- Email: [email protected]
Laboratory modelling of erosion damage by vortices in slurry flow
HYDROM-04408; No of Pages 8 Hydrometallurgy xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Hydrometallurgy journal homepage: www.elsevier.com/locate/hydromet
Laboratory modelling of erosion damage by vortices in slurry flow L.J.W. Graham a,⁎, J. Wu a, G. Short a, C.B. Solnordal a, C.Y. Wong a, G. Brown b, O. Celliers b, D. Whyte b a b
CSIRO Mineral Resources, Melbourne, VIC, Australia Technology Delivery Group, Alcoa Refining, WA, Australia
a r t i c l e
i n f o
Article history: Received 30 September 2015 Received in revised form 8 July 2016 Accepted 20 July 2016 Available online xxxx Keywords: Erosion Vortex flow Fluid dynamics
a b s t r a c t Erosion damage caused by suspended particles in slurries leads to production loss and on-going maintenance costs. Such damage is common in flow equipment used in slurry transport, including processing equipment in alumina refineries. CSIRO has been conducting research under AMIRA P931 “Multiphase Flow Erosion” projects from 2006 to 2013, under sponsorship funding support from Alcoa, BHP Billiton, Rio Tinto Alcan, Vale and Pentair Valves (Tyco Flow). CSIRO has a continuing focus on building knowledge and methods to improve prediction of the service life of flow equipment under erosion conditions, and to develop strategies to reduce erosion through altered flow design. Interestingly, none of the case studies requested by the industry partners involved simple impingement erosion or sliding bed erosion. The emphasis has been on solving erosion problems using the principles of the underlying multiphase fluid dynamics, which called for an in-depth treatment of non-uniform flows. Consequently the current study was very different from the usual treatment of erosion (which focuses on direct impingement simply because it is easier to model and measure) and instead addressed the much more common industrial problem of localised erosion. By using fluid dynamics modelling, experimental visualisation and quantitative measurements of erosion scars, several fluid dynamic mechanisms have previously been identified as causing severe erosion attacks. These included erosion by vortices, by flashing and by various non-uniform flows. Observations of accelerated wear have shown that vortex erosion is present in many flow geometries critically important in sponsors' plants, e.g. pipe work around a valve, protrusions in a pipe and many conventional engineering designs. This paper focuses on vortex erosion in a variety of flow situations and examines the fluid dynamics and consequent erosion. © 2016 Published by Elsevier B.V.
1. Introduction Severe erosion attack can occur in mineral process plants where complex fluid dynamics occurs. Reducing the maintenance costs associated with erosion damage is of significant interest to the minerals industry. Hence, CSIRO has been conducting fluid dynamics based erosion research under AMIRA P931 “Multiphase Flow Erosion” projects from 2006 to 2013, under sponsorship funding support from Alcoa, BHP Billiton, Rio Tinto Alcan, Vale and Pentair Valves (Tyco Flow) as well as continuing research activity in this area. Vortex flows are one example of this and have been discussed in the literature by Brown (2002) in the context of a blanked tee in pipe work and Graham et al. (2010) who examined the flow around obstacles and the consequent erosion. A good summary of the fundamentals of vortex flows caused by obstacles is given by Simpson (2001).
⁎ Corresponding author. E-mail address: [email protected] (L.J.W. Graham).
As far as erosion due to vortex action around obstacles is concerned, there are several recent literature examples, particularly from the practical case of erosion of the base of bridge piers where the erosion action due to vortices generated from obstacles is significant. There is also interest in the vortex erosion phenomena in the areas of heat transfer where the obstacle (or “dimple”) is used in heat exchangers to promote turbulence. Escauriaza and Sotiropoulos (2011) used a detached eddy simulation approach to model the flow around a surface mounted pier. A new model was developed to examine the erosion of the surface around the pier. Their model was found to qualitatively reproduce the erosion effects observed, however the computed rate of erosion was found to be less than from observations. Another recent example from the literature is Kirkil and Constantinescu (2015) who used experimental and numerical methods to examine the classic case of a cylindrical obstacle in a channel. Their numerical approach is interesting as a Direct Numerical Simulation was used for the low Reynolds number case (16,000) whereas a detached eddy simulation was used for the higher Reynolds number of
http://dx.doi.org/10.1016/j.hydromet.2016.07.013 0304-386X/© 2016 Published by Elsevier B.V.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
2
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
50,000. The numerical results agreed very well with the PIV measurements. No erosion modelling was attempted. Zhao et al. (2014) used numerical methods to examine the heat transfer and erosion behaviour of tubes used in heat exchangers which were fitted with dimples to enhance heat transfer. It was found that it was possible to optimise the tube design and vortex generators to get a good compromise between heat transfer and erosion performance. The preceding examples demonstrate that the study of erosion due to vortices generated by obstacles is a topic of considerable interest. Vortices are also generated in other flow situations where the flow is required to change direction as demonstrated below. A tee section is commonly encountered in mineral processing plants where the flow is required to be sent into one of two possible directions where the unused flow path could, for example, be closed by a knife gate valve. Blanking plates are also used for more long term blockage of the unused leg. Under some circumstances it is known that a vortex can develop in the tee which leads to significant erosion on the blanking plate or valve blade (Brown, 1999; Brown, 2002). The generic flow geometry is shown in Fig. 1. This flow has previously been the subject of a Computational Fluid Dynamics (CFD) study by Brown (1999) and Brown (2002). The CFD results in these papers showed that a vortex was present at the position of the maximum erosion on the blanked end provided some swirl was present in the inlet flow. A photograph of a full scale eroded blanking plate is shown in Fig. 2(a). No erosion modelling was presented in these papers, although paint was used to visualise the erosion pattern in a separate investigation in the laboratory at CSIRO (Fig. 2(b)). This paper presents examples of the erosion due to vortex action around obstacles and that caused by a blanked tee. Three examples are presented: erosion on a pipe wall caused by a cylinder protruding from the wall; erosion on a cylinder in cross-flow caused by a protrusion on the cylinder, and erosion in and around a blanked tee-junction. Quantitative measurements of the surface erosion are made and CFD analyses of the erosion around the obstacles are presented. Fig. 2. (a) Full scale example of erosion in tee blank, (b) paint erosion model using the paint modelling technique from Wu et al. (2011).
2. Experimental and CFD methods 2.1. Slurry flow loop for erosion tests The erosion tests were conducted in a slurry flow loop, schematically represented in Fig. 3. The flow loop consists of a 3000 L agitated tank, a Warman 4 × 3 AH slurry pump, a magnetic flowmeter and appropriate connecting pipe work in NB 50 mm pipe. The sample under investigation was arranged in a straight, vertical length of pipe. Normally a straight length of pipe of at least 20 diameters was provided before the sample, and the rig could be configured for up to four samples simultaneously if required. Cylindrical samples of the same material as the sample under investigation were tested at the same time (Fig. 3(a)). In this way they acted as a control sample undergoing direct impact erosion, for comparison with the sample under investigation. 2.2. Erosion measurement
Fig. 1. Generic blanked tee geometry. The blanked end may be due to a valve being closed for example.
All erosion measurements were made using a Sheffield Discovery II coordinate measurement machine (CMM) as shown in Fig. 4. The repeatability of measurements made using the CMM was of the order of ±3 μm. The sample under consideration was placed in the CMM both before and after exposure to slurry, and a map of the erosion extent was created by determining the difference between the two profiles. More detail of the procedure has been published elsewhere (Wong et al., 2015).
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
3
Fig. 3. (a) Schematic diagram, and (b) photograph of slurry flow loop, showing location of the sample under investigation as well as the control sample.
2.3. Vortex erosion caused by a protruding obstacle Slurry flow through a pipe with an obstacle protruding from its inner wall was studied as shown in Fig. 5. The erodible sample of interest in this case was a short section of pipe made from two pieces of aluminium (6061-T6) with a 53 mm diameter section machined out of each piece. This allowed access for the touch probe of the CMM to determine the profile of the pipe inner surface. The obstacle was a 16 mm diameter cylindrical stainless steel pin inset into the pipe and protruding approximately half way across the pipe cross-section. Using stainless steel for the obstacle ensured that it did not erode significantly during the test programme, since the erosion of interest was on the pipe wall around the obstacle, and not the obstacle itself. A replaceable plug contoured
to the shape of the pipe wall was also made to allow for good access for the CMM probe, as shown in Fig. 6. A 1 mm diameter spherical CMM probe was used for all tests. Table 1 shows the nominal test conditions.
Flow
Fig. 5. Pipe geometry with cylindrical obstacle, mounted on coordinate measurement machine. Arrow indicates flow direction.
Fig. 4. Sheffield Discovery II coordinate measurement machine (CMM) at CSIRO Fluids Engineering laboratory.
Fig. 6. Measuring the uneroded condition of the experimental test section. The obstacle has been removed and a plug has been inserted to allow for probe access in the immediate vicinity of the obstacle.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
4
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
Table 1 Test conditions. Parameter
Details
Pipe Obstacle Material Solids type Solids d50 Fluid Solids volume concentration Velocity
ID nominal 53 mm, 150 mm in length OD 16 mm, height 26.5 mm, stainless steel Aluminium (for rapid tests) Unimin Silica Flour 100G 25 μm approx. Water 12% (v/v) 4.5 m s−1
rig as it was known from previous work (Brown, 2002) that a degree of swirl was required to develop the vortex on the blanked end. The swirler was arranged such that the vanes turned through 90° over a distance of 2 pipe diameters. Experiments with the swirler removed showed that the vortex was naturally generated in the tee. However, by using the swirler, a more stable and reproducible vortex could be generated. This was particularly important for comparing the results to CFD simulations where a consistent boundary condition is desirable. 2.5. CFD methods
Fig. 7. Schematic diagram of experimental blanked tee. Arrows indicate flow direction.
A cylinder sample of identical material (6061-T6 aluminium) to the pipe piece was also used as a control sample (see Fig. 3(a)) for simultaneous erosion tests. This allowed for a comparison between the direct impact type of erosion on the cylinder and that caused by vortices around the sample of interest. Other obstacles were considered, including a cable tie around the control cylinder. The subsequent erosion this caused to the cylinder surface was studied. The results of these experiments are discussed in Section 3. 2.4. Erosion due to vortex in a blanked tee A slurry flow loop containing a blanked tee section was designed and manufactured from 316 L stainless steel pipe. A schematic representation of the tee is shown in Fig. 7. A swirler was incorporated into the
The CFD model for slurry flow through a pipe with an obstacle protruding from its inner wall used a combined Eulerian-Lagrangian approach to modelling the flow. The model was implemented using commercial software ANSYS-CFX 13.0. The effect of turbulence was simulated using the Baseline (BSL) Reynolds Stress Model to better capture the vortices expected in the flow. The fluid was modelled as pure water at room temperature, while the particles were tracked separately using Lagrangian particle tracking techniques. One way coupling was assumed between water and particles, as simulations with two-phase coupling showed no difference in the predicted flow fields. Solid surfaces were modelled using a no-slip wall condition. The inlet to the flow domain was specified as a Dirichlet boundary with medium turbulence intensity and uniform velocity of both water and particles, while the outlet was specified to be a constant pressure boundary with pressure equal to 101,325 Pa. The mesh was created using ANSYS-Workbench and was an unstructured tetrahedral mesh with wall inflation, and was made up of approximately 1,000,000 elements. The geometry extended 10 diameters upstream and downstream of the protruding obstacle. Convergence of the fluid flow was achieved within 200 iterations. 100,000 particle tracks were then calculated to visualise the particle behaviour. The flow and particle motion were used to help understand the erosion mechanisms. 3. Results and discussion 3.1. Vortex erosion caused by a protruding obstacle It is known that vortex erosion occurs in plant situations and it would be useful to have a ranking test for materials under the conditions of vortex erosion. The horseshoe vortex that typically occurs around the base of a protrusion has the advantage of being readily reproducible and geometrically simple. It is thus a good candidate geometry for erosion tests and the flow structure and erosion caused by this geometry are investigated in detail here. The test slurry was made from tap water and 12% by volume silica flour (d50 approximately 25 μm) at ambient conditions. The superficial velocity was 4.5 m s−1. Fig. 8 shows the erosion obtained after 2 h of exposure to slurry flow. It is seen that the maximum erosion is 0.11 mm and this maximum
Fig. 8. Experimentally determined erosion distribution after exposure to slurry flow for 2 h.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
5
Fig. 9. Experimentally determined erosion distribution after exposure to slurry flow for 170 h.
occurs close to the junction between the obstacle and the pipe wall. At the end of the tests, after 170 h of exposure to slurry flow, the maximum erosion is 3.3 mm as shown in Fig. 9. The 170 hour case also shows the formation of secondary erosion scars behind the first large erosion scar. These can also be seen in the photograph of the erosion in Fig. 10. It appears that the formation of the relatively deep primary erosion scar causes additional flow disturbance, which then leads to the formation of the secondary erosion scars. It is considered that the erosion shown above is a consequence of vortex action. The CFD results (discussed later) show the existence of a horseshoe vortex which is expected for this kind of flow around a surface mounted obstacle. The ratio between the erosion due to the vortex and that due to direct impact on the control cylinder sample under the same flow conditions is shown in Fig. 11. The ratio is approximately 3 after two hour exposure to slurry flow. Also shown is the ratio between the erosion due to the horseshoe vortex and the erosion on the pipe wall away from the obstacle, where it is unaffected by the horseshoe vortex. Of interest here is that the erosion due to the horseshoe vortex can be of the order of 80 times the erosion on the pipe wall itself. CFD results are presented in Fig. 12 using the pristine geometry (time = 0 h). Fig. 12 shows a plan view of the velocity distribution around the obstacle at a distance of 0.5 mm from the pipe wall. The image shows the expected characteristics of flow around a cylinder. Specifically, a stagnation point exists immediately upstream of the obstacle, water accelerates around the obstacle with velocities as high as 8 m s−1, and there is a recirculation region behind the obstacle (where flow is expected to be unsteady). Fig. 13 shows a plot of the surface of constant swirl vector which shows the development of the horseshoe vortex around the obstacle. So, the CFD predicts the vortex responsible for the accelerated erosion, but quantitative erosion rate prediction from the CFD model was
Fig. 10. Photograph of the erosion at 170 hour exposure.
significantly higher than observed experimentally and is still a ‘workin-progress’. Fig. 14 shows the particle paths from the CFD. The particles decelerate as they approach the stagnation point at the front of the obstacle, then accelerate to greater than 9 m s−1 as they pass around it. It is this acceleration together with increased particle impact angle due to the vortex that contributes to the substantial erosion scar around the obstacle, since erosion rate is related to particle impact velocity by a power law. The predictability of the vortex in the horseshoe case suggests that it may be useful as a ranking test for material response to erosion under vortex conditions. Current erosion tests, such as the ASTM G65 dry sand rubber wheel, lack a connection with the actual mechanism of erosion under industrial conditions. To reduce the erosion due to the vortex, a change must be made to the geometry to reduce the vortex action. Changes in geometry can be tested both numerically and experimentally to assess the potential for erosion reduction.
3.2. Surface imperfections The purpose of these experiments was to investigate another protrusion, mounted around a cylinder in cross-flow. The protrusion was 1.5 mm proud of the test piece and was a cable tie affixed to the cylinder. 316 L stainless steel and 27Cr white iron samples were used. 25 μm silica particles were the erodant. The superficial velocity was 3.4 m s−1. A photograph of the stainless steel sample before erosion is shown in Fig. 15.
Fig. 11. Solid line - ratio of experimentally determined maximum erosion due to vortex and the maximum erosion on the control cylinder. Dashed line - ratio of the experimentally determined maximum erosion due to the vortex and the maximum erosion on the pipe wall away from the vortex.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
6
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
3.3. Vortex in a blanked tee Aluminium samples were tested in the blanked tee geometry, with 6% by volume silica flour (d50 approximately 25 μm) in water at ambient conditions. Fig. 18 shows a photograph of an eroded tee blank sample from a test run at bulk slurry velocity of 6 m s−1. Fig. 19 shows the corresponding CMM measurements. These experimental results show a similar pattern to the full scale example of Fig. 2. Particles under the action of such a vortex produce more damage than from the direct impact erosion, as confirmed by the measurements that the ratio of such vortex erosion over the reference cylinder is 1.6– 3.5 over a variety of tests including aluminium and 316 L stainless steel target materials. It was also observed that erosion occurred on the tee itself in locations near the pipe junction as shown in Fig. 20. A photograph of this erosion is shown in Fig. 21. Again, reducing or removing the vortex action may reduce this erosion.
4. Conclusion Fig. 12. CFD results showing slurry velocity distribution 0.5 mm from pipe wall surface.
Fig. 16 shows the experimentally determined erosion scar on the cylinder surface of the white iron sample where the white iron shows increased erosion in the vicinity of the cable tie/cylinder junction. The ratio of the increased erosion to the control cylinder erosion is 2.8 for this case. Fig. 17 shows the results for the 316 L stainless steel cylinder with a cable tie obstacle at bulk slurry velocity 3.4 m s−1. For this case the ratio between the vortex erosion and the cylinder erosion was 3.5. The erosion pattern is qualitatively similar to the white iron case.
A variety of flow patterns have been investigated which share the characteristic of a flow disturbance resulting in vortices. These include obstacles on surfaces and a vortex generated in a blanked tee. Each of these flows shows an increase in erosion relative to that obtained by direct impact of particles which is a consequence of the vortex action. The typical range is 1.6–3.5 times the direct impact erosion as measured on the control cylinders used as a reference in each erosion test. On a pipe wall, the erosion due to vortex action can also be approximately 80 times the background erosion in our tests. The predictability of the vortex in the case of the horseshoe pattern suggests that it may be useful as a ranking test for material response to erosion under vortex conditions. That would be an improvement on current erosion tests, such as the ASTM G65 dry sand rubber wheel,
Fig. 13. CFD results showing surface of constant swirl vector equal to 1000 s−1, showing the horseshoe vortex wrapping around the front of the obstacle, and the trailing vortices behind.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
7
Fig. 16. Experimentally determined erosion distribution on the surface of a cylinder in cross-flow. Cylinder is made from 27Cr white iron with a cable tie obstacle. Bulk slurry velocity is equal to 3.4 m s−1, exposure time is 482 h. Silica flour particles, d50 approx. 25 μm.
which lack any connection to the actual mechanism of erosion under industrial conditions. A change in the fluid dynamics design of these and other geometries showing vortex flows may offer a means of reducing the vortex action and thus providing a reduction in erosion. Tests can be done on small scale model systems which can be used to quantify the expected improvements in erosion. Acknowledgements The CSIRO authors are grateful for the support of AMIRA Projects P931 and P931A. Sponsors of these projects are Alcoa, BHP Billiton, Rio
Fig. 14. CFD results showing particle paths hitting and flowing around the obstacle, coloured by (a) particle size, and (b) particle velocity. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 15. Cable tie on cylinder in pipe sample, pre-erosion.
Fig. 17. Experimentally determined erosion distribution on the surface of a cylinder in cross-flow. Cylinder is made from 316 L stainless steel with a cable tie obstacle. Bulk slurry velocity is equal to 3.4 m s−1, exposure time is 209 h. Silica flour particles, d50 approx. 25 μm.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
8
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
Fig. 21. Erosion damage on tee. Arrow indicates flow direction. Fig. 18. Photograph of sample with 48 hour exposure at 6 m s−1. Silica flour particles, d50 approx. 25 μm.
Tinto Alcan, Vale and Pentair Valves (Tyco). Laboratory support has also been supplied by Dean Harris and the CSIRO Workshops and the authors are grateful for their support. References
Fig. 19. CMM measurements of vortex erosion corresponding to Fig. 18.
Brown, G.J., 1999. Erosion Prediction in Slurry Pipeline Tee-Junctions. Second International Conference on CFD in the Minerals and Process Industries. CSIRO, Melbourne, Australia, pp. 237–242. Brown, G.J., 2002. Erosion prediction in slurry pipeline tee-junctions. Appl. Math. Model. 26 (2), 155–170. Escauriaza, C., Sotiropoulos, F., 2011. Initial stages of erosion and bed form development in a turbulent flow around a cylindrical pier. J. Geophys. Res.-Earth Surf. 116. Graham, L.J.W., Lester, D.R., Wu, J., 2010. Quantification of erosion distributions in complex geometries. Wear 268 (9–10), 1066–1071. Kirkil, G., Constantinescu, G., 2015. Effects of cylinder Reynolds number on the turbulent horseshoe vortex system and near wake of a surface-mounted circular cylinder. Phys. Fluids 27 (7). Simpson, R.L., 2001. Junction flows. Annu. Rev. Fluid Mech. 33, 415–443. Wong, C.Y., et al., 2015. Slurry erosion of surface imperfections in pipeline systems. Wear 336–337, 72–85. Wu, J., et al., 2011. An effective modeling tool for studying erosion. Wear 270 (9–10), 598–605. Zhao, X.B., Tang, G.H., Ma, X.W., Jin, Y., Tao, W.Q., 2014. Numerical investigation of heat transfer and erosion characteristics for H-type finned oval tube with longitudinal vortex generators and dimples. Appl. Energy 127, 93–104.
Erosion location
Fig. 20. Tee geometry showing location of erosion on the tee itself.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
Contents lists available at ScienceDirect
Hydrometallurgy journal homepage: www.elsevier.com/locate/hydromet
Laboratory modelling of erosion damage by vortices in slurry flow L.J.W. Graham a,⁎, J. Wu a, G. Short a, C.B. Solnordal a, C.Y. Wong a, G. Brown b, O. Celliers b, D. Whyte b a b
CSIRO Mineral Resources, Melbourne, VIC, Australia Technology Delivery Group, Alcoa Refining, WA, Australia
a r t i c l e
i n f o
Article history: Received 30 September 2015 Received in revised form 8 July 2016 Accepted 20 July 2016 Available online xxxx Keywords: Erosion Vortex flow Fluid dynamics
a b s t r a c t Erosion damage caused by suspended particles in slurries leads to production loss and on-going maintenance costs. Such damage is common in flow equipment used in slurry transport, including processing equipment in alumina refineries. CSIRO has been conducting research under AMIRA P931 “Multiphase Flow Erosion” projects from 2006 to 2013, under sponsorship funding support from Alcoa, BHP Billiton, Rio Tinto Alcan, Vale and Pentair Valves (Tyco Flow). CSIRO has a continuing focus on building knowledge and methods to improve prediction of the service life of flow equipment under erosion conditions, and to develop strategies to reduce erosion through altered flow design. Interestingly, none of the case studies requested by the industry partners involved simple impingement erosion or sliding bed erosion. The emphasis has been on solving erosion problems using the principles of the underlying multiphase fluid dynamics, which called for an in-depth treatment of non-uniform flows. Consequently the current study was very different from the usual treatment of erosion (which focuses on direct impingement simply because it is easier to model and measure) and instead addressed the much more common industrial problem of localised erosion. By using fluid dynamics modelling, experimental visualisation and quantitative measurements of erosion scars, several fluid dynamic mechanisms have previously been identified as causing severe erosion attacks. These included erosion by vortices, by flashing and by various non-uniform flows. Observations of accelerated wear have shown that vortex erosion is present in many flow geometries critically important in sponsors' plants, e.g. pipe work around a valve, protrusions in a pipe and many conventional engineering designs. This paper focuses on vortex erosion in a variety of flow situations and examines the fluid dynamics and consequent erosion. © 2016 Published by Elsevier B.V.
1. Introduction Severe erosion attack can occur in mineral process plants where complex fluid dynamics occurs. Reducing the maintenance costs associated with erosion damage is of significant interest to the minerals industry. Hence, CSIRO has been conducting fluid dynamics based erosion research under AMIRA P931 “Multiphase Flow Erosion” projects from 2006 to 2013, under sponsorship funding support from Alcoa, BHP Billiton, Rio Tinto Alcan, Vale and Pentair Valves (Tyco Flow) as well as continuing research activity in this area. Vortex flows are one example of this and have been discussed in the literature by Brown (2002) in the context of a blanked tee in pipe work and Graham et al. (2010) who examined the flow around obstacles and the consequent erosion. A good summary of the fundamentals of vortex flows caused by obstacles is given by Simpson (2001).
⁎ Corresponding author. E-mail address: [email protected] (L.J.W. Graham).
As far as erosion due to vortex action around obstacles is concerned, there are several recent literature examples, particularly from the practical case of erosion of the base of bridge piers where the erosion action due to vortices generated from obstacles is significant. There is also interest in the vortex erosion phenomena in the areas of heat transfer where the obstacle (or “dimple”) is used in heat exchangers to promote turbulence. Escauriaza and Sotiropoulos (2011) used a detached eddy simulation approach to model the flow around a surface mounted pier. A new model was developed to examine the erosion of the surface around the pier. Their model was found to qualitatively reproduce the erosion effects observed, however the computed rate of erosion was found to be less than from observations. Another recent example from the literature is Kirkil and Constantinescu (2015) who used experimental and numerical methods to examine the classic case of a cylindrical obstacle in a channel. Their numerical approach is interesting as a Direct Numerical Simulation was used for the low Reynolds number case (16,000) whereas a detached eddy simulation was used for the higher Reynolds number of
http://dx.doi.org/10.1016/j.hydromet.2016.07.013 0304-386X/© 2016 Published by Elsevier B.V.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
2
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
50,000. The numerical results agreed very well with the PIV measurements. No erosion modelling was attempted. Zhao et al. (2014) used numerical methods to examine the heat transfer and erosion behaviour of tubes used in heat exchangers which were fitted with dimples to enhance heat transfer. It was found that it was possible to optimise the tube design and vortex generators to get a good compromise between heat transfer and erosion performance. The preceding examples demonstrate that the study of erosion due to vortices generated by obstacles is a topic of considerable interest. Vortices are also generated in other flow situations where the flow is required to change direction as demonstrated below. A tee section is commonly encountered in mineral processing plants where the flow is required to be sent into one of two possible directions where the unused flow path could, for example, be closed by a knife gate valve. Blanking plates are also used for more long term blockage of the unused leg. Under some circumstances it is known that a vortex can develop in the tee which leads to significant erosion on the blanking plate or valve blade (Brown, 1999; Brown, 2002). The generic flow geometry is shown in Fig. 1. This flow has previously been the subject of a Computational Fluid Dynamics (CFD) study by Brown (1999) and Brown (2002). The CFD results in these papers showed that a vortex was present at the position of the maximum erosion on the blanked end provided some swirl was present in the inlet flow. A photograph of a full scale eroded blanking plate is shown in Fig. 2(a). No erosion modelling was presented in these papers, although paint was used to visualise the erosion pattern in a separate investigation in the laboratory at CSIRO (Fig. 2(b)). This paper presents examples of the erosion due to vortex action around obstacles and that caused by a blanked tee. Three examples are presented: erosion on a pipe wall caused by a cylinder protruding from the wall; erosion on a cylinder in cross-flow caused by a protrusion on the cylinder, and erosion in and around a blanked tee-junction. Quantitative measurements of the surface erosion are made and CFD analyses of the erosion around the obstacles are presented. Fig. 2. (a) Full scale example of erosion in tee blank, (b) paint erosion model using the paint modelling technique from Wu et al. (2011).
2. Experimental and CFD methods 2.1. Slurry flow loop for erosion tests The erosion tests were conducted in a slurry flow loop, schematically represented in Fig. 3. The flow loop consists of a 3000 L agitated tank, a Warman 4 × 3 AH slurry pump, a magnetic flowmeter and appropriate connecting pipe work in NB 50 mm pipe. The sample under investigation was arranged in a straight, vertical length of pipe. Normally a straight length of pipe of at least 20 diameters was provided before the sample, and the rig could be configured for up to four samples simultaneously if required. Cylindrical samples of the same material as the sample under investigation were tested at the same time (Fig. 3(a)). In this way they acted as a control sample undergoing direct impact erosion, for comparison with the sample under investigation. 2.2. Erosion measurement
Fig. 1. Generic blanked tee geometry. The blanked end may be due to a valve being closed for example.
All erosion measurements were made using a Sheffield Discovery II coordinate measurement machine (CMM) as shown in Fig. 4. The repeatability of measurements made using the CMM was of the order of ±3 μm. The sample under consideration was placed in the CMM both before and after exposure to slurry, and a map of the erosion extent was created by determining the difference between the two profiles. More detail of the procedure has been published elsewhere (Wong et al., 2015).
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
3
Fig. 3. (a) Schematic diagram, and (b) photograph of slurry flow loop, showing location of the sample under investigation as well as the control sample.
2.3. Vortex erosion caused by a protruding obstacle Slurry flow through a pipe with an obstacle protruding from its inner wall was studied as shown in Fig. 5. The erodible sample of interest in this case was a short section of pipe made from two pieces of aluminium (6061-T6) with a 53 mm diameter section machined out of each piece. This allowed access for the touch probe of the CMM to determine the profile of the pipe inner surface. The obstacle was a 16 mm diameter cylindrical stainless steel pin inset into the pipe and protruding approximately half way across the pipe cross-section. Using stainless steel for the obstacle ensured that it did not erode significantly during the test programme, since the erosion of interest was on the pipe wall around the obstacle, and not the obstacle itself. A replaceable plug contoured
to the shape of the pipe wall was also made to allow for good access for the CMM probe, as shown in Fig. 6. A 1 mm diameter spherical CMM probe was used for all tests. Table 1 shows the nominal test conditions.
Flow
Fig. 5. Pipe geometry with cylindrical obstacle, mounted on coordinate measurement machine. Arrow indicates flow direction.
Fig. 4. Sheffield Discovery II coordinate measurement machine (CMM) at CSIRO Fluids Engineering laboratory.
Fig. 6. Measuring the uneroded condition of the experimental test section. The obstacle has been removed and a plug has been inserted to allow for probe access in the immediate vicinity of the obstacle.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
4
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
Table 1 Test conditions. Parameter
Details
Pipe Obstacle Material Solids type Solids d50 Fluid Solids volume concentration Velocity
ID nominal 53 mm, 150 mm in length OD 16 mm, height 26.5 mm, stainless steel Aluminium (for rapid tests) Unimin Silica Flour 100G 25 μm approx. Water 12% (v/v) 4.5 m s−1
rig as it was known from previous work (Brown, 2002) that a degree of swirl was required to develop the vortex on the blanked end. The swirler was arranged such that the vanes turned through 90° over a distance of 2 pipe diameters. Experiments with the swirler removed showed that the vortex was naturally generated in the tee. However, by using the swirler, a more stable and reproducible vortex could be generated. This was particularly important for comparing the results to CFD simulations where a consistent boundary condition is desirable. 2.5. CFD methods
Fig. 7. Schematic diagram of experimental blanked tee. Arrows indicate flow direction.
A cylinder sample of identical material (6061-T6 aluminium) to the pipe piece was also used as a control sample (see Fig. 3(a)) for simultaneous erosion tests. This allowed for a comparison between the direct impact type of erosion on the cylinder and that caused by vortices around the sample of interest. Other obstacles were considered, including a cable tie around the control cylinder. The subsequent erosion this caused to the cylinder surface was studied. The results of these experiments are discussed in Section 3. 2.4. Erosion due to vortex in a blanked tee A slurry flow loop containing a blanked tee section was designed and manufactured from 316 L stainless steel pipe. A schematic representation of the tee is shown in Fig. 7. A swirler was incorporated into the
The CFD model for slurry flow through a pipe with an obstacle protruding from its inner wall used a combined Eulerian-Lagrangian approach to modelling the flow. The model was implemented using commercial software ANSYS-CFX 13.0. The effect of turbulence was simulated using the Baseline (BSL) Reynolds Stress Model to better capture the vortices expected in the flow. The fluid was modelled as pure water at room temperature, while the particles were tracked separately using Lagrangian particle tracking techniques. One way coupling was assumed between water and particles, as simulations with two-phase coupling showed no difference in the predicted flow fields. Solid surfaces were modelled using a no-slip wall condition. The inlet to the flow domain was specified as a Dirichlet boundary with medium turbulence intensity and uniform velocity of both water and particles, while the outlet was specified to be a constant pressure boundary with pressure equal to 101,325 Pa. The mesh was created using ANSYS-Workbench and was an unstructured tetrahedral mesh with wall inflation, and was made up of approximately 1,000,000 elements. The geometry extended 10 diameters upstream and downstream of the protruding obstacle. Convergence of the fluid flow was achieved within 200 iterations. 100,000 particle tracks were then calculated to visualise the particle behaviour. The flow and particle motion were used to help understand the erosion mechanisms. 3. Results and discussion 3.1. Vortex erosion caused by a protruding obstacle It is known that vortex erosion occurs in plant situations and it would be useful to have a ranking test for materials under the conditions of vortex erosion. The horseshoe vortex that typically occurs around the base of a protrusion has the advantage of being readily reproducible and geometrically simple. It is thus a good candidate geometry for erosion tests and the flow structure and erosion caused by this geometry are investigated in detail here. The test slurry was made from tap water and 12% by volume silica flour (d50 approximately 25 μm) at ambient conditions. The superficial velocity was 4.5 m s−1. Fig. 8 shows the erosion obtained after 2 h of exposure to slurry flow. It is seen that the maximum erosion is 0.11 mm and this maximum
Fig. 8. Experimentally determined erosion distribution after exposure to slurry flow for 2 h.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
5
Fig. 9. Experimentally determined erosion distribution after exposure to slurry flow for 170 h.
occurs close to the junction between the obstacle and the pipe wall. At the end of the tests, after 170 h of exposure to slurry flow, the maximum erosion is 3.3 mm as shown in Fig. 9. The 170 hour case also shows the formation of secondary erosion scars behind the first large erosion scar. These can also be seen in the photograph of the erosion in Fig. 10. It appears that the formation of the relatively deep primary erosion scar causes additional flow disturbance, which then leads to the formation of the secondary erosion scars. It is considered that the erosion shown above is a consequence of vortex action. The CFD results (discussed later) show the existence of a horseshoe vortex which is expected for this kind of flow around a surface mounted obstacle. The ratio between the erosion due to the vortex and that due to direct impact on the control cylinder sample under the same flow conditions is shown in Fig. 11. The ratio is approximately 3 after two hour exposure to slurry flow. Also shown is the ratio between the erosion due to the horseshoe vortex and the erosion on the pipe wall away from the obstacle, where it is unaffected by the horseshoe vortex. Of interest here is that the erosion due to the horseshoe vortex can be of the order of 80 times the erosion on the pipe wall itself. CFD results are presented in Fig. 12 using the pristine geometry (time = 0 h). Fig. 12 shows a plan view of the velocity distribution around the obstacle at a distance of 0.5 mm from the pipe wall. The image shows the expected characteristics of flow around a cylinder. Specifically, a stagnation point exists immediately upstream of the obstacle, water accelerates around the obstacle with velocities as high as 8 m s−1, and there is a recirculation region behind the obstacle (where flow is expected to be unsteady). Fig. 13 shows a plot of the surface of constant swirl vector which shows the development of the horseshoe vortex around the obstacle. So, the CFD predicts the vortex responsible for the accelerated erosion, but quantitative erosion rate prediction from the CFD model was
Fig. 10. Photograph of the erosion at 170 hour exposure.
significantly higher than observed experimentally and is still a ‘workin-progress’. Fig. 14 shows the particle paths from the CFD. The particles decelerate as they approach the stagnation point at the front of the obstacle, then accelerate to greater than 9 m s−1 as they pass around it. It is this acceleration together with increased particle impact angle due to the vortex that contributes to the substantial erosion scar around the obstacle, since erosion rate is related to particle impact velocity by a power law. The predictability of the vortex in the horseshoe case suggests that it may be useful as a ranking test for material response to erosion under vortex conditions. Current erosion tests, such as the ASTM G65 dry sand rubber wheel, lack a connection with the actual mechanism of erosion under industrial conditions. To reduce the erosion due to the vortex, a change must be made to the geometry to reduce the vortex action. Changes in geometry can be tested both numerically and experimentally to assess the potential for erosion reduction.
3.2. Surface imperfections The purpose of these experiments was to investigate another protrusion, mounted around a cylinder in cross-flow. The protrusion was 1.5 mm proud of the test piece and was a cable tie affixed to the cylinder. 316 L stainless steel and 27Cr white iron samples were used. 25 μm silica particles were the erodant. The superficial velocity was 3.4 m s−1. A photograph of the stainless steel sample before erosion is shown in Fig. 15.
Fig. 11. Solid line - ratio of experimentally determined maximum erosion due to vortex and the maximum erosion on the control cylinder. Dashed line - ratio of the experimentally determined maximum erosion due to the vortex and the maximum erosion on the pipe wall away from the vortex.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
6
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
3.3. Vortex in a blanked tee Aluminium samples were tested in the blanked tee geometry, with 6% by volume silica flour (d50 approximately 25 μm) in water at ambient conditions. Fig. 18 shows a photograph of an eroded tee blank sample from a test run at bulk slurry velocity of 6 m s−1. Fig. 19 shows the corresponding CMM measurements. These experimental results show a similar pattern to the full scale example of Fig. 2. Particles under the action of such a vortex produce more damage than from the direct impact erosion, as confirmed by the measurements that the ratio of such vortex erosion over the reference cylinder is 1.6– 3.5 over a variety of tests including aluminium and 316 L stainless steel target materials. It was also observed that erosion occurred on the tee itself in locations near the pipe junction as shown in Fig. 20. A photograph of this erosion is shown in Fig. 21. Again, reducing or removing the vortex action may reduce this erosion.
4. Conclusion Fig. 12. CFD results showing slurry velocity distribution 0.5 mm from pipe wall surface.
Fig. 16 shows the experimentally determined erosion scar on the cylinder surface of the white iron sample where the white iron shows increased erosion in the vicinity of the cable tie/cylinder junction. The ratio of the increased erosion to the control cylinder erosion is 2.8 for this case. Fig. 17 shows the results for the 316 L stainless steel cylinder with a cable tie obstacle at bulk slurry velocity 3.4 m s−1. For this case the ratio between the vortex erosion and the cylinder erosion was 3.5. The erosion pattern is qualitatively similar to the white iron case.
A variety of flow patterns have been investigated which share the characteristic of a flow disturbance resulting in vortices. These include obstacles on surfaces and a vortex generated in a blanked tee. Each of these flows shows an increase in erosion relative to that obtained by direct impact of particles which is a consequence of the vortex action. The typical range is 1.6–3.5 times the direct impact erosion as measured on the control cylinders used as a reference in each erosion test. On a pipe wall, the erosion due to vortex action can also be approximately 80 times the background erosion in our tests. The predictability of the vortex in the case of the horseshoe pattern suggests that it may be useful as a ranking test for material response to erosion under vortex conditions. That would be an improvement on current erosion tests, such as the ASTM G65 dry sand rubber wheel,
Fig. 13. CFD results showing surface of constant swirl vector equal to 1000 s−1, showing the horseshoe vortex wrapping around the front of the obstacle, and the trailing vortices behind.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
7
Fig. 16. Experimentally determined erosion distribution on the surface of a cylinder in cross-flow. Cylinder is made from 27Cr white iron with a cable tie obstacle. Bulk slurry velocity is equal to 3.4 m s−1, exposure time is 482 h. Silica flour particles, d50 approx. 25 μm.
which lack any connection to the actual mechanism of erosion under industrial conditions. A change in the fluid dynamics design of these and other geometries showing vortex flows may offer a means of reducing the vortex action and thus providing a reduction in erosion. Tests can be done on small scale model systems which can be used to quantify the expected improvements in erosion. Acknowledgements The CSIRO authors are grateful for the support of AMIRA Projects P931 and P931A. Sponsors of these projects are Alcoa, BHP Billiton, Rio
Fig. 14. CFD results showing particle paths hitting and flowing around the obstacle, coloured by (a) particle size, and (b) particle velocity. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 15. Cable tie on cylinder in pipe sample, pre-erosion.
Fig. 17. Experimentally determined erosion distribution on the surface of a cylinder in cross-flow. Cylinder is made from 316 L stainless steel with a cable tie obstacle. Bulk slurry velocity is equal to 3.4 m s−1, exposure time is 209 h. Silica flour particles, d50 approx. 25 μm.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013
8
L.J.W. Graham et al. / Hydrometallurgy xxx (2016) xxx–xxx
Fig. 21. Erosion damage on tee. Arrow indicates flow direction. Fig. 18. Photograph of sample with 48 hour exposure at 6 m s−1. Silica flour particles, d50 approx. 25 μm.
Tinto Alcan, Vale and Pentair Valves (Tyco). Laboratory support has also been supplied by Dean Harris and the CSIRO Workshops and the authors are grateful for their support. References
Fig. 19. CMM measurements of vortex erosion corresponding to Fig. 18.
Brown, G.J., 1999. Erosion Prediction in Slurry Pipeline Tee-Junctions. Second International Conference on CFD in the Minerals and Process Industries. CSIRO, Melbourne, Australia, pp. 237–242. Brown, G.J., 2002. Erosion prediction in slurry pipeline tee-junctions. Appl. Math. Model. 26 (2), 155–170. Escauriaza, C., Sotiropoulos, F., 2011. Initial stages of erosion and bed form development in a turbulent flow around a cylindrical pier. J. Geophys. Res.-Earth Surf. 116. Graham, L.J.W., Lester, D.R., Wu, J., 2010. Quantification of erosion distributions in complex geometries. Wear 268 (9–10), 1066–1071. Kirkil, G., Constantinescu, G., 2015. Effects of cylinder Reynolds number on the turbulent horseshoe vortex system and near wake of a surface-mounted circular cylinder. Phys. Fluids 27 (7). Simpson, R.L., 2001. Junction flows. Annu. Rev. Fluid Mech. 33, 415–443. Wong, C.Y., et al., 2015. Slurry erosion of surface imperfections in pipeline systems. Wear 336–337, 72–85. Wu, J., et al., 2011. An effective modeling tool for studying erosion. Wear 270 (9–10), 598–605. Zhao, X.B., Tang, G.H., Ma, X.W., Jin, Y., Tao, W.Q., 2014. Numerical investigation of heat transfer and erosion characteristics for H-type finned oval tube with longitudinal vortex generators and dimples. Appl. Energy 127, 93–104.
Erosion location
Fig. 20. Tee geometry showing location of erosion on the tee itself.
Please cite this article as: Graham, L.J.W., et al., Laboratory modelling of erosion damage by vortices in slurry flow, Hydrometallurgy (2016), http://dx.doi.org/10.1016/j.hydromet.2016.07.013