YENEPOYA INSTITUTE OF TECHNOLOGY

Department of Mechanical Engineering

Seminar presentation
on
“IMAGE BASED CHARACTERIZATION OF
ALETRANTIVE FUEL COMBUSTION WITH
1
MULTIFUEL BURNERS”
Guidance by: Presented by:
Prof. J W PINTO Chethan P A
Assistant Professor, 4DM14ME013
22-04-2018
Mech. Dept.,
YIT, Moodbidri
2 CONTENTS
 INTRODUCTION
 BASICS FOR IMAGE PROCESSING AT MULTIFUEL BURNERS
 NEW METHODS FOR ALTERNATIVE FUEL STREAKLINE
DETECTION
 RESULTS
 CONCLUSION
 REFERENCE

22-04-2018
 3 INTRODUCTION
 In the field of industrial combustion processes, the number of
multifuel burners has increased significantly during the past
years.
 These multifuel burners can use alternative fuels like plastic,
tire chips, or animal meal in arbitrary high fractions, besides
fossil primary fuels such as brown coal.
 Due to declining availability of fossil fuels, the use of these
low-cost, low-ranked, and in large quantities available
alternative fuels provides a great savings potential.
 At the same time, the increased use of alternative fuels can
Fig. 1: Example of a multifuel burner.
not only reduce costs but also reduce emissions.
 Alternative fuels differ in their combustion characteristics
basically from fossil fuels.

22-04-2018
 4 BASICS FOR IMAGE PROCESSING
AT MULTIFUEL BURNERS
Multifuel Burners
 The primary fossil fuel is added by the coaxially arranged
coal channel.

 For alternative fuel, two supply channels are available.

 In this application, a mixture of plastic and tire chips are

fed to channel 1 and animal meal to channel 2.

 The flame shape can be influenced by the coaxially

arranged swirl air channel.

 The exit angle of the alternative fuel can be influenced by Fig. 2: Schematic description of the front of a
the pressure of a vertical air flow, called pneumo deflector. multifuel burner. 22-04-2018
5 Cont…

 The combustion behaviour of individual fuel

portions can be retraced from a lateral view.

 The animal meal is not included in this

figure.

 The alternative fuel mixture of channel 1

leaves at the centre of the burner mouth and
forms the fuel streakline.

 The blue line in Fig. 3 describes the average

fuel streakline. Fig. 3: Lateral schematic view of a multifuel burner with
leaving fuel.
22-04-2018
6
Cont…
Infrared Camera System

 The different degrees of transmission of

combustion gases and solid particles gives the
limited visibility of fuel within the flame.

 Relevant combustion gases are water vapour

(H2O), carbon dioxide (CO2), and carbon
monoxide (CO).

 Relevant solid particles are soot or dust.

Fig.4: Comparison of the views to a multifuel burner
with a camera system using the (a) visual, (b) near-
infrared, and (c) long-wavelength infrared range.
22-04-2018
 Scene
7 Cont… I

A
Image acquisition
Image processing G

E
P
Image pre-processing
R
 Image-processing methods can be used
O
for automatic analysis of camera C
monitored scenes. Image analysis/feature
E
extraction
S
S

I
Parameter calculation
N

Scene interpretation

22-04-2018

Fig. 5 Steps of image processing.

 8 Cont…
Experimental Setup

 Images of a multifuel burner were acquired at the

rotary kiln head of an industrial cement plant.
 The thermal conversion of raw material to clinker
takes place in a rotary kiln with preheater.
 Dropping and incompletely burned fuel ends up in
the reaction area of the raw materials and directly
affects the chemical conversion process and also the
corresponding final product.
 A long-wavelength infrared camera (Pyroinc
380LF) with a resolution of 384 × 288 pixels and a
Fig. 6: Infrared images with different settings of a
frame rate of 50 Hz from the company DIAS
multifuel burner. (a) and (b) gives the variations of
Infrared was used for the image acquisition. pneumo deflector and variation of swirl is given in (c)
and (d). 22-04-2018
 9
NEW METHODS FOR ALTERNATIVE
FUEL STREAKLINE DETECTION Image pre-processing
Image

Average Fuel Streakline ROI parameter Reduction to ROI

Image analysis

 Robust detection of the average fuel streakline is Recursive time averaging

proposed basis for the extraction of further image

Invert image
based parameters.

 Streaklines are the loci of points of all particles, Parameter peak

detection
Peak determination

which originated from the same initial position. Parameter extraction

Gauss estimation

 A first evaluation of the fuel streakline is

obtained by detecting the dark tail of the
Column wise mean Column wise standard
alternative fuel close to the burner mouth. value deviation

Fig. 7:Flow diagram for the evaluation of the

22-04-2018
fuel tail close to the burner mouth.
10 Cont…
Particle Detection

 This method allows the detection of fuel in the areas

of a compact fuel tail especially close to the burner
mouth.

 To be able to determine the location of the

alternative fuel also in areas further away from the
burner.

Fig. 8:Flow diagram for the particle detection22-04-2018

of alternative fuel.
11 Cont…
Combining Average Fuel Streakline and Particle
Detection

 The average fuel streakline and particle detection are combined in this section for
the location of alternative fuel.
 A large number of detected particles rise the influence of the particle detection, and
the average fuel streakline has no longer influence on the final result.

22-04-2018
 12 RESULTS
Generation of Reference Data
 A software tool has been developed, which allows to
set manually the fuel streaklines.

 Due to the large amount of image data, 56 image

sequences during an experiment with varying
settings of the multifuel burner were selected and are
evaluated by experts.

 The average fuel streakline over all expert

streaklines is used as reference data for each
sequence.
Fig. 9:(a) Reference data for fuel streakline estimation. (b)
Comparison of the average fuel streakline of the new methods
(yellow) and the reference data (blue). 22-04-2018
 13 Cont…
Results Compared to Reference
Data
 The deviation is measure by the size of the area
between the two streaklines.

 Deviation to reference streaklines (area

between two streaklines) for automatically
detected fuel streaklines (blue) and for the
average deviation of all experts’ streaklines
(red; red area: minimum and maximum
deviation in the reference data) .

 The dots reflect the analysed sequences, the

lines interpolate the time between the Fig. 10: The comparison with reference data
examined sequences. 22-04-2018
14 Cont…

Fig. 11:Burner settings during reference data sequences.

22-04-2018
15 Cont…
Table 1: Comparison of cumulative deviation of the different methods

Method Cumulative deviation

Reference data 13 472.1

Average fuel streakline 14 635.8
Particle detection 35 253.0
MAP estimation 20 193.3
Adapted MAP estimation 12 994.0
22-04-2018
16 CONCLUSION
 The new camera-based methods for automatically detecting the fuel streakline allow an
early detection of changes in the combustion characteristics of alternative fuel.
 Even under difficult visual conditions as in cement production, the detection of the fuel
streakline is possible and deviates only slightly from the evaluation of experts.
 The current combustion behaviour of the alternative fuel are derived based on the fuel
streakline and new image features.
 Therefore, it is possible to facilitate a constant operation with higher fractions of
alternative fuels.
 A long-term measurement campaign at an industrial plant.
 It will be focused on the analysis of varying alternative fuel properties and on the
parameterization and applicability of the process control concepts.

22-04-2018
17 REFERENCES
[1] A. Rahman, M. Rasul, M. Khan, and S. Sharma, “Impact of alternative fuels on the
cement manufacturing plant performance: An overview,” Procedia Eng., vol. 56, pp. 393–
400, 2013.
[2] A. K. Amiruddin and S. Sapuan, “Burner flame image analysis techniques,” in Proc.
Nat. Conf. Softw. Eng. Comput. Syst., 2007.
[3] J. Ballester and T. Garcia-Armingol, “Diagnostic techniques for the monitoring and
control of practical flames,” Prog. EnergyCombust. Sci., vol. 36, pp. 375–411, 2010.
[4] W. B. Baek, S. J. Lee, S. Y. Baeg, and C. H. Cho, “Flame image processing and
analysis for optimal coal firing of thermal power plant,” in Proc. Int. Symp. Ind. Electron,
2001, pp. 928–931.
[5] W.-B. Horng, J.-W. Peng, and C.-Y. Chen, “A new image-based real-time flame
detection method using color analysis,” in Proc. Int. Conf. Netw., Sensing Control, 2005,
pp. 100–105.
22-04-2018
18 Cont…
[6] C. Duchesne, J. J. Liu, and J. F. MacGregor, “Multivariate image analysis in the process
industries: A review,” Chemometrics Intell. Lab. Syst., vol. 117, pp. 116–128, 2012.
[7] H. Yu and J. F. MacGregor, “Monitoring flames in an industrial boiler using multivariate
image analysis,” AIChE, vol. 50, no. 7, pp. 1474–1483, July 2004.
[8] G. Lu, G. Gilabert, and Y. Yan, “Vision based monitoring and characterisation of
combustion flames,” in Proc. J. Phys. Conf. Ser. XIII, 2005, vol. 15, pp. 194–200.
[9] H. Chen, X. Zhang, P. Hong, H. Hu, and X. Yin, “Recognition of the temperature condition
of a rotary kiln using dynamic features of a series of blurry flame images,” IEEE Trans. Ind.
Informat., vol. 12, no. 1, pp. 148–157, Feb. 2016.VOGELBACHER et al.: IMAGE-BASED
CHARACTERIZATION OF ALTERNATIVE FUEL COMBUSTION WITH MULTIFUEL
BURNERS 597
[10] J. Smart, G. Lu, Y. Yan, and G. Riley, “Characterisation of an oxycoal flame through
digital imaging,” Combustion Flame, vol. 157, pp. 1132–1139, 2010.
22-04-2018
19

THANK YOU

22-04-2018