Brent, Glad to see you are working hard to develop your FEA skills. Keep in mind what you have said concerning the necessity for a trained engineer working on the float design to have his FEA work reviewed multiple times by a professional engineer during the refinement of the model for a proposed float design. A similar situation was described in the NASA Tech article that reviewed the development of the complicated submersible hull. Some hints that might help your make your acrylic cylinder modeling closely reflect reality: 1. You should consider modeling the top hatch surface as it will be on your actual installation as a domed hatch has more surface area. 2. The top and bottom metal surface should be rings instead of solid surfaces. While there is minimal deflection of these surface, you may find some unusual interaction with the acrylic cylinder. 3. O-ring sealing surfaces will create additional stresses that will interact with those found in the acrylic and metal surfaces. 4. On reflection, there should be metal rods tying the upper metal surfaces with the lower metal surface. Metal rods between the two will apply just enough force to seal the upper and lower acrylic surfaces yet restrain additional down force experienced from increasing water pressure. 5. Question: Did you restrain the acrylic cylinder in the horizontal direction at the interface between the metal and acrylic surfaces? I believe this is what your color images reflected. This would be close to the middle case in Stachiw that Jon referenced but you would need to add the internal raised metal rim to better see the focused multiple stresses that would result in this case. If you go for the first case in Stachiw, the above mentioned metal rods would hold the cylinder in position but you would probably see the same deflection (compression) across the whole length of the cylinder instead of the compound bended shape that I depicted in my Flickr sketches. 6. Just because you get results in your FEA modeling that compare favorably with a proven FEA analysis does not mean that every other model you do will have the same consistency and accuracy (the examples of your float engineer and the sub modeling support this conclusion). Look forward to your future efforts with FEA but ask that you place a disclaimer with your work posted to PSUBs concerning its validity and accuracy. R/Jay Resepectfully, Jay K. Jeffries Andros Is., Bahamas Save the whales, collect the whole set. -----Original Message----- Jon, I think that's a excellent plan. I was happy to have a real world destructive test to compare my original K-250 dome FEA work with. My results came out very close, and since I have a stronger dome model then the real deal, I suspect if I ran the data with a model of my dome, that came from a digital scan of the dome, I might be able to come even closer. I'm working on making a digital scanner currently. I'm aware that my original dome could be a fair bit different then the one the Captain had destroyed. But on the other hand the might be really close to the same. It was just some data I could work on, to see if I was in the same ball park. I've also been comparing my results to other FEA data on sub parts and assemblies I have found or been given. I know your plane flat viewport calculator is quite conservative, but I started out using that to compare my results some time ago. Now that I have better data on the material properties for Acrylite GP, Polycast, Plexiglas, and ASME PVHO-1 minimums and FEA skills, I'll run some more windows that fit the viewport calculator parameters, to see what we can learn. I know the CosmosWorks FEA software I'm using is able to do more advanced calculations. But I need to pull data from all sources I can at this point. Cheers, Brent |