Building a software workflow around MFEM

Discussion in 'General Computing' started by detourdog, Sep 7, 2020.

  1. detourdog
    Joined: Sep 2020
    Posts: 25
    Likes: 0, Points: 1
    Location: Western Massachusetts, USA

    detourdog Junior Member

    Hey All,

    I'm the curious type and I'm thinking I will start applying my curiosity to boat design.
    I have a strong technology background and I'm gearing up to start contemplating a boat. I believe the modern way to consider boats is with Finite Element Analysis. I believe this can be used to be analyze and the boat performance as well as the quantities needed for production planning.

    I have read through Jorge's epic Infusion Plan thread so I believe I will be vacuum infusing laminates and testing and determining strengths. I also really like to understand what I'm doing so that is why I'm doing this the hardway.

    The question is where do I start. I have identified a BSD licensed software package called MFEM.
    https://mfem.org

    Which I believe will be the tool I will use for finite analysis. I have a couple of existing CAD programs that I use (not on the boat software list).

    My plan is to make Symbols for the various elements that I want to finitely analyze and export symbols for processing. I was going to start with the catamaran parameters spreadsheet in the spreadsheet library to see if I can get something working.

    My intention is to develop a library of symbols for my future use.
    I have been reading this document which I found for free.

    ISSC 2009 Committee V.8 SAILING YACHT DESIGN
    If I start to feel good about my approach I assume the purchasing of ISO 12215-x:2019 is the legitimate way to find the current accepted formulas and values.

    Does this sound like a reasonable approach?
    Do free sources of current scantling forumulas, etc exist?

    Thanks,

    Ben
     
  2. detourdog
    Joined: Sep 2020
    Posts: 25
    Likes: 0, Points: 1
    Location: Western Massachusetts, USA

    detourdog Junior Member

    Talking to myself in case any other lurkers are curious... I forgot I had a copy of Principles of Yacht design by Lars and Rolf Eliasson. It seems to be old enough that the ISO rules weren't adopted but they are referenced and a road map for compliance is there. This should keep me busy for months.
     
  3. TANSL
    Joined: Sep 2011
    Posts: 7,369
    Likes: 699, Points: 123, Legacy Rep: 300
    Location: Spain

    TANSL Senior Member

    First of all, welcome to the forum. I hope you find in it all the help and support that you are looking for.
    I sincerely believe that it is not worth it that you work for months with that book. Its formulas, very useful when there was nothing else, 20 or 25 years ago, may have become obsolete and totally superseded by the Classification Societies and by ISO 12215, which, in a few years, have improved and completed their regulations.
     
  4. Ad Hoc
    Joined: Oct 2008
    Posts: 7,773
    Likes: 1,678, Points: 113, Legacy Rep: 2488
    Location: Japan

    Ad Hoc Naval Architect

    Welcome to the detourdog

    What brings you to this conclusion?

    What do you mean by 'symbols'.
     
  5. detourdog
    Joined: Sep 2020
    Posts: 25
    Likes: 0, Points: 1
    Location: Western Massachusetts, USA

    detourdog Junior Member

    It's not strongly held but if I had to defend the statement. I would say that the Polynesian or double canoe form was developed over large time spans and other hull forms have been developed in between. They all used modern techniques for their development. Would be the simple answer.

    The long philosophical reply could be like this.
    The tool of finite element analysis is very efficient. I can use the same computer and software for iterating hull forms, predicting resource usage during hull construction, and analysis of the hull under usage. If someone was starting out today in our modern times it would make sense to understand this tool.

    Once it's understood it could be rejected.
    I'm thinking in the classical lisp sense where a symbol was an electronic signal representing an object.
    I'm settling for an organizational tool in my CAD program that will provide a way to assign attributes shapes and tracking them in a database.

    Essentially I'm trying to develop a publishing process. I want to work on ideas in the real world and assign the ideas to symbols and finally output the relevant symbols as GCODE for output on my CNC and appropriate material quantities.

    I see a symbol as the interchange of developed ideas through software packages.

    In one sense when I a output the symbols from my CAD program I want to run them through MFEMlike software package to get analysis I could never do with paper pencil.

    Why do I want to do this? I don't know. I have found myself with the time and space so I stopped asking why.

    I must say I was a lurker for a very long time before I understood how to use this site. I think it will be good.

    Thanks for all the comments,

    Ben
     
  6. Ad Hoc
    Joined: Oct 2008
    Posts: 7,773
    Likes: 1,678, Points: 113, Legacy Rep: 2488
    Location: Japan

    Ad Hoc Naval Architect

    And how do you plan to validate and verify your output?
     
  7. detourdog
    Joined: Sep 2020
    Posts: 25
    Likes: 0, Points: 1
    Location: Western Massachusetts, USA

    detourdog Junior Member

    Suppose currently I don’t trust FEM and I don’t understand real life. I can make predictions using FEM and apparently verify hull forms with model boats in water. I have also read about testing laminates in the Gougon book and other places. I was surprised also to find YouTube videos of people testing laminates with sophisticated and unsophisticated methods.

    Trial and error got s this far right?
     
  8. Ad Hoc
    Joined: Oct 2008
    Posts: 7,773
    Likes: 1,678, Points: 113, Legacy Rep: 2488
    Location: Japan

    Ad Hoc Naval Architect

    And how do you do that?
     
  9. detourdog
    Joined: Sep 2020
    Posts: 25
    Likes: 0, Points: 1
    Location: Western Massachusetts, USA

    detourdog Junior Member

    I'm starting to think you are trolling me but I will be polite and answer this question.

    If I believe something is a good idea I can draw it in CAD, output the symbols from my drawing and run them through FEM process.
    Take the output which I will assume are performance predictions.

    Take an actual scale model of hull form from the previous step and see if it performs as expected.

    This is what I assume the process is. This is why I asked the question here...
     
  10. Ad Hoc
    Joined: Oct 2008
    Posts: 7,773
    Likes: 1,678, Points: 113, Legacy Rep: 2488
    Location: Japan

    Ad Hoc Naval Architect

    Perhaps because you are uncomfortable in answering questions based upon your curiosity...But as a simple rule of thumb...those unable to answer questions on the points they themselves raise, always default to calling the person seeking classification of their questions, a troll, merely to avoid answering the questions and thus, changing the narrative.

    There are a lot of assumptions you're making, and yet in the same breath suggesting this is in itself sufficient??

    Question.
    Do you understand what FEM is?
     
  11. detourdog
    Joined: Sep 2020
    Posts: 25
    Likes: 0, Points: 1
    Location: Western Massachusetts, USA

    detourdog Junior Member

    Yes, I'm stating simple assumptions

    Sorry for implying you may be trolling but I couldn't see where you were going. My big fear is that everyone is going to tell me to subscribe to AutoCad:)

    I'm happy to answer honest question because I have no problem admitting I have no idea what I'm doing.

    I'm making simple assumptions because I'm trying to confirm I know what I'm talking about.
    That is a fair question and since I'm unsure of the answer myself I will let you be the judge.

    I believe FEM to be a body of mathematics born from calculus. I hope to use it as a software tool to rapidly understand what is happen in 3-d space.
    I want to try to understand the statistical interaction between different materials. I believe that I can use the tool predict the properties of laminate structures.

    I'm an autodidactic so everything I learn is through experience. Thank you for your help and guidance.
    Ben
     
  12. Ad Hoc
    Joined: Oct 2008
    Posts: 7,773
    Likes: 1,678, Points: 113, Legacy Rep: 2488
    Location: Japan

    Ad Hoc Naval Architect

    It would take far too long to explain all the nuances of FEM. But suffice to say it is all an "approximation".
    The objective is to reduce the amount of approximations into something more "expected".

    Take a simple circle, and discretise the circle into 6 elements:

    upload_2020-9-10_20-21-25.png

    It is a hexagon, not a circle. Now discretise into 12 elements... better... but it is not a circle.

    No matter how many times you divide the circle into smaller and smaller elements, it is still only an approximation - mathematically.

    So now imagine a complex 3D structure...just imagine how much "approximation" is required to model a 3D structure??!!. Since when the model is loaded, what 'shape' does it adopt?....the shape, is also like the circle, an approximation. Ergo, if you cannot define the shape of deformation correctly the result will not be accurate, it will be an approximation. And the more complex the structure the more approximations occur.

    Thus you need to validate the model you create against a real physical model to ascertain how far 'away' from reality are the 'assumed' approximations.

    And it is nonsense to use FEM to obtain the stress between layers of fibre etc in campsites. Why... because it is impossible to verify. Since the presence of a strain gauge between fibres of an actual laminate influences the result. Ergo, how can you assume it to be correct... you can't.

    I could go on.. but you get the picture.

    FEM is just a "tool"...an advanced an complex tool, but it only produces usable result if you understand its limitations and how/where to validate the model with the real physical world. Otherwise it is just a pretty picture and nothing else.
     
  13. detourdog
    Joined: Sep 2020
    Posts: 25
    Likes: 0, Points: 1
    Location: Western Massachusetts, USA

    detourdog Junior Member

    Yes, I agree. I see that my thread title could lead one to think I was putting more faith in FEM than is warranted. My thread title though refers to MFEM which is a specific software package that I'm going to evaluate for my usage of FEM.

    Thank you for that explanation. I think technology is more useful when it's not exceptional. If the additional analysis using FEM analysis is built into the process it becomes transparent to the user. Having it as defined part of the platform creates new oppertunites.

    If I ever get to the point of building a boat it will have a large embedded sensor network in the hull:)
     
  14. detourdog
    Joined: Sep 2020
    Posts: 25
    Likes: 0, Points: 1
    Location: Western Massachusetts, USA

    detourdog Junior Member

    Are there drastic changes between the initial ISO 12215 and ISO 12215 2019?

    I have no problem buying copies of the standard once I feel that it I'm going to use what I'm working on.

    I don't mind if this initial work is not perfect. I'm just trying to explore as far as practical. I would be content working with any standard that is freely available and covers craft under 24m and multihulls.

    I see you offer classes in an area I'm very much interested in. I'm way off the beaten path for software tools and not likely to change:)


    Here is the CNC in the facility I'm building. This software workflow is for this location.

    [​IMG] [​IMG] [​IMG]

    The overall software workflow will manage every aspect of production.
     

  15. rxcomposite
    Joined: Jan 2005
    Posts: 2,752
    Likes: 608, Points: 113, Legacy Rep: 1110
    Location: Philippines

    rxcomposite Senior Member

    Design will not manage workflow. You can build a flowchart but that's it. So far I see the discussions center around the "tools" to be used such as FEM, CAD, CNC, ISO guidelines,ect.

    To design a workflow, you need a basic understanding of PERT CPM and a grasp on how a boat is built during each stage. Basically a time and motion study or guidelines. Then you can subdivide your workplace real state according to the workflow.

    Workflow can be grouped into:
    1. Laminating Area- Hull and deck and small parts
    2. Cutting and trimming area- Cutting of flanges, lamination excesses, flat sheets
    3. Assembly area- Where the parts are assembled to form the boat. This is the messy part where the goey resin is lugged about the boat which was in pristine condition when it came out of the mold.
    4. Outfitting- Where holes are cut to fit parts (windows, cleats, hatches, ect) and Engine installation (mast and rigging if a sailboat)

    You must have a lifting mechanism to lift the hull out of the mold and a large cradle to move the boat around during the process.

    I may still have some data on production process.
     
Loading...
Forum posts represent the experience, opinion, and view of individual users. Boat Design Net does not necessarily endorse nor share the view of each individual post.
When making potentially dangerous or financial decisions, always employ and consult appropriate professionals. Your circumstances or experience may be different.