Optimizing Turbines with CFD

About Designcraft

Trond Gudmundsen is a technical engineer with a background in blacksmithing and construction welding in Oslo, Norway. He started his consultancy office Designcraft in 2010 and works with his customers to design metal construction and general product development. Currently, he is working on a project for DeepRiver AS, a company that is active in the research and development of power plants that produce electricity from slow/fast moving river, ocean, and tidal currents.

In this project, Trond is implementing design solutions to improve the efficiency of the turbines using fluid mechanics simulations results obtained from SimScale.

Learn more
Trond Gudmundsen

SimScale Customer support is positively surprising, because usually the customer pays a lot of money but needs to wait days or even weeks for support reply. SimScale service has a very clear structure, that one account manager and one support engineer are my only points of contact to listen to my request, and positively keep me updated.

Trond Gudmundsen



Virtual Testing of Turbine Designs to Complement Physical Prototyping

DeepRiver AS needed to test the energy efficiency of its turbine design called the Pilot Turbine to determine which design solutions would improve its product. The physical prototype was based on an old water wheel concept, which was full-scale tested in Klapeida, Lithuania.

Although paper calculations gave a promising efficiency and the physical test also showed good results, it was still important to deeper understand the behavior of water through the turbine construction; for example, how the pressure is distributed on the turbine blades.

The project then moved to a critical stage, where modifications and improvements to the design had to be implemented in the technical drawings for the construction of the final turbine version.


CFD Simulations to Optimize Water Turbine Designs

For the validations, Trond considered investing in on‐site testing with an on‐premise CFD simulation software license, however, the costs proved to be too high. While searching for a more optimal solution, Trond learned about SimScale, a powerful, cost‐efficient, numerical analysis tool that could be used to verify each of the design changes before the company started full production.

With SimScale, Trond was able to set up and run CFD simulations to better understand the local conditions and analyze parameters, including flow rate, turbulence, frequency, and rotations effect on flow.

Firstly, Trond ran 2 simulations on the Pilot Turbine. Physical tests and simulations matched each other’s conclusion, and both made it obvious that the turbine construction should be more streamlined.

Later, incremental changes to the design were made. Each version was tested using simulation until the final version was ready. For the simulations, Snappy hex mesh with a base mesh box was used to create the “water” body and regional refinements around the object to be studied.

The meshing took about 35 minutes with 16 cores, and each simulation took approximately 1,5 hour on 32 cores.

Incompressible fluid flow with “k‐omega SST” (k–ω) turbulent model was chosen. The whole construction to be analyzed was about 3 x 3 x 6 meters, and a quarter of the construction model was simulated to save computing time.


Same Turbine Effectiveness with 30% Size Reduction

The simulation results helped Trond to better understand how to capture pressure to obtain more energy from the water flow, and how to avoid unnecessary turbulence and drag. The simulations have also contributed to increasing the effectiveness of the turbine design.

By far, although the optimized turbine construction is reduced by 30% in size, has almost half the number of turbine blades and a smaller diameter of the turbine, the new design is still as effective as its predecessor.

With on-premises simulation software, it took 1 hour to prepare the geometry, 1 hour to set up the simulation, and up to 49 computing hours to complete one run. With SimScale, on the other hand, the processing time was dramatically decreased to 1,5 hours in real time on 32 cores (48 hour computing time on 32 cores).

Furthermore, with the ability to run multiple simulations at the same time, Trond was able to use up to 150 computing hours in just 2 working days. The progress bar on the platform, as well as e‐mail notifications also gave him a clear overview of ongoing and finished work, and let Trond better control the pace of client projects.

Right now, our senior CFD support engineer Pawel is collaborating with Trond, whenever he needs technical assistance. We are looking forward to seeing the finished product of this new turbine design produced by DeepRiver AS.