Turbomachines, and centrifugal pumps in particular, are of major importance to contemporary society. Everyone deals with them directly or indirectly in one's daily life. Examples are pumps for boiler units, drinking water supply, energy generating plants and process industry. Centrifugal pumps have been in use for many years. Therefore, the innocent reader could assume that further research is no longer required in this field, under the slogan “everything has been investigated and solved before”. However, this is not correct. Developments are still continuing rapidly, for example to smaller dimensions, higher speeds and efficiencies and new areas of application.
These developments are stimulated by improved manufacturing and computer simulation techniques. A technique like CAD-CAM makes it possible to manufacture a new design quickly and economically. By simulating the flow in centrifugal pumps, the designer can optimize these using computers instead of through actual tests. This approach is widely accepted in aircraft design (“computer wind tunnel”). In contrast, new centrifugal pumps are usually developed as modifications to existing designs by a trial-and-error method, which is time-consuming and expensive. The new approach based on flow simulations is especially suitable for designing engineered centrifugal pumps that must meet specific customer specifications. Then it is frequently necessary to deviate from existing types of centrifugal pumps: new concepts need to be developed.
The objective of the research at the
Turbomachinery Laboratory of the Department of Mechanical Engineering at the
The approach of the research consist of a combination of
Difficulties that arise in this approach are (1) complex equations, (2) geometries and (3) unsteady flows.
The approach is based on the fundamental conservation laws of physics. Based on an asymptotic analysis the equations for the main flow can be simplified. This main flow is inviscid, irrotational and incompressible. This means that the main flow is an unsteady potential flow. Viscous effects are only important in boundary layers along walls and in wakes behind the impeller blades. These assumptions are valid around the best efficiency point.
The geometries are very complex, due to their three-dimensional doubly-curved shape of the impeller blades (and sometimes diffuser blades). The volute also has a complicated shape.
The rotation of the impeller with respect to the volute causes the flow to be unsteady, especially at off-design conditions of the pump. Usually the flow in both parts is considered separately. This simplifies the problem, but it is becoming clear that a correct description of the interaction between impeller and volute is very important.
Current research projects that are described here deal with:
For measuring velocities and pressures in impeller channels and volutes a test-rig is available in the Turbomachinery Laboratory. With this test-rig it is possible to directly measure the relative velocity in the impeller channels, since the equipment rotates with the impeller. Velocities are measured using “Laser Doppler Velocimetry” (LDV). Using two laser beams the two components of the velocity are measured that are in the plane perpendicular to the laser beams. The relative flow field in a number of two-dimensional impellers has been determined. These showed that the flow field is accurately described by the potential flow model. Currently, the test-rig is being adapted to three-dimensional measurements. This is especially important for mixed-flow impellers. To this end, an impeller has been manufactured from plexi-glass. The principle of the modified test-rig is that by measuring two components of the velocity in two different planes, the three-dimensional velocity field is obtained.
For reasons of efficiency and computing times a completely new system has been developed for simulating numerically the flow in impeller-volute configurations. The system is called COMPASS, an acronym for Centrifugal Or Mixed-flow Pump Analysis System. It is based on the finite-element method with substructuring and implicit Kutta conditions as special features. With this system unsteady potential flows in impeller and volute are computed. Thus the interaction between impeller and volute is properly accounted for. To this end the computational domain is divided into blocks with a cubical topology. Blocks in impeller and volute are separated by a conical surface along which the impeller blocks slide. In this way the rotation of the impeller is simulated.
The system is easy to use, since the geometry of impeller and volute are described parametrically. The mesh generation is also automated. Finally, the post-processing of velocity and pressure field is integrated into the system. An example of such a visualization of the pressure on the impeller blades is shown below.
An important quantity to the designer of a pump is its efficiency under various operating conditions. In order to predict the efficiency the viscous losses need to be quantified. An inviscid model can not directly give these. Yet it is possible to estimate the various sources of loss, based on the inviscid core velocities in the pump. To this end, viscous loss models were adapted from the literature.
Here some results are shown of the
simulations in an industrial mixed-flow pump of Flowserve in
Note that to obtain these performance curves, unsteady computations were performed, and the results were subsequently averaged over time.
An important flow phenomenon in pumps is cavitation. Herewith the inception of bubbles is meant, that arise when the local pressure drops below the vapour pressure of the liquid being pumped. This drop in static pressure is caused by the acceleration of the fluid around the impeller nose. Downstream in the impeller these bubbles cavitation implode as a result of the higher pressure. This can cause damage to the blades, the so-called cavitation erosion, which reduces their life expectancy. Furthermore, these implosions increase the noise level. Cavitation performance is usually expressed in an NPSH-Q graph. NPSH is Net Positive Suction Head, the stagnation pressure at the inlet necessary such that no cavitation occurs and Q is the flow rate. When the actual inlet pressure and the NPSH-characteristics are known, the operating range and the life expectation can be predicted. A centrifugal pump should be designed so that over the complete operating range the NPSH value does not exceed the available inlet stagnation pressure.
Current research deals with the prediction of cavitation inception on the basis of completely three-dimensional potential flow computations with COMPASS. This is the correct approach, contrary to commonly used quasi three-dimensional models that require empirical coefficients. This research is funded by Flowserve Hengelo and NOVEM. The figure below shows the result of such a computation for a mixed-flow impeller. In addition measured NPSH-inception values are shown. Both visual and acoustic measurements were made. It is clear that over a wide range the cavitation inception characteristics are correctly predicted.
This opened the way for parametric optimization studies in which the shape of the nose profile is modified. An example is shown below, demonstrating the advantageous effect of an asymmetrical profile.
Currently, the cavitation model is being extended in order to compute, for inlet pressures below inception values, the shape of the cavitation sheet.
During the past few years major advances have been made in predicting the hydraulic performance of pumps. Especially complete impeller-volute configurations can now be analysed, contrary to other research groups that only study components of pumps with their viscous computations. Therefore industry shows increasing interest in the hydraulic analysis system COMPASS. This system will be further improved by refining the hydraulic models. In first instance the models for leakage flow, disc friction, mixing losses and three-dimensional boundary layers will be considered. Additional experimental work will be done on the velocity and pressure fields in mixed-flow pumps. COMPASS will also be used in optimization studies of efficiency, cavitation etc.
A permanent factor in this research is the close cooperation with industry. The importance of this cooperation lies in the coordination of the direction of the research and in the transfer of the developed technological knowledge to companies.
Currently, advanced design methods are being developed that are based on inverse-design and optimization methods.
In inverse-design methods, the duty of the impeller is specified in terms of flowrate, angular velocity, head, some “blade loading” and meridional geometry. From these, the blade geometry is computed; the blade curvature is thus the result.
The inverse-design methods and the optimisation methods are described in detail in the Ph.D. thesis of Remko Westra.
CFD for pump design: a tutorial
Tutorial
presented at the 5th International Symposium on Pumping Machinery, 2005 ASME
Fluids Engineering Division Summer Meeting and Exhibition, 19-23 June,
M.Sc Thesis of Robin van de Vondervoort, April 2015
Industrial cooperation is done via TurboValley.