»No easy problems left to solve«

[/B]Dejan Radosavljevic: As with the general[ds_preview] industry trend towards designing more efficient ships everyone is focusing on utilizing latest technology advances to maximize vessel performance while minimizing fuel consumption, and implicitly reducing vessel impact on the environment. Part of this trend, and arguably a trend of its own, is the move towards simulations at full scale and under simulated conditions that are closer to reality, such as ship performance in waves. Out of twelve presentations within the marine track, if we exclude the two general ones dealing with best practices in simulation, nine out of the remaining ten included simulations at full scale and five were looking at some aspects of performance in a sea state. This move towards more complex simulations (which brings them closer to reality) included one utilizing the fully coupled FSI (Fluid Structure Interaction) solver within STAR-CCM+ to simultaneously look at propulsive performance and stresses within blades of a large propeller.

And looking at Siemens’ vision, where the fully automated design exploration becomes the standard method to achieve best designs, it was good to already see a number of presentations adopting this approach showing companies are recognizing the value this can bring to simulation.

Where is the CFD market heading, is there a next »big thing« on the horizon?

Radosavljevic: A lot of research is continuing in the background to find better, more accurate, more efficient numerical methods that would speed up the moves towards the »full scale under real operating conditions« goal. Everyone is hoping for a method that will significantly speed up simulations, but I am not aware of anything on the horizon showing promise to be successfully commercialized in the near term.

For me personally, the next »big thing« would be for the industry as a whole, including ship owners and the IMO, to accept that today you can achieve much better vessel designs if you focus on full scale simulations, and apply automated design exploration. For the same cost of running one set of tests on a single design in a towing tank at model scale it would be possible to run hundreds or even thousands of design variants, at full scale, guaranteeing to end up with a better design.

What does the integration of CD-adapco into Siemens mean for the industry?

Radosavljevic: An uncomfortable truth about modern engineering is that there really are no easy problems left to solve. In order to meet the demands of the industry, it is no longer good enough to do »a bit of CFD« or »some stress analysis«. Complex industrial problems require solutions that span a multitude of physical phenomena, which often can only be solved using simulation techniques that cross several engineering disciplines.

Siemens’ acquisition of CD-adapco means that tools such as STAR-CCM+ become part of a much wider ecosystem of predictive engineering analytics tools (the Simcenter Portfolio), based around the vision of producing a »Digital Twin«, a fully functional digital prototype that simulates all aspects of the real life physical performance or a design.

To improve the design software, you need validation data and advocate an open approach …

Radosavljevic: Unfortunately there are several facets to this issue, which is why it is proving difficult to crack. For data to be usable for software validation it has to be detailed, fully compatible, and of good quality. We need to first start with detailed geometry of ship hull and appendages, including the propeller. This is already a sticking point as most of the shipyards or specialist designers view such information as their intellectual property and are not willing to share it. Then there is the issue of acquiring high-quality data under acceptable environmental conditions. This usually means contracting highly reputable companies to instrument vessels and spend (potentially considerable) time on board trying to achieve good measurements. Not to mention the costs.

And while these measurements happen regularly, the party that invested in them has done so for their own reasons. So, finding all the above-mentioned elements and parties coming together with common consent has proven to be close to impossible, so far. Some data was generated through closed Joint Industry Projects where partners would share the cost in order to gain access to data, but the industry as a whole is finding it difficult to access good quality data for validation at full scale. Thankfully, this has finally changed a couple of months ago, when Lloyd’s Register held a full scale CFD validation workshop. Perhaps this will help change the minds of other stakeholders as the results of the LR workshop clearly showed maturity of CFD.

Last year you said, shipowners were still too conservative to accept full scale CFD. Has that changed already?

Radosavljevic: With over 150 years of tradition in how you design ships any change will always be slow. And it is painfully so. Shipyards and designers have already recognised benefits of CFD and want to move much more towards modern simulation technology, but claim shipowners still demand, and only trust, towing tank tests. Owners tend to justify this by the lack of confidence in CFD. A couple of other things could also contribute to those views such as a certain level of misunderstanding of fundamental physics, lack of communication and perception of risk.

Yet accuracy of CFD is well understood and when CFD is applied properly it can match accuracy of measurements (which also, by the way, could yield completely wrong data if not done properly). Automotive and aerospace industries have been relying on CFD simulations for much longer than marine and today could not function without it. It is fully trusted and embedded within design process in those industries, yet, the same fundamental physics but different industry, the level of trust among those impacting the process is low.

The recent workshop hosted by Lloyd’s Register offered data for blind testing of CFD simulations of a self-propelled vessel at full scale under three different speeds. It was the industry’s first blind test of this kind and exactly what was missing as already discussed earlier. 22 simulation sets were submitted from 15 different countries, different people using different CFD codes. And the outcome? The full report and the data sets are available for public access at LR website, but I’ll just mention that the standard deviation for ship speed calculation, across all three speeds and all simulations, was 3%. And now, with the availability of this data companies can fine-tune their processes and best practices and make it better. Once the majority of shipowners lend their full support to simulation-based approach to ship design it will truly open the doors to innovation.