700
600
Temperature(K)
500
400
300 Experimental Steam(K)
Simulated Steam(K)
200
Improved Steam(K)
100
0
0 10 20 30 40 50 60 70
Time(mins)
Fig.2: comparative steam temperature against time for burning fuel mix 90:10.
79
Modeling and Simulation of some Parts of a Fluidized Bed Combustor
700
600
Temperature(K)
500
400 Experimental Tube Surface (K)
300
Simulated Tube Surface(K)
200
100 Improved Tube Surface(K)
0
0 20 40 60 80
Time(mins)
Fig.3: comparative tube surface temp against time for burning fuel mix 70:30.
iv. Comparative Combustor Wall and burning rate of 0.015 kg/s.
Temperature Experimental combustor wall temperature
Figure 4 shows comparison between the indicated a higher value when compared
experimental and simulated combustor with simulated value; this is because of
wall temperature when burning fuel mix over simplification of parameters of
ratio of 50:50 (lignite: poultry droppings) refractory brick lining of the combustor.
in the fluidized bed combustor. The data Comparative statistical analysis carried out
obtained from theoretical model forms the between the experimental and simulated
bases of experimentation. The noticeable combustor wall temperature shows R2 of
variations in experiment were then 0.978 and error of 1.02 % [8] (see
incorporated for an improved model. Wall Appendix XIV). The modified model for
temperature is needed for combustion the combustor wall temperature was found
control, restart process and evaluation of to reach 338 K (maximum). The improved
heat losses to the walls. The experimental model given as:
and simulated combustor wall temperature Tw = 324.95 + 1.64t − 0.077Tb − 0.005Tt
were found to have reached maximum … 15
values of 335 K and 331.4 K respectively,
at an average steam flow rate of 0.016 kg/s
345
340
335
330
Temperature(K)
325
320
315 Experimental Combustor Wall(K)
310 Simulated Combustor Wall(K)
305
300 Improved Combustor Wall(K)
295
290
0 20 40 60 80
Time(mins)
Fig.4: comparative wall temperature against time for burning fuel mix ratio (50:50).
80
I. I.Ozigis, H. Dandakouta and G. Egbo
81
Modeling and Simulation of some Parts of a Fluidized Bed Combustor
Appendices
Appendix I: Predicted Combustion of 100:0 (Lignite: Poultry Droppings).
S/ Time Air, Fuel, Steam, Improved Bed Steam Wall Tube
N (min) kg/s kg/s kg/interval Bed (K) (K) (K) (K) (K)
1 0 0.11 0.009 8 314.24 303 303 303 303
2 5 0.112 0.0095 8.5 452.84 361.6 303.3 303.2 305.4
3 10 0.1135 0.0095 9.05 621.79 423.5 304.9 303.8 312.2
4 20 0.1164 0.010 10.3 777.28 557.0 314.8 306.3 339.3
5 30 0.12 0.010 11.5 931.25 703.0 335.9 310.6 385.1
6 40 0.1230 0.012 12.6 993.55 860.9 370.2 316.9 452.2
7 50 0.127 0.013 13.95 1044.43 1029.7 419.4 325.3 543.7
8 60 0.13 0.015 15 1194.96 1208.6 485.7 335.9 665.4
82
I. I.Ozigis, H. Dandakouta and G. Egbo
83
Modeling and Simulation of some Parts of a Fluidized Bed Combustor
84
I. I.Ozigis, H. Dandakouta and G. Egbo
Appendix XII: Comparative Steam Temperature using Simulated and Experimental Results
Paramet Valu ANOVA Collineari Diagnostic Residual
er e ty
Paramete Sum Coefficie T- Sig. Parameter Mean
r of nt Stat F Square
Square
R2 0.98 Regressi 3301.5 Constant -99.89 -9.25 0.00 Regression 1650.7
8 on 6
F- 291. Residual 45.35 Experime 0.252 6.88 0.00 Residual 9.07
Statistic 8 nt 4 1
Sig.F 0.00 Simulatio 0.068 5.47 0.00
n 9 3
Std error 2.4
Appendix XIII: Comparative Tube Temperature using Simulated and Experimental Results
Parameter Valu ANOVA Collinearity Diagnostic Residual
e
Paramete Sum Coefficien T-Stat Sig. Paramete Mean
r of t F r Square
Square
R2 0.93 Regressi 3273.2 Constant -80.24 -7.833 0.00 Regressi 1636.6
8 on 1 on
F- 111. Residual 73.7 Experiment 0.172 4.906 0.00 Residual 14.7
Statistic 0 4
Sig.F 0.00 Simulation 0.090 5.359 0.00
3
Std error 3.84
85
Modeling and Simulation of some Parts of a Fluidized Bed Combustor
Appendix XIV: Comparative Wall Temperatures using Simulated and Experimental Results
Paramet Value ANOVA Collinearit Diagnostic Residual
er y
Parameter Sum of Coefficien T- Sig. Paramete Mean
Square t Stat F r Square
R2 0.978 Regressio 3341.71 Constant -514.12 -39.1 0.00 Regressio 1670.8
n n
F- 1618. Residual 5.2 Experime 0.852 16.07 0.00 Residual 1.033
Statistic 2 nt
Sig.F 0.00 Simulatio 0.874 10.7 0.00
n
Std error 1.02
86