Computational Power Leads to Improved Ship Designs

In the last 15 years, computational fluid dynamics (CFD) software has made big leaps. What was once merely considered a ‘numerical towing tank’ is now an integral part of ship design. Many shipyards and naval architecture companies have an in-house CFD department, while others – including ship owners – contract an external company for this very specific work.

One such company is Van Oossanen Fluid Dynamics, based in Wageningen, the Netherlands. Maritime Holland spoke to Niels Moerke, their managing director.

Moerke says: “Our company was established by Pieter van Oossanen, who saw the potential of CFD at a very early stage. After working for 20 years at the Marin research institute, which is also now a provider of CFD services, he set up his own company in 1992. In the early 2000s, we tried a variety of different software packages and settled on Fine/Marine, a code developed at the University of Nantes. Van Oossanen was actually the first commercial user of Fine/Marine and we are still a beta-tester and co-developer of the software. Van Oossanen does a lot of naval architecture for yachts, and because these yachts are often one-offs, it has provided an enormous amount of full-scale validation of our CFD work in the past ten years.”

Code language

In CFD software, there are various approaches: some are based on a potential-flow code, which calculates the pressure on the hull, and derives the wave profile and pressure resistance from that. The frictional resistance is then added on top by calculation. Other CFD packages use a Reynolds Averaged Navier Stokes Equations (RANSE) solver, which also calculates the viscous effects in the flow, taking into account the friction due to the viscosity of the water.

Niels Moerke explains: “We have found that the viscous resistance on the hull is anything but uniform, and there is even an interaction between viscous effects and pressure effects. When we tried to optimise a bulbous bow with a potential flow code, and later checked this with a more accurate RANSE code, we found out that we would have done it differently if we had used the RANSE code from the start. That’s why we always work on full scale and with a RANSE solver, even though it takes more computing time and is therefore more costly.”

The flatwater resistance of ships has, for many years, been accurately predicted with CFD software, but Van Oossanen Fluid Dynamics increasingly works on more complex questions. Many ships are, nowadays, not optimised for a single draught and speed, but for a range of different draughts, speeds and even sea states. Niels Moerke calls this broad spectrum optimisation. This corresponds a lot better with the operational profile of the vessel than the seatrial condition alone, which used to be the single reference point. A bulbous bow for example can be optimised for a single speed, but you can also try to find the best compromise. A vessel can be sailing 30 per cent of the time at 17 knots in ballast condition and 70 per cent of the time at 15 knots in loaded condition. A bulb optimised for either one of these conditions will not save as much fuel on a yearly basis as one which takes both conditions into account, weighed to their share in annual fuel consumption.

Seakeeping analyses

A typical seakeeping study involves the comparison of two or three different versions of a hull shape in an identical sea state. The CFD software creates a ‘laboratory’ condition, showing exactly the effect on ship motions and total resistance for each of the design options. When the effects are known, it becomes possible to optimise in order to achieve a certain outcome. The results of a seakeeping study are usually presented in a video, showing the motions of each ship, as well as graphs representing the desired output, such as vertical accelerations or total resistance (composed of flatwater resistance and added resistance due to ship motions). But is such a seakeeping analysis reliable?

Moerke confirms: “We have validated this method by making predictions for a model in the towing tank, and then compared the motions of the computer model with those of the scale model. The outcome was exactly the same, except for some disturbance in the towed model due to vibrations in the carriage.”

Does this make the towing tank unnecessary? For a simple flatwater resistance run, perhaps yes, but for a statistical analysis over a wide range of speeds, wave angles and sea states, the towing tank is still the most cost-effective solution. Seakeeping studies in CFD are, for the moment, mostly limited to regular waves. To see what a ship does in an irregular sea, with waves coming simultaneously from different angles and the ship moving in all six degrees of freedom (pitching, heaving, swaying, surging, yawing and rolling), it still requires too much computing power to be done in a realistic timeframe. Towing tanks are also the preferred option for manoeuvring trials, such as zig-zag tests and turning circles. For naval customers, you can even do manoeuvring trials in waves. CFD is not there yet, although validations on zig-zag tests are currently being performed by Van Oossanen.

Early insight

The advantage of CFD, compared to model tests, is that it is cost-effective, has a short lead-time, and can often be done earlier in the design cycle, allowing for alterations to be done rather than just checking that nothing unexpected happens. Another advantage is that it gives more insight into what happens below the waterline. While a paint-stripe test in a towing tank allows us to visualise the waterlines on the hull, a CFD analysis also shows what happens at a distance to the hull, and shows exactly the places on the hull where the frictional and pressure resistance are greatest. Identifying these spots allows the opportunity to do something about them, and often a large accumulated saving can be obtained with what seems like minor tweaks here and there. This can go from bilge keel or shaft strut alignment, to the exact shape of the bulb, skeg or the bowthruster scallops. Model tests in a towing tank are less suitable for this kind of work because the water is basically too viscous at model scale. The effect of this is that the boundary layer (the layer of water dragged along with the ship) is relatively thicker on model scale than in full scale, which may lead to sub-optimal results.

Aerodynamics

Not all CFD work at Van Oossanen Fluid Dynamics is focused beneath the waterline. For clients in the superyacht industry, sometimes wind studies are done. These are used to optimise the shape of the superstructure (in agreement, of course, with the designer) to create pleasant aft decks or sundecks, free of excessive draught or turbulence. It can sometimes be as easy as adding a strategically placed wind deflector. For naval customers, the company has conducted studies to determine the flow of the exhaust gases, in order to avoid interference with helicopter landings on a ship. It is required that the helicopter blades rotate in air with a rather uniform temperature, and exhaust gases can ruin this.

Ongoing research

One very exciting project currently being undertaken is CFD work for team RISE, of the University of Delft. This student team is building what they hope to become the world’s fastest rowing boat, sailing entirely on two hydrofoils. Van Oossanen is no stranger to hydrofoils as they also carry out the CFD work for their sister company, Hull Vane, producers of the hydrofoil used to reduce resistance and ship motions on displacement vessels. Computational Fluid Dynamics has played an essential role in the development of this maritime innovation, and it is certain more innovations will come from the insight provided by CFD. The Van Oossanen group spends about 30 per cent of the working hours in unpaid R&D work every year, often based on suggestions from within the team.

Aside from the significant investment in both hardware and software, the toughest part in the provision of CFD is finding the right people. A good CFD specialist needs a solid combination of programming skills and thorough hydrodynamic knowledge, and even the right person needs extensive training. That is why it is unsurprising that CFD work is often outsourced to specialist companies.

Tom Oomkens

This article was previously published in Maritime Holland edition #3 – 2016.