by EJFoeth » Fri Oct 16, 2020 7:56 am
> Though I suspect by now a lot of that empiric work as moved onto the digital landscape?
Hull forms, propellers are CNC-milled, many details 3D-printed, but fitting out is still the model makers job. We recently developed a new autonomous submarine model but can't find any pics online...
> However, we're still plugging away with wind tunnels and models in motorsports as with increasing computing power, the CFD demand becomes more detailed CAD models, which means the average case solve time really hasn't changed that much (in fact, it's gone up!)! And in any given day in the wind tunnel, we can bang through dozens of iterations, each being subjected to a full ride height map.
One of the reasons towing tanks and cavitation tunnels still exist. And when you are going to unsteady conditions (maneuvering and sea keeping) CFD will quickly become prohibitively expensive (for non-navy work, that is). CFD has become critically important in the design stage though and is gaining ground quickly in most aspects of design.
> Point being, for what we're doing we still see the point of physical models and wind tunnels even in light of massive increases in computing power and CFD over the past 20 years.
There are many difficulties with CFD with capturing critical flow details, plus, you need high-quality measurements to validate your CFD codes with. People are moving to so-called scale resolved-turbulence models, meaning that you try capturing as many of the smaller structures in the flow at an enormous computational cost. Capturing all flow details on a propeller remains difficult, especially when you include fluid-structure interaction for estimating blade tonals (acoustics) or deformations (the latter are becoming increasingly more important with composite propellers/rotors).
> Though I suspect by now a lot of that empiric work as moved onto the digital landscape?
Hull forms, propellers are CNC-milled, many details 3D-printed, but fitting out is still the model makers job. We recently developed a new autonomous submarine model but can't find any pics online...
> However, we're still plugging away with wind tunnels and models in motorsports as with increasing computing power, the CFD demand becomes more detailed CAD models, which means the average case solve time really hasn't changed that much (in fact, it's gone up!)! And in any given day in the wind tunnel, we can bang through dozens of iterations, each being subjected to a full ride height map.
One of the reasons towing tanks and cavitation tunnels still exist. And when you are going to unsteady conditions (maneuvering and sea keeping) CFD will quickly become prohibitively expensive (for non-navy work, that is). CFD has become critically important in the design stage though and is gaining ground quickly in most aspects of design.
> Point being, for what we're doing we still see the point of physical models and wind tunnels even in light of massive increases in computing power and CFD over the past 20 years.
There are many difficulties with CFD with capturing critical flow details, plus, you need high-quality measurements to validate your CFD codes with. People are moving to so-called scale resolved-turbulence models, meaning that you try capturing as many of the smaller structures in the flow at an enormous computational cost. Capturing all flow details on a propeller remains difficult, especially when you include fluid-structure interaction for estimating blade tonals (acoustics) or deformations (the latter are becoming increasingly more important with composite propellers/rotors).