The heat transfer degradation as a function of time at various gas approach ..... D.P (2002), Fundamentals of Heat and Mass Transfer, 5th edition, John ... Schlichting, H. (1979), Boundary Layer Theory, 7th edition, McGraw-Hill, New York.
13th International Heat Transfer Conference Sydney, Australia, 2006
IHTC13-661
EXPERIMENTAL INVESTIGATION OF PARTICULATE FOULING IN WASTE HEAT RECOVERY FROM THE ALUMINUM INDUSTRY E Næss, T Slungaard, B Moxnes1 and O Sønju Norwegian University of Science and Technology, Trondheim, Norway 1 Hydro Aluminium, Øvre Årdal, Norway
Abstract The main objective of the present study is to establish criteria for energy recovery from dust-laden off-gases from aluminum electrolysis cells. No earlier measurements have been reported in the literature for this application. An experimental investigation of particulate fouling from a real industrial gas stream on bare and finned tubes in crossflow has been performed. The gas stream is an exhaust gas from an aluminum smelting plant, containing approximately 200 mg/Nm3 dust particles with a cut size of ca. 0.5 μm. A small test section consisting of a bare tube and an annular-fin tube were used for the tests. The heat transfer degradation as a function of time at various gas approach velocities and temperatures was recorded for experiments with duration from 40 to more than 2000 hrs. A simplified data reduction procedure provided quantitative information on the effect of deposit formation. Fouling resistances increased with time during the first period of operation. Asymptotic fouling resistances were obtained for superficial flow rates above ca 10 kg/m2s, but did not stabilize for the lower flow rates. The net fouling rates decreased with increasing flow velocities. The annular-fin tube experienced a slightly higher fouling rate than the unfinned tube. Visual observations of the deposit showed that the major part of the particle deposition was on the downstream side of the tubes. For the annular-fin tube, the flow passages between the fins were completely blocked by deposits in the shadow region of the tube. On the upstream side, fluffy deposits were obtained both on the tube and fin surfaces at low velocities. At higher velocities, the deposit became harder, but significantly thinner.
1. Introduction Present technologies for aluminum production using the Hall-Heroult (‘prebaked’ anode) technology has a total energy consumption of about 13 kWh/kg produced aluminum, of which approximately 50% is lost as sensible heat through the pot linings and to off-gases extracted through the aluminum cell hoods. The heat loss to the off-gases represents roughly half of the total sensible heat loss, i.e. about 3.25 kWh/kg aluminum (Grjotheim et al, 1993). An aluminum plant with a typical production capacity of 200 000 tons/year thus generates hot exhaust gases with a theoretical energy content of ca. 650 GWh/yr. The actual recoverable heat is lower, depending on the gas temperature. In present practice without heat recovery, relatively large amounts of off-gases are used (in the order of 100 Nm3/kg produced aluminum), yielding gas temperatures in the range of 100-200°C. An increase of this temperature is possible with new cell hood shielding designs, significantly increasing the availability and usability of the waste heat. The off-gases contain gaseous pollutants and dust from the aluminum cells, which must be cleaned in bag filters and gas scrubbers. Due to temperature limitations of the gas cleaning process (maximum ca 140°C in the filters), heat recovery should take place upstream of the filters for the most efficient utilization. This means that the heat recovery units will be operating in a dust-laden
environment, with the potential of severe heat transfer degradation due to dust deposition on the heat transfer surfaces (fouling). The objective of the present study was to investigate the fouling characteristics of candidate geometries for heat recovery from the dust-laden off-gases from the aluminum cells, and to provide quantitative data regarding the heat transfer deterioration due to fouling deposit buildup at different gas velocities in order to enable the thermal-hydraulic design of heat recovery heat exchangers. In order to obtain data suitable for evaluation of actual situations, the experiments were performed at an aluminum smelting plant, using real off-gases.
2. The Fouling Process The fouling process from a particulate gas stream to a cooled surface is complex, and involves the consideration of the combined effects of particle transport to the surface, particle adhesion to the surface and particle removal from the surface (Beal, 1983). In addition, the deposit morphology need be considered in order to quantify the thermal properties of the deposit layer. Particle transport to the surface is controlled by diffusion (small particles), combined inertia and diffusion (intermediate sized particles), and inertia/impaction (large particles). An excellent review of the fluid flow mechanisms of particle transport to the wall is given by Papavergos and Hedley (1984). The particle transport may be enhanced by the temperature gradients (thermophoresis), which may dominate particle transport to the wall for small particles, as discussed by e.g. Talbot et al (1980) and Rosner (1986). Once a particle hits the wall, it may adhere, slide along the surface or bounce off. The mechanisms controlling this process are not well understood, hence the available information is largely empirical (see e.g. Beal (1983), Epstein (1988) and Isdale (1991)). The process of removal of deposited particles from the surface is influenced by a number of factors: Forces adhering the particles to the surface such as van der Waal, electrostatic, thermophoretic, gravitational and surface tension forces, and forces opposing the adhesion forces, most importantly the shear and lift forces exerted by the flowing gas, inertial impaction from larger particles in the flowing gas, and gravity forces. Last, but not least, the properties of the deposited layer determines the effective thermal conductivity as well as the binding forces of the deposit. Among the more recent studies, Temu et al (2001a, 2001b) performed a numerical analysis of the deposition of polydisperse distributions of ferro-silica particles on a tube in crossflow and compared their calculations with experimental data obtained in laboratory environments. However, their study only considered the initial fouling process where the deposit thickness was small (and had no effect on the gas/surface interface geometry), and the simulations did not predict the deposition process downstram the flow separation point. Although a significant body of both theoretical and empirical work exists for predicting the effect of the various mechanisms involved in the fouling process, one is still dependent on experimental data from the actual plant system under consideration in order to obtain reliable quantitative data on the net deposition, and the effect on heat transfer degradation.
3. Experimental setup and data reduction 3.1 Experimental setup Aluminum cell off-gas was extracted iso-kinetically from the centre of the 1000 mm diameter horizontal main gas channel leading from the aluminum cells to the bag filters. The gas was led to the test section through a 100 mm diameter pipe, insulated and equipped with external heating wires in order to maintain the gas temperature. The main test section was rectangular, with a crosssectional dimension of 200x90 mm, and mounted vertically with the gas directed downwards.
Upstream the test tubes, the sidewalls of the test section were equipped with externally wrapped electric heating wires capable of increasing the off-gas temperature from the main channel temperature of approximately 120°C and up to about 180°C, when desired. Downstream the test section an adjustable speed fan ensured constant flow rate. The off-gas was then led back into the main gas channel. The test section consisted of individual ‘modules’ of length 1000 mm mounted in series, each containing one active test tube. The test tube geometries considered in the present work were: • •
A cylindrical tube with 36 mm outside diameter; see Table 1 for dimensions. A circular tube with annular fins as shown in Figure 1; see Table 1 for dimensions.
Both tubes were singular tubes, i.e. there were no adjacent dummy tubes to simulate flow in a tube bundle. Thus, the unfinned tube behaved like a single tube in a confined channel. However, the annular-fin tube only had a 9 mm clearance between the fin-tip and the channel wall, hence the flow conditions around the tube were relevant for flow across a tube row. Each test tube was cooled by compressed air at ambient temperature flowing internally in the tube. The tubes were also equipped with in-tube blockages in order to increase the effective coolant heat transfer coefficient. Each of the cooling air lines were equipped with mass flow meters and thermocouples registering the air inlet and outlet temperatures, providing data for heat duty calculations. In addition, the offgas temperature was measured immediately upstream of each test tube, and the tube wall temperatures were measured at mid-position at both the front and rear stagnation points. For the finned tube the wall-mounted thermocouple were positioned at midspace between adjacent fins. Main test tube dimensions Bare tube Annular-fin tube Tube outside dimension Ø36 Ø36 Active tube length 200 200 Fin geometry Annular Fin dimension Ø72 Number of fins 15 Fin thickness 2.0 Tube/fin material C.S. C.S. Blockage factor 0.4 0.46
Table 1
mm mm mm mm
Blockage factor: Fraction of total flow area blocked by test tubes. C.S. Carbon steel
Figure 1
Annular-fin circular tube
-
A data acquisition system collected and recorded data from the test tubes at 60-second intervals. In addition to these data, the off-gas flow rate and a signal giving qualitative information about the dust concentration variations were recorded. The dust concentration and the particle size distribution were measured manually. 3.2. Data reduction The apparent heat transfer coefficient (happ) was calculated from the recorded data using Equation (1), with Tw being the arithmetic mean of the measured upstream and downstream wall temperatures: happ
⎡ Atot ⋅ (Tg − Tw ) ⎤ =⎢ − Rw ⎥ ⎢⎣ M& ⋅ c p ⋅ (Ta ,in − Ta ,out ) ⎥⎦
−1
(1)
Here, M& is the cooling air flow rate, Ta,in and Ta,out are the cooling air inlet and outlet temperatures, Atot is the total external side heat transfer surface, Tg is the off-gas temperature and cp is the cooling air mean specific heat capacity. Rw is the thermal resistance in the tube wall between the position of the thermocouples and the outer surface. Making the normal assumption that the external side heat transfer coefficient is uniform, and introducing the simplification that the fouling factor is uniformly distributed, the apparent fouling factor (Rf) may be determined from Equation (2) (cf. Incropera and DeWitt, 2002). −1
happ
⎡1 ⎤ ⎛ A + A fin ⋅η ⎞ = ⎢ + R f ⎥ ⋅ ⎜ tube ⎟ Atot ⎣ h0 ⎦ ⎝ ⎠
(2)
h0 is the external side convective heat transfer coefficient, Rf is the apparent fouling factor, Atube is the exposed tube external heat transfer surface (between the fins), Afin is the heat transfer surface of the fins, Atot is the total external heat transfer surface (Atot=Atube+Afin), and η is the fin efficiency. At startup, the tubes were clean and unfouled, and the convective heat transfer coefficient, h0, was determined for each experimental run with data from the early phase using Equation (2) and with Rf=0. As a quality assurance measure, the values of the heat transfer coefficient determined in this manner were compared to values obtained using the same experimental setup under clean conditions, using hot air instead of aluminum cell off-gas. The agreement was found acceptable, and the hot air tests also showed good agreement with literature correlations for the heat transfer coefficient, e.g. Zukauskas and Ziugzda (1985) for unfinned tubes and PFR Engineering (1976) for finned tubes. For the bare tube, Afin=0 and Rf was found directly. For the annular-fin tube, however, the fin −1 efficiency, η, is a function of he = ⎛⎜ 1 + R f ⎞⎟ in addition to the fin geometry and fin material thermal ⎝
h0
⎠
conductivity. Hence, an iterative procedure was required in order to determine the fouling factor. In the present study, the fin efficiency was calculated using the closed-form formulation suggested by Schmidt (1945), as follows:
η=
tanh ( m ⋅ b ) m⋅b
(3)
b=
⎞ ⎡ ⎛ D f ⎞⎤ d0 ⎛ D f ⋅⎜ − 1⎟ ⋅ ⎢1 + 0.35 ⋅ ⎜ ⎟⎥ 2 ⎝ d0 ⎠ ⎣ ⎝ d0 ⎠ ⎦
(4)
where m=
2 ⋅ he k f ⋅tf
;
Here, kf is the thermal conductivity of the fins, tf is the fin thickness, Df is the fin diameter, and do is the base tube outside diameter. An uncertainty analysis was performed based on the observed variability of the primary variables (temperatures and flow rates), indicating a 95% confidence interval of δRf/Rf ≈ ±8 %, where δRf is the uncertainty in the fouling factor. During the early test phase (time less than ca. 25 hrs) the uncertainty was higher.
4. Results and Discussion 4.1 Initial considerations The particle size distributions were measured using a cascade impactor in both the main gas channel and the test section, and are presented as mass distributions in Figure 2. The agreement was fairly good, with a tendency of having less of the larger particles in the test section. For both positions, approximately 70% (by mass) of the particles were in the submicron range. The particle concentration was measured in the test section at various intervals, and varied between 135 and 280 mg/Nm3, with an arithmetic average of 195 mg/Nm3. This was somewhat lower than measured in the main gas channel, where the concentration was measured to be in the range 240350 mg/Nm3, with an arithmetic average of 300 mg/Nm3. Part of the discrepancy is explained by the lower concentration of larger particles in the test section; however, the overall agreement was considered acceptable. The off-gas acid dew point was measured to be approximately 42°C. In actual heat recovery conditions the heat exchanger wall temperatures would be kept above this temperature, partly to avoid increased particle sticking properties, but mainly because of the increased corrosion conditions. Therefore, the wall temperatures for all test tubes were kept well above the dew point temperature. The off-gas temperature was kept in the range 110-130°C for all but one experiment, and the tube wall temperature was kept at approximately 60-70°C. 1.0
Test section Main gas channel
Mass fraction smaller than [-]
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.1
1
10 Particle diameter [µm]
Figure 2
Particle size distributions
100
4.2 General fouling behavior Sample plots of typical heat transfer reduction are shown for both geometries in Figure 3 for a superficial gas mass flux of 7.6 kg/m2 s. The reduction of the apparent heat transfer coefficient was observed to be significant for both geometries. The clean surface heat transfer coefficients were comparable in magnitude (within 10% of each other), for this experiment. The fouling factors extracted from the same experimental series are shown in Figure 4. The behavior of the unfinned and finned tubes were different than shown in Figure 3, the finned tube having a significantly higher effective fouling factor. The reason for this apparent discrepancy was the increase of the fin efficiency with increasing fouling factor, which partly compensated for the reduction in heat transfer conductance from the gas to the fin surface due to fouling. For this particular experiment, the annular-fin tube fin efficiency increased from 0.78 at clean conditions and to 0.9 after approximately 1 000 hours of operation. The reasons for the difference in fouling factor development were also evident from visual observations of the deposits. Both test tubes were observed to have a thin deposit on most of the heat transfer surface, the deposit on the fins being slightly thicker than on the tube surface. However, at the downstream side of the tubes, massive particle deposits were observed, in the low wall shear force region, also roughly corresponding to the ‘shadow’ region sheltered from the sweeping and impaction of larger particles, as indicated in Figure 5. On the surface outside the massive deposit region, the deposit consisted of a thin (< 0.1 mm), hard layer close to the wall, and a fluffier outer deposit which could easily be removed by blowing pressurized air across it. The deposit in the massive deposit region also had a loose consistence, making removal easy.
1.00
happ/happ,0 [-]
0.75
0.50
0.25 Unfinned tube Annular-fin tube 0.00 0
300
600
900
1 200
1 500
Duration [hrs]
Figure 3 Heat transfer reduction versus time for a superficial off-gas mass flux of 7.6 kg/m2 s. happ,0 is the apparent heat transfer coefficient at clean conditions, i.e. with Rf=0.
0.025
Fouling resistance, Rf [m2K/W]
0.02
0.015
0.01
0.005
Annular-fin tube Unfinned tube 0 0
300
600
900
1 200
1 500
Duration [hrs]
Figure 4
Fouling factor versus time for a superficial off-gas mass flux of 7.6 kg/m2 s
Region of massive deposit
Figure 5
Region of massive deposits (both geometries).
4.3 Effect of gas velocity Fouling factor development at different superficial mass fluxes varied significantly, as illustrated for selected experimental runs with the annular-fin tube in Figure 6. The qualitative trends applied also for the smooth tube geometry. At superficial gas mass fluxes up to about 10 kg/m2 s (corresponding to narrow gap maximum velocity of about 22 m/s), the fouling factor increased monotonically with increasing time for the entire experiment duration (up to 2 000 hrs), however, showing a falling rate behavior. At higher mass fluxes, the fouling curve approached an asymptotic value, the asymptote being reached sooner and also being lower with increasing mass fluxes. For the initial period, taken as the first 10 hours of each experimental run, where particle resuspension is often assumed negligible, the fouling rates were found to be fairly independent of the gas mass flux. The differences observed at later stages can, therefore, be interpreted to be the combined results of the deposition rate, counteracted by the shear forces exerted on the deposit layer, and the impaction erosion caused by larger particles. At higher mass fluxes the shear forces increase, as do the inertia of larger impacting particles, effectively decreasing the net particle
deposition rates on the exposed surfaces. This observation is in accordance with other works, e.g. Abd-Elhady (2003), who studied experimentally the effect of gas velocity on the impaction removal rate of a particulate deposit. 0.035
Fouling factor [m2K/W]
0.030 4.5 kg/m2s 0.025 0.020 0.015
7.2 kg/m2s
0.010
11.0 kg/m2s 15.2 kg/m2s
0.005 0.000 0
500
1 000
1 500
2 000
Duration [hrs]
Figure 6
Fouling factor versus time for the annular-fin test tube at various superficial mass fluxes
The surfaces in the ‘shadow’ region of the tubes all experienced massive deposits, independent of the gas mass flux. Hence, in this region the heat flux from the gas was practically zero. An asymptotic limit of the surface cleaning effect by the fluid shear and the particle impaction is limited to maintain the unshaded part of the heat transfer surfaces shown in Figure 5 clean. The potentially completely clean surface areas were estimated from simple geometric considerations to be 58% for the bare tube total area and 70% for the annular-fin tube. Ignoring two-dimensional conduction effects in the fin, the maximum asymptotic value of happ/happ,0, where happ,0 was the apparent heat transfer coefficient at clean conditions, should then approximately correspond to the values stated above. However, the unfinned tube was relatively thick-walled, rendering this assumption too simplistic. A numerical calculation was performed in order to assess the circumferential heat conduction in the tube wall. The resulting asymptotic heat transfer to the tube, assuming no heat transfer to the external surface of the tube downstream the flow separation point, taken as 105° from the front stagnation point (Schlichting, 1979), was calculated to happ/happ,0 ≈ 0.78. The agreement with observations was good for the annular-fin tube geometry for high mass fluxes, as shown in Figure 7. For the unfinned tube, the simplistic estimate of happ/happ,0=0.58 was seen to be too conservative. However, the inclusion the two-dimensional heat conduction effects on the downstream side of the tube (i.e. happ/happ,0=0.78) yielded acceptable agreement with the experimental data. As a consequence, it can be argued that finned tubes may be attractive in heat recovery applications with dusty gases, since it was demonstrated that it was possible to keep the swept part of the tube and fin surface clean. A further increase in the asymptotic happ/happ,0-ratio may also be possible by using finned tubes with a more streamlined tube shape, minimizing the ‘shadow’ region on the downstream side of the tube. Examples of such geometries are e.g. oval or elliptic tubes.
1.00
0.78
happ/happ,0 [-]
0.75
0.70 0.58
0.50
0.25
Unfinned tube, m''=9.6 kg/m2s Annular-fin tube, m''=15.2 kg/m2s
0.00 0
200
400
600
800
Duration [hrs]
Figure 7 Asymptotic heat transfer reduction for large superficial gas mass fluxes. happ,0 is the apparent heat transfer coefficient at clean condition. 0.0040
Tgas=452 K, Twall=362.5 K Tgas=402 K, Twall=342 K
Fouling factor [m2K/W]
0.0035
2
Rf,10=0.0032 m K/W
0.0030 0.0025 0.0020 2
Rf,10=0.0027 m K/W 0.0015 0.0010 0.0005 0.0000
0.0
2.5
5.0
7.5
10.0
12.5
15.0
Duration [h]
Figure 8 Initial fouling-factor development for the annular-fin tube at different temperatures. Gas mass flux 9.6 kg/m2s.
4.4 Effect of gas temperature The thermophoretic deposition velocity for submicron particles is roughly proportional to the temperature gradient term (-∇T/T), see e.g. Talbot et al (1980). A rough comparison of the measured fouling factors after 10 hours of operation for the annular-fin tube at comparable fluid velocities, but different gas and wall temperatures are shown in Figure 8. Taking the lower curve as
reference, it was observed that at a 17% increase of the temperature gradient term (Tg-Tw)/Tm, where Tm is the average film temperature [=(Tg+Tw)/2], yielded roughly 18% increase in Rf. This is in agreement with the theoretical value, and is indicative of the initial deposition process being strongly influenced by thermophoresis. It is, therefore, to be expected that the initial deposit growth rate further increases for applications with higher off-gas temperatures. The difference in fouling factor development diminished as the deposit thickness increased. However, after about 200 hours of operation the fouling factors became practically identical for the present experiments. This was explained by the reduced effect of thermophoresis due to fouling deposit buildup, combined with a probable weakening of the adhesion forces in the deposit layer, enabling a more efficient particle removal by impaction or sweeping of larger (inertia-controlled) particles.
5. Conclusions The experimental program showed that asymptotic fouling behavior was possible for both finned and unfinned tubes in crossflow. The major part of the deposit was formed on the rear side of the tubes, where shear forces were low and which were protected from impaction of larger particles. Asymptotic fouling factors were dependent on the gas mass flux, and limiting values for the heat transfer reduction were established for the finned tubes. Thermophoresis played a significant role in the deposition process, as was also observed from visual inspections. Whether a full-scale heat recovery unit can operate without aided particle removal (steam or pressurized air blowing, low frequency vibrations, mechanical hammering etc) depends on the acceptability of high off-gas mass fluxes required. As shown, superficial mass fluxes in the range of approximately 10 kg/m2 s were required in order to obtain asymptotic fouling conditions with acceptable levels of surface area compensation. However, the narrow gap flow velocities were high (in the order of 20 m/s), leading to potentially high pressure drops and material erosion due to particle impaction. The tubes have, however, been exposed to the gas stream continually for several thousand hours, showing no signs of material erosion on either the fin surfaces or tube surfaces. Further reduction of the asymptotic fouling factor may be obtained by shaping the tubes more aerodynamic, thus reducing or eliminating the ‘shadow’ (wake) region on the downstream side of the tubes, where particles accumulate.
6. References Abd-Elhady, M.S., Rindt, C.C.M., Wijers, J.G. and van Steenhoven, A.A. (2003), Removal of Particles from a Powdery Fouled Surface Due to Impaction, ECI Conf. on Heat Exchanger Fouling and Cleaning: Fundamentals and Applications, USA. Beal, S.K. (1983), Particulate Fouling of Heat Exchangers, in Fouling of Heat Exchanger Surfaces., ed. Bryers, R.W., United Engineering Trustees Inc., pp. 215-234. Epstein, N. (1988), Particulate Fouling of Heat Exchanger Surfaces: Mechanisms and Models, in Fouling Science and Technology, eds Melo, L.F., Bott, T.R. and Bernardo, C.A., Kluwer Academic Publishers, Dordrecht, pp. 143-164. Grjotheim, K., Huglen, R. and Kvande, H. (1993), Principles of Energy Balance, Thermochemistry and Theoretical Energy Consumption, in Introduction to Aluminium Electrolysis, 2nd edition, eds. Grjotheim, K. and Kvande, H., Aluminium Verlag, Dusseldorf, pp.9-34. Incropera, F.P. and DeWitt, D.P (2002), Fundamentals of Heat and Mass Transfer, 5th edition, John Wiley & sons, New York.
Isdale, J.D. (1991), Prediction of Fouling on Finned Tube Heat Exchangers, Proc. European Conf. on Finned Tube Heat Exchangers, Stuttgart, March 21. Papavergos, P.G., and Hedley A.B. (1984), Particle Deposition Behaviour from Turbulent Flows, Chem. Eng. Res. Des., vol. 62, pp.275-295. PFR Engineering (1976), Heat Transfer and Pressure Drop Characteristics of Dry Tower Extended Surfaces, PFR Engineering Systems Report BNWL-PFR-7-100, Marina Del Ray, Ca, USA Rosner, D.E. (1986), Transport Processes in Chemically Reacting Flows, Butterworths, Boston. Schlichting, H. (1979), Boundary Layer Theory, 7th edition, McGraw-Hill, New York. Schmidt, T.E. (1945), The Heating Capacity of Finned Tubes, Bull. Int. Inst. Refrig., Annex G-5. Talbot, L., Cheng, R.K., Scheffer, R.W. and Willis, D.R. (1980), Thermophoresis of Particles in a Heated Boundary Layer, J. Fluid Mech., vol. 101, pp. 737-758. Temu, A., Næss, E. and Sønju, O.K. (2001a), Theoretical study of Particle Deposition onto a Cylinder in Cross-Flow, in: Bott, T.R., Watkinson, A.P. and Panchal, C.B. (eds), Proc. Int. Conf. on Mitigation of Heat Exchangers and its Economic and Environmental Implications, pp.393-400,Begell House, USA. Temu, A., Næss, E. and Sønju, O.K. (2001b), Experimental study of Particle Deposition onto a Cylinder in Cross-Flow, in: Bott, T.R., Watkinson, A.P. and Panchal, C.B. (eds), Proc. Int. Conf. on Mitigation of Heat Exchangers and its Economic and Environmental Implications, pp.401-408, Begell House, USA. Zukauskas, A., Ziugzda, J. (1985), Heat Transfer of a Cylinder in Crossflow, Hemisphere Publishing, USA.
EXPERIMENTAL INVESTIGATION OF PARTICULATE FOULING IN WASTE HEAT RECOVERY FROM THE ALUMINUM INDUSTRY E Næss , T Slungaard, B Moxnes1 and O Sønju Norwegian University of Science and Technology, Trondheim, Norway 1 Hydro Aluminium, Øvre Årdal, Norway
Abstract The main objective of the present study is to establish criteria for energy recovery from dust-laden off-gases from aluminum electrolysis cells. An experimental investigation of particulate fouling from an industrial gas stream using bare and finned tubes in crossflow is reported. The gas was an aluminum smelting plant exhaust gas, containing approximately 200 mg/Nm3 dust particles with a cut size of ca. 0.5 μm. A small test section consisting of a smooth tube and an annular-fin tube were used for the tests. The heat transfer degradation as function of time at various gas approach velocities and temperatures were recorded for experiments with duration from 40 to more than 2000 hrs, and quantitative information on the fouling behavior was obtained. Fouling resistances increased with time during the first few hundred hours of operation. Asymptotic fouling resistances were obtained for flow rates above ca. 10 kg/m2s, but did not stabilize for the lower flow rates. Net fouling rates decreased with increasing flow velocities. The annular-fin tubes experienced highest fouling factors. Visual observations of the deposit showed that the major part of the particle deposition was on the downstream side of the tubes, completely blocking the interfin flow passages in the shadow region of the tubes. On the upstream side, a fluffy deposit was observed both on the tube and fin surfaces at low velocities. At higher velocities, the deposit became harder, but much thinner.