A Discussion With Kevin Schneider On Inventor Fusion

Today at AU 2010 I spent some time with Kevin Schneider, Product Manager of Inventor Fusion. Kevin and I have been discussing Fusion online and offline every since it was first released. In fact, I think that just about the only thing we ever talk about.

Inventor Fusion Technology Preview 4 was released a few weeks ago and it has become quite clear to me that Autodesk is getting ready to make a product out it. Here I am not talking about the Change Manager part of Inventor Fusion. In fact, in TP4 Autodesk did not enhance the Change Manager a lot. I am referring to the standalone Fusion direct modeling application. They have refined it to a point where is has come quite close to a product that is ready to be shipped. This aligns perfectly with the announcement that Buzz Kross made yesterday about Autodesk going after the DIY market with Fusion (see “Autodesk’s Plans For Inventor Fusion“).

I asked Kevin what kind of experience Autodesk had in the DIY space. Kevin replied, “I have spent some time with customers over the past year. Some of the ease of use features that we added were a direct result of working with non-CAD professional users“. I asked whether he thought someone with zero CAD experience could be able to use Fusion. Kevin replied, “I think its easier than anything that is out there today. Is it easy enough? No. We have a lot more to do. When I look at Fusion today I see that its functional but not beautiful. I think we can do so much better and we will“.

The enhancements added in TP4 were mainly of three kinds: (1) Ease of use, (2) Ability to import data from other sources, and (2) Wiring it up with simulation technologies. Kevin told me, “We want to make Fusion very easy to learn and use. We have found that users are using Fusion for top-down conceptual design. They start with an empty canvas and have no idea how many parts they are going to design. They sketch, boolean, split and do all kinds of stuff. Its a completely free user experience“.

Autodesk has added basic surfacing support to Fusion. They also turned on surface import support for existing translators, including CATIA V5, Parasolid, Alias Design, IGES, Rhino, etc. SolidWorks support is not in TP4 and Kevin tells me that they are working on it.

Regarding simulation Kevin said, “Analysts are not CAD professionals. They need tools that are easy to use. They need to import data from anywhere. They need model simplification which in fact is the first thing that they do to a model before analyzing it. Fusion TP4 now comes with a simplification wizard with which a user can defeature a model automatically using just a few mouse clicks. One of the problems faced by analysts is that not all the data they get is high quality and sometimes won’t stitch into a solid. Fusion now has tools to repair and clean up data. They also want to do mid planes and 2D cross sections for simulation. The surfacing tools added to Fusion allow that“. Yesterday I wrote, “Fusion can and will be a direct competitor to SpaceClaim Engineer“. I guess you can see why.

Yesterday I had mentioned that Autodesk will be shipping Inventor Fusion free with AutoCAD. To make it easier for users, they will now be able to copy and paste 2D and 3D geometry between AutoCAD and Fusion. Kevin said, “We found a huge number of AutoCAD customers who wanted to move to 3D for mechanical design but were reluctant to do so for a number of reasons, the main one being having to learn a whole new way of history based parametric modeling. They have taken to Fusion really well. The thing with Fusion is that you only have to learn a few new things. But the good thing is that they all apply if you decide to take the next step and move to full blown Inventor. So it makes a big leap into a set of smaller steps“.

The current TP of Fusion will run till May 2011. That’s six months. If you are an AutoCAD 2D user thinking of going 3D, I strongly urge you to try out Inventor Fusion. Heck, if you are any CAD user and want to try out some really nice stuff that Autodesk has been working on, check out Inventor Fusion. You don’t need Inventor to be installed on your computer.

I will end with an Autodesk video of the previous TP of Fusion.

  • Shyamalroy

    Deelip, I watched the Fusion demo – extremely impressive.

    One question – how did the user decide that the flange would need 8 holes and not 4 or 6?

    • The way it works now, FEA does not cook up a solid model for us. It needs a solid model to begin with and then tells us if it will work or not.

  • Shyamalroy

    Thanks. That’s what I thought.

    By the way that was the same question posed by an engineer at British Society of Mechanical engineering meeting in UK when Autodesk demonstrated Mechanical Desktop several years ago.

    FEA does not cook up a model – it tells what is wrong with a model. Engineering before modeling tells how to get the right model to start with.

    I will be very impressed if Fusion had some basic pre-modeling analysis capability. That would minimize “build-test-fix” cycles and prevent design dead ends.

    I think we have discussed this before.

  • Shyamalroy

    Thanks. That’s what I thought.

    By the way that was the same question posed by an engineer at British Society of Mechanical engineering meeting in UK when Autodesk demonstrated Mechanical Desktop several years ago.

    FEA does not cook up a model – it tells what is wrong with a model. Engineering before modeling tells how to get the right model to start with.

    I will be very impressed if Fusion had some basic pre-modeling analysis capability. That would minimize “build-test-fix” cycles and prevent design dead ends.

    I think we have discussed this before.

    • In the 3D world, the scope of doing pre modeling analysis is limited, since the analysis needs the 3D model to begin with.

      • Shyamalroy

        Deelip I totally disagree with you.

        90 percent of mechanical Engineering is done in 2D.

        • Interesting. So are you saying that things like FEA, CFD, etc. all fall in the 10%? So what’s in the 90%?

  • Shyamalroy

    FEA is applied to analyze a design after it has been done. It is an analysis tool not an engineering design tool. CFD is applicable in less than .001% of design situations.

    Following is a partial list of mechanical engineering challenges that are/can be solved with 2D geometry:

    Alternate internal combustion engine design
    Automotive components design
    Belt systems design
    Boat and Ship Structure Design
    Calculation library development
    Car seat mechanism design
    Composites design
    Conveyor system design
    Design for six-sigma quality
    Earth moving machinery design
    Electro-Mechanical designs
    Fluid container design
    Force vector diagrams
    Forging process calculations and validation
    Forward and inverse kinematics
    Four wheel steering mechanism design
    Hydraulics driven machinery design
    Layout analysis
    Linkage and mechanism designs
    Logic driven machinery design
    Masonry Bridge Evaluation
    Mechanism space claim analysis
    Mechatronic design
    Motion control system design
    Opto-mechanical tolerance analysis
    Packaging design
    Packaging machinery design
    Paper machinery Design
    Pre-manufacturing tolerance validation
    Pre-modeling tolerance analysis and allocation
    Pressure vessel Design
    Pump Design
    Servo driven machinery design
    Steering mechanism design
    Strength of materials analysis
    Structural Engineering
    System level mechanism design
    Supply chain manufacturing capabilities validation
    Valve design

    • I think you are confusing invention with engineering. Maybe invention is too strong a word. But I think you get my point.

      • Shyamalroy

        Deelip I am not confusing Inventing with Engineering.

        ALL new designs start because of an invention or a set of product functional requirements that are typically set by marketing and business needs. Engineering is applied to ensure the functional requirements are met.

        The real question is whether the user verifies the functional requirements before committing to form or just starts blindly creating the form and then hope FEA will help them to make sure the design will function as required.

        It is not a 2D or 3D issue or an Invention vs Engineering issue.
        The issue facing any company today is whether to engineer first and design later or to do the design first and engineer later. The consequences of adapting any of the two philosophies can be substantial and therfore require careful consideration.

        I am not an expert on modeling but I can sense that with the advent of powerful technologies that are now available in products like Fusion, the modeling aspect of the design has progressed by quantum leaps. If it was possible to perform even rudimentary pre-modeling engineering to estable critical parmeters that could be used to influence the model behavior, I beleive the user would enjoy an exponenetial productivity gain and product cost and quality advantage.

        • I get your point. I guess I could quickly create a simple mechanism of an “alternate internal combustion engine”, constrain it using assembly constraints and do a motion analysis on it in 3D. I could even make it parametric and drive the parameters using equations and all kinds of fancy stuff. Later when I am done with the basic schematic I could then flesh out the parts by creating the real geometry for them. The thing is that 3D geometry creation is not that big a deal nowadays. So it would make more sense to create a parametric solid model and do the calculations directly with it instead of messing around with schematics.

        • Nainar

          I am reminded of one of the automotive guy’s request when I was working for a motion simluation software company. He asked the same question, when I dont even have an idea of a solid model, how can I create it to do dynamic analysis? He wanted a way to represent kinematics schematically and he will assign mass/intertia properties or play around till he gets what he wants. Then it is the job of the designer to design the solid part meeting that mass properties and then iterate back and forth. In my understanding this is a fantastic way to innovate/invent but how much percentage of this would fall in, either 90% or 10% is questionable.

  • Shyamalroy

    You got it.

  • Shyamalroy

    Nainar it is very efficient to solve and validate kinematic and physics functions in 2D.

    I recall the effective sologan “Make a Working Model Before You Make A Solid Model” that took Working Model 2D software to the best selling dynamic solver status in the world.

    Crudely put – pre-modeling 2D simulation is good for seeing how the product will work quickly and easily, 3D models are great to see what the product looks like before submitting to post-modeling analysis using an external software.

  • Binsar

    Mr. Roy’s argument sounds like saying a bicyle is the best mode for mobility.
    It is light weight
    It can be repaired by almost anyone
    It can be used on road and off road
    no fuel required
    no polution
    it can be carried in hand when there is no road etc etc… but still people prefer motorcycles and cars…. šŸ™‚

    • Nainar

      BiCycle is the most green transportation one can think of and has been serving humanity for ages. And it was a precursor to all the new mobility that is being seen.

      He has been very clear and he is not alone. You talk to any one who does kinematic/dynamic analysis they dont care about draft angle, corner reliefs which are all manufacturing data, that has to be precise geometrically.2D Kinematics was liked by most of the analysts to do pre modeling analysis, come up probable mass properties and then start with solid modeling. When you don’t even know where to start with, going back to the basics always gives the right direction. That is what Shyamlroy means. The catch is that one needs to invent, try new things. Copycats dont do such things. Guys who increase CC of an engine just by increasing bore dia without worrying about involved dynamics don’t need to do this.

  • Guest

    You could parametrize the model and perform FEA for several configurations of the solid model. 4 holes, 6 holes, 8 holes. big deal.