Anda di halaman 1dari 17

SCALING PARAMETERS FOR PFBC CYCLONE SEPARATOR

SYSTEM ANALYSIS
Antonia Gil, Luis M. Romeo and Cristbal Corts
CIRCE, Centro de Investigacin del Rendimiento de Centrales Elctricas
Centro Politcnico Superior. Mara de Luna, 3.
50015 Zaragoza (Spain)
email: antgilma@posta.unizar.es, luismi@posta.unizar.es, tdyfqdb@posta.unizar.es

ABSTRACT
Laboratory-scale cold flow models have been used extensively to study the behaviour
of many installations. In particular, fluidized bed cold flow models have allowed developing
the knowledge of fluidized bed hydrodynamics. In order for the results of the research to be
relevant to commercial power plants, cold flow models must be properly scaled.
Many efforts have been made to understand the performance of fluidized beds, but up
to now no attention has been paid in developing the knowledge of cyclone separator systems.
CIRCE has worked on the development of scaling parameters to enable laboratory-scale
equipment operating at room temperature to simulate the performance of cyclone separator
systems. This paper presents the simplified scaling parameters and experimental comparison
of a cyclone separator system and a cold flow model constructed and based on those
parameters. The cold flow model has been used to establish the validity of the scaling laws for
cyclone separator systems and permits detailed room temperature studies (determining the
filtration effects of varying operating parameters and cyclone design) to be performed in a
rapid and cost effective manner. This valuable and reliable design tool will contribute to a
more rapid and concise understanding of hot gas filtration systems based on cyclones.
The study of the behaviour of the cold flow model, including observation and
measurements of flow patters in cyclones and diplegs will allow characterising the
performance of the full-scale ash removal system, establishing safe limits of operation and
testing design improvements.

INTRODUCTION
Hot gas cleaning plays an essential role in the development of new power generation
technologies such as Integrated Gasification Combined Cycle (IGCC) or Fluidized Bed
Combustion (FBC). The removal of solid particles from the combustion flue gases is essential
in order to maintain gas turbine working conditions and particle emissions in safe limits. In
particular, in PFBC power stations the entrainment of bed particles may lead to erosion and
fouling in downstream equipment. Special importance is the damage of gas turbine blades. In
addition, fly ash can also produce corrosion due to metal alkali compounds.
Cyclone separator systems offer nowadays one of the best solution for removing
particles from high temperature high pressure installations. Combustion gases from
pressurized beds are an example of this harsh environment, and cyclones are, nowadays, the
only solution commercially available for these power stations. These systems are simple, low
cost and maintenance with relatively high collection efficiency. Its main disadvantages are the
complex hydraulic behaviour and a efficiency decrease for small particles (below 5 m). In
Escatrn PFBC power plant, the hot gas filtration equipment is a two-stage process performed
in nine streams between the fluidized bed and the gas turbine. The cyclones are high
efficiency, Van Tongerens type, with a tangential inlet, cylindrical body, conical base, and an
axial outlet for clean gases and outlet port for solid particles in the lower part. The solid
extraction bin has been replaced by a dipleg (similar to those found in catalytic cracking) and
a suction nozzle through which collected particles are evacuated along with some amount of
transport gas. In contrast with other devices, such as series of pressure-tight lockhooppers, the
solid extraction by pneumatic conveying improves cyclone efficiency and allows reliable
handling and cooling of ash particles with low cost.
Ash and combustion gases exit the pressurized bed at nearly 800 C and 11 bar(g). The
gas and solid mass flow rates depend on load, coal and sorbent characteristics and other
operating variables, in particular those related with fluidisation. These parameters have a
strong influence on the separation efficiency of the cyclones, and there are not well
established theories able to achieve an accurate and complete prediction. Operating
experience at Escatrn have shown sintered deposits and unsteadiness in the dipleg and the
suction nozzle that modify cyclone separation efficiency and affect the cyclone performance
and the capacity of ash conveying lines.
The possibility of achieving analyses to establish the influence of different variables in
a real installation is limited due to a non-controlled operating conditions and a lack of data to

obtain conclusions. Laboratory-scale cold flow models have been used extensively to study
the behaviour of many installations. In particular, fluidized bed cold flow models have
allowed developing the knowledge of fluidized bed hydrodynamics. In order for the results of
these researches to be relevant to commercial power plant, cold flow models have to be
properly scaled. In the present study, scaling parameters have been developed to build a
dipleg and cyclone cold flow model of a PFBC power plant. The cold flow model permits
detailed room temperature studies, such as determining the filtration effects of varying
operating parameters and cyclone designs.
The study of the behaviour of the cold flow model, including observation and
measurements of flow patterns in cyclone and dipleg, will allow characterising the
performance of the full-scale ash removal system, establishing safe limits of operation, testing
design improvements and determining the filtration effects of varying operating parameters.
This paper presents the most relevant scaling parameters for a cyclone separator system and
an experimental comparison of a PFBC cyclone separator system and a cold flow model
constructed based on those parameters.

CYCLONE SCALING PARAMETERS


Collection efficiency and cyclone pressure drop are the most important variables in
cyclone behaviour. Criteria for cyclone scaling parameters, based on maintain collection
efficiency, have been proposed by several researchers (Cheremisinoff and Cheremisisnoff,
1986; Dirgo and Leith, 1986; Svarovsky, 1981, 1986; Leith and Litch, 1972; Abrahanson et
al., 1978). It is generally assumed the necessity of maintain, at least, Stokes number in order
to maintain collection efficiency. Cheremisinoff and Cheremisinoff (1986) developed the
scaling parameters by analysing the forces that influence a particle within the cyclone. They
proposed to maintain Froude, Fr = gD v in2 , and Reynolds numbers, Re = dv s g , and a
ratio between densities and lengths, equation 1.

D
= f Fr , Re, s

g d p

(1)

For the laminar regime of flow (Re<2), they suggested ignoring the inertia of the medium and
hence they concluded that the main variables affecting collection efficiency were Froude and
Stokes numbers, Stk =

g d p v

D g

. Svarovsky (1981) has also proposed to maintain Froude

and Stokes numbers. He has developed these parameters from the equation for accelerated,
three-dimensional particle motion with the main assumption of Stokes regime, equation 2:
Re p =

d p g vr
g

<1

(2)

Other authors only mention the importance of maintaining Stokes number especially at low
solid concentrations (Dirgo and Leith, 1986), or suggest that is the Stokes number the main
variable in cyclone efficiency (Svarovsky, 1981; Leith and Licht, 1972). Finally, Chao (1982)
for a dilute flow as the freeboard of a PFBC power plant, has suggested as scaling parameters
the Stokes and Froude number and a particle Reynolds number. The last number is necessary
when the flow regime is different from Stokes regime. Although these authors have proposed
different theories for cyclone scaling, none of them has verified their proposals in a real
installation. So, as first approximation cyclone behaviour depends on the following variables,
figure 1

f ( g , s , g , d p 50 ,Vin , D , g ,C s _ in )

(3)

1-

D
Vin
Cs
s
g
g
dp

Figure 1. Parameters affecting cyclone behaviour

Where dp50 has been selected as variable to take into account the particle size distribution at
cyclone inlet. Selecting vin, g and dp50 as independent variables in order to identify
dimensionless groups:
4

gD g DVin D s C s _ in

f 2 ,
,
,
,

g
p 50
g
s
in

(4)

where Fr = Vin2 gD is the Froude number. With the approximation of a flow regime with low
relative velocity for the particles in the cyclone and considering the influence of the
aerodynamic forces by means of the Stokes number (which is a combination of three of the
previous numbers, equation 5).

g DVin d p 50 s s d p250Vin

Stk =
= D
g
D
g
g

(5)

Finally, equation (4) is rearranged to read:

C s _ in

f Fr , Re, St , s ,

g
s

(6)

Froude number takes into account the relation between gravitational and inertia forces.
When particle diameter is small, less than 10 m, some authors neglect this number due to the
fact that gravitational forces are small compared with inertia forces (Mothes and Lffler,
1985). In our case, more than 80% of the particles are bigger than 10 m so, it seems
necessary to maintain Froude number.
Reynolds number is not considered to analyse cyclone collection efficiency.
Generally, its influence is near negligible due to Reynolds number is high enough. According
to experiments at different temperatures and pressures, from 1 to 6 bars, and from 293 to
1123C, Reynolds influence is significant for values from 103 to 105. The tendency observed
is that Reynolds higher than 105 does not influence cyclone behaviour (Morweiser and
Bohnet, 1996).
Most of the authors take into account Stokes number to analyse cyclone collection
efficiency. Stokes number relates the inertia forces with aerodynamic forces in a flow field.
Stokes number and cyclone efficiency are closely related. Generally, researchers consider
laminar flow or assume Stokes law (Rep<1) for particle resistance. Also they assume constant
particle diameter and low particle concentration (< 5g/m3) in order to neglect the influence
between particles. There are several problems related with Stokes conservation between real
cyclone and cold flow model. Firstly, there is a particle interaction when solid concentration
is bigger than 5 g/m3. Secondly, there is a particle size distribution (PSD) at real cyclone inlet,
so the maintenance of Stokes is difficult, due to it would be necessary to maintain Stokes for
each particle size and scale the whole PSD. To solve this problem, researchers usually work

with Stokes number for dp50, but it does not give an exact idea of the PSD conservation. It is
possible to have a wide PSD, and a narrow PSD with the same dp50 and the cyclone behaviour
would be completely different. Finally, from a practical point of view, it is impossible to
achieve a scaled PSD. In Escatrn PFBC less than 1% of particles are smaller than 0.4 m,
that is the limit for Rep<1. In the scaled model, only 20% of the particles are smaller than
10m that is the limit for Rep<1 in this case so, not to maintain Stokes number will introduce
a small error for scaling cyclone efficiency. The error comes from the difference between the
scaled and real dp50 in cold flow model, instead of 16.5 m, it is 23 m. Moreover, the
influence of this error is smaller due to the high particle concentration (near 300 g/m3). In the
case of high particle concentrations, the assumptions of considering Stokes influence are not
realistic (i.e. isolated particle, Stokes law or laminar flow) because of particle interactions so,
its influence is reduced. Also, according Cheremisinoff and Cheremisinoff (1986) the
influence of Stokes is important in laminar regime of flow where the inertia of the medium
may be ignored. In other case, they concluded the influence of Reynolds and a ratio of
densities, equation (1).
Up to now, the cyclone studies have been based on small particle concentrations.
According Abrahamson et al. (1978) for high particle concentration, bigger than 10 g/m3, the
effect of particle agglomeration is dominant. Hoffman et al. (1991 and 1992) studied the
effect of solid concentration, up to 150 g/m3, on cyclone behaviour. They strengthen that there
were no theories that could explain the effect of concentration and they concluded the
influence of solid concentration in collection efficiency and cyclone behaviour. Finally,
Wheeldon and Burnard (1987) in experiments in Grimethorpe (6-12 bar, 640-910C) observed
its influence when collection efficiency increased with solid concentration. As a conclusion, it
seems reasonable to consider solid concentration in cyclone behaviour, especially for high
solid loads, due to the interaction between particles and agglomeration, that modify the
theoretical influence. Escatrn PFBC particle concentration at cyclone inlet is 290 g/m3, much
bigger than 10 g/m3 considered as high particle concentration and bigger than 150 g/m3 of
Hoffman et al. (1991 and 1992) tests. For these reasons, it is reasonable to maintain particle
concentration as a scaling parameter. Moreover, the high particle concentration and particle
diameter support the influence of gravitational forces by means of Froude number.
Scaling parameters reduce to Froude number and particle concentration. It could be
also desirable include the Stokes number although scaling particle size distribution could be
impossible to achieve from a practical point of view. Table 1 lists the geometric and operating

parameters that were used to calculate the values of the cyclone scaling parameters, and the
dimensions of the 1/5 scaled cyclone. The real cyclone mean particle diameter, dp50, has been
taken from design data, since to it is not possible to obtain a particle sample from the cyclone
inlet. Table 2 shows a comparison between the values of the dimensionless scaling parameters
for the two cyclone systems.

Table 1. Comparison between Escatrn PFBC and laboratory cold flow model parameters of
the cyclone separator system
REAL CYCLONE

LABORATORY COLD

SEPARATOR SYSTEM

FLOW MODEL SYSTEM

D (mm)

1000

200

T (K)

1030

293

P (bar a)

11.14

3.22

g (kg/m s)

4.25e-5

1.82e-5

g (kg/m3)

3.8

3.8

s (kg/m3)

2800

2800

dp50 (m)

40

23

Vin (m/s)

30

13.4

Cs_in (g/m3)

290

290

Mair (kg/s)

10.6

0.19

Mash (g/s)

800

14

Table 2. Comparison of Escatrn PFBC and cold flow dimensionless scaling parameters

Froude number
Reynolds number
Stokes number
Density ratio
Solid
concentration

Fr = vin2 gD
Re =

Stk =

g Dv in

s d p250 vin
s

C s _ in

D g

REAL CYCLONE

LABORATORY COLD

SEPARATOR SYSTEM

FLOW SYSTEM

91.74

91.79

2.68 e+6

5.60 e+5

3.16

5.45

736.8

736.8

104 e-6

104 e-6

DIPLEG SCALING PARAMETERS


The collected particles removed from the PFBC cyclone system flow down a dipleg
and are transported by means of pneumatic conveying thought a suction nozzle in the bottom
of the dipleg. There are no specific models of this flow inside the dipleg, so an approximate
analysis, resorting to reasonable and simple models of the physical situation, is appropriate.
Flow patterns inside the dipleg may be expected to possess the following features:

Particles slid down on the dipleg wall, following a helical path of variable pitch.

The gas flow generally follows two paths: near the wall, it moves downwards, with
tangential and axial velocity components, whereas it is reversed at the leg core,
forming an upward flow.
As first approximation dipleg behaviour depends on the following variables
f ( g , s , g , Ddl , L ,Vax ,Vt , g , C sol )

(7)

In this case, selecting, L, g and Vax as independent variables in order to identify


dimensionless groups and rewriting
Ddl s
Vax Vax2 L gVax

,
,
,
,C ,
f
L g sol Vt gL

(8)

A 1/5 scaled model has been built for the dipleg. This factor has been chosen to simulate the
whole PFBC cyclone system. First dimensionless group is a ratio between dipleg dimensions
and it is maintained. Particle-solid densities ratio is also conserved since it was also a
dimensionless group in cyclone scaling. Solid concentration as solid- air flow through dipleg
is maintained with the assumption of cyclone efficiency is similar. Velocities ratio
dimensionless number is equal to the tangent of the sliding particle velocity. Particles
tangential velocity is supposed to be proportional to the tangential velocity calculated with
Alexander equation (Alexander, 1949):
Vt r n = cte

(9a)

0.3

T
0.14

n = 1 1 0.67 Ddl
*

283

(9b)

Particle axial velocity is proportional to the gas velocity though the dipleg and could be
estimated as
Vax =

ma _ nozzle

g Asn

(10)

The conservation of the transport air flow makes the sliding angle to be conserved. This takes
into account the effects of vortex extending due to suction. This effect could modify the
dipleg flow patterns, so it is important to maintain it constant:
n

Vax
m sn 1 Ddl
m A D
=

= sn in dl
g Asn Vin D
Vt
mair Asn D

(11)

The small difference in the values of exponent n in Alexander equation modifies slightly the
sliding angle between PFBC cyclone and cold flow model. This difference could be estimated
as a 1.6 % and can not be reduced since n depends on the temperature. Froude number is
calculated as a relation between axial velocity, dipleg length and gravity. Because of the
conservation of transport air, Froude number is maintained. Finally, Reynolds number could
not be conserved due to gas viscosity is fixed by the cold flow model temperature.
Table 3 lists the dipleg geometric and operating variables. They have been used to
calculate the dipleg scaling parameters and dipleg erection. Table 4 summarised the
dimensionless parameters of the dipleg, the values of the scaling parameters were closely
matched between the cold flow model and the Escatrn PFBC except for Reynolds number.

Table 3. Comparison between Escatrn PFBC and laboratory dipleg cold flow model
parameters
REAL DIPLEG

LABORATORY COLD
FLOW MODEL DIPLEG

Ddl (mm)

400

80

L (m)

9.15

1.83

T (K)

1030

293

P (bar a)

11.14

3.22

g (kg/m )

3.8

3.8

s (kg/m3)

2800

2800

Mair (kg/s)

10.6

0.19

Mnozzle (g/s)

163

2.9

Mash (g/s)

800

14.0

Table 4. Comparison of dipleg Escatrn PFBC and cold flow model dimensionless scaling
parameters

Dipleg dimensions

Density ratio

Solid concentration

Reynolds number

LABORATORY COLD

REAL DIPLEG

FLOW MODEL DIPLEG

Ddl

22.0

22.0

736.8

736.8

75 e-3

74 e-3

0.568

0.559

105.3

105.4

1.96 e+5

4.08 e+4

C sol = M ash
Vax

Velocities ratio
Froude number

ESCATRON PFBC

M air

Vt

Fr = Vax2 gL

Re =

L gV ax

CYCLONE AND DIPLEG COLD FLOW MODEL


Figure 1 shows cyclone and dipleg built with the similarity criteria aforementioned
and have linear dimensions which are one-fifth those of the real PFBC dimensions. Figure 2
shows the cold flow model device that is made up of the following components:
- Two air-pressure supplies: main (L1) and a secondary line (L10) for
fluidisation of the particle-discharging hopper.
- Pressurized solid storage tank (T1) and variable speed rotary valve connected
to a Venturi nozzle (L3).
- Primary cyclone inlet (L4), air with a design solid concentration of 290 g/m3.
- Experimental primary cyclone (T2), equipped with a PMMA dipleg (T4).
- Secondary cyclone (T3) with a solids-collecting bin (T7).
- Two sedimentation chambers with fabric filters for complete collection of
particles and cleaning of the exhaust air (T5 and T6).
- Control of airflow by flow meters and valves downstream the sedimentation
chambers (I1 and I2)

10

Figure 1. Scaled cyclone dimensions

S
h
De
D
H

Dimensions
a/D
0.460
b/D
0.203
De/D
0.307
S/D
0.891
h/D
1.310
H/D
3.795
B/D
0.399
L/D
9.022

Figure 2. Laboratory cold flow model cyclone separator system

T P

ash
hopper

hopper
fluidisation

P
rotary feeder

clean air

filter 2
secondary
cyclone

control
pressure valve

Pin = 2.21 bar(g)

primary
cyclone

T P

Tin = 20 C

clean air

ash flow= 50 kg/h


ash extraction
dipleg

air flow= 680 kg/h


P

filter 1

suction
nozzle

11

CYCLONE AND DIPLEG EXPERIMENTAL COMPARISON


Experimental scaling verification is based on a comparison of pressure drop. Cyclone
pressure drop has a strong influence on collection efficiency. The impossibility to take data of
collection efficiency at PFBC power station to compare with cold flow model has made us to
validate the scaling parameters based on cyclone pressure drop. For pneumatic transport as in
dipleg, pressure drop plays an essential role in its analysis. Probes have been placed in several
points though dipleg, and partial and total dipleg pressure drop have been obtained for
verification.
The data used in the following scaling comparisons were taken at operating conditions
after a periodic overhaul of the Escatrn PFBC, with cyclones revised and cleaned of sintered
deposits. The data used in the scaling comparison were taken at a single operating condition
and in steady- state operation. The PFBC power plant load was approximately 90% of the full
load with a standard deviation of 1 MW. Cold flow model pressure drop data were taken
using pressure transducers with a range of 0/62 mbar for cyclone pressure drop and 62/+62, 37/+37, -37/+37 and 5/5 mbar for dipleg. The data from Escatrn PFBC were obtained with
pressure transducer with a range of 0/490 mbar for cyclone pressure drop, ad 98/+98 mbar
for dipleg pressure drop. Data uncertainty is about 0.4% of span.
PFBC cyclone pressure drop transducers are installed in five of the nine cyclones.
Data from pressure drop varies from a maximum of 199.2 mbar to a minimum of 141.0 mbar.
Table 5 shows the PFBC data and its comparison with cold flow model data.
PFBC pressure drop data shows discrepancies between real cyclones. Most probably, a
combination of ashes and gases non-homogeneity at cyclone inlet could cause these
discrepancies. Another cause to explain this effect is cyclone fouling. As it will be proved in a
next paper (Romeo et al., 1999), fouling causes a reduction in cyclone pressure drop. A
different fouling between cyclones could be the cause of discrepancies in real data.
In order to compare and validate the cold flow model, it is necessary to scale down the
Escatrn PFBC cyclone pressure drop data or scale up the cold flow model data. The
dimensionless variable for the pressure drop is
P

1 Vin2
2

C s _ in

= f Fr , s ,
g s

(12)

The right-hand term of the equation (12) is conserved due to the maintenance of
dimensionless numbers that affect cyclone behaviour. So, to scale pressure drop data it is

12

necessary take in account the velocity ratio to the power of two, it means to operate by a
factor of five.
The agreement between Escatrn PFBC pressure drop data and cold flow model data
scaled up is excellent, as indicated in table 5. For the PFBC cyclones, the pressure drop has an
average of 168.0 mbar and a standard deviation of 20.3 mbar, the 99% of the data would be in
the range of 127.4/208.6 mbar. In the cold flow model, the average is 148.5 mbar with a
standard deviation of 9.0 mbar, so the 99% of the data would be in the range of 124.0/166.5
mbar. The latter range of data is approximately inside the former one. This agreement
provides a verification of the scaling proposed above.

Table 5. Comparison between PFBC and cold flow model cyclone pressure drop data
Maximum

Minimum

Average

Standard
deviation

Cyclone 1

191.4

175.7

182.5

7.00

Cyclone 3

199.2

190.3

194.7

3.83

Cyclone 5

174.9

167.7

172.3

2.95

Cyclone 7

150.8

142.9

147.7

2.89

Cyclone 9

145.4

141.0

143.1

1.72

194.7

143.1

168.0

20.5

34.43

24.06

29.73

1.79

172.2

120.3

148.5

9.0

Average of five PFBC


Cyclones
Cold flow Model
Cyclone
Cold flow Model
Cyclone Scaled data

Tables 6 and 7 show also a comparison between PFBC and cold flow model dipleg
pressure drop data. Three zones are observed in the dipleg, an upper one where the pressure
drops strongly. An intermediate region where the pressure drop is negative and particles are
going down vertically. Finally, the last zone near the suction nozzle where the pressure drops
due to pneumatic transport. These three zones are observed both, in the PFBC and the cold
flow model, so that qualitatively the behaviour or the diplegs are similar, although a
difference in values is also observed. Table 6 shows the data at real and cold flow model
dipleg.

13

Table 6. Comparison between PFBC and cold flow model dipleg pressure drop data
Maximum

Minimum

Average

Standard
deviation

PFBC dipleg pressure drop 1

63.2

60.6

61.8

1.02

PFBC dipleg pressure drop 2

- 3.44

- 4.94

- 3.94

0.54

PFBC dipleg pressure drop 3

22.3

16.3

18.1

2.31

Cold flow model dipleg pressure drop 1

2.50

Cold flow model dipleg pressure drop 2

- 0.18

Cold flow model dipleg pressure drop 3

0.04

In table 7 the scaled data for the cold flow model is compared with real data. Scaling
up has been done in the same manner as cyclone scaling up, i.e. the velocity ratio to the power
of two. In spite of the similar tendencies in pressure drop, the scaling is not as good as the
cyclone scaling. In this case the effect of varying Reynolds number is affecting the agreement.
Taking into account the Reynolds influence and multiplying by 5 (PFBC and cold flow model
Reynolds relation) the agreement of pressure drop in the two upper zones is excellent.
Possibly, a combination of suction and ash deposition is the responsible of discrepancies in
the lower zone pressure drop data. In addition, the effect of fluidisation air in PFBC dipleg
bottom has not been taken into account in the cold flow model constructed, and it could
modify the measurements or the behaviour in this zone. Further studies are necessary to
explain this discrepancy.

Table 7. Comparison between PFBC and cold flow model dipleg pressure drop data
Scaled data and
PFBC data

Cold flow model

Scaled data

data

Reynolds
influence

Pressure drop 1

61.8

2.50

12.5

62.5

Pressure drop 2

- 3.94

- 0.18

- 0.9

- 4.5

Pressure drop 3

18.1

0.04

0.2

1.0

14

CONCLUSIONS
A 1/5-scale model of the Escatrn PFBC cyclone system has been constructed based
on scaling parameters. Comparisons of cyclone pressure drop from the cold flow model and
Escatrn PFBC indicates that the cyclone behaviour of the two cyclones is similar. Because of
cyclone pressure drop is one of the most important parameters in collection efficiency, it is
assumed the cyclone efficiency would be maintained in both systems. This point remains
open due to the impossibility to validate the cold flow model results at the real system.
An analysis of the main variables in PFBC cyclone dipleg has been done. This study
has not been addressed before. It has been impossible to maintain all the scaling parameters
that influence dipleg behaviour. Reynolds number has not been maintained due to cyclone
scaling determinate the value of some variables in dipleg behaviour. Reynolds influence has
been taken into account to validate the cold flow model data. Comparison of dipleg pressure
drop from the cold flow model and Escatrn PFBC show a good agreement through the
dipleg. In the suction nozzle some discrepancies has been observed. The reason for these
discrepancies could be the different behaviour of the ash conveying lines in Escatrn PFBC
and the sedimentation chamber in the cold flow model. Further studies are needed to fully
understand fluid flow around suction nozzle.
The cold flow model is revealed as an important tool to optimize and understand the
cyclone system behaviour. It is also useful to know the influence of different operational
variables. At present, these studies are being carried out.

ACKNOWLEDGEMENTS
This research has been fully supported by ENDESA, S.A. Mr Alfonso Ruiz, director
of Escatrn PFBC power station, Dr. Emilio Menndez, head of the R&D department of
ENDESA, S.A., and Mr. Diego Martnez, head of the R&D department at Escatrn are
gratefully acknowledged for making possible the project and for all the facilities provided.
Escatrn power plant personnel are also acknowledged for their useful assistance.

NOMENCLATURE
Asn

suction nozzle area (m2)

Cs_in

solid concentration at cyclone inlet (g/m3)

Csol

solid concentration at dipleg (g/m3)

cyclone diameter (m)


15

Ddl

dipleg diameter (m)

dp

particle diameter (m)

dp50

particle diameter at 50% of the PSD (m)

Fr

Froude number

gravity acceleration (m/s2)

dipleg length (m)

ma_nozzle

air flow through the suction nozzle (g/s)

Mair

air flow at cyclone inlet (kg/s)

Mash

ash flow at cyclone inlet (kg/s)

pressure (bar)

Re

Reynolds number

Rep

particle Reynolds number

Stk

Stokes number

temperature (K)

Vax

axial velocity at dipleg (m/s)

Vin

inlet cyclone velocity (m/s)

vr

relative velocity (m/s)

Vt

tangential velocity at dipleg (m/s)

Greek Leters

cyclone efficiency (%)

gas density (kg/m3)

gas viscosity (kg/m s)

solid density (kg/m3)

REFERENCES
Abrahamson J, Martin CG, Wong KK (1978) The Physical Mechanisms of Dust
Collection in a Cyclone. Transactions of the Institute of Chemical Engineers, Vol. 56, pp.168177.
Alexander R (1949) Fundamentals of Cyclone Design and Operation. Procedures of
the Australian Institute Min. Metall, n 152, pp.203-208.
Chao BT (1982) Scaling and Modelling. Handbook of Multiphase Systems. Edited by
Gad Hetsroni, Hemisphere Publishing Co, pp.(3)44-(3)48

16

Cheremisinoff NP, Cheremisinoff PN (1986) Particulate Capture from Process Gas


Streams in Encyclopedia of Fluid Mechanics, edited by Cheremisinoff NP, Vol. 4, pp. 12171279
Dirgo J, Leith D (1986). Design of Cyclone Separators in Encyclopedia of Fluid
Mechanics, edited by Cheremisinoff NP, Vol. 4, pp.1281-1306
Hoffman AC, Arends H, Sie H (1991) An Experimental Investigation Elucidating the
Nature of the Effect of Solids Loading on Cyclone Performance. Proceedings of the Filtration
Society, Filtration & Separation, Vol. 2, pp.188-193
Hoffman AC, van Santen A, Allen RWK, Clift R (1992) Effects of Geometry and
Solid Loading on the Performance of Gas Cyclones. Powder Technology, Num. 70, pp.83-91.
Leith D, Litch W (1972) The Collection Efficiency of Cyclone Type Particle
Collectors. A New Theoretical Approach. Air Pollution and Its Control, AIChE Symposium
Series, Num.26, Vol.68, pp.196-206
Morweiser and Bohnet (1996) Influence of Temperature and pressure on Separation
Efficiency and Pressure Drop of Aerocyclones in High Temperature Gas Cleaning, edited by
E.Schmidt. Institut fr Mechanishe Verfahrenstechnik und Mechanik.
Mothes H, Lffler R (1985) Motion and Deposition of Particles in Cyclones. German
Chemical Engineering, Num.8, pp.223-233
Romeo LM, Gil A, Corts C (1999) Improving Hot Gas Filtration Behaviour in PFBC
Power Plants. Paper accepted to be presented at 15th International FBC Conferences
Svarovsky L (1981) Solid Gas Separation in Handbook of Powder Technology, edited
by Elsevier ,Vol 3, pp.33-52
Svarovsky L (1986) Solid-Gas Separation in Gas Fluidization Technology, edited by
Geldart D, John Wiley & Sons, pp.197-217
Wheeldon JM, Burnard GK (1987) Performance of Cyclones in the Off-Gas Path of a
Pressurized Fluidized Bed Combustor. Proceedings of the Filtration Society, Filtration &
Separation, Vol. 3, pp.178-187

17

Anda mungkin juga menyukai