Computational fluid dynamics (CFD) is both a science and an art – in the right hands it has the power to optimise vessel performance, increase efficiency, troubleshoot design problems and much more. The CFD team in Wärtsilä Propulsion are blowing out 25 candles on their cake this year, so we’re celebrating with a look back at some lesser-known facts about CFD to see how far it’s come. Settle back with a slice of cake and get ready to learn a thing or two about this game-changing discipline.
Nowadays we think nothing of having computers that are small enough to fit in our pockets, but 25 years ago the computing power available was on nothing like the same scale. “The first workstations we used for CFD had 300 MB RAM, a 4 GB hard drive and a very heavy old-school monitor,” explains Norbert Bulten, Senior Product Performance Manager, who has been working on CFD at Wärtsilä since the field was in its infancy. “In 1997 I was using a single workstation, but now we have a cluster of about 1,600 computers.”
“Thanks to this computing power the simulations we run are now 100–250 times more complex than when we first started. We also run multiple projects simultaneously,” Norbert continues. “Every simulated model is divided into small parts called cells, with smaller cells used wherever more detail is needed. We used to run simulations using 200 thousand cells, but now 20 million cells would be a routine calculation – and up to 50 million wouldn’t be unheard of. Each cell shows the velocity in three directions, the pressure and how that affects the flow, and the more cells we use, the more accurate our simulation will be. The art of CFD is putting the smallest cells where the most is happening to ensure the very best result.”
Back in 1997 Norbert first used CFD to design a waterjet inlet duct. “Once we’d succeeded with a static duct design we moved onto thrusters, waterjets and propellers, improving performance as we went. Now we have moved from component level to system level, considering the vessel as a whole and how it behaves in the water. This holistic approach achieves even greater performance improvements for our customers, helping them to minimise fuel consumption and contributing to the decarbonisation of shipping.”
CFD predicts flow phenomena to analyse how individual components work together, allowing greater optimisation than would be possible individually. “For example, we can dramatically improve the way azimuth thrusters interact with the ship’s hull,” highlights Norbert. “We haven’t improved the thrust performance of our tilted thrusters by changing the angle of the propeller shaft by 8°, but their interaction with the ship is dramatically better. This is a game changer if you look at overall ship efficiency – a small change, made possible by CFD, that delivers up to 20% higher effective thrust.”
Designing the perfect propeller isn’t easy, and it used to be you could only test your design with a model propeller in a tank once you were done. This was time-consuming and expensive, and didn’t allow for multiple iterations to optimise the final design. “Now it is much simpler,” shares Norbert. “We put the propeller geometry into the CFD software and we have a performance curve ready in a couple of hours rather than the days it used to take. The designer can run CFD for 10 or 20 variants, then use this information to squeeze out every last percentage of performance gain.”
CFD predicts flow phenomena to analyse how individual components work together, allowing greater optimisation than would be possible individually.
The same principle goes for whole-ship design. “Our OPTI-Design process looks at every aspect including energy-saving devices and other efficiency-boosting solutions to really push the limits of what’s possible,” says Norbert. “Finding the perfect balance of low noise, comfort and high efficiency would be impossible without CFD.”
Throughout the ship design process there are all kinds of checks and balances to find any issues, but occasionally a problem will slip through. If this only comes to light during the sea trial you need to act fast. “We’ve often stepped in to help find the solution even when it hasn’t been our responsibility,” shares Norbert. “For example, there was one case where the propeller showed unexpected vibration levels. Naturally, the customer came to us first. In this case our years of design experience was not enough to explain the behaviour so we made a detailed CFD simulation of the vessel with propeller, rudder and shaft line brackets. CFD showed it was the inflow causing the problems, due to poorly aligned brackets and not the propeller itself. Together with the customer we were then able to develop a modification to solve the problem.”
We used CFD to design a new waterjet inlet duct, which resulted in a top-performing vessel.
In another case the shipyard didn’t use the right waterjet inlet duct design. “We ran some quick simulations and showed them how, with minimal welding and grinding, they could fix the problem,” recounts Norbert. “It was a subtle difference but it was enough to get the performance over the design threshold and the vessel ready to be handed over to the customer. For the sister vessel we then used CFD to design a new waterjet inlet duct, which resulted in a top-performing vessel that could accelerate from 0–55 knots (approx. 100 km/h) within one minute.”
For those who are used to relying on model-scale testing, it can be a leap to believe in CFD analysis. “CFD is about confidence – for it to be valuable, people need to believe it works,” says Norbert. “That makes our 25-year successful track record very important. Another way we can help people to understand and visualise the results of CFD calculations is with virtual reality (VR), which shows the phenomena that are happening to the vessel. Once we’ve performed the CFD analysis it is just a simple post-processing step for us, but putting on the VR glasses really helps people to see where the flow is going for themselves. They can look around the vessel, releasing particles everywhere and creating their own understanding.”
People can experience first-hand how the water is flowing and affecting their vessel in a way that is clear and easy to understand.
“Virtual dive inspections using VR glasses can also show things which would otherwise be impossible to demonstrate,” adds Norbert. “In real life a diver couldn’t go anywhere near a moving vessel, but in our VR setup we can switch the propellers on and swim through them. People can experience first-hand how the water is flowing and affecting their vessel in a way that is clear and easy to understand.”
Model-scale measurements have been the industry standard for over 100 years and CFD has predominantly been used in the early design stage, before the measurements are carried out. But as computing power has increased and CFD has advanced, it is now often doing a better job than model-scale measurements – and it’s doing it faster and cheaper. “The Reynolds-scale effect in physics shows the difference in results that can be seen between model and full-scale,” explains Norbert. “If a customer wants to test in a model basin we can provide the curve to measure and the model tests will confirm it. But we can also perform full-scale simulations, meaning there is no need to do the model-scale tests at all. The more we learn, the more we discover where the assumptions of model-scale have got it wrong. For example, with a vessel with a duct it’s crystal clear that model basins don’t accurately represent what’s happening. I can show the theory behind it. That’s the nice thing about fluid dynamics – water hasn’t changed in 100 years so the physics remains the same!”
Norbert is clearly passionate about what CFD has brought to the industry and what it can offer in the future. “Back in the 1990s I thought I’d do CFD for maybe five years and then move on – now suddenly it’s 25 years later! During that time we’ve not only managed to clearly demonstrate CFD’s success in the field, we’ve gone further and done it faster than anyone thought possible. I can’t wait to see what the next 25 years will bring!”