A while back, I had some meetings in New York and Atlantic City with a prominent real estate developer to discuss the construction of a rather substantial motoryacht. Even in those days he was a hard guy to get into a room to talk, dividing his time as he did between his Fifth Avenue penthouse and a grand oceanside residence in Palm Beach. I can only imagine it’s even harder today: Manhattan, Palm Beach, Washington D.C. Anyway, during each of our conversations, he would inquire several times about the design fee. Each time, I’d answer with the same number. Eventually, seemingly exasperated, he exclaimed, “That can’t be. After all, the computer does all the work, right?”
I rolled my eyes. If only.
That was more than 30 years ago. And although we have most definitely made technological advances since then, when it comes to testing hulls to predict performance, not much has changed. Computers still do not design boats. People do. Computers simply assist in the process. After all, CAD stands for computer-aided design.
And even though our current commander-in-chief might embrace Hollywood’s futuristic depiction of glass-clad labs where secret agents summon glitzy designs of submarines, aircraft or top-secret gizmos on demand, the fact remains that, despite the ever-increasing utility of new technologies, the evolution of the tools used to confirm hull design has been grinding along at a comparatively glacial pace.
Consider: The Scottish shipbuilding company William Denny and Brothers opened the first commercial ship model basin in 1883. Since then, nearly all advances made in hull, propeller and appendage design have been tested in commercial or private model basins or in specialized model tanks, such as cavitation tanks.
In the intervening years, software development has continued to mature as surface- and solid-modeling has become more and more accurate. Many applications have achieved commercial success, among them predictive and developmental programs for hydrodynamics, velocity performance, finite element analysis, and the current fair-haired prodigy of the digital design world, computational fluid dynamics (CFD). Each helps designers predict what will happen in real-world conditions, but clients continue to believe that a computer can now design the perfect hullform. But can it?
Several naval architects told me this phenomenon has struck them as well, including those a generation older and a generation younger than I.
“Tank-testing is very interesting for the clients,” said Ronno Schouten, manager of design at de Voogt Naval Architects (Feadship’s naval architectural firm), which uses CFD to predict seakeeping and tank testing for final analysis. “Watching the designers and technicians at work in a model basin is both captivating and stimulating, but watching the analysis of CFD live in the design office can be interesting to them as well. [Regardless], we use CFD [to predict] seakeeping, but it’s backed up for final analysis by tank-testing.”
Quite often, model tests show results that disagree with design software predications. Donald Blount learned this before starting his own naval architectural firm, while serving as head of the U.S. Navy’s Combatant Craft Department at the David Taylor Model Basin.
“Kids go to college and study CFD,” he said, “but what bothers me is that they know how to enter the equations and code numbers into a computer, but the whole secret of the thing is [to] actually represent—in detail—the shape that they’re studying. It makes a big difference in the distributions, the geometric representations. For some features, you don’t need a lot of elements to describe that shape, and for [others], you need more than you could believe to get a realistic answer.
“Do you have to do something in the water or can you do it digitally?” he continued. “The answer is, You can do it in the water, and some things you can do digitally very well, and some you have to be an expert in understanding the wet part before you can get good results from CFD and a digital approach.”
Computer geeks call this concept GIGO (garbage in, garbage out), referring to the quality of data input for any given prediction. The more accurate the data that is input into the machine, the more accurate the result will be.
Naval architects agree that having a wealth of good data on file is extremely important to any testing regimen. CFD can be used for calculations that used to be the sole province of wind tunnels. Nowadays, the software can handle flow calculations for such factors as air conditioning, superstructure design and placement of exhaust vents.
“It’s all about reducing design-cycle time,” Schouten said. “After all, model tanks, wind tunnels and computer time are quite expensive, so the key is to integrate the available tools to find a balance that works best to go from initial to final design as accurately and cost-effectively as possible.”
In addition to the cost of the robust hardware needed to grind out predictions, the software itself is quite pricey, with licenses for each program running in the many thousands of dollars (per workstation, per year).
Nowhere in the recreational marine industry has computational design advanced farther than in America’s Cup racing. Today’s remarkable AC catamarans—and their equally remarkable cost—illustrate what happens when one takes computational design to the extreme on a per-boat basis. The amount of data collected in each race is staggering, and all of it is vital to refine not only the overall design, but also each and every operational parameter as well. The result is a designer’s most coveted scenario: data from full-scale testing under actual operational conditions.
So, how long will model basins be used before they go the way of the horse and buggy? Were I a betting man, I’d say don’t count them out just yet.
“CFD is not a proven technique for predicting seakeeping characteristics of vessels moving through the water in waves,” Blount said. “That’s probably the next big hurdle that needs to be done, but part of that hurdle is, How do you model a random or irregular sea mathematically? You can do this in a towing tank, but the mathematical representation of that is, I think, maybe five years or more away before you should even have balls enough to spend millions of dollars on a boat [that’s] been designed entirely by using CFD and digital techniques for something that is supposed to perform well in rough water.”
For more information: devoogtnavalarchitects.nl dlba-inc.com; marin.nl; oossanen.nl
CASE STUDY: Computational Fluid Dynamics vs. Model Testing
A deep-rooted—and healthy—skepticism guides engineers in their quest to balance human safety with trends in equipment and materials. To that end, naval architects have cautiously embraced computational fluid dynamics (CFD) as a means to reduce design-cycle time, especially in the early stages of the design process, while asking whether the results are as accurate as those achieved by using model testing.
Donald Blount, whose company designed the 220-foot (67-meter) record-speed-setting Fincantieri Destriero, had a client so intrigued by the subject that, prior to making a build decision for a 262-foot (80-meter) high-speed motoryacht, funded an exhaustive study to compare the predictability of the two methods.
Employing the testing facilities at SSPA in Göteborg, Sweden, the study examined three hullforms: double chine, round bottom with spray rail, and single chine. Each would be tested at two displacements and as many as three longitudinal center of gravity positions. This program ensured that a wide range of data would be collected to compare the hulls. “The objective of this investigation,” the study states, “was to identify a hullform that achieved suitably low resistance…to satisfy an endurance requirement while remaining dynamically stable…to satisfy a top speed requirement.”
After running the models in each of its various permutations, some of the same test conditions were compared to results from a CFD application—which, the study stated, “demonstrated that it could satisfactorily differentiate the bare hull resistance between several candidate hullforms,” and noting, “The absolute magnitude of the resistance predictions is generally within three percent [of] the experimental data.”
Blount says the results of the two prediction methods indicated some difference in speed when dynamic instability was likely to occur.
“I personally prefer experimental testing, since experimental results are immediate, confirming you have met design expectations or not,” he said. “If expectations are not met, one can readily alter a model to evaluate alternative solutions.
“An analytical model representing hull geometry requires a very experienced hydrodynamicist for input to CFD software to obtain realistic results,” he added. “Inexperienced CFD users are dangerous, as they might not recognize the validity or reality of output predictions, which can vary due to poorly distributed elements for hull geometry.”