Traditional pump design methods have relied on testing physical prototypes. However, creating a digital prototype and using computational fluid dynamics can enable rapid investigation of multiple designs over a range of conditions, writes Elspeth Mosedale

Pumps are the unsung heroes of many aspects of our lives. They enable our water and sewage systems, ventilation systems, cooling in energy plants and many more applications.

Pump design engineers face multiple challenges when creating new, better products. Pumps have to operate well over a range of conditions, meet government efficiency regulations and run reliably over long lifetimes with minimal downtime for repairs.

Traditional design methods have relied heavily on testing physical prototypes. This can be expensive and time consuming, with multiple cycles of design/re-design and testing for each new product. Creating a virtual prototype enables rapid investigation of multiple pump designs over a range of possible operating conditions, which not only speeds up the design process but also reduces the costs associated with physical testing.

Taco Comfort Solutions is a manufacturer of heating and cooling equipment, accessories and systems. All pump manufacturers need to meet more stringent government-legislated efficiency requirements by 2020 and in order to meet these requirements, Taco uses STAR-CCM+ software from Siemens PLM Software in its design process.

STAR-CCM+ enables Taco to create a digital prototype of its design and simulate the three-dimensional flow in the pump using computational fluid dynamics (CFD).

Design and simulation

Taco uses STAR-CCM+ as a ‘virtual lab’ prior to creating and testing physical prototypes. Peter Vandal leads Taco’s CFD design operations as a product engineer. He guides us through a typical design process.

After importing the geometry from 3D design software SolidWorks, all of the meshing, simulation and analysis is then performed in STAR-CCM+, as shown in Figure 1. For each pump design the CFD simulation gives a full, three-dimensional representation of the flow and pressures within the pump housing.

Figure 1. A pump is imported into STAR-CCM+
Figure 1. A pump is imported into STAR-CCM+ where it is meshed and simulated

“We can take readings at any point in the model, like virtual probes,” explains Vandal. “For example, we can pull out the static pressure or total pressure values at the same points as the probes used in the lab testing, so we can directly compare the STAR-CCM+ results with the real-world application.”

This validation of the CFD model gives the team a high level of confidence in the CFD results, along with the predictive insight into performance characteristics such as head, efficiency and BEP flow rate.

Once Vandal and his team have three or four designs which yield acceptable performance in STAR-CCM+, they move on to building and testing their physical prototypes. Even with their ability to rapidly prototype (via 3D-printing) the stereolithography (SLA) impellers, each one costs up to $2500 to build and test. Exploring the performance of numerous digital prototypes in STAR-CCM+ brings significant cost and time savings to Taco, enabling a high probability of success in a single phase of prototype testing, to ensure a short development time window.

Simulating the complete curve

Vandal and his team run steady state simulations on each design, testing multiple flow points to establish the best efficiency point (BEP) flow rate.

In an ideal world, pumps would always run near the BEP but this is not always the case. To prevent field problems and meet customer needs, it is important to look at off-design flow to be sure the pump will be robust over the complete operating curve.

Vandal describes three distinct sections: “On the left hand side of the pump curve you can get an excessive droop, which is not acceptable. The BEP lies in the center and on the right hand side you need to examine the operating conditions at high flow rates.”

Taco performs unsteady simulations in STAR-CCM+, aiming to examine the entire pump curve.

“We initialize the CFD model with a steady state run at BEP flow rate, then change the boundary conditions so the flow changes towards the extreme flow rate.”

Vandal shared one example of a case which showed an underperforming design on the left hand side of the performance curve: “At off design flow rates the pump flow can become highly unsteady. If there is too much drop in pressure on the left hand side of the pump curve there could be problems with operating a parallel pumping setup.

“I was able to run an unsteady simulation in STAR-CCM+ at 10 per cent and 50 per cent of the BEP flow rate (GPM), to quantify the amount of droop between the maximum head point (50 per cent) and near shut-off head (10 per cent).

“I used the same method to simulate alternative designs before we built the SLA impellers for lab testing. The designs which went forward for testing were those which showed good results not only at BEP but also at low flow rate, with a smaller drop in pressure on the left side of the performance curve.”

Streamlining workflows

Using STAR-CCM+ for virtual testing has boosted Taco’s design workflow and reduced the costs associated with physical testing. As Vandal comments: “When I think back a few years to the first pumps I worked on, we would order parts for testing, they would be bad and we would have to do the whole design, ordering, and testing process again.

“Now that we are using STAR-CCM+ we are cutting out this iterative ‘start over’ process and speeding up the complete design cycle”.

He notes how the use of the software has improved over recent years. “When we first started using STAR-CCM+ we did not have much lab test data to correlate with the CFD results, so we did not know the best model settings (such as roughness, or leakage flows) to use. We now have a library of lab test data which validates the STAR-CCM+ results, and know the best model settings to use to get accurate results.”

And Taco is determined to streamline its design process further still. “We are always looking for ways to use the software faster,” says Vandal. “Any time we spend a couple of hours on something we will say ‘how can we do this in 50 minutes?’ We want to work smarter with the software, not harder.”

Peter and his team have worked closely with a dedicated support engineer to streamline their simulations by developing a quick multi-design testing process, using STAR-CCM+’s part-swapping capability. Once the model has been set up and run in STAR-CCM+, part-swapping enables Vandal to replace just the impeller or volute with an alternative design, while the CFD model settings and boundary conditions remain unchanged, as in Figure 2.

Figure 2. Part swapping allows
Figure 2. Part swapping allows quick change of the impeller

STAR-CCM+ then rapidly regenerates the mesh, and computes a new solution. “We use a standardized naming convention in SolidWorks, so when I want to change one part in the STAR-CCM+ model I literally hit the replace part button and run a new simulation,” says Vandal. “We love working this way, as we can test multiple cases one after the other using the same mesh and simulation settings. This not only makes it quick to run multiple simulations, but also gives us confidence that the results will be consistent and repeatable.”

Taco is aiming to build on the knowledge garnered over the last two years. Chief technology officer Greg Case says: “We are relying on our strategic use of CFD simulation to help us beat our competition to market with better performing pumps, and increase market share”.

Vandal addes: “Essentially, we want to eliminate the feedback loop between lab testing and the design stage, so that each pump design has as few lab tests as possible. We want to ‘explore digitally and confirm physically’ rather than the other way around.”

Elspeth Mosedale is Technical marketing Engineer with Siemens PLM Software. www.plm.automation.siemens.com