Liukkonen.pdf
(
929 KB
)
Pobierz
International Flame Research Foundation
The Finnish and Swedish National Committees
Finnish – Swedish Flame Days 2013
P
ARTICLE
C
AM
-ONLINESYSTEMFORMONITORINGASH
PARTICLESINENERGYBOILERS
Mika Liukkonen*
A
, Yrjö Hiltunen
A
A)
YM Modeling and Research Oy, Viestikatu 7, FI-70600, Kuopio
* Corresponding author, mika.liukkonen79@gmail.com
ABSTRACT
Fuel quality is playing an important role in energy business, as the energy producers are
facing a number of challenges originating from the large variability in the energy
content and elemental structure of fuel. Fluidized bed is a good example of an energy
process in which monitoring the condition and performance is increasingly important,
especially when challenging solid fuels like biomass are used. The coarseness of the bed
material should be monitored, because it provides indirect information on problematic
phenomena such as sintering or agglomeration of the bed material. Nonetheless, it is
well-known that, although often performed indirectly by sieving the bottom ash of the
plant, conducting sampling and measurements in a large-scale CFB is usually difficult
and laborious, and therefore it is seldom a regular operational routine. In the paper we
present a simple, low-cost online system called ParticleCam for estimating the
properties of bed material using online images taken from the bottom ash. The operation
of the ParticleCam is based on digital image data, which is refined to a user-friendly
form. After capturing an image, the related pixel data is processed and analyzed to
extract desired features such as average particle size or the amount of unburned
material. In this paper we demonstrate that the system is able to automatically detect
changes in the grain size of bottom ash, using real ash samples from a gasification plant.
Keywords: Monitoring; Agglomeration; Sintering; Ash; Grain size; Fluidized bed
1
Introduction
Fuel quality is playing an ever more important role in energy business, as the producers
are facing a number of challenges originating from the large variability in energy
content and elemental structure of fuel. This is especially true when solid,
heterogeneous fuels such as biomass fuels are used. Large variations in the fuel quality
and co-combustion of challenging and possibly low-grade fuels, added to the growing
demands for process efficiency and the efforts to reduce harmful emissions, are the
primary reasons for today’s need for monitoring energy conversion processes more
exhaustively than ever before.
- 1 -
International Flame Research Foundation
The Finnish and Swedish National Committees
Finnish – Swedish Flame Days 2013
Fluidized bed is a good example of an energy process in which monitoring the condition
and performance is increasingly important, especially when challenging solid fuels like
biomass are used (Liukkonen et al., 2011a, 2011b, 2012a, 2013). It is well known that in
fluidized bed gasification, for example, the use of difficult fuels such as biomass can
lead to severe problems such as ash sintering or agglomeration. It is likely that these
kinds of problems could be detected by monitoring the grain size of the bed material.
Furthermore, despite the fact that the main components of the bottom ash of a CFB
gasifier using biomass are bed materials (i.e. sand and limestone), small amounts of
solid impurities such as glass and metal pieces, and carbon can also exist (Palonen &
Nieminen, 2004). Detection of changes in the amounts of these harmful components
could likewise be useful.
In fluidized beds the coarseness of the bed material should be monitored, because it
provides indirect information on problematic phenomena such as agglomeration of the
bed material. Nevertheless, the possibilities for performing sampling and measurements
in industrial-scale CFBs are generally rather restricted, because of limited accessibility
and safety precautions (Redemann et al., 2008). Because there are not any direct
methods available for measuring the coarseness of the bed material, this is often
performed indirectly by monitoring the bottom ash which is removed from the process.
This is generally achieved by sieving the bottom ash manually, which is laborious,
expensive and susceptible to errors. Also discontinuous sampling systems have been
developed for bench-scale fluidized bed combustors (Hofmann et al., 2005; Binder &
Werner, 2006), but in real industrial conditions these applications would require
frequent and regular maintenance to work properly.
Another way of monitoring the material coarseness in fluidized beds is to use
computational tools, soft sensors, which rely on existing measurement data and which
could compensate or replace expensive and difficult measurements. These kinds of
computational tools have been developed for estimating the bed quality (Liukkonen et
al., 2010, 2013), but they have inherent disadvantages such as the need for a sufficient
amount of example data of sufficient quality (i.e. manually acquired sieving data), for
which they are also prone to many errors. Preferably, the coarseness of the material
removed from the process should be monitored directly and online, after which the
estimate could be used as an indirect indicator for changes in the bed material. In earlier
experiments, we have managed to demonstrate that theoretically it could be possible to
monitor the grain size of the bottom ash of an industrial-scale fluidized bed boiler using
an automated camera system (Liukkonen et al., 2012b).
Additionally, there are also other properties than ash coarseness which can prove to be
of interest when monitoring energy conversion processes. One of these is the amount of
unburned fuel present in the ash. The usual target for detecting unburned carbon in
energy plants is the fly ash (See e.g. Batra et al., 2008; Clifford, 2008; Liu et al., 2010).
Other approaches for estimating the amount of unburned carbon in the fly ash has been
modeling by using physical models (Pallarés et al., 2009), data mining (Jin & Fu, 2010)
or a combination of these. Moreover, detection of unburned carbon content in the solid
or gaseous combustion products is, of course, also possible in laboratory conditions
(Bartoňová et al., 2007; Wagner et al., 2008). In summary, it seems that there is not a
widely-recognized method to monitor the amount of unburned fuel in the bottom ash of
fluidized beds, but at the same time it has to be noted that we have gained extremely
- 2 -
International Flame Research Foundation
The Finnish and Swedish National Committees
Finnish – Swedish Flame Days 2013
promising results in earlier experiments aimed at monitoring the amount of unburned
carbon particles (Liukkonen et al., 2012c).
In the paper we present a simple, low-cost online system for estimating the properties of
bed material using online images taken from the bottom ash. ParticleCam system is
based on digital image data, which is refined to a user-friendly form. The main parts of
the system, which can be placed in a strategically chosen location, e.g. over the bottom
ash conveyor of an energy boiler, are a systems camera and a measurement PC, which
handles the capturing of images and processes and analyzes the digital image data. After
capturing an image, the related pixel data is analyzed to extract desired features such as
average particle size or the amount of unburned material in the present condition. In the
present paper we introduce the newest results showing that the system is able to detect
material coarsening in the bottom ash.
2
Material and methods
2.1
Hardware
The prototype of the system includes two main components, a systems camera and a
measurement PC, which are placed in a protective, sealed and pressurized case designed
to resist harsh and dirty industrial conditions. By keeping the layout and components as
simple as possible and by using efficient computational processing of images, we have
been able to maintain the material costs of the system at a very low level, which makes
it economically efficient.
Some application possibilities of the ParticleCam system are presented in Figure 1. The
images taken by the system and the calculated grain size information can be shown in
the control room PC, for example. Another possibility would be to transfer the grain
size information directly to the automation system of the plant.
Figure 1. Potential application possibilities of the ParticleCam system.
- 3 -
International Flame Research Foundation
The Finnish and Swedish National Committees
Finnish – Swedish Flame Days 2013
2.2
Software
The main software components of the ParticleCam system are image acquisition
software to control the camera and automate image capturing, filtering software to
select the images of sufficient quality, analysis software to process and analyze the
selected images, and monitoring software to present the calculated information (See
Figure 2 for an example).
Capturing and analyzing images from an ash conveyor is relatively difficult due to
many real-world challenges. The environment is dusty and dirty, the illumination can be
poor, particles are small and overlap with each other, material is in layers, moving
structures of the conveyor can appear in the image zone, and so on. Fortunately there
are many potential image processing tools which can be used for preparing digital
images for the measurement of the features and structures they reveal (Russ, 2011), and
computational processing of images makes it possible to conquer the prevailing
challenges. Despite this, the software components usually require some case-specific
tailoring which depends on the individual characteristics of the target process.
Figure 2. An embedded monitoring application showing a trend of material coarsening.
2.3
Material samples
Ten manually sieved ash material samples from a fluidized bed gasification plant were
used to validate the image analysis. In other words, the mass distribution of each sample
was known a priori. A set of 100 images was taken from the ten bottom ash samples
consisting of different material distributions (i.e. different degrees of coarseness; See
Table 1). In other words, ten images were taken from each sample. The procedure was
such that the sample material was mixed carefully with a plastic spoon before capturing
each image, so that each image was taken using different and differently organized
- 4 -
International Flame Research Foundation
The Finnish and Swedish National Committees
Finnish – Swedish Flame Days 2013
material grains as a target. Therefore, the calculated grain size for each sample is an
average of ten different images, which improves the statistical sampling.
Table 1. The mass distributions of the ten ash material samples. The sieve size fractions are
organized from finest to coarsest.
#1 #2 #3 #4 #5 #6 #7 #8 #9 #10
Fraction 1 (finest)
0.50 0.10 0.20 0.10 0.03 0.03 0.05 0.02 0.04 0.01
0.20 0.80 0.17 0.08 0.10 0.33 0.07 0.05 0.01 0.13
Fraction 2
Fraction 3
0.10 0.05 0.06 0.50 0.23 0.24 0.70 0.03 0.15 0.12
0.02 0.03 0.36 0.20 0.50 0.17 0.08 0.10 0.64 0.30
Fraction 4
Fraction 5 (coarsest)
0.18 0.02 0.21 0.12 0.13 0.23 0.11 0.80 0.16 0.44
SUM
1
1
1
1
1
1
1
1
1
1
In order to validate the analysis results we need, of course, an index to which they can
be compared. For this purpose the following index of coarseness (CI) was calculated for
each sample:
=
∑
(
)
∗ ()
(1)
in which N is the total number of individual sieve size fractions used (N = 5 in this
case), Cut
Up
and Cut
Low
represent the upper and lower cut limits of the sieve size
fraction at stake (in millimeters), respectively, and m is the measured mass-% of the size
fraction at stake (See Table 1). Thus, the larger the value of CI, the coarser is the
material sample it represents. The calculated coarseness indices can be seen in Table 2.
Table 2. The calculated coarseness indices (CI) of samples.
#1
#2
#3
#4
#5
#6
#7
#8 #9 #10
0.23
0.17
0.34
0.29
0.36
0.34
0.27
0.65 0.40 0.49
3
Results and discussion
This far we have reached extremely promising results which show that the system is
capable of indicating changes occurring in material coarseness with a good accuracy.
The experiment in which material coarsening is demonstrated artificially using the
bottom ash of a gasifier is presented in Figure 3. In Figure 4, the average particle size
index calculated from digital images is compared to the coarseness index (CI) of the
samples.
A pilot system of ParticleCam has been already developed and can be used for
diagnosing the quality of ash produced in energy conversion processes. The system
includes embedded advanced image processing software which can be tailored for many
- 5 -
Plik z chomika:
mmarmour
Inne pliki z tego folderu:
Bajamundi.pdf
(209 KB)
Brink.pdf
(656 KB)
Gunarathne.pdf
(886 KB)
GW_Konttinen.pdf
(666 KB)
HaoWu.pdf
(542 KB)
Inne foldery tego chomika:
images
Presentations
Zgłoś jeśli
naruszono regulamin