CFX streamlines pumps design
CFD Simulations for Pump Design
Optimizing the design of pumps requires that their operation be understood in some detail. Experimental methods and past experience are undoubtedly important, but the most effective way to study pump performance is through Computational Fluid Dynamics(CFD).
CFD simulation of a pump provides a complete picture of its operation, and allows engineers to identify areas where there is recirculation, flow detachment or where cavitation will occur. More importantly, the engineer can establish the causes of the behaviour, and with this knowledge, can reliably and confidently direct design improvements or operating strategies. In this way, the design can be optimised to give reduced energy consumption, lower head loss, minimisation of blade erosion, prolonged component life and better flexibility of the system, before the prototype is even built.CFD relies on the numerical solution of the Navier-Stokes equations, the set of partial differential equations (PDEs) that describe fluid flow. In a CFD model, the region of interest, a pump casing for example, is subdivided into a large number of cells which form the grid or mesh. In each of these cells, of which there may typically be 300,000, the PDEs can be rewritten as algebraic equations that relate the velocity, pressure, temperature, etc. in that cell to those in all of its immediate neighbours. The resulting set of equations can then be solved iteratively, yielding a complete description of the flow throughout the domain. Powerful graphical post-processors then display the results in an easily understandable way. The basic CFD tools that are needed to simulate pumps have been available in general-purpose commercial CFD software for many years. However, ensuring true success with CFD requires that it must be integrated within the design environment, and until recently, lack of integration has been a major barrier to the widespread uptake of the technology. This article discusses the integration of AEA Technology’s CFX-TASCflow software into the design cycle for industrial pump manufacturers and illustrates the progress which has been achieved in the streamlining of the design process.
Trends in CFD integration
AEA Technology has been addressing the issue of CFD integration for several years, and has developed specific turbomachinery-oriented software tools to facilitate problem definition for its CFX-TASCflow CFD solver. For bladed components, AEA Technology’s interactive CAD (computer aided design) software tool, CFX-BladeGen, provides the essential link between blade design, advanced fluid analysis and manufacturing. Incorporating decades of extensive turbomachinery design and analysis expertise into a user-friendly graphical environment, CFX-BladeGen is used to create 3-D blade geometries (see Figure 1), enabling the engineer to design new blades or re-design existing types to achieve new design goals. CFX-BladeGen can handle the design of a variety of rotating and stationary bladed components, including axial and radial blades, for applications such as: inducers, pumps, compressors, turbines, expanders, turbochargers, fans and blowers.
This October, AEA Technology has also released CFX-VoluteGen, which complements CFX-BladeGen by providing a cost-effective, application-specific CAD environment for parametric design of volutes and scrolls. With CFX-VoluteGen, configurations for volutes and scrolls are easily selected and parametrically adjusted by the designer. Area control methodology is applied to interactively design the components with design verification accommodated in a 3-D viewing environment. Figure 2 shows a typical graphical user interface of the software. The user can select from a number of pre-configured cross-sections and specifies the dimensions, number of sections, number of discharges (one or more), and the sizing method. CFX-VoluteGen then generates the complete volute, greatly reducing the tedious CAD work required and making drawing production significantly simpler.
In CFX-VoluteGen, an optional semi-automatic meshing package allows preliminary designs to be quickly and easily prepared for analysis in CFX-TASCflow. The output from CFX-BladeGen meanwhile can be imported into CFX-Turbogrid, which also performs mesh generation. Combined, CFX-BladeGen and CFX-VoluteGen allow blade, volute and scroll geometries to be created and modified with the minimum of effort, and then to be exported for rapid CFD analysis. The design engineer can therefore include CFD as an integral part of the design process and model the fluid dynamic behaviour of turbomachinery long before a working model is created. Once an acceptable design is achieved, it can then be exported directly to the CAD environment for design and manufacture, thereby ensuring that perfectly consistent data are used at all stages, from design and analysis through to production.
Having performed the CFD analysis of the flow, it is vital that as much information as possible be extracted from the results. Ideally, the post-processor should be able to provide conventional three-dimensional views of the flow in the device, as well as on blade-to-blade and meridional planes (Figure 3). The details of the flow are invaluable in understanding overall performance, but as global performance measures allow comparison of different designs, it is also essential that quantitative results such as efficiencies, losses, flow rates and head rise can be calculated.
CFX in action - Predicting pump performance
Once the pump geometry has been specified and a mesh has been created, the flow equations need to be solved. The nonlinearity of the momentum conservation equations makes this difficult, especially for the highly swirling flows encountered in rotating machinery. Where the majority of commercial CFD codes, which use ‘segregated’ solvers, would find it difficult and computationally expensive to obtain the solution, CFX-TASCflow employs an extremely efficient coupled solver approach. This treats the pressure and all components of the velocity field as a single system, and solves them simultaneously. The result is much greater stability in the solution and significantly shorter run times, particularly on the large meshes which are required in real engineering applications. The practical benefit of this is that more design variations can be investigated in a given time.
The most challenging aspect of simulating pumps is representing the interactions between the rotor and stator. CFX-TASCflow includes three different multiple frame of reference algorithms for this: frozen rotor, stage, and transient interaction. The frozen rotor method employs a quasi-steady algorithm where the rotor and stator are modelled at a fixed (frozen) position relative to each other. Rotational terms are included in the moving frames, but transient effects are neglected. This provides an efficient method for the calculation of interactions between impellers and casings (volutes), and is also a viable option for compact machines with small distances between rotor and stator. However, several calculations for different frozen positions are required to obtain a true representation of the pump performance. The stage method also employs a quasi-steady algorithm where the discrete fluxes passed from the rotor to stator grid and vice versa are circumferentially averaged. This is most applicable to axial machines, compressors and pumps with return or outlet diffusers.
Transient rotor-stator interaction is the most realistic approach, but requires the highest computational effort. The true transient behaviour between rotors and stators is simulated using a sliding grid technique, in which the relative positions of the components are re-calculated at each time step. This is essential for cases where unsteady-state effects are important, for instance when investigating instabilities, transient loadings or transient heat transfer to blades.
CFX-TASCflow has an extensive track record in the turbomachinery industry, and is used by over a thousand engineers world-wide. One such user is the Danish Grundfos Group, which manufactures 8 million pumps each year, and has 50% of the global market for circulation pumps. Recently, in order to meet continued customer demand for high performance and to retain its leading position in the market, Grundfos decided to extend its range of pumps by building a new, larger pump which could operate at higher flow rates. The company had no experience of designing pumps within the proposed parameters, so engineers decided to use CFX-TASCflow as an integral means of transferring their knowledge and experience of smaller pumps to this larger application. Further motivation to use CFD was provided by the requirement that, to ensure the final production pump would be competitively priced, the number of design prototypes would be limited to two, compared to the usual six for completely new designs.
Grundfos’ engineers took the design of an existing pump and scaled it up to the new size and new flow range. Six different designs were created and, as the product was a multi-stage centrifugal pump, the impeller and the crossover had to be modelled. CFD simulations were performed on each design so that the flow curve, the efficiency curve and the power consumption could be verified. Once these calculations had been completed, two of the designs were chosen and further amendments were made. Two prototypes were then manufactured and tested, the results confirming the predicted performance curves. Finally, one of the two designs was chosen and further CFD simulations were run to determine performance under different conditions.
In Scotland, Weir Pumps have also implemented CFX-TASCflow in their design process. A typical application for them included confirmation of the final design of a mixed-flow bowl pump with an open impeller and vaned diffuser. Weir Pumps used CFX-TurboGrid to generate the meshes and the frozen rotor technique to model the interaction between the impeller and the diffuser.
In order to test the sensitivity of the machine to tip clearance, Weir performed calculations for a variety of tip gaps. Figures 4 and 5 show CFD predictions of the flow through the pump impeller. Results for the analysis of the whole machine were compared with experiment, both for the gross parameters of head versus flow and for more detailed measurements of local axial velocity from LDA data. The calculations showed that separation off the suction side of the diffuser blade was responsible for the bulk of the losses in the machine at low flow rates.
Ideally the pump should have a constantly rising head flow curve across the range of operation. In the original design the head flow curve was seen to be unstable at reduced flow, leading to problems when operating machines in parallel or while bringing the machine up to load. By looking at the individual component characteristics it was possible to identify a steeply rising loss characteristic for the diffuser in a region where the impeller characteristic was relatively flat.
CFX-TASCflow confirmed that the final design was improved by reducing the impeller best efficiency flow to give a steeper gradient in the impeller performance curve and a reduced power requirement across the affected range. This work clearly showed the importance of component interaction in the performance of rotating machinery, and demonstrated that CFD can be relied upon to predict this accurately.
To find out more about the CFX software, why not join us at one of AEA Technology’s European seminars. This winter we will hold seminars on 2nd February 2000 in Stockholm and 8th February 2000 in Amsterdam. Invited speakers from the industry will discuss the implementation of CFX in their design environment, and rotating machinery technologists from AEA Technology will demonstrate the ease of use of the CFX software. For more information, please contact Nathalie Hamill on: +44 1235 436541, e-mail nathalie.hamill@aeat.com.
