Long Beach lessons: Simulate in 1D/3D, add test for better decisions

Queen Mary at dusk

What a difference a year makes! At the 2013 NX CAE Symposium, Siemens introduced its LMS acquisition to long-time NX CAE, NX Nastran and Femap users; this year, LMS’ products turned up in nearly every presentation, highlighting the important integration of 1D and 3D simulation with test and PLM. Rather than trying to explain what LMS’ portfolio brings to the table, Siemens let its customers explain how it all fits together for both model-based engineering and the enterprise as a whole.

Presenters spoke about their particular design challenges and integrating simulation earlier and more often into their processes, moving away from digitizing bend-and-break to actually exploring design alternatives early enough for them to have an impact on the final decisions. One company* highlighted the use of LMS Imagine.Lab Amesim in proposal writing, for very early what-if analyses. That’s an awesome idea: Why offer to do something that’s not economically or technically possible? Why spend a lot to figure this out, with more complicated 3D tools, if the contract might not ultimately be awarded to you?

Other speakers covered Model-Based System Engineering (MBSE), a concept which has an official definition but is still a fuzzy, ill-defined idea for many. Everyone can agree, though, that MBSE is a worthy approach, saving time and money and improving quality — if only we can get there. MBSE models are complex and take time to build. One speaker, therefore, suggested his company’s approach: building high-level models to keep “on the shelf”, ready for use when needed. His point: adding details and modifying models is faster than building them from scratch, leading to greater agility and customer responsiveness.

The return on investment in CAE is a function of this reuse, but also of increasing familiarity with the tools offset by the growing complexity of the systems being modeled; that’s too hard to quantify, so most organizations no longer seem to be trying. We used to hear a lot of “didn’t build and crash 5 prototypes so save thousands of dollars” — not this time. Today, CAE is used to be more creative, validate assumptions, reach new opportunities and do it all more quickly. Companies are setting performance targets for their products and then modeling and simulating until they get as close as possible, in a multidomain world that encompasses cost, function, safety, manufacturability and many other factors. These iterations enable engineering teams to make the best possible decisions by knowing the impact of any particular change, well before it’s made in the physical world.

Siemens PLM didn’t make any earthshaking announcements during the event, but did give a further glimpse into its simulation and test product strategy — and it’s just what you’d expect, given the breadth of the portfolio.  Jim Rusk, Sr. VP Product Engineering Software told us that we should think of NX as an innovation platform; design, simulation, mechatronics, manufacturing line design and so on, are all apps in that environment. NX 10, available in December, includes

  • multiphysics improvements such as structural/thermal interactions with 1-way & 2-way coupling
  • enhancements to let users solve coupled problems on the same mesh, with common element types, properties, boundary conditions, and solver controls
  • a new expression management system for describing complex boundary conditions, and
  • adaptive meshing for improved solution speed and accuracy

among many other things, like a touch interface — go here for a more comprehensive list.

Mr. Rusk laid out a timetable for the integration of NX and LMS tools, but was clear that this was not intended to exclude third-party solutions: “Openness is key, and has been part of our strategy all along. We have good, longstanding relationships with other [commercial] providers.”

Back to the Siemens portfolio. Siemens continues to bring LMS and NX together, and is devoting a lot of resources at areas like complex acoustics and fluid-structure interactions. You may recall that LMS bought Samcef just before Siemens bought LMS; in NX 10, Siemens introduces a new NX CAE environment for the Samcef solver to model composite delamination and calculate ply stresses due to vibration. Even better: integration between NX, Samcef and Fibersim, so that composite models designed in Fibersim can be analyzed in Samcef. The overall plan builds on the LMS Virtual Lab / NX CAE workflows introduced in 2013 (NX 9) and extends out to 2018 (NX 13), and includes the integration of acoustics, MBD, durability and NVH. Said Rusk, “when we’re done, we’ll have market leading capabilities.”

I think Siemens PLM already has market-leading capabilities in many areas, but they’re not yet well integrated where they cross brands. That’s only to be expected — LMS and Siemens co-existed but were not particularly closely aligned before the acquisition and it takes time to bring together such established tool sets. The goal, however, is clear: create a closed-loop, systems-driven product development environment, which defines, validates and tracks performance requirements to make sure that the product that hits the street/shelf/skies is what the customer wants.

One user said it better than I can: “No software will do everything. The ones to choose will work well together and be best at what they do.”  Siemens is serious about this, investing in best-of-breed solutions and knitting them together, while making the technology better, more capable and easier to use.

* Siemens PLM asked me not to identify which customer presenter said what. Take a look at the agenda to see, in general, who spoke.

Image above is of the Queen Mary, the venue for the 2014 Siemens CAE & Test Symposium. If you’re in the Long Beach, CA area, the Queen Mary is worth a visit. Much of it is open to the public for free but a $10 tour can make her come alive for you.

Note: Siemens graciously covered some of the expenses associated with my participation at the event but did not in any way influence the content of this post.

MSC’s Apex rising to simulation’s challenges

What do we most often hear, when talking to people about why they don’t do simulation, or don’t do more simulation? It’s usually one of more of three things:

  1. It’s too hard and experts are too expensive (or not needed constantly, so it’s difficult to justify having them on the payroll).
  2. It doesn’t fit naturally into the product development workflow, with simulation tasks having to branch off periodically because they take too long to be carried out in the main stream — but that means results come back on an earlier iteration and someone has to figure out how to apply them to the current design iteration.
  3. CAD models don’t translate well or easily to simulation; some fiddling is always needed. Even when CAD models are decent, they may share more information with supply chain partners than is desired, so a dumbed-down version has to be created specifically for simulation. Then that version has to be managed … Nightmare.

Yet we all agree simulation can save time and money, and create a more desirable and fit-for-purpose product. So what’s the holdup? Why the disconnect between what user companies want and developer companies deliver?

Part of the answer lies in the fact that the codes in use are, in general, decades old and developed in the time of punch cards and tape. The workflows around those products could only adapt to current user interface and compute trends to a limited extent; a major rethink is needed to move the industry forward.

Enter direct modeling for simulation. Now analysts, who aren’t CAD jockeys, could manipulate geometry, try out alternatives, perhaps even take part in the early stages of the design process. Take that a step further and improve the CAD/mesh/solve/examine workflow and you get to MSC’s new Apex product.

MSC Software, one of the granddaddies of simulation, essentially took itself out of the limelight a couple of years ago to rethink its product lines and refocus the company. CEO Dominic Gallello brought his Autodesk vibe of talking to customers about their projects, problems and bottlenecks to the task, meeting with analysts and managers around the world to discover what MSC could do better, differently, to fundamentally change the simulation workflow. Apex, the result of this exploration, has actually been on the market for a while, as customers took the product on shakeouts, suggested further improvements and new directions — and now it’s finally ready for broader public view.

Yesterday MSC announced that Apex Modeler is available now, and that Apex Structures should be out by the end of 2014. I’ve seen presentations and canned demos, so this is a very company-directed perspective, but here’s what I know:

  • Apex is visually different from other CAE solutions. This screenshot show simulation results (the structure’s natural frequencies):

Apex Structures 3

Notice how much screen space is devoted to the model. Icons and menu panes appear when needed, and the visuals are bold and unmistakable. This user interface is a critical part of MSC’s approach to making simulation easier. Apex’s UI includes video tutorials, workflow instructions and at-mouse guidance that MSC says makes Apex faster to learn and easier to use — the company says customers are seeing work completed in half a day that used to take 3-4 days, and that the typical user’s time to productivity is faster than with other methods.

  • Apex is more than a spiffy new UI. MSC’s research found that nearly 80% of the simulation task’s time was spent on going from CAD to mesh, a largely non-value-added activity. Apex’s meshing engine, the company says, can be 50 times faster, employing Parasolid-based direct modeling technology for geometry cleanup and idealization.
  • Pre-solve-post may be a thing of the past. Apex enables users to stay within the environment to maintain associativity between geometric and analytical data; geometry changes also edit the mesh — there is no separation between the two. MSC Nastran is integrated into Apex, so the user can validate models and run simulations from the same user interface. Rather than cleaning up a model, setting up the solver and then post-processing the results, it’s all in a single environment. I’m not sure about this, but it appears that you can’t create a model that can’t be solved — you no longer need to prep the model, run the job to see if there are errors, fix those, do it again; all of that iteration seems to be unnecessary with Apex.
  • MSC is building Apex around a concept I’ve seen before, though carried out far more elegantly here: solve only the parts of the model that have changed. They’ve devised what they call Computational Parts, which allow users to analyze an assembly and then run only incremental analyses to allow fast, agile what-if studies. In future releases, Computational Parts can also be idealized, meaning that sensitive intellectual property can be protected, with only the relevant part-to-part connecting data exposed.
  • Apex isn’t intended to replace any current MSC offering, but to run along side them. MSC sees Apex as the platform for applications yet to come, with the long-term goal of providing a multi-user, multi-discipline, and multi-fidelity workspace. Right now, computational parts and assemblies support linear representations; over time, this will grow to include thermal, non-linear, motion and other physics.
  • These first releases of Apex are targeted at MSC’s core automotive and aerospace markets, where Apex is complimentary to current workflows. Over time, MSC sees Apex expanding to new customer types, markets and sizes, where it will be a stand-alone solution.

It’s been interesting watching MSC try to keep the wraps on Apex while hiring people, trademarking names and submitting patents (16 filed so far). I’m excited that it’s finally here and we can start taking a closer look. Tune in to MSC’s Apex launch webcasts on Thursday, October 2, 2014 and Thursday, October 9 –register at www.mscapex.com– and tell me what you think.

Autodesk in the big leagues with Nastran

Remember a couple of months ago, when Autodesk acquired NEi Software and didn’t announce much of anything at all about it? They were trying out a new strategy, where they do the deal and work on the product and strategy and later go public with plans and announcements. Today was that day: the launch of Autodesk Nastran, a standalone version of what was NEi Nastran, and Autodesk Nastran In-CAD, the CAD-embedded version, along with a reveal of a bigger, broader strategy for simulation.

Autodesk Nastran IN-CAD

First, for (former) NEi customers: Autodesk Nastran is the version you’re used to, available for desktop and cloud solving. Autodesk Nastran In-CAD is what you may have known as NEi Works, the SolidWorks-embedded version — but now expanded to also include Inventor. (Apparently, the In-CAD project was already underway when the acquisition happened; NEi knew it needed to diversify beyond SolidWorks but couldn’t really call an Inventor version “Works”. Hence, In-CAD.) Much of the NEi team joined Autodesk and continues to work on expanding the product.

Both products are available for purchase as of today, and are priced at “around $10,000″ on a perpetual basis (no term subscriptions).

Then, for the bigger Autodesk community: If you’re a Simulation Mechanical customer, you may have noticed that you can choose a solver during your setup (and you’ve had that option since shortly after the NEi acquisition took place). If you’re comfortable with the traditional solver, by all means keep using it. Autodesk has no plans to make it go away. If you’re familiar with Nastran, try that. Your choice.

I still need to process a long day of briefings, but here are my main take-aways right now:

  • We can agree or disagree with how NEi’s customers were notified of the acquisition, but taking time to figure out how Nastran fits into Autodesk’s overall offering was smart. It’s better to come out later with a well-defined strategy than gleefully announce something half-baked.
  • Paraphrasing here, but Autodesk’s vision is to create simulation tools that are productive and personal, targeted to roles (designer, engineer, analyst specialist) without dumbing down or losing capability.
  • The simulation platform is intended to be open (multi-CAD and multi-CAE), flexible, desktop and cloud, mobile devices where appropriate — in other words, taking advantage of new technologies that competitors may have a harder time adopting because of legacy code design decisions.
  • Autodesk has been working on this strategy since the early 2000s, with the acquisition of PlassoTech, and has learned something with each integration. The people, channel and technology implications are managed more carefully.
  • Adding Nastran isn’t a competitor-killer. It’s a way to get Autodesk and its message of removing bottlenecks, simplifying workflows and creating targeted solutions to people who’ve heard of Nastran (or NEi) and want to add its proven capabilities to what they do, whether they’re currently using simulation or not.
  • The Nastran products are meant to augment current solutions, let people who don’t have access to the hardware or competitor software licenses do simulations (perhaps in the cloud), for a reasonable price.

Finally, Autodesk sees that the world is, fundamentally, multiphysics. And that has a lot of implications, such as needing the physics to model FEA, CFD, plastics, composites, linear and non-linear, dynamic and static, and so on. It also means creating licensing mechanisms that work under many different conditions, making exploration, optimization and design of experiments affordable in terms of software licensing and hardware utilization.

As part of the Nastran announcements, Autodesk also gave more details on Simulation Flex (fka Autodesk Sim 360 Pro with Local Solve — a truly awful name), which includes Simulation Mechanical (static stress, linear dynamic analysis and mechanical event simulations), CFD (fluid flow and thermal simulations) and Robot Structural Analysis (structural and AEC code checks). All of this, plus Moldflow, is in Simulation Moldflow Flex. 

The point here isn’t the laundry list of capabilities in each package. It’s the flexing between desktop and cloud. For an annual subscription, buyers get unlimited use of the products on their desktops as well as a set number of cloud credits that are drawn on for simulations in the cloud. Some things make more sense on the desktop, others make more sense to run remotely so that the desktop is available for current tasks. You decide.

One last thing. Nastran plays across many different verticals;, so don’t let the mechanical gizmos in the marketing materials distract you. The Nastran offering continues to be relevant for auto, aero, shipbuilding, architectural, civil and many other users.

Image of Autodesk Nastran In-CAD courtesy of Autodesk.

Note: Autodesk graciously covered some of the expenses associated with my participation in the briefings but did not in any way influence the content of this post.

Altair to acquire VisSim for embedded systems design

EarningsAltair today announced that it is acquiring Visual Solutions, makers of the VisSim graphical block diagram language for modeling and simulating complex dynamic systems, including embedded systems.

Visual Solutions says it has over 250,000 users worldwide, scientists and engineers in process control, aerospace, mechatronics, electric motor control, pulp and paper, nuclear, wind and hydro power, data communications, economics, HVAC, and biomedical applications, who rely on VisSim’s for real-time execution of high fidelity models.

As more users move towards modeling and simulating the controller along with the plant being controlled, and automatic generating the controller codes, technologies like VisSim become more critical. VisSim itself lets users define and simulating large-scale complex dynamic systems; VisSim add-ons extend this functionality to include real-time hardware-in-the-loop prototyping and control, C code generation, and the modeling, simulation, and building of embedded systems.

I did a bit of research into this area several years ago and learned that this workflow, from simulation through to code, is critical. Block diagram tools like VisSim (as well as LabView, MapleSim and MATLAB/Simulink, among others) let users drag and drop blocks into a work space and join them to create algorithms. These algorithms are simulated and, once everything meets requirements, used to generate C code for embedded chips. The magic is that users are simulating at the system level and can quickly debug before ever generating code for their control systems. Easier to create, faster to simulate, less error prone in execution.

Terms of the transaction were not disclosed, but it is expected to close by the end of July.

DS adds SIMPACK’s multi-body sim to its offering

EarningsDassault Systèmes today announced that it has acquired SIMPACK (formerly INTEC), developer of the SIMPACK general purpose multi-body simulation software that’s been used for decades to analyze the motion, coupling forces and stresses on mechanical and mechatronic systems.

SIMPACK AG is a cool little company that I’ve followed for 15 years. CEO and founder Alex Eichberger spun INTEC out of the German Aerospace Centre, where SIMPACK’s early development took place.  He and his team (including my old boss from Computervision, Bob Solomon) have grown SIMPACK globally, and outwards from aerospace into road and railway vehicles, engine simulation, wind turbine design, vehicle NVH and fatigue studies, and more. Today, SIMPACK says it has more than 130 customers, including many global brands. Transaction details were not disclosed.

Not surprising, SIMPACK will be operated under the SIMULIA brand. Dr. Eichberger said in the press release about the deal that SIMPACK and DS  “have similar technology-driven cultures and a shared long-term commitment to scientific excellence. We will join forces and technologies to provide superior virtual and physical experiences, expanding industry solution experiences for transportation & mobility, aerospace & defense, and other industries.”

My take: SIMULIA knows the gaps in its offering and is aggressively going after the best technologies to build out its portfolio. SIMPACK is solid technology whose broader adoption was hampered by the company’s size — it’s very hard for a relatively small German company to compete on a global scale with the likes of MSC Software and Altair, who have bigger budgets, presence and reach. At this level of sophistication, customers want local training and support, hard for a smaller company to provide. SIMULIA has done a good job with earlier acquisitions, integrating them into the fold where it make sense but recognizing the technical expertise that made them acquisition targets in the first place, and leaving them alone to do what they do best.

I have a call with SIMULIA to discuss this later today and will update after the call.  Nope. DS canceled the call because it’s in the quiet period preceding its earnings announcement. Don’t read anything sinister into this — they probably don’t want to run into trouble by foreshadowing the earnings. We’ll learn more soon enough.