Dec 062011

It’s pretty common to see application stories, where engineering software vendors try to demonstrate why their software is wonderful.

I find application stories interesting, but they don’t really tell me what I want to know: What is the right tool for the job? Or rather, what are the right tools for the job?

I started thinking this way maybe 28 years ago, when I’d see different types of manufactured goods, and ask myself if the CAD software I was using could handle that kind of design.

Even up to 10 years ago, the answer has too often been “no.” Fortunately, things have gotten better—but choosing the right tools for the job hasn’t gotten particularly easier. There are just so many options today.

In the sprit of understanding engineering software tools, I’d like to start a thought experiment: Looking at particular products, and considering what toolsets would be best for their design.

Not that I actually know what the best toolsets would be. I’m not that smart. But I do know a lot of smart people—on both the user and vendor sides of the markets—read this blog.  So, consider this a request for feedback. I’d like to hear from software vendors about their tools. And from people who have real-world design/engineering experience with a particular type of product and the relevant tools. I’ll gather what I learn, and write a follow-up article.

Inspired by an upcoming webinar, the first product type I’d like to look at is a golf club head.

Golf Lessons

This Wednesday (December 7, 11:00 AM CST ), Pointwise, Intelligent Light and the University of Tennessee at Chattanooga SimCenter are presenting a free webinar that illustrates the various steps of the complete computational fluid dynamics (CFD) process typically followed in aerodynamic analyses of realistic geometries.

They’ll be creating meshes with Pointwise, and with tools developed at the UTC SimCenter. Steady and unsteady CFD solutions will be computed on a distributed memory LINUX compute cluster with TENASI, a UTC SimCenter parallel-unstructured Reynolds averaged Navier-Stokes code. Post-processing will be performed using FieldView by Intelligent Light.

For this webinar, they’re using a pretty well-known type of geometry: a golf club head (a wood.)

I assume they chose a golf club head as an example because it’s interesting, and it lets them demonstrate what their tools can do—not because their software is the only (or even best) choice for analyzing golf club head designs. (I’m thinking it’s possibly massive overkill.  But there’s nothing wrong with that, is there?)

What tools does it take to design a golf club head (specifically, a wood?)

The USGA has a set of rules that govern the design of golf clubs (and their heads.) They cover all kinds of arcane details, including everything from the geometry of grooves to the volume of heads.

The goal of club designer is to work within those constraints, maximizing range and accuracy, providing as much forgiveness for swing variations and errors as possible, while making an aesthetically desirable product. (What, you don’t think golfers care about aesthetics?)

The fact that different number woods have differing face angles would suggest that that an ideal CAD program for golf club head design might have parametric capabilities. The USGA rules provide a set of constraints that also hint at using a parametric CAD program.

Still, not all CAD systems can effectively use the USGA constraints as parameters. For example, one of the rules is that a wood head’s volume must not exceed 460 cubic centimeters. For many CAD systems, it’s simply impossible to drive geometric dimensions using volume. It gets worse: The USGA limits the moment of inertia of a club head to 5900 g-cm2. See if you can plug that constraint into most CAD systems.

Woods are aerodynamic clubs, designed to be swung fast. While a wood’s face may be flat, not much else about it is. This implies that an ideal CAD system would have the ability to handle class-A surfaces. Certainly with G2, and possibly with G3 surface continuity.

With any product that’s aerodynamic in design, it’s a given that CFD should be in the design toolset. At least, if you want to compete with the market leaders. If you really wanted to complicate the analysis, you could optimize for under-water shots, for when players need to hit their balls out of  a water hazard. (Or you could add many-body dynamics analysis for when they need to hit out of a sand trap.)

It’s also a given that FEA should be in the bag of tricks, to optimize strength and stiffness within the USGA geometric constraints.

Modal, vibration, and acoustic analysis might make sense too (though these might imply analyzing a full club, not just a club head.) Modal response and vibration figure into performance, but sound figures into aesthetics. Guess which is more important? Karsten Solheim built a golf club empire based on the sound his putter made when hitting a ball: Ping.

Beyond CAD and CAE, there’s the issue of optimization. To do real justice to the problem of golf head design requires going beyond the “red is bad, green is good” school of static FEA thinking. It requires going to the Pareto frontier, to find the set of optimal design solutions. There are a number of interesting tools available to help you get there.

Chances are that, if you’re going to do a truly rigorous design of a golf club head, you might want to model a golf swing. For that, you’ll need a computer algebra system. And, since you’ll eventually want to do testing with physical prototype, you’ll probably want an instrumentation/data acquisition system, to capture and use test data.

And you thought golf clubs were simple.

Well, golf clubs look simple, at least. But they require real engineering, based on real science. That implies that they require serious engineering software tools. I don’t think you can get away with using SketchUp and AutoCAD LT. (As nice as they are for some things.)

The toolset for designing commercially competitive wood heads (which are not usually made out of wood anymore) includes CAD/CAID, meshing, FEA, CFD, post-processing, optimization, math, instrumentation, and probably a half-dozen things I’ve forgotten. I’m not counting manufacturing tools, because, at least for wood heads, most are produced by foundaries using investment casting.

While I could tell you, off the top of my head, what toolsets I think might work well for this design problem, I’m far more interested in hearing what toolsets you think would work best. If you have some thoughts, either leave a comment, or write me a note at


  • Blake Courter

    Fun article, Evan.

    Although it might be career-limiting to say it, I am not a golfer.

    I assume that in order to get a golf club to sell, it must perform well. I suspect that look, feel, brand, and celebrity endorsements are ultimately more important, although some of those factors are likely derived from performance. My marketing brain suspects that fancy-sounding technology sound bites can greatly increase profitability. Did the Nike Air really cause basketball players to jump higher? Was it a brilliant product/market fit?

    But to answer your question, if I wanted to compete in a high-performance market, I would want to have the best team I could afford using the best tools for the job. CAE would come first. Ideally, CFD would drive the overall geometry shape directly, so the resulting surface geometry would drive the overall shape of the head. Specifically, the hard constraints such as the angle of the striking face, approximate volume, etc., and resulting surface geometry with low drag would result. Some mesh-to-surface workflow might be helped through a reverse engineering tool such as Geomagic. If CFD-generated surfaces aren’t practicable, then you’re looking and a multivariate design optimization on surface control points. You might be able to get away with that in a history-based modeler, but I would probably turn to an aerodynamic surface heavy-lifter like Multi-Surf.

    Once you have the CFD geometry optimized, you’re looking at a three pronged problem. ID, materials, and FEA converge. Let’s back off of ID for a moment. Although the outer surfaces are fixed, the internal geometry probably has many degrees of freedom. It seems to me that the next step is to wire the geometry to the simulations using an MDAO tool like ANSYS Workbench, Esteco Modefrontier, Noesis Optimus, or perhaps even manage everything from the top down in Comet. For geometry, I’m partial to direct modeling, so I can dimension and optimize one model many different ways through various simulations without having to create many different constraint regimes to juggle my different simultaneous modeling intentions. To my knowledge, SpaceClaim* is the best product for the geometry part of the equation once the tricky surfaces are defined.

    Along the way, I would intimately involve industrial designers. Assuming that the geometry is mostly generated from engineering, I would have them refining the styling and producing deliverables I could use to sell the idea to potential customers and partners. For that, a Rhino or Modo would probably be the best fit, along with a really fast renderer like Keyshot so they can quickly work up different color and style combinations.

    But I am not an expert in precisely what tools are best, which is why I would surround myself with the best talent and let them use the tools they need. 1-2 engineers and a great designer or outsourced design firm should do it. What I would not do is force everybody do use a big box store CAD system that has a long laundry list of features but isn’t excellent at anything in particular. In fact, I probably wouldn’t bother documenting the design more than necessary to get it mass produced, which is surprisingly little nowadays.

    The process I outlined isn’t something I invented to answer your question. I’m extrapolating some workflows my customers have shown me. Although I’m still no expert engineer and designer, this is how I’ve seen it done by the best.


    * I’m an employee of SpaceClaim Corporation

  • Alex

    Hi Evan,

    I wanted to run a story idea by you about a new 3D design software from Dassault Systèmes, FashionLab, that is gaining traction within the fashion community and she thought you may be interested in the story. FashionLab has received the backing of such luminaries as Julien Fourniè, Jonathan Riss and François Quentin who have described developing their design in 3D as “game changing.” Here’s a quick video to give you an idea of how fashion designers use FashionLab:

    The best way I can explain the technology is using a Project Runway analogy. In each episode, the designers dutifully scribble down their sketches on a tablet and then build their garments based on that 2D sketch. As you know, sometimes designers get in trouble because of the way a particular material is draped or the look of it from a non-head-on angle. FashionLab would enable the designers to not only view and design their garment from a 360 degree angle and also simulates the drape and look of the garment based on the materials the designer is intending to use. This helps designers avoid putting together garments that are not ready for the runway and finding that out only after spending hours putting it all together.

    Julien Fourniè’s story, in particular, is interesting because he was able to take a design from start to Fashion Week in only a few weeks. He was in a time crunch for the all-important event and was able to design, construct and walk a garment down the runway in an abbreviated amount of time because of his use of FashionLab.

    If you’re interested in learning how 3D technology is altering the fashion world, I’d be happy to put you in touch with a Dassault Systèmes spokesperson if you’d like to learn more about the technology. Let me know if you’d like to move forward or have any questions.


    Alex Parrella
    Account Executive | fama PR
    Liberty Wharf | 250 Northern Avenue | Suite 300 | Boston, MA 02210
    p: 617-986-5021 | m: 617-602-6160 | e:
    Twitter: @AJParrella

    Boston’s #1 company to work for 2 years in a row
    And the only PR firm named to the BBJ’s “Best Places to Work” 5 years in a row (2007-2011)

  • Kevin Quigley

    The one factor – probably the overriding factor – in all this tool selection is cost. Cost of both selection and procurement of the “tool”, cost of using the tool, and cost of development time (in outlay and loss of sales through delays).

    All this sounds fantastic – FEA, CFD, ID, etc etc….now tot up the cost of all that software, tot up the cost of the skill needed to drive it (pardon the pun), and now try to evaluate the ROI on that investment.

    I know nothing about Golf, but I know a lot about other sporting goods. CAD companies and associated businesses like to think these products will be designed virtually to the nth degree but the reality is that state of the art equipment requires tweaking in the field.

    Do you think you win (say) the Tour de France on bikes that have never been physically tested and refined? CAD tools are a good starting point – and valuable for helping to identify refinement paths – but they will never replace the physical prototype where function is critical.

    Many of the world leaders in sporting goods and specialist areas are small companies who simply cannot justify spending tens of thousands per seat on a CAD system – it is FAR more cost effective to use a generic system, allied to skilful people who can make parts and skilful people who can test those parts.

    Bottom line – the most important tool is the person running the project.

  • Patrick Archer

    when I decided to tackle on golfing I thought it was just as simple as hitting the ball. Baseball was more easier than it look rather than playing golf. Since you need accurate precision and focus in order to win.

  • Marsha Gold

    Creating a simple golf club requires a lot of work and effort. And also it requires a lot of engineering skills and experience. Plus you cannot just really on CAD alone for you need some other third party software to finish it.

  • Sherry Bray

    @Marsh Gold

    Yes I have to agree at your point since there are some instances were in one application could not suffice our need. And we need another one to accomplish and do the finishing touches of it.

  • hj ruasiah

    I wanted to run a story idea by you about a new 3D design software from Dassault Systèmes, FashionLab, that is gaining traction within the fashion community and she thought you may be interested in the story.