International Journal of Fluid Machinery and Systems DOI: http://dx.doi.org/10.5293/IJFMS.2016.9.3.277 Vol. 9, No. 3, July-September 2016
ISSN (Online): 1882-9554
Original Paper
Numerical Prediction of Inlet Recirculation in Pumps
Andrej Lipej
1and Duško Mitruševski
21
Faculty of technologies and systems
Na Loko 2, 8000 Novo mesto, Slovenia, andrej.lipej@fts-nm.si
2
SM Pumps,
Ljubljana, Slovenia, dusko.mitrusevski@gmail.com
Abstract
The development of heavy-duty process pumps, usually based on various design criteria, depends on the pump’s application.
The most important criteria are Q-H, efficiency and NPSH characteristics. Cavitation due to inlet recirculation is not often one of the design criteria, although many problems in pump operation appear because of inlet recirculation, when the operation range is from 0.5-0.8 Qopt.
The present paper shows that steady state CFD analysis of inlet recirculation can give quite good results for the design of new hydraulic shapes, which have been developed to expand operating range and to minimize the harmful influence of recirculation at part load. In this paper, the structures of inlet recirculation are presented, as well as detailed shapes of vortices between the blades for various operating regimes, axial velocity distribution at the impeller inlet, the experimental results of NPSH and efficiency characteristics of an existing and newly designed pump.
Keywords: process pump, CFD, cavitation, flow recirculation.
1. Introduction
Flow recirculation at the inlet of centrifugal pumps may cause different (harmless) effects [1], such as noise, vibration, erosion damage and large forces on the impeller. It can also cause cavitation due to recirculation [2]. The chance of damage is heavily dependent on the suction energy level [3], specific speed of the pump, the NPSH margin in the pump and the nature of the flow provided to the suction piping. According to experience, low suction energy pumps are not susceptible to damage from suction recirculation. Solid particles and corrosives can increase damage during suction recirculation, similar to with classic cavitation, even if the NPSH characteristic is good.
A lot of papers dealing with experimental methods [4] or CFD analysis of classic cavitation problems [5, 6] due to NPSH characteristics can be found in the literature on the subject. However, there are not a lot of papers dealing with the numerical analysis of cavitation due to recirculation.
This article presents a detailed numerical analysis of recirculation at the inlet of an impeller and between impeller blades. In normal situations, cavitation arises at the suction side of an impeller blade because certain conditions cause pressure reduction, which can be equal to evaporation pressure. In most cases, the development of centrifugal pumps is based only on the conditions concerning cavitation along the walls of blades. Cavitation due to an inlet vortex is analysed very rarely. However, this type of cavitation may cause material damage, particularly on the pressure side of the blades, which are usually not problematic due to cavitation. Flow in the pump usually has unsteady behaviour and if some special unsteady phenomena have to be analysed, it is not possible to avoid the use of unsteady CFD analyses [7, 8].
Designing the inlet geometry of heavy-duty process pumps is important in defining an operating range free of recirculation at partial flow. Cavitation due to recirculation reduces the level of pump reliability and the danger of impeller damage is substantial, although NPSHa is much higher than NPSHreq of the pump. The operating conditions of pumps in a system dictate the technical solution of the casing and impeller inlet geometry. Under part load operation, a Q << Qopt recirculation vortex occurs at the impeller inlet.
Recirculation free operating range of the pump depends on:
• Specific speed nq,
• Suction specific speed SS,
• Inlet geometry of the casing / impeller.
ReceivedApril 19 2016; accepted for publication June 2 2016: Review conducted by Xuelin Tang. (Paper number O16013C) Corresponding author: Andrej Lipej, Ass. Professor, Andrej.lipej@fts-nm.si
Designing an impeller shape with a wide recirculation free operating range is not an easy task. Moreover, testing the recirculation range and intensity on model pumps on the test rig is a demanding procedure. Computational fluid dynamic combined with experience, and by testing result statistics, could be a very effective method for designing a recirculation free range for pumps with different specific speeds (nq).
2. Theoretical Overview
Cavitation and recirculation in centrifugal pumps have a significant effect on pump performance and pump lifespan.
Recirculation at part flow, at an impeller inlet, directly influences the reliability of the pump and limits the operating range.
Damage can appear on the pressure side of the impeller blade because of cavitation due to recirculation (Fig. 1).
Fig. 1 Damage of impeller of split casing pump because of cavitaiton of recirculation
An appropriate hydraulic design for a pump should prevent many of the negative effects of cavitation due to recirculation. The basic criteria for a recirculation free range is suction specific speed, SS, of the pump.
SS = nQ 0.5/ NPSH 0.75 (1)
SS is defined for BEP and 3% NPSHreq. Normal values of SS are:
• SS = 160-220 for axial inlet impellers,
• SS = 220-280 for suction impeller with axial inlet.
The design criteria for SS is usually more important than the criteria for a high level of efficiency, but for a recirculation free range, suction specific speed should be optimized with the value of maximal allowed NPSHreq of the pump.
Other important parameters (Fig. 2) for a recirculation free range are as follows:
• Impeller inlet diameter D1,
• Impeller hub diameter Do,
• Impeller blade inlet angles.
In general, improving NPSHreq and higher SS increase the recirculation range and limit the pump’s safe operating range.
Fig. 2 Meridional cross section – best efficiency point (left), position of inlet recirculation at part load (right)
Pump recirculation can cause surging and cavitation even when the available NPSHa exceeds the supplier's published NPSHreq by a considerable margin. Suction recirculation typically produces a loud crackling noise in the pump. Recirculation noise is of greater intensity than the noise from low NPSH cavitation and makes a random knocking sound. Discharge recirculation will
D
1
D
produce the same characteristic sound as suction recirculation, with the exception that the highest intensity is noticed at the discharge volute or diffuser.
3. Existing Hydraulic Geometry
Cavitation damage due to recirculation happens very often in pump operation. One example is a water-cooling pump with wide operation range and where the operating point often reaches the recirculation range [9].
Basic pump data:
• Water cooling split casing pump nq = 40,
• Medium: Water 62°C,
• SS = 206 (for half flow rate).
The pump operates in the range from 0.7-1.0 Qopt. Although available NPSHa in the system is higher (Fig. 3) than NPSHreq of the pump, cavitation due to recirculation can damage the impeller blades (Fig. 1).
Fig. 3 Dimensionless Q-NPSH characteristic of the existing pump
Damage due to recirculation cavitation on the existing pump are presented in Fig. 1. Usually, damage is only near the suction side of the impeller blade. In this case damage is seen at the pressure side of the impeller blades [9]. From the existing cavitation characteristics there was no cavitation damage expected for this pump.
4. Design of New Geometry Using CFD Analysis
The main task here was making a hydraulic shape of a new pump impeller with less recirculation at the inlet and to avoid cavitation due to recirculation. Some changes to existing impeller geometry have been made to obtain improved cavitation characteristics, and some new impellers were designed. The final geometry is presented in Fig. 4.
The main difference between the existing and final geometry is generally in the reduction of suction specific speed. To achieve this goal, a slightly different inlet diameter was implemented, and changes were made to the inlet angles of the impeller blades.
The consequences of the above-mentioned changes are noticed in a wider operating range without inlet recirculation, higher overall efficiency and worse NPSH characteristics near the best efficiency point.
In Table 1, a comparison between old and new geometry for important NPSH and recirculation criteria, is shown.
Table 1. Comparison of NPSH and recirculation criteria
Pump nq SS Do/D1
Existing 40 206 0,51
New 41 191 0,56
Fig. 4 Three views of the computational domain – inlet chamber, impeller and spiral casing
For the newly designed geometry, a computational grid (Fig. 5) was generated in order to find out the energetic and cavitation characteristics, and to predict the location, intensity and operating regimes for recirculation appearance.
The quality of the computational grid is an important condition for accurate numerical flow analysis results, particularly in the case of dominant decelerating flow in the pump [10, 11]. Grid refinement is very important near the impeller walls. Besides grid refinement, special attention has been paid to the grid quality parameter - y+, on mesh orthogonality and on expansion and aspect ratio. In the case of the analysed pump, the y+ in almost the completely computational domain is between 20 and 50. Just in a very small area, which is not relevant to the accuracy of the results, y+ exceeds 50 (Fig. 6).
Fig. 5 Computational grid – complete pump and leading edge of the impeller
The computational grid was made separately for the inlet chamber, impeller, spiral casing and inlet pipe. Special attention, in terms of the mesh quality, was paid to the impeller grid generation (Fig. 5) and this is the main reason that the number of elements in the impeller is more than half of the complete computational grid.
Computational grid data:
• Total number of elements – 3500000
• Impeller number of elements – 2000000
• Spiral casing number of elements – 600000
• Inlet chamber number of elements – 700000
• Inlet pipe – 200000.
Fig. 6 Distribution of y+
In the most important part of the computational domain in the impeller, the expansion ratio near the wall is 1.5 (Fig. 5) and the maximal aspect ratio is around 150. The size of the first layer height, near the wall, is 1.0 10-4 m, and the number of inflation layers is 10.
Fig. 7 Boundary conditions
The boundary conditions are presented in figure 7. At the inlet the flow rate is defined, at the outlet the average pressure and so called ‘frozen rotors’ for sliding interfaces were used.
Numerical analysis in the pump (Fig. 8) has been performed for different flow rates and some of them are presented in Table 2.
Table 2. Operating regimes
An important issue in CFD analysis is the turbulent model used. In our case, the k-omega SST turbulent model was used [12].
Fig. 8 Flow distribution in inlet chamber and spiral casing
A new hydraulic design was developed to enable a wider operating range without recirculation and to reach the operating range around BEP. CFD analyses have been done for more partial flow rates when Q << Qop.
Operating point Flow rate Q/Qopt
A 0.26
B 0.33
C 0.46
D 0.56
E 0.62
F 0.65
Fig. 9 Inlet xial velocity component for six operatin points
In Fig. 9 the distribution of the axial velocity component at the impeller inlet cross section, perpendicular to the axis of rotation for six operating points, is presented. It is obvious that at OP A, a huge region of back flow is present. Back flow is no longer present only at OP F. The same results are presented in Fig. 10, where stream lines and velocity vectors are presented at a meridional cross section through the impeller. At OP A, a strong vortex near the shroud is seen. At the OP B, C, D the vortex becomes weaker and at OP E and F almost disappears.
Fig. 10 Inlet recirculation vortex at vertical cross section for six OP
The main reason for inlet recirculation is not in the fact that pressure increases from inlet to outlet, but in the fact that the pressure field at the inlet is not symetrical. In one way impeller blades also distort the pressure field near the leading edge.
If damage occurs at the hidden side (pressure side) of an impeller and cannot be seen without the use of a special experimental device, the most probable cause is suction recirculation. Classic cavitation damage usually occurs on the visible (suction side) side of the impeller, close to the leading edge. Using CFD analysis, it is possible to locate the recirculation zone quite accurately and to find out the intensity and size of the vortex region (Fig 10 and Fig. 12).
A
F
A
Fig. 11 Absolute (blue) and relative (black) velocity vectors near the leading edge of the impeller – OP A (left), OP F (right)
Inlet recirculation at part load is the only reason for cavitation damage at the pressure side of an impeller. At this operating point, the flow velocity angles in front of the leading edge are not optimal (Fig. 11).
Vortices distribution between the impeller blades is shown in Fig. 12. In comparison with Fig. 9 and Fig. 10, it can be seen that at this cross section, vortices already disappear at OP E. This is important for cavitation characteristics due to recirculation.
Cavitation occurs only if the intensity of the vortex is very high and not always when vortices are present. From the pressure distribution, it can be seen that the vortex intensity is high enough for cavitation only at OP A and OP B.
Fig. 12 Recirculation between impeller blades for six operating points: A, B, C, D, E and F
When the flow rate is less than BEP, inlet velocity distribution from hub to crown is not uniform. Velocity distribution near the hub remains almost the same as near optimal flow rate. For a radius bigger than half of the inlet channel, the velocity drops significantly, and this is the reason for the back flow and vortex occurrences. The axial velocity component distribution at the impeller inlet is presented for three operating points, in Fig. 13.
Usually, for the accurate prediction of pump characteristics, unsteady analysis is required, but in this case, deciding on steady state calculations was appropriate, because we just wanted to improve cavitation characteristics due to inlet recirculation and did not want to obtain accurate absolute values for efficiency or distribution of losses in each part of the pump. The relative comparison between the old and newly designed pumps shows satisfactory results. The numerical predicted efficiency is higher than the experimental one by around 5%, because the mechanical and volumetric losses were not taken into account. Similar results with approximately the same computational mesh quality were presented in reference [10].
A
E
Fig. 13 Impeller inlet axial velocity component distribution for three operating regimes
If flow is unsteady like in this case, the difference between steady state and unsteady numerical analysis can be quite significant, particularly in comparison with energetic characteristics. In our case, when vortex appearance is analysed, the qualitative difference between steady state and unsteady numerical results is not so important. In Fig. 14, the vortices between the impeller blades for steady state and unsteady analysis are presented. For unsteady analysis the length of the time step is Δt=0.0005 s.
Fig. 14 Recirculation between impeller blades zone for OP A – steady state (left), unsteady (right)
The newly designed pump has better Q-H characteristics. The head at part load is higher (Fig. 15) and the influence of inlet recirculation is much smaller than in the case of the old pump geometry (Fig. 16).
Fig. 15 Dimensionless Q-H characteristic of the new pump compared with existing one
Fig. 16 Dimensionless NPSH characteristic of the new pump
A small recirculation zone is visible at operating point D, but real problems with cavitation can start only at operating point A, which is not a very important operating regime. The newly designed pump can operate at a much wider operating regime without cavitation due to recirculation in comparison with the old pump, where harmful recirculation appears much earlier, around Q/Qopt=0.7. The comparison of the efficiency distribution between existing an new pump is presented in Fig. 17.
Fig. 17 Comparison of efficiency between existing and new pump
5. Conclusion
Each impeller design has its specific recirculation characteristics, which can be improved only with new hydraulic design. The present paper shows the basic characteristics of an existing pump and consequences of cavitation damage caused by inlet recirculation. To make improvements in recirculation characteristics, it is necessary to understand the flow conditions at various operating regimes, which are presented in this paper. Before any geometrical changes to the impeller are carried out, the prediction of suction specific speed and checking the blade tip/hub ratio are necessary. The improved pump was obtained after an optimization procedure and the results of the new geometry show better cavitation characteristics and smaller recirculation range at part load operation. The cavitation characteristics of the pump were improved using CFD analysis.
There are also some other possible improvement interventions, which are capable to minimize the recirculation consequences:
• Increasing output capacity,
• bypass between the discharge and the suction side,
• harder and better impeller material.
All above-mentioned corrections can reduce the intensity of the cavitation damage.
At the beginning of the development process, it is necessary to have reliable information about recirculation zones. To diagnose the location and intensity of vortices for different operating regimes, it is important to define some essential cross- sections for vortex presentations. The majority of published papers about inlet recirculation deal with theoretical or experimental methods and do not describe the 3-D flow inside the pump channels.
In this paper, we have shown that using steady-state calculations gives us enough information to improve the geometry. This is important, because in the optimization procedure, we have to analyse a large number of different geometries, and using unsteady calculations, this would be too time-consuming for industrial projects. Usually, we have well-defined criteria in the optimization
methods, which can be presented quantitatively, but in our case, the criteria are more qualitative.
Significant locations of flow conditions and the nature of their presentations are recommended in this paper. All the presented results can help developers to make pumps with low intensity inlet recirculation and consequently with less cavitation. Each step of the CFD procedure is explained in order to obtain reliable and acceptable results for industrial projects.
To achieve better pump characteristics due to inlet recirculation, a change of suction specific speed was induced through a reduction of the impeller eye diameter and some changes to the inlet angles of the impeller blades. The consequences of the implemented changes are noticeable in a wider operating range without inlet recirculation, higher overall efficiency by about 4%, and slightly worse NPSH characteristics near the best efficiency point.
Nomenclature
Do D1 H Hopt N NPSH NPSHa
Impeller hub diameter [m]
Impeller eye diameter [m]
Head [m]
Optimum head [m]
Rotational speed [rpm]
Net positive suction head [m]
Available net positive suction head [m]
NPSHreq NPSHo%
NPSH1%
NPSH3%
Q Qopt SS
Required net positive suction head[m]
Net positive suction head at 0% head drop [m]
Net positive suction head at 1% head drop [m]
Net positive suction head at 3% head drop [m]
Flow rate [m3/s]
Optimal flow rate [m3/s]
Suction specify speed [-]
References
[1] Florjančič, Trouble-shooting Handbook for Centrifugal Pumps, Turboinštitut, Ljubljana, 2008.
[2] Duško Mitruševski, Cavitation due to recirculation – important criteria for design of heavy-duty process pumps, 6th IAHR International Meeting of the Workgroup on Cavitation and Dynamic Problems in Hydraulic Machinery and Systems, Ljubljana, 2015.
[3] Allan R. Budris, How to avoid damage from internal ‘suction recirculation’, WaterWorld Magazin.
[4] Wang, L., Li, F.C., Dong, Y., Cai, W.H., Su, W.T., Study of the Performance characteristics of Centrifugal Pump with Drag- reducting Surfactant Additives, International Journal of Fluid Machinery and Systems, Vol. 4, No. 2, April-June 2011, 223-228.
[5] Song, P., Zhang, Y., Xu, C., Zhou, X., Zhang, J., Numerical Studies on Cavitation Behaviour in Impeller, of Centrifugal Pump with Diffrernt Blade Profiles, International Journal of Fluid Machinery and Systems, Vol. 8, No. 2, April-June 2015, 94-100.
[6] Sedlar, M., Sputa, O., Komarek, M., CFD Analysis of Cavitation Phenomena in Mixed-Flow Pump, International Journal of Fluid Machinery and Systems, Vol. 5, No. 1, January-March 2012, 18-29.
[7] Hatano, S., Kang, D., Kagawa, S., Nohmi, M., Yokota, K., Study of Cavitation Instabilities in Double-Suction Centrifugal Pump, International Journal of Fluid Machinery and Systems, Vol. 7, No. 3, July-September 2014, 94-100.
[8] Kobayashi, K., Chiba, Y., Computational Fluid Dynamics of Cavitating Flow in Mixex Flow Pump with Closed Type Impeller, International Journal of Fluid Machinery and Systems, Vol. 3, No. 2, April-June 2010, 113-121.
[9] SM Pumps - Internal report 23/2012.
[10] Cheah, K.W., Lee, T.S., Winoto, S.H., Unsteady Analysis of Impeller-Volute Interaction in Centrifugal Pump, International Journal of Fluid Machinery and Systems, Vol. 4, No. 3, July-September 2011, 349-359.
[11] Škerlavaj, Aljaž, Titzshkau, M., Pavlin, Rok, Vehar, Franci, Mežnar, Peter, Lipej, Andrej. Cavitation improvement of double suction centrifugal pump HPP Fuhren. Proceedings of the 26th IAHR Symposium on Hydraulic Machinery and Systems, 19-23 August 2012, Beijing, China, (IOP Conference Series, ISSN 1755-1315, vol. 15, part 2, 2012). London: Institute of Physics, 2012.
[12] ANSYS CFX Solver Theory Guide, ANSYS, Inc., Southpointe, 2600 Ansys Drive, Canonsburg, PA, Release 17.0, January 2016.
