Digitalization
Recreating Reality
Higher efficiency through computational fluid dynamics
Dr. Roland Jester-Zürker, Head of Fluid Mechanics at Voith Hydro, explains how Voith experts are utilizing computational fluid dynamics to better understand and improve the operational efficiency of hydropower turbines and generators.
Successful operation of new hydropower plants depends on accurately predicting how machinery and fluids will interact before the plant is built. Dr. Roland Jester-Zürker, Head of Fluid Mechanics at Voith Hydro, uses computational fluid dynamics (CFD) to research and plan optimal plant design and maximize operational performance.
Can you explain what CFD is and how it is used at Voith?
Essentially, a group of skilled engineers uses a combination of powerful computer hardware and software to predict and analyze fluid flow inside our machinery. CFD enables us to reliably investigate and evaluate complex physical relationships very early in the design process. It helps us to achieve high levels of performance, reliability and efficiency for our products. We can virtually test things such as the effects of geometric variations, performance limits during operation or the stresses placed on components due to the pressure of fluid flow. Being able to carry out detailed, accurate analyses on the computer has also led to a major reduction in physical model testing, which reduces design lead time and makes the process faster and more cost-effective for our customers.
Are physical model tests still important?
Absolutely. Validating numerical results of challenging flow regimes like off-design condition is crucial to achieve the required prediction accuracy. We work closely with our colleagues in the hydraulic lab at the Brunnenmühle research and development center in Heidenheim. Close collaboration between the teams is a big advantage – it enhances prediction accuracy, which speeds up the planning process and improves product performance.
Is all the expertise in-house?
To accelerate development, we look to the future with external partners such as the Institute of Fluid Mechanics and Hydraulic Machinery at the University of Stuttgart. With their support, we recently completed one of the largest computational investigations ever undertaken on a Francis turbine. It focused on part-load operation and its effect on water flow through the draft tube. Our main objective was to derive an improved CFD process to increase prediction accuracy, which is a challenging task in the case of part-load operation. It was an intensive effort involving extremely refined grid models with over 300 million nodes, using the computational power of 2,000 processors working in parallel.
Do you also investigate generators?
Generators are just as important, and we look at them intensively. Because they are electromagnetic machines and not designed with fluid mechanics in mind, fluid flow investigation is more complicated, but we’ve built a process that simulates air flow in an entire generator. This allows us to look at how to cool the machines properly, minimize energy loss and optimize heat transfer. It also enables us to improve overall machine behavior, extend its lifetime and minimize overall costs.
How is CFD evolving and what changes are on the horizon?
One of the most important assumptions in CFD is turbulence modeling. To date, there is no universal turbulence model, but we expect that scale-resolving approaches will improve prediction accuracy. Increasing the level of detail and including multi-physics aspects will also be a task for the near future. On a more general level, computing power has improved considerably since we started using CFD around 30 years ago. This has given us faster and more accurate results that have helped reduce design-cycle times. In the coming decades, we expect this accuracy and speed to improve further still.
Turbine flow
Predicting how machinery and fluids interact ensures smooth operation.
Virtual testing
Enables performance, reliability and efficiency to be improved.
Water flow analysis
The analysis, carried out in partnership with the University of Stuttgart, investigated how part-load turbine operation affects water flow through the draft tube at a hydropower plant.
Over 2,000 processors and highly refined grid models with over 300 million nodes were used.
The result:
An improved CFD process to increase flow prediction accuracy.