Very nicely explained. Every university student taking their first fluid dynamics or CFD course should see this video. I also agree with your statements of asking CFD vendors to prove the accuracy. A Proof of Concept exercise is a perfect way to do this. CFD is indeed complex and it starts with the meshing, which requires a lot of knowledge and can be very time consuming. There are fortunately a handful of companies that are making great strides in simplifying the process. One that is very accurate and has developed a marine specific template that works very well can be viewed in our channel.
I love those videos, but when seeing how many different factors have to be considered it's now clear why there are no deep dive CFD calculations in Watersports like Kayak, Standup padlling hulls etc... but your videos are great to understand where the different strengths of certain boarddesigns will be.
What to say - excellent topic! I appriciate yours presentation with so much love for topic like CFD! In general for my topic in hydraulic and researching deep reinforcement learning control a heaving plate of breakwater this dynamics model of the wave - plate interaction based on an in-house COMPUTATIONAL FLUID DYNAMICS (CFD) solver. Thank you for Accuracy and how to achieved DMS! It is obvious how The Fast Fourier Transform (FFT) analyses indicate that heaving motion of the active - controlled plate can reduce the amplitude of fundamental component of transmited wave, compared to fixed plate. Yours presentation with big knowledge can be direction - path how to prove the potential of CFD with DRL (that is utilized to find the optimal strateg for wave dissipation) actively dissipation ocean waves. 👍👌
Hygenberg said "When I die I would have two questions to God, why Turbulence and why relativity, and I really think he has the answer to the second one"
I loved this video! I've never taken a CFD class, however, this video made me understand in 13 min what CFD is. Btw, your presentation skills are neat! I will defend my thesis tomorrow and your presentation style will help me as a reference.
f1 engineers do use CFD to predict flow, then wind tunnels with a scaled model and even flovis paint on cars to the real flow ! so on this exemple i think they know how accurate a simulation is and how it impact the workflow !
But what is CFD ultimately trying to calculate? Is it just trying to find the *drag* of a certain hull shape at certain speeds? I ask because that's what I'm most interested in, but I hope it can take into account the rotating propeller too, and all the water swirling around it.
Absolutely. Drag of hull shape is most common, with the swirling propeller included. That is also common. One of the hardest problems people face with CFD is not imagining wide enough. Basically, if you can imagine it, we can probably achieve it in CFD. Other things CFD can do: - A free falling lifeboat entering the water. - A submarine diving - A maneuvering study to predict the turning circle of a ship at different speeds - Wind resistance on a jackup rig - Want to place a sensitive sonar sensor on the bottom of your hull? CFD can find a location with minimal turbulence. - Want to minimize your fuel burn on the next ship? CFD can run through hundreds of hull shapes to optimize the hull. - Something a little more low key? Maybe just a trim optimization study to save 2% on your resistance. These are all examples of CFD projects that were completed in the past. (Some were more expensive than others.) The best part of working with CFD engineers is that you can mix and match and customize to get the exact type of project to fit your budget and needs.
@@DatawaveMarineSolutions Thank you! I hope you can do a follow up vid about towed tanks. That way we can contrast CFD with towed tank tests. I'm really interested in how the accuracy and cost compares. E.g., it would seem that towed models are much cheaper and easier to do, but how accurate are they?
I've heard your explanation of computational fluid mechanics well. By any chance, can you recommend me a program that I need to work on? And can you make a video of the program? That'll be a very useful resource.
I have started a series of videos on programming from scratch to solving Navier-Stokes equations in matlab you can watch those videos on my channel. This course will cost you 20K if you purchase on other online platforms,I am providing the videos absolutely FREE of cost.
I know. I am preparing several videos targeted towards a wider audience. I plan to intersperse them with the CFD videos to ensure everyone gets some useful content.
Not that I'm aware of. Pipe vibration may change with flow rate, but that is probably due to the vibrations of the pump which generated that flow rate. The more common way to measure flow rate is to use a Venturi meter. It is a small section of pipe where the pipe diameter contracts temporarily. We then place one pressure sensor just before the contraction, and another at the middle of the contraction. Based on the pressure difference between the two locations, we can calculate the fluid speed and flow rate.
Depends on the simulation. But for example: take air flowing through a duct for HVAc. The engineer species the pressure, air velocity, and air temperature on each boundary. That simulation might predict how airflow splits between ducts and check the temperature at each outlet. There are other settings as well. I'm oversimplifying.
*_...Einstein [_**_03:20_**_] failed in Special Relativity too because his assumptions were self-contradictory..._* *_...we did something like CFD for stellar interiors with Runge-Kutta, but I noticed we should've tried x(y) as well as y(x), to look for regions where solutions naturally diverge..._*
I'm thinking of learning to use OpenFOAM to design a solar powered boat, probably a 12-15m trimaran. The low power requirement requires low resistance and the large top roof area (40-50m²) makes me think windage is going to be a problem. What is your opinion on openFoam vs commercial software? How hard is it to learn to do this? I'm good with computer but an amateur and not too great at math and have little knowledge of engineering. I don't need absolute truths, just a design that works as best as possible for the application, so a "relative truth" to find / optimize a design that works. Or is this all crazy? PS: Thanks for the video!
I have used openFOAM commercially, and it is not easy to learn. I often say that free software is not free. It requires an intensive time and research investment. On the one side, I encourage people to learn about CFD and gain wider exposure. On the other side, I have to caution people that a lot more goes into CFD than just the software. Nothing in the software tells you the correct settings for each particular CFD problem. If you get the wrong settings, CFD simulations can be wrong by 1000% or more. Essentially useless. Instead, I would recommend looking into a program called Michlet. This is a simplified resistance estimate program that you can learn more about here: dmsonline.us/tools/applications/ And of course, for safety reasons, my official answer is that you should have an engineer design the vessel. But I respect that not everyone can afford that. Some critical safety issues that benefit from engineering design are the hull structure and vessel stability.
There are very many options for research. I suggest you look into meshing algorithms and strategies. The mesh is still the single biggest factor to determine quality of CFD simulations. It is a very manual process. A good mesh requires a CFD engineer to understand the expected flow patterns before even running the simulation. We could benefit from a more automated approach to meshing. For example, the boundary element method (BEM) is an alternative to RANS. This is less accurate, but far faster. BEM simulations complete in seconds. RANS takes hours. BEM requires a mesh of the boundaries in the domain. Imagine only meshing surface boundaries and then run a BEM simulation from that mesh. The computer then uses the velocity gradients from that BEM simulation to calculate a field function for mesh density in the volume mesh of the RANS simulation. That type of approach could greatly help with meshing and simulation quality.
@@DatawaveMarineSolutions Thank you Dr. But If I can have an example of topic which is not done yet I will be happy. Now I have no idea about the topic to chose. Thank you Dr
This typically includes extended training in different analysis techniques for both internal and external fluid flows. Internal flows = pipes. External flows = air or water flow around objects (car, airplane, boat, building, etc.) Different techniques that you learn about include analytical, numerical, and experimental methods. Best way to describe this: you need to learn the words that form the mathematical language for fluid analysis. As an example, say I wanted to determine the lift and drag force on a wing section. Methods that I know how to use include: 1. Lookup coefficients for similar NACA foil. 2. Calculate velocity potential and pressures using Jukowsky transform (analytical method) 3. Use a panel code to get coefficients. (numerical: boundary integral method.) 4. Use vortex lattice method to determine lift. (numerical: boundary integral method.) 5. Use RANS CFD (numerical: RANS) 6. Create a physical model and test in a wind tunnel (experimental method). And then within fluids analysis, we each tend to specialize within our industry. For example, I specialize working with free surface physics: the interface between air and water on boats. An aeronautical engineer specializes in compressible flow: air flow speeds near the sound barrier.
Can I ask something? I study physics B.Sc . Is it possible to become or work as chemical engineer if I do M.Sc and PhD in Computional fluid dynamics ? Can I design Chemical reactors or Nuclear reactors Or it is better to have Chemical engineer B.Eng and M.Eng too ?
Not really sure about the best strategy for your career. But I do know that I have industry colleagues with a PhD in chemical engineering, and now they specialize as a programmer for CFD software. So the options are there. I do know engineers use CFD for chemical and nuclear reactor design. And they usually need to encode a lot of that information by hand. (For example, the software doesn't know how much heat gets released from a specific chemical reaction. You need to specify those values.) Hope that helps a little.
"I don't think I've once seen a general solution to the Navier Stokes eq"??? Dude you lost me - You're the "expert" on fluid dynamics and it sounds like you don't know that this is a Millennium problem. NOBODY has FOUND a solution - and there's an active field in math investigating if there exists one to begin with...