Linked simulation tools provide powerful automated design facility
28 February 2016
Improving pump efficiency has become a core task for Grundfos' design engineers; to aid them in this effort, they have been using finite element analysis since the 1980s and computational fluid dynamics since the 1990s.
Originally, FEA and CFD were used for research and troubleshooting, but the company now uses simulation for product design
and has subsequently linked together a chain of simulation tools
into an automated design loop, called PumpIt, that enables engineers to investigate hundreds of designs without manual intervention.
An optimisation routine explores the entire design space and identifies optimal designs based on criteria set by the Grundfos engineers. PumpIt provides a user interface that allows engineers to specify the type of component they want to design, such as an impeller, guide vane or volute. Then PumpIt initiates an automated design optimisation loop that calls simulation tools
, including Ansys CFD software
, to explore the design space.
In a recent example, Grundfos engineers used PumpIt to drive Ansys CFX CFD simulation to optimise the hydraulic surfaces of a new pump design. They increased the hydraulic efficiency of the pump by up to 2 percent while extending the maximum efficiency level over a wider range of flow rates. The team used Ansys Mechanical FEA software
to optimise the pump's structure to ensure fatigue life targets are met, while minimising weight and manufacturing costs.
A key goal of the optimisation process is to improve a pump's peak efficiency. Delivering a relatively flat performance curve with high efficiency levels over as wide a range of flow rates as possible is another. The flow rate of the pump depends on the installation; a relatively flat performance curve can deliver high levels of efficiency for many applications; it also reduces cavitation, thereby increasing pump life.
Another important goal in pump design is to meet component structural requirements with the minimum amount of material. Minimising material usage reduces manufacturing costs and decreases component weight. In turn, lighter pump components allow the use of less expensive bearings
while also reducing noise and vibration.
PumpIt in action
Recently, the Grundfos engineering team used PumpIt to design the multi-stage hydraulics
for a new pump, developing parametric models of all pump components to define hydraulic geometries of the surfaces in contact with the pumped fluid. The design objectives were to maximise hydraulic efficiency and deliver maximum efficiency over the widest possible range of flow rates. The PumpIt tool used design of experiments (DoE) to create a series of iterations that explored the design space for each component. PumpIt then generated the geometry for each design iteration and issued a call to the CFD software
to simulate each iteration.
The initial DoE included about 40 designs. It determined which parameters played the most important roles in the simulation, and it delineated broad ranges of the most promising values for these parameters. The parameters and values were then used as the starting point that automatically generated additional design iterations for CFD evaluation. The optimisation routine evaluated the results of each design iteration and then performed additional iterations based on these results. Each iteration moved the design closer to the efficiency objectives.
To obtain high-fidelity results for an optimised design, engineers used up to 48 cores to perform analyses on a high-performance computing cluster. The cluster had more than 1,000 cores using more than 8 terabytes of random access memory and 50 terabytes of high-speed storage running on the Lustre parallel file system. Using this system, the Grundfos engineers were able to simulate hundreds of design iterations overnight.
The next step was to evaluate the manufacturability of the most promising designs. The Grundfos team considered how easily each design could be produced with several alternative production technologies. To do this, they evaluated the geometric parameters that are required for each production method. In this case, they decided to stamp the components from stainless steel sheet metal. The performance statistics for the best designs were displayed in the PumpIt multi-dimensional solution visualiser that can be configured to display any outcome variable.
After optimising the hydraulic design with CFD, Grundfos engineers performed structural analysis with Ansys Mechanical to ensure that each component would meet fatigue life requirements while keeping cost and weight as low as possible. They mapped the hydraulic pressure determined by CFD simulation onto the finite element analysis to specify loading conditions that accurately match the complex hydraulic pressure distribution. To minimize computational cost and accommodate an acceptable mesh density, cyclic symmetry was applied. Fatigue life was evaluated using the notch stress method. The team also used sub-modelling to evaluate critical areas of the component with a high level of detail without greatly increasing simulation time.
Engineers configured the automated workflow to vary the welding
parameters so they could determine how to optimise the welding
process. The team also used structural simulation to predict the potential for failure in any given lifespan, the only other way of obtaining this information being an expensive and lengthy physical testing programme.
Grundfos engineers performed several sensitivity studies on input variables, including welding
thickness, air gap and flow points, to determine the robustness of the design with respect to fatigue life of components. By providing a statistical distribution of fatigue life for each input variable, simulation made it possible to improve both component quality and reliability. On other structural components, topology optimisation was used at the concept level of the design process to arrive at a design proposal that was then fine-tuned. This replaced time-consuming and costly design iterations, reducing development time and overall cost while improving design performance.
With simulation, Grundfos engineers have significantly improved the hydraulic efficiency of their new pump. Compared with a traditional prototype-based design process, the multiple physics simulation process reduced overall design time for the new pump design by 30 percent and saved approximately $400,000 in physical prototyping
Contact Details and Archive...