I was fortunate enough to be invited back to COFES, the Congress on the Future of Engineering Software, for a second year. While last year I was just a n00b and spent (probably too much) time just getting the lay of the land, this year I made much better use of the experience.
The theme of this year’s event was Correcting 2020 Vision: what should we be planning for in six years?
What’s It All About
COFES is kinda like the Edges Only brownie pan.
With the Edges Only brownie pan, every piece has a delicious, crispy, chewy edge bit. That’s the best part, right? (Click image for source.)
Let me explain. Over fifteen years ago the conference organizers, Brad Holtz and Joel Orr, were at a conference at which the sessions were so-so and all the good interaction was between the attendees in the hallways. So they decided to start their own conference that was all hallway discussions. This is the same principle behind the Edges Only brownie pan because after all, the best parts of a brownie are the chewy bits around the edges.
That’s not to say that COFES doesn’t have keynote presentations but they’re more like conversational lubricants that get everyone warmed up for the smaller roundtables and vendor/analyst suites where interested parties get together for freeform discussion on a variety of topics.
Tech Soft 3D Annual Customer Meeting
Again counting my good fortune, the kind folks at Tech Soft 3D invited me to their customer briefing which was held prior to the formal start of COFES. Their briefing featured four trends that are strongly influencing their business.
- When it comes to platforms it’s desktop, web, and mobile (not or). While there certainly are user interface issues to be resolved when it comes to interacting with 3D data on a mobile device, all indicators are that’s where people want to see their data. Adding mobile into the platform mix also adds new choices for programming language.
- Big Data – Large is the New Medium. Because 3D datasets are only getting larger, the push toward mobile devices forces you to make choices about where to do computations, how to manage bandwidth usage, and how to manage data in general.
- Sharing is Caring. One of the ultimate purposes of 3D data is to share it and re-use it across the organization. Tech Soft’s 3D PDF and JT allow this to happen.
- Regarding 3D PDF in particular, they are seeing tremendous growth in its use. They cited several reasons for this including i) the price point is below your IT department’s radar, ii) you get great bang for your buck, iii) there is a lot of customer pull for 3D PDFs, iv) being an ISO standard means no worries, and v) it just works.
- Model based design is gaining traction. Design is becoming synonymous with CAD.
Eli Ettia Keynote
Architect Eli Ettia spoke on the topic of Engineered Architecture (EA) during his special session address. His main point was that the current way buildings are designed is out-of-date and holds the entire industry hostage. According to him, the fraction of urban dwellers who live in slum conditions will double in the 40 years between 2010 (one third) and 2050 (two thirds). Without serious change to the architectural process, we cannot improve this situation.
Ettia’s EA process is a way to bridge the art and technology of architecture, turning vision into reality, while being 30-50% less costly, having 30-60% faster time to market, being more environmentally friendly, and lower lifecycle cost.
Design for the proposed Kostabi World Tower, designed by Eli Ettia Architects to be 2,000 feet tall. Image from Eli Attia Architects. I like the mesh-like external appearance.
Ettia is also very interested in circular forms in architecture which he prefaced by a quote from Sioux Chief Black Elk: “The world always works in circles and everything tries to be round.” In the photo below of one of his slides, he illustrates how a round exterior form will suffer lower structural wind loads.
For a CFD guy, it was comforting to see the efficacy of round structures demonstrated via streamlines.
Paul Saffo on Foresight for Innovators
Paul Saffo is a managing director at DISCERN and his presentation was entitled Cultivating Foresight: Seeing Clearly Amidst Exponential Change. One of his main observations is that change is never linear. Instead, progress bumbles along relatively flatly until there’s an exponential spike. He identified seven tips for helping understand change.
- You must identify your biases because they arbitrarily narrow your vision. [I tend to see meshes everywhere.]
- Identification of driving forces and their nature is key. Are they constants (things you only notice when they change), cyclic, or novelties?
- Look for cross impacts on the driving forces. Those can be dampening, accelerating (such as the influence of the Apollo program on development of the microchip), or amplifying (like a rogue wave or solar flare).
- Don’t forget to consider wild cards: low probability, high impact events.
- Never mistake a clear view for a short distance. Events take time.
- Cherish failure.
- Embrace uncertainty.
Geometric Modeling Kernels circa 2020
C3D Labs hosted a technology suite on the requirements for geometry kernels in the year 2020. This 30-minute event started out considering the issues forced upon kernels by cloud-based CAD including:
- parallelism, whether that be multi-thread or multi-process
- serialization of data
- the user experience and its dependence on whether computations take place on the server or client and the available bandwidth
- data storage
At least one person in this session was convinced that the main driver behind CAD in the cloud was the availability of more compute power than you can have locally.
I personally believe that mobile CAD (which is not the same as CAD in the cloud) has some serious user interface issues to overcome with respect to creating and interacting with 3D content on a mobile device.
Of course, no session on CAD would be complete without CAD file translation rearing its ugly head. And, of course, no one saw any solution in sight. As one participant said, “Legacy happens.”
Simulation 2020
Beyond CAE‘s Dennis Nagy hosted an analyst briefing on what engineering simulation will be like six years from now.
This conversation had a little trouble getting out of the gate while folks debated exactly what was meant by engineering simulation and whether tools developed in spreadsheets count. [For the purposes of the discussion, I wanted to narrow the definition to computational, physics-based simulation with it’s associated “big data” component.]
Regardless, the issue boiled down to this as vocalized by intrinSIM‘s Joe Walsh. The international business environment is getting much more competitive. Everyone is “lean” or using their favorite acronym quality effort or have outsourced whatever they can to save cost. Businesses now need new ways to lower cost, lower risk, and increase innovation. (Or as one participant put it, what motivates people to spend money is if they can decrease cost, increase revenue, or meet a regulatory obligation.)
How are businesses going to accomplish these things? By understanding their product behavior as early in the design process as possible. And how will they do that? With simulation. The good news is that simulation has moved from a Kevin Bacon style six degrees of separation from the CEO to one degree (the CIO and/or CTO).
The questions to ask are what are the roadblocks to expanded use of engineering simulation? A main cause identified was that tools are too general. There is no design and analysis tool for truck axles; the engineer must make a generalized tool work for that purpose.
Of course, cost of software and hardware is a factor too. But it’s not clear whether specialized tools should cost less (because they’re a subset of the general tool) or more (because by being specialized they deliver quite a bit more value to the end user.)
Reality as Context
Monica Schnitger hosted an analyst briefing entitled Data Capture Grows Up: Reality as Context. The increasing use of 3D scanning (see Google’s Project Tango) is generating reams of point cloud data. [Which to me is just a mesh without connectivity.] How about if we were able to use point clouds directly as geometry?
There are significant challenges to making this a reality including data set size and storage, feature (edge) detection, object detection (e.g. this is a door, that is an ear), and simplification via meshing.
PLM Component Technology
The fine folks at Siemens PLM Software hosted a technology suite to address trends that are pushing components like geometry kernels and constraint managers to new places. These trends are the cloud, mobile, additive manufacturing (leading to mass personalization), collaborative design, and STL and point cloud as geometry.
The first thing I learned in this session is that whenever Jon Hirschtick (Onshape) talks, people listen. In fact, when someone asks a question people turn first expectantly to Jon to see if he’ll answer. What he pointed out what that this new cloud/mobile landscape is quite different from what we have gotten used to doing (developing for primarily Windows PCs). As he put it, to build a cloud-based app you have to make a “staggering” number of decisions.
Just a few of the considerations here are whether the mobile device is for viewing only or authoring. Right now, the cost for computing in the cloud is high relative to storage. If that changes, so changes the business model.
Where Is Your Software?
In It’s 2020: Do You Know Where Your Software Is? analyst Chad Jackson addressed the issues of cloud-based software.
The move toward cloud-based applications is driven by benefits such as accessibility (anwhere, anytime by anyone), lower on-site computer hardware requirements, and robust disaster recovery.
Things are not all rosy, however. There are some sticky issues regarding data ownership and development and support of add-ins for cloud apps.
When asked whether everything will be in the cloud by 2020, several participants were adamant that yes, that would be the case. The “new generation” of users won’t understand the desktop.
This is all predicated by the assumption that bandwidth performance will grow faster than that of CPUs (which it is expected to do by a factor of 10).
George Dyson’s Keynote
Technology historian George Dyson’s keynote was so enthralling I took no notes. But he walked us through the development of modern computing including photographs of the actual researchers, their devices, and artifacts of their work such as logs.
To get an idea of what this was like, watch Dyson’s TED Talk from 2003.
We were all fortunate enough to receive a copy of his book, Turing’s Cathedral, which is on my to-read short list.
In Summary
COFES is one of those events where everyone else in the room is guaranteed to be smarter than you. I hope this brief summary has persuaded you of that fact.
You may have noticed that I’ve shared with you mostly the tender brownie from the middle of the pan. I ate all the chewy hallway bits myself.
What Others Thought
- Kenneth Wong’s COFES Prelude: Time to Break the Code to Rebuild It?
- Ralph Grabowski’s What I Learned at COFES 2014
- [Added 10 May 2014] Technology Resistance by Denis Morais
COFES 2015 will be held in Scottsdale, AZ on 16-19 April.
RSS - Posts
First, while I’m happy architects are considering fluid dynamics in their design, the streamlines shown on those shapes is very misleading from how the actual flow would behave on those shapes. Particularly, the smooth, attached flow around the circular shape grossly misrepresents the actual unsteady motion of the fluid that would occur at the Reynolds number around a building. We only need to look to the Tacoma Narrows Bridge collapse, which in fact had a much more aerodynamic shape than a circle yet suffered catastrophically. I point this out partly because the oscillating flow over blunt objects is an introductory topic in fluid dynamics and partly because of my OCD.
To the issue of cloud computing. My general the sense from articles, blogs, presentations, etc about cloud computing that it’s a really solution looking for a problem. And to a point, a reaction to the move of so many other application domains towards enabling a broader sense of mobility (as in people accessing information from any device anywhere). Which could be a good thing but is there really a “need” in the CAE world for this? Are companies and engineers really screaming for a solution? I’m not so sure. And if they are, it’s in a very different context than what everyone sees cloud computing currently to be. To me it feels like a “all the cool kids are doing it so we should too!”.
In my opinion, the #1 barrier to cloud computing is data security. This is particularly the case with the current mood about personal data security. Second is cost. This issue has been discussed by others more educated than I about cloud computing cost but the $/cpu cycle only makes up a small portion of the overall.
Unless there’s a dramatic change in how we envision utilizing it, I don’t see it gaining much traction. I’m all for increase our ability to be more efficient but it currently feels like a square peg in a round hole.
Let the Luddite comments begin.
What I hear is that people are anticipating a growth in computational simulation – more analyses and bigger analyses. Also, there is an assumption that more computing capacity will be needed to handle that increased simulation load. A further assumption is that this expanded computing capacity cannot be hosted locally by every organization that needs simulation. Therefore, the compute resources will be remote. Hence, the cloud.