Sie sind auf Seite 1von 15

Wear 261 (2006) 715729

Numerical and experimental investigation of the relative erosion severity between plugged tees and elbows in dilute gas/solid two-phase ow
Xianghui Chen a, , Brenton S. McLaury b , Siamack A. Shirazi b
b

Alion Science & Technology, 8502 A N. 128th E. Ave., Owasso, OK 74055, USA Department of Mechanical Engineering, University of Tulsa, 600 South College Avenue, Tulsa, OK 74104, USA Received 29 April 2005; received in revised form 12 January 2006; accepted 24 January 2006 Available online 28 February 2006

Abstract Elbows, that are integral part of piping systems, are vulnerable in erosive environments. Plugged tees are used in industrial practice to replace elbows when erosion is expected. However, it remains unclear that plugged tees reduce erosion. Thus, this paper investigates the relative erosion severity between plugged tees and elbows for dilute gas/solid two-phase ow where the pressure is close to the atmospheric pressure. A computational uid dynamics (CFD) based erosion prediction model was applied to predict the relative erosion severity. Experimental tests were conducted to verify the simulation results obtained for gas/sand ows. The ratio of erosion at the end of the plugged section to that in an elbow was found to approach a constant value for a range of conditions. A correlation is presented that provides the ratio of erosion of the outer downstream corner of the plugged tee to that in an elbow. The signicant effect of sand loading on the relative erosion severity was discussed. The inuence of phase density (e.g. water/sand ow) was preliminary explored. 2006 Elsevier B.V. All rights reserved.
Keywords: Solid particle; Erosion; Plugged tee; Elbow; CFD; Gassolid ow

1. Introduction Erosion caused by entrained solid particles in piping systems is a concern for many industrial practices. Elbows, that are an integral part of piping systems, are vulnerable in erosive environments. Wangs study [1] shows that increasing the elbow radius is an effective way to reduce the sand erosion damage. When it is unfeasible to use a long radius elbow due to space considerations, plugged tees are usually put in service rather than standard elbows to reduce erosion. As sketched in Fig. 1, the physical conguration of plugged tees dictates that plugged tees are subject to erosion damage when solid particles are present. Therefore, knowledge of the relative erosion (severity) between plugged tees and elbows provides a guideline for appropriate application of these two geometries. The relative erosion is dened as the maximum erosion in the plugged tee divided by the maximum erosion in the elbow.

Corresponding author. Tel.: +1 918 274 9398; fax: +1 918 274 9399. E-mail address: xchen@alionscience.com (X. Chen).

Elbows are a typical subject of erosion studies primarily because of the broad applications and susceptibilities to erosion. For instance, erosion in elbows was experimentally analyzed by different investigators [27]. Several empirical correlations and mechanistic models were developed to calculate erosion in elbows for gas/solid and gas/liquid/solid ows [3,5,7,8]. Recently, the computational uid dynamics (CFD) approach has been widely applied for erosion prediction in elbows [1,914]. Erosion studies in plugged tees are rare in the literature, except preliminary discussions found in references [3,7,10]. Thus, very limited information is available in the literature regarding the relative erosion between plugged tees and elbows. Bourgoyne [3] compared measured erosion between plugged tees and elbows for gas/solid ow with high sand volume concentrations (about 0.12%). As will be discussed in this paper, Bourgoynes observations based on high sand loadings are very different from the ndings of this study for dilute gas/solid ow. Thus, the purpose of this study is to provide deeper understanding of the relative erosion between plugged tees and the elbows with the primary focus on dilute air/sand ows. A computational uid dynamics (CFD) based erosion prediction model was applied to numerically predict erosion in plugged tees and

0043-1648/$ see front matter 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2006.01.022

716

X. Chen et al. / Wear 261 (2006) 715729

Nomenclature a A b B CD dp D epar eper E E f E E SC E f E


SC

A E SC ER f( ) A F B F D F P F FS I L L0 mp n p r Re u U p V V0 w x y z

empirical constant in Eq. (10) empirical constant in Eq. (9) empirical constant in Eq. (10) body force vector (N) drag coefcient particle diameter (m) pipe diameter (m) particle impingement restitution coefcient in tangential direction particle impingement restitution coefcient in normal direction original prediction of end region prediction of end region due to the rst impingement original prediction of side and corner region prediction of side and corner region due the rst impingement prediction of side and corner region with adjusted particle recirculation erosion ratio dened by Eq. (9) (kg/kg) particle impact angle and pipe material properties dependent correlation added mass force (N) buoyancy force (N) drag force (N) pressure gradient force (N) particle shape coefcient unity matrix plugged section length of the plugged tee particle terminal distance (m) mass of particle (kg) empirical constant used in Eq. (9) pressure (Pa) elbow radius (m) Reynolds number uid velocity uctuation components (m/s) uid velocity vector (m/s) particle velocity vector (m/s) particle impingement speed (m/s) empirical constant in Eq. (10) empirical constant in Eq. (10) empirical constant in Eq. (10) empirical constant in Eq. (10)

uid density (kg/m3 ) surface tension (Pa m) stress tensor (N)

Fig. 1. Schematic sketch of the plugged tee and elbow.

elbows for a broad range of ow conditions. Additionally, experimental erosion tests were conducted in air to evaluate the simulation results. Based on simulations and experimental data, the relative erosion between two regions of the plugged tees and elbows are discussed. Further experiments were performed to preliminarily investigate the sand volume concentration effect on the relative erosion severity. The inuence of uid properties is also discussed via the CFD erosion simulations in plugged tees and elbows in water/sand ow. Bourgoynes observations of erosion for water/sand ow qualitatively support the CFD erosion predictions. 2. CFD prediction 2.1. CFD-based erosion prediction model This paper describes only the framework of the erosion prediction model. Further details are given by Chen et al. [10]. Predicting erosion by employing this model is a three-step process: the continuous carrier uid ow eld simulation; particle tracking using a Lagrangian approach; and erosion calculation using information on particle impingements on the wall. The single-phase CFD-based erosion prediction model is implemented in the commercially available CFD code, CFX-4. The stringent assumption of this model is that particleparticle interactions are negligible. This assumption implies that the model is designed for dilute systems. The rst step is to solve the continuous carrier uid ow equations. The continuity and momentum equations employed by CFX-4 are given in Eqs. (1) and (2), respectively [15]: ) = 0 + (U t ) (U U ) = B + (u u + ) + (U t (1)

Greek letters contribution factor of particle recirculation impingements par standard deviation of epar per standard deviation of eper uid viscosity (Pa s) particle impingement angle ( ) 0 empirical particle impingementangle in Eq. (10) ( )

(2)

X. Chen et al. / Wear 261 (2006) 715729

717

where the stress tensor, , is given by p + ( U )T = I+ U (3)

Table 1 Erosion model empirical constants Empirical constant Material Carbon steel A 0 a b w x y z n 15 38.4 22.7 1 3.147 0.3609 2.532 1.73 1559B0.59 109 Aluminum 0.388 107 10 34.79 12.3 5.205 0.147 0.745 1 1.73

After obtaining the ow eld, the particle trajectories are simulated. Assuming particles not to affect the ow eld, one-way coupling between sand particles and the carrier uid is applied in the current model. It is noted that the one-way coupling method is suitable only for low solid loading. Clift et al. [16] proposed the governing equation of particle motion in the uid based on Newtons second law: mp p dV D + F P + F B + F A =F dt (4)

D, F P, F B and F A represent the drag force, pressure gradient F force, buoyancy force and added mass force, respectively. Chen et al. [10] gave full descriptions of these forces. Once a particle impinges the pipe wall, the accompanied energy lost during the impingement process must be taken into account to determine the reected particle trajectory. Grant and Tabakoff [17] performed sand particle impingement experiments on aluminum specimens in the wind tunnel where the sand particle diameter is 200 m and the air velocity is 110170 m/s. The experimental data indicates that the restitution coefcient is dominantly affected by the particle impact angle, . Grant and Tabakoff treated the particle rebound dynamics of the particles in a statistical sense and postulated the mean values of the coefcients of restitution (epar and eper ) and the corresponding standard deviations (par and per ): epar = 0.998 1.66 + 2.11 2 0.67 3 eper = 0.993 1.76 + 1.56 2 0.49 3 par = 2.15 5.02 2 + 4.05 3 1.085 4 per = 0.0005 + 0.62 0.535 2 + 0.089 3 (5) (6) (7) (8)

The relevant empirical constants in Eqs. (9) and (10) are summarized in Table 1. Divided by the pipe material density and the local grid cell area, the erosion ratio can be converted to the local wall thickness loss or the so-called local penetration. 2.2. CFD prediction of the relative erosion between plugged tees and elbows As previously discussed, long radius elbows are more erosion resistant than standard radius (r/D = 1.5) elbows. In industrial practice, plugged tees rather than standard radius elbows are applied in an erosive environment if space is not available for long radius elbows. Thus, it is meaningful to make comparisons of erosion between plugged tees and elbows. The physical congurations of plugged tees and elbows are sketched in Fig. 1. A plugged length, L, of 1.5D is used in this study. By applying the above described CFD-based erosion prediction procedure, erosion predictions were performed in plugged tees and elbows for the ow conditions tabulated in Table 2. It should be noted that the sand volume concentration for these ows is less than 0.0015%. Thus, the ows are considered as dilute gas/sand ows to which the CFD-based erosion prediction model is applicable. As a sample, detailed predictions in the 0.0254 m diameter plugged tee and elbow are presented for air ow with a mean velocity of 45.72 m/s and a sand diameter of 150 m. A sample of predicted particle trajectories (of 10 particles) is shown in Fig. 2. As expected, the majority of the particles released at the pipe inlet impinge the inlet projected area of the plugged tee and the elbow. For this ow condition, the particle inertia dominates the particle motion. As shown in Fig. 2(a), particle recirculation characterizes the particle motion in the plugged tee. As will be discussed, the potential consequence of particle recirculation is concentrated erosion at a certain area of the plugged tee. In practice, the cap (the end surface) of the plugged tee can be made with a desired thickness. It is also more economical to make the
Table 2 Simulation matrix of erosion prediction in plugged tees and elbows Pipe diameter (m) Air velocity (m/s) Particle diameter (m) Sand ow rate (kg/s) 0.0254, 0.0508, 0.1016, 0.2032 15.24, 30.48, 45.72 50, 100, 150, 220, 300 3.0e-4

Chen et al. have demonstrated the signicance of applying the concept of stochastic particle rebound. Impingement information, such as impact speed and impact angle, is gathered as particles hit the wall. Using this information, the erosion ratio can be calculated. The erosion ratio is dened as the mass loss of the pipe wall due to erosion divided by the mass of particles impacting the wall. According to Ahlert [18] and McLaury [19], the erosion ratio is given by
n ER = AFs V0 f ( )

(9)

where ER is the erosion ratio (kg/kg) and V0 is the particle impingement velocity; A is a material dependent coefcient. For the carbon steel, A is a function of the material Brinell hardness, B. Fs is a particle shape coefcient; Fs = 10 for sharp (angular), 0.53 for semi-rounded, or 0.2 for fully rounded sand particles; n is an empirical constant and takes the value of 1.73; and f( ) is particle angle dependent function which is determined by f ( ) = a 2 x + b,
2

for 0

cos2

sin(w ) + y sin + z, for > 0

(10)

718

X. Chen et al. / Wear 261 (2006) 715729

Fig. 2. Predicted sample particle trajectories in the plugged tee and elbow.

end surface thicker than the other sections of the plugged tee. Hence, two regions in the plugged tee are dened (see Fig. 3(a)) to clarify the erosion in the corresponding regions: end region and side and corner region. The predicted erosion patterns are presented in Fig. 3. It should be noted that erosion is given as penetration (mm/kg), which represents the pipe wall local thickness loss per unit mass of sand loading. Fig. 3(a) indicates that the maximum erosion occurs at the end surface of the plugged tee (end region). Considerable erosion is also observed at the side and corner region. As shown in Fig. 3(b), high erosion occurs in regions corresponding to the downstream projection of the inlet area for on the elbow. For this sample case, the maximum penetration of the plugged tee (in the end region) is about half of that in the elbow, which means the relative erosion between the plugged tee end region and the elbow is about 0.5. For the ow conditions listed in Table 2, the maximum penetrations were determined from the simulation results for the end region and side and corner region of the plugged tees, respectively. Dividing these maximum penetrations by the maximum penetration in the elbows yields the relative erosion for both the end region and side and corner region of the plugged tee. The results are presented in Figs. 47 for four pipe diameters as a function of particle diameter. It is clear that the relative erosion in the end region is strongly affected by the pipe diameter and par-

Fig. 4. Predicted relative erosion between the plugged tee and the elbow with 0.0254 m diameter.

ticle diameter. Depending on the actual ow conditions, a peak value of the relative erosion appears. The trends of the relative erosion in the end region also tend to converge to a constant of 0.5 as the particle diameter becomes sufciently large. Similar

Fig. 3. Predicted erosion pattern in the plugged tee and elbow of air/sand ow.

X. Chen et al. / Wear 261 (2006) 715729

719

Fig. 5. Predicted relative erosion between the plugged tee and the elbow with 0.0508 m diameter.

Fig. 6. Predicted relative erosion between the plugged tee and the elbow with 0.1016 m diameter.

trends of the relative erosion of the side and corner region can be observed, but typically approaches a smaller value than the end region as the particle diameter increases. This information is very useful since erosion occurring at the side and corner region determines the service life of the plugged tee. 3. Experimental study 3.1. Description of experimental set-up Air was used as the carrier uid and was near standard conditions in the test cell. As schematically shown in Fig. 8, air with the injected sand ows vertically upward into the test section and was discharged horizontally. Erosion tests were performed on test cells representing a plugged tee and an elbow with 25.4 mm diameter, as shown in Fig. 9. The relative plugged length of the plugged tee (L/D) is 1.5, and the elbow radius (r/D) is 1.5. Elbow and plugged tee specimens are made of aluminum. Specimen 1 of the plugged tee covers the plugged end surface. Specimens 24 have a width of 6.4 mm and are placed on the center plane of the plugged tee test cell. The elbow specimen is a 6.4 mm 6.4 mm bar that was bent to match the radius of curvature of the elbow and is also place along the center plane. Sand was injected through the sand injector at a constant rate during each test.

Experiments of mass loss measurements indicate that erosion of specimens 3 and 4 of the plugged tee is negligible as compared to specimens 1 and 2 of the plugged tee and the elbow specimen. Thus, local thickness loss measurement of specimens 1 and 2 of the plugged tee and the elbow specimen was measured to obtain the relative erosion between both regions of the plugged tee (end region and side and corner region) and the elbow. Local thickness loss proles were measured using a prolometer (shown in Fig. 10). The accuracy of the prolometer is 0.1 m. Before and after the test, the specimen surface proles were recorded by the prolometer at the same location. The difference of these two surface proles determines the local thickness loss provided that a measurement reference is given. V-shaped scratches are drawn on the specimen surface. The cross-sectional prole (depth and shape) of the scratches are recorded by the prolometer. It is assumed that the bottom of a scratch does not erode, so the bottoms of the scratches serve as the prolometer measurement reference locations. Due to the highly sensitive nature of the local thickness measurement, the surface proles before and after the test must be taken at the identical location to yield meaningful thickness measurements. Thus, two scratches (major scratch and minor scratch) with a certain angle were introduced on the specimen surface for each measurement (see Figs. 11 and 12). The surface proles were taken across these two scratches with the prolometer reading line perpendicular

720

X. Chen et al. / Wear 261 (2006) 715729

Fig. 8. Schematic of experimental set-up for erosion test.

both edges of specimen 2. For the elbow specimen, measurements were made at seven locations in 10 increments along the elbow specimen curve starting 20 downstream of the elbow inlet, as seen in Fig. 12. 3.2. Experimental erosion test Simulation results shown in Figs. 47 reveal that the particle diameter is the dominant factor that affects the relative erosion between the plugged tees and elbows. Thus, experimental erosion tests were conducted to primarily validate particle diameter effects. In the current study, three different average particle size groups were used: 5075, 125175 and 212275 m. The average particle diameters of these groups are about 65, 150, and 250, respectively. The gas velocity was 45.72 m/s. The test conditions are summarized in Table 3. As a sample, the prolometer measurement of specimen 1 of the plugged tee is presented in Fig. 13 for the case of particle diameter = 150 m. From Fig. 13, the maximum erosion is 40 m/kg. Experiments show that the joint corner of specimen 2 of the plugged tee experiences high erosion. Note that specimens 1 and 2 of the plugged tee correspond to the end region and side and corner region of the plugged tee, respectively. For the elbow, the maximum penetration is detected at measurement location

Fig. 7. Predicted relative erosion between the plugged tee and the elbow with 0.2032 m diameter.

to the major scratch. Thus, the measurement location can be uniquely located using the distance between two scratches. In the experiment, the uncertainty of the measurement location is about 2.54 105 m. As shown in Fig. 11(a), measurements were taken along the centerline from the bottom (ow discharge side) toward the top of specimen 1 of the plugged tee. As shown in Fig. 11(b), measurements were taken along the center line for

Fig. 9. Test cell and specimens of the plugged tee and elbow.

X. Chen et al. / Wear 261 (2006) 715729

721

Fig. 10. Prolometer and surface prole measurement. Fig. 12. Measurement locations and scratches of the elbow specimen. Table 3 Conditions of experimental erosion tests in the plugged tee and elbow Pipe diameter (m) Mass of sand injected (kg) Test time (min) Fluid velocity (m/s) Sand volume concentration (%) Particle diameter (m) 0.0254 1.0 80 45.72 5.1e-4 65, 150, 250

5 which is about 60 downstream of the elbow inlet. Using the measured maximum penetrations in specimens 1 and 2 of the plugged tee and the elbow specimen, the relative of both the end region and side and corner region are obtained. 4. Results analysis 4.1. Validation of simulation results In Fig. 14(a), experimental data shows that the relative erosion between the plugged tee end region and the elbow is almost a constant having the value of 0.5. For sufciently large particle sizes (greater than 150 m), the relative erosion is accurately predicted. While for smaller particle sizes, the relative ero-

sion is over-predicted. Careful analysis of the simulation results (including erosion pattern and particle deposition rate) indicates that particle recirculation intensity becomes stronger with decreasing sand size, which results in more concentrated erosion in the end region of the plugged tee. When particle size increases, the erosion on the end surface is more uniformly distributed. This analysis suggests that there is better agreement between experimental and simulated results when erosion caused by the particle recirculation is negligible. The particle recirculation seems to be numerically exaggerated in the simulations for small particle sizes, which results in the overpredictions of the relative erosion. The comparisons in Fig. 14 and the discussions are made based on the experimental data and simulation results obtained for the plugged tee and elbow with 0.0254 m diameter. Figs. 4(a)7(a) consistently show that the relative erosion between the end region of the plugged tee and the elbow approaches a constant value of about 0.5 when the particle diameter becomes sufciently large. Meanwhile, a particle diameter dependent peak value of the relative erosion is observed in Figs. 5(a)7(a). Following the previous argument, the peaks may be due to the unrealistic predictions of the particle recirculation.

Fig. 11. Measurement locations and scratches of specimens 1 and 2 of the plugged tee.

722

X. Chen et al. / Wear 261 (2006) 715729

Fig. 13. Sample measurements to obtain local thickness loss on specimen 1 of the plugged tee using prolometer.

Excellent agreement is observed in Fig. 14(b) between the experimental data and simulations. One question that may arise here is why the over-prediction shown in Fig. 14(a) for small sand sizes does not appear in Fig. 14(b). The predicted erosion patterns for relevant ow conditions were carefully examined.

The predicted maximum erosion at the side and corner region is located exactly at the joint corner, which is consistent with experimental results. Erosion at the corner is primarily caused by the rst impingement of the particles. Thus, the predictions in Fig. 14(b) are free of the numerical discrepancy of particle recirculation that is included in Fig. 14(a). Again, it is pointed out that this discussion is based on the numerical and experimental results and observation of the erosion obtained in the plug tee and elbow with 0.0254 m diameter and an air velocity of 45.72 m/s. Figs. 5(b)7(b) show that the relative erosion is not a monotonic function of the particle diameter. Depending on the actual ow conditions, the maximum erosion of the side and corner region may be detected in the plugged section of the plugged tee rather than at the corner. For instance, Fig. 15 shows the predicted erosion pattern for the ow with air velocity of 45.72 m/s, particle diameter of 50 m, in 0.0508 m diameter pipe. It can be deduced that the localized erosion in the

Fig. 14. Predicted and measured relative erosion between the plugged tee and the elbow.

Fig. 15. Numerical exaggeration of particle recirculation causes localized erosion in the plugged section of the plugged tee.

X. Chen et al. / Wear 261 (2006) 715729

723

Fig. 16. Predicted relative erosion between the plugged tee and the elbow with 0.0254 m diameter due to rst impingement of particles.

Fig. 17. Predicted erosion between the plugged tee and the elbow with 0.0508 m diameter due to rst impingement of particles.

plugged section (not on the end surface) is due to the same particle recirculation that is responsible for the over-predictions of erosion on the end surface. Thus, exaggerated particle recirculation also results in the irregularity of the trends shown in Figs. 5(b)7(b). 4.2. Development of a simplied model The particle recirculation characterizes the particle motion in the plugged section of plugged tees. Fig. 14(a) shows that the numerically exaggerated particle recirculation predictions result in over-predicted erosion in the plugged tees. The rst impingement of particles and impingements resulting from recirculation are two contributors to erosion in plugged tees. The contribution due to each must be distinguished if it is deemed that particle recirculation in the plugged section is being exaggerated in the simulations. Figs. 1619 present the relative erosion that is due to the particle rst impingement only. Figs. 1619 indicate that the relative erosion due to the rst impingement of particles is a monotonic function of the particle diameter and is weakly affected by the gas velocity. As shown in Figs. 16(a)19(a), the relative erosion of the end region increases with the particle diameter and becomes a constant value of 0.5 when the particle is sufciently large. This constant

is also observed in Figs. 4(a)7(a). More importantly, this constant value is experimentally conrmed, as shown in Fig. 16(a). The difference between the measurement and simulations is contributed by particle recirculation. It is clear that the contribution of particle recirculation decreases with increasing particle diameter. Figs. 16(a)19(a) show similar trends of the relative erosion between the plugged tee end region and the elbows. Based on these ndings, it is recommended in this study that the relative erosion between the end region of plugged tees and elbow is a constant having the value of 0.5. From Figs. 16(b)19(b), the relative erosion due to the rst impingement of particles for the side and corner region decreases with increasing particle diameter and approaches a constant value that is diameter dependent. In order to effectively account for the contribution of the particle recirculation on erosion in the side and corner region, an empirical particle recirculation impingement contribution factor is introduced. The purpose of this contribution factor is to adjust the amount of erosion resulting from particle recirculation, since the effect is being over-predicted. A combination of the experimental and predicted results of relative erosion between the plugged tee end region and the elbow are used to determine this factor. The experiments show that a constant value of 0.5 can be used for the relative erosion for all particle sizes of interest. Predicted deviations

724

X. Chen et al. / Wear 261 (2006) 715729

Fig. 18. Predicted relative erosion between the plugged tee and the elbow with 0.1016 m diameter due to rst impingement of particles.

Fig. 19. Predicted relative erosion between the plugged tee and the elbow with 0.2032 m diameter due to rst impingement of particles.

examining the end region and the elbow from this constant are used to determine the particle recirculation contribution factor as follows: =
f E 0.5 E f E E E E

(11)

f is the E is the prediction shown in Figs. 4(a)7(a); E E where E prediction due to the rst impingement of particles, as shown in Figs. 16(a)19(a). The particle recirculation contribution factor indicates the ratio of actual erosion due to particle recirculation to the overall predicted erosion of particle recirculation. Particle recirculation in the plugged section results in erosion both in

Fig. 20. Relative erosion between the plugged tees (side and corner region) and the elbows.

X. Chen et al. / Wear 261 (2006) 715729

725

the side and corner, and the end regions. Statistically, the particle recirculation contribution factor should be similar. Thus, it is assumed that the end and side and corner regions have the same particle recirculation contribution factor. Thus, the overpredicted erosion due to particle recirculation in the side and corner region can be approximately compensated by . The relative erosion in this region is adjusted using :
A f f SC SC SC E SC E =E + (E )

5. Discussion 5.1. Particle tracking in the plugged tees Good agreements between predictions and data shown in Fig. 14 suggest that the ow eld is reasonably predicted in both the plugged tee and the elbow. However, the discrepancy in Fig. 14(a) for small particles implies that particle tracking in the plugged tee needs to be improved. Particle tracking is determined by the particle motion governing equation and the particle-wall rebound model provided that the carrier uid ow simulation is sufciently accurate. In the current study, the stochastic particle rebound model by Grant and Tabakoff [17] and the non-stochastic particle rebound model by Forder et al. [20] are compared to investigate the effects of the particle rebound model on the particle tracking and resulted erosion. The non-stochastic particle rebound model by Forder et al. is expressed as follows:

(12)

A is the adjusted relative erosion, E SC the prediction where E SC f presented in Figs. 4(b)7(b), and ESC is the prediction due to rst particle impingement, as presented in Figs. 16(b)19(b). The gas velocity has little inuence on the adjusted relative erosion, which is also demonstrated by experimental data listed in Table 4. Thus, the average value of the results obtained from Eq. (12) for three velocities can be used to represent the adjusted relative erosion which is plotted in Fig. 20. It is clear that the relative erosion between the side and corner region of plugged epar = 1 0.78 + 0.84 2 0.21 3 + 0.028 4 0.022 5 tees to elbows strongly depends on the particle diameter and pipe diameter. As shown in Fig. 20, the trends can be closely corre(15) lated by the following simplied model in the dimensionless form: 106 dp L L0 0.0852 0.0526 ln 0 + 0.4404 ln ln , dp < 65 m + 3 . 6964 dp D dp A ESC = (13) 106 dp L0 L0 5 + 4.0 10 + 0.1054 ln ln + 1.9268, dp 65 m 0.0994 ln D dp dp where L0 (m) is the distance for a particle to reach its terminal velocity in stagnant atmospheric air. L0 is determined in the current study by using the simplied particle motion equation: mp U | p d 2 C |V dV D + F A = p D p ) p U =F (V dt 8 p mp dV 2 dt

eper = 0.988 0.78 + 0.19 2 0.024 3 + 0.0027 4

(16)

(14)

= 0 and CD is the drag coefcient. Chen et al. [10] where U describe how to determine CD . Good agreement is observed in Fig. 20 between the adjusted predictions obtained by using Eq. (12) and the predictions from Eq. (13). Similar trends of relative erosion between the plugged tees (side and corner region) and the elbows are also observed for different pipe diameters. Apparently, the CFD results based on Eq. (12) and the predictions by the proposed model (Eq. (13)) for the diameter of 0.0254 agree very well with the experimental data. Fig. 20 suggests that the introduction of the particle recirculation impingement contribution factor is a reasonable approximation method to account for erosion due to particle recirculation in the plugged tees. It should be noted that the recommended value of 0.5 for the end region and Eq. (13) are strictly applicable to gas/sand ows where the carrier uid properties are comparable to the atmospheric air and dilute gas/solid systems.

For the air/sand ow with air velocity of 45.72 m/s and sand diameter of 50 m, the predicted particle trajectories in the plugged tee by the stochastic and non-stochastic models are compared in Fig. 21. As shown in Fig. 21(a), severe particle recirculation in the plugged section is predicted by the nonstochastic model. Some particles recirculate repeating almost identical paths more than one hundred times. These repeating impingements at the same location cause highly localized erosion in the plugged tee. The unrealistic particle recirculation predicted by the non-stochastic particle rebound model is greatly improved by the stochastic model, as shown in Fig. 21(b). The explanation is that the particle rebound process is literally an event of stochastic nature. However, the particle rebound models show little effect on particle tracking and predicted erosion in the elbows where no particle recirculation impingements are expected. Fig. 22 shows that the non-stochastic particle rebound model gives predictions that drastically differ from experimental results. The predictions are substantially improved by applying the stochastic particle rebound model. In spite of the signicant improvement resulting from the application of the stochastic particle rebound model, discrepancies still exist in Fig. 14(a). It is noted that one-way coupling is used for particle tracking and the predictions assume no particle

726

X. Chen et al. / Wear 261 (2006) 715729

Fig. 21. Sample particle trajectories in the plugged tee predicted by the stochastic and non-stochastic particle rebound models.

interactions. The discrepancies shown in Fig. 14(a) suggest that these embedded assumptions may not suitable for the plugged tees even when the sand loading is very low since considerable particle recirculation occurs. It is expected that the predictions in the plugged tees can be further improved by considering the particle interaction effects in the particle motion governing equation and using two-way coupling particle tracking. 5.2. Sand volume concentration effects The simulations and experimental data were obtained from air/sand ow with very low sand volume concentration, 0.0005%. These results show that erosion in plugged tees is less than but comparable to erosion in elbows. However, based on experiments with high sand loadings, Bourgoyne [3] stated that erosion in plugged tees is less than in elbows by two orders of magnitude. In order to verify the inuence of sand volume concentration, further experimental erosion tests were conducted in both the plugged tee and the elbow with 0.0254 mm diameter with three different sand volume concentrations where the highest one is comparable to the concentration used by Bourgoyne. The test conditions and results are shown in Table 4.

From Table 4, it is clear that the relative erosion severity is strongly affected by the sand volume concentration. High sand volume concentration results in reduction of erosion in both the plugged tee and the elbow due to strong particle to particle interactions. However, the reduction of erosion in the plugged tee is more drastic than in the elbow. It is also observed that under high concentrations, the erosion on specimen 1 is even less than on specimen 2 of the plugged tee, which is the opposite to low concentration experiment. These observations may suggest two particle interference zones (see Fig. 23) in the plugged tee. One is near the end surface where the reected particles interact with impinging particles. The second zone is at the joint section of the two pipes where the incoming particles collide with the outgoing particles from the plugged section due to particle reection, gravity, and ow recirculation. These two particle interference zones result in negligible erosion on the wall of the plugged section. Thus, the erosion reduction of the plugged tee is more strongly affected by the sand volume concentration than the elbow. In depth investigations of particle interaction in both the plugged tees and elbows for high sand volume concentrations are needed to better understand the erosion phenomena under such conditions.

Table 4 Effects of sand loading on the relative erosion severity Region End End End End End End Side and corner Side and corner Side and corner Side and corner Side and corner Side and corner Air velocity (m/s) 30.48 45.72 30.48 45.72 30.48 45.72 30.48 45.72 30.48 45.72 30.48 45.72 Sand volume concentration (%) 7.5e-4 5.0e-4 6.0e-3 4.0e-3 5.0e-2 3.3e-2 7.5e-4 5.0e-4 6.0e-3 4.0e-3 5.0e-2 3.3e-2 Relative erosion (plugged tee/elbow) 0.534 0.517 0.048 0.051 0.011 0.010 0.095 0.118 0.058 0.052 0.032 0.033

Fig. 22. Effects on particle-wall rebound model on the predicted relative erosion between the plugged tee (end region) and the elbow.

X. Chen et al. / Wear 261 (2006) 715729

727

5.3. Fluid property effects The above analysis and developed simplied model are made based on the experimental data and simulation results obtained from air/sand ow. CFD simulations of erosion were performed for water/sand ow to illustrate how the uid properties affect the relative erosion severity between the plugged tees and elbows. Figs. 24 and 25 present predicted sample particle trajectories and erosion patterns in both the plugged tee and the elbow for the case of water velocity of 3 m/s and sand diameter of 150 m. As compared to Fig. 2(b), Fig. 24(b) shows that particles closely follow the ow streamlines in the elbow, due to the high drag force exerted on the particles by water. Particles are also driven by the inertia force to cross the ow streamlines and impinge the elbow at the exit region. This causes the maximum erosion to occur at the exit region of the elbow, as shown in Fig. 25(b). Further simulations of erosion in elbows for water/sand ow reveal that the location of maximum erosion in the elbow is weakly inuenced by ow parameters including

Fig. 23. Schematic of sand particle interference in the plugged tee.

Fig. 24. Predicted particle trajectories in water.

Fig. 25. Predicted erosion pattern in the plugged tee and elbow of water/sand ow.

728

X. Chen et al. / Wear 261 (2006) 715729

ow velocity, pipe diameter and particle diameter. These observations were experimentally conrmed by Blanchard et al. [2]. Flow streamline redirection is more drastic in plugged tees than in elbows. As seen in Fig. 24(a), more particles deviate from the streamlines and impinge the plugged tee at the joint corner of the plugged tee. Comparing Figs. 21(a) and (b), greater erosion is observed at the joint corner of the plugged tee than at exit region of the elbow. Erosion was predicted for water/sand ow in both plugged tees and elbows using pipe diameters and particles diameters listed in Table 2. Simulations consistently show that plugged tees experience greater erosion than elbows. The predictions are qualitatively supported by Bourgoynes [3] observations that when only liquid (i.e. water) and sand are present, the plugged tee is less resistant than the long radius or short radius elbows. Particle motion and resulting erosion in both plugged tees and the elbows are primarily determined by two mechanisms: drag force and particle inertia. Particle inertia dominates the particle motion with low density and viscosity uid (e.g. air). The redirection of the uid in the plugged tees has little effect on the particles. Consequently, the particles directly impinge the end surface of the plugged tee before they pass through the plugged tee geometry transition. When the uid density and viscosity increase, the drag force exerted on particles by the carrier uid has greater effect on particle motion. Consequently, fewer particles impinge the end surface of the plugged tee and more particles tend to impact the joint corner. Drag force is the dominant mechanism for water/sand ow. The particle impingements primarily take place at the joint corner with greater erosion than in the elbow. Similarly, in air/sand ow, the effect of drag force increases when the particle diameter decreases, which results in increasing erosion at the joint corner of the plugged tees (side and corner region). This analysis is conrmed by Fig. 20. 6. Conclusion This study concerns the relative erosion severity between plugged tees and standard elbows (r/D = 1.5). A CFD-based erosion prediction model was applied to predict erosion in plugged tees and elbows for a wide range dilute air/sand ows. Experimental erosion tests were conducted in the plugged tee and elbow with 0.0254 m diameter to validate the simulation results. Both simulations and data show that considerable erosion occurs at the end and side and corner regions of the plugged tees. Based on the analysis of the CFD simulations and experimental data, a constant value of 0.5 is recommended for relative erosion severity between the end region of plugged tees and elbows. A simplied model in dimensionless form was developed to predict the relative erosion severity of the side and corner region. It is shown that the sand loading signicantly affects the relative erosion severity. Experiments under high sand volume concentration indicate that erosion in plugged tees is about two orders of magnitude less than erosion in elbows. This study demonstrates that the relative erosion severity is signicantly affected by the carrier uid properties. Simulations

in water/solid ows indicate that plugged tees experience greater erosion than elbows. Bourugoynes [3] experimental observations qualitatively conrm the numerical simulations. Thus, the uid properties of an erosive environment must be carefully examined to determine whether to adopt plugged tee to redirect the ow. Further investigation should be conducted to incorporate uid properties in the simplied model for the prediction of relative erosion severity. Acknowledgement The authors would like to express their gratitude to the member companies of the Erosion/Corrosion Research Center (ECRC) at The University of Tulsa for supporting this study. References
[1] J. Wang, S.A. Shirazi, A CFD based correlation for erosion factor for longradius elbows and bends, ASME J. Energy Res. Tech. 125 (2003) 26 34. [2] D.J. Blanchard, P. Grifth, E. Rabinowicz, Erosion of a pipe bend by solid particles entrained in water, ASME J. Eng. Ind. 106 (1984) 213 217. [3] A.T. Bourgoyne, Experimental study of erosion in diverter systems due to sand production, in: Proc. SPE/IADC-18716, Louisiana, February 28March 3, 1989. [4] R. Eyler, Design and analysis of a pneumatic ow loop, M.S. Thesis, West Virginia University, West Virginia, 1987. [5] Q.H. Mazumder, Development and validation of a mechanistic model to predict erosion in single-phase and multiphase Flow, Ph.D. Dissertation, The University of Tulsa, Oklahoma, 2004. [6] R. Mishra, S.N. Singh, V. Seshadri, Study of wear characteristics and solid distribution in constant area and erosion-resistant long-radius pipe bends for the ow of multisized particulate slurries, Wear 217 (1998) 297 306. [7] M.M. Salama, An alternative to API 14E erosional velocity limits for sand laden uids, ASME J. Energy Res. Tech. 122 (2000) 71 77. [8] B.S. McLaury, S.A. Shirazi, Is API RP 14E reliable for predicting an erosional production velocity when sand production is anticipated? in: Proc. ETCE/OMAE Joint Conference, Paper No. ETCE00-PR054, New Orleans, LA, February 1417, 2000. [9] B. Bozzini, M.E. Ricotti, M. Boniardi, C. Mele, Evaluation of erosioncorrosion in multiphase ow via CFD and experimental analysis Wear 255 (2003) 237245. [10] X. Chen, B.S. McLaury, S.A. Shirazi, Application and experimental validation of a computational uid dynamics (CFD) based erosion prediction model in elbows and plugged tees, J. Comput. Fluids 33 (2004) 12511272. [11] X. Chen, B.S. McLaury, S.A. Shirazi, A comprehensive procedure to estimate erosion in elbows for gas/liquid/sand multiphase ow, ASME J. Energy Res. Tech. 128 (2006) 19. [12] J. Fan, X. Yao, K. Cen, Experimental and numerical investigation of a new method for protecting bends from erosion in gas-particle ow, Wear 251 (2001) 853960. [13] A. Keating, S. Nesic, Particle tracking and erosion prediction in three-dimensional bends, in: Proc. of ASME FED (Fluids Engineering Division) Summer Meeting, FEDSM2000-11249, Boston, June 1115, 2000. [14] R.J.K. Wood, T.F. Jones, J. Ganeshalingam, N.J. Miles, Comparison of predicted and experimental erosion estimates in slurry ducts, Wear 256 (2004) 937947. [15] AEA Technology, CFX-4.4 solver, Oxfordshire, United Kingdom, 2000.

X. Chen et al. / Wear 261 (2006) 715729 [16] R. Clift, J. Grace, M. Weber, Bubbles, Drops and Particles, Academic Press, 1978. [17] T. Grant, W. Tabakoff, Erosion prediction in turbomachinery resulting from environmental solid particles, J. Aircraft 12 (1975) 447471. [18] K. Ahlert, Effects of particle impingement angle and surface wetting on solid particle erosion of AISI 1018 steel, M.S. Thesis, The University of Tulsa, OK, 1994.

729

[19] B.S. McLaury, Predicting solid particle erosion resulting from turbulent uctuation in oileld geometries, Ph.D. Dissertation, The University of Tulsa, OK, 1996. [20] A. Forder, M. Thew, D. Harrison, A numerical investigation of solid particle erosion experienced within oileld control valves, Wear 216 (1998) 184193.

Das könnte Ihnen auch gefallen