11

This question is neither scientific nor technical but more career related.

I am at a junction in my professional life where I need to make a decision with regard to the future of my career. At the moment I am an R&D manager and my job is very management-oriented (supervising, problem identification) so I asked my own manager about changing my functionality to go on a more technically oriented quest.

Following my request, he asked me a simple question that made me understand how unaware I am about the computational science market. He told me: "OK, no problem, but tell me precisely, what is it that you wanna do?". It's a simple question, right? But it's so difficult for me to find an answer, not because I don't know what I would like to do, but because in my mind I have an image of what I want to do and I can't put it in words. I'll tell you why.

I have a master's degree in computational mechanics and a PhD in computational fracture mechanics. What I want to do must include modeling and simulation and programming at the same time. But I can't describe it this way to my manager, I have to be specific about the job in industrial jargon. I'll give you an example. If I tell him: "Hey, listen, I have a background in CFD (Computational Fluid Dynamics)" he would understand and he will easily find me a mission in that area.

I am aware of the "fourth industrial revolution" and new automation such as "automation in engineering simulation", so I am looking for careers in line with future demands. Is automation in engineering simulation a thing? What are the jobs for computational mechanics engineers in the context of industry 4.0?

P.S.

For those who might say: "Hey dude listen, what you are looking for is software development" I would reply that's right, but, because of my computational background, most software development teams remain skeptical about my software development skills, although I am a skilled programmer.

Anton Menshov
  • 8,672
  • 7
  • 38
  • 94
Dude
  • 580
  • 3
  • 14
  • 5
    I don't understand -- your boss knows perfectly what "computational fluid dynamics" as a job specialization means, but is completely lost when confronted with "computational mechanics"? Can't you just say "the same kind of things, but for solids and machine components rather than fluids"? – Federico Poloni Oct 17 '21 at 18:30
  • 3
    Also, it may be difficult for your boss to find a job in this area within your current company, but I don't think your field is a dead end. It's not like this "industry 4.0" revolution will make all previous jobs obsolete. Industries will still need to make planes and car engines and bridges. – Federico Poloni Oct 17 '21 at 18:34
  • 14
    What is ‘Industry 4.0’? Sounds like something Wired made up. – A rural reader Oct 17 '21 at 20:07
  • 1
    "...most of software development teams remain skeptical with regard to my software development skills, although I am an skilled programmer." Perhaps one way to get around this is to make numerous contributions to open source CFD packages. – NNN Oct 18 '21 at 08:43
  • 3
    @Aruralreader: See https://en.wikipedia.org/wiki/Fourth_Industrial_Revolution - the term appears to have been introduced in Germany. Also called "smart industry" in the Netherlands. – J W Oct 18 '21 at 19:26
  • 1
    @JW: Thank you, I learned something new today. – A rural reader Oct 18 '21 at 20:42
  • You mention changing your "functionality". Do you mean changing your duties/focus within your current position or do you mean finding a new position, be it at this company or elsewhere? – J W Oct 19 '21 at 14:10
  • 1
    The whole "Industry 4.0" thing is about merging machines with databases so that manufacturing can do itself mostly all-by-itself. It's about getting machines to talk to other machines so that they can work together without humans. It's got nothing to do with CFD unless a fluid model is something that can help decide what a machine should do while it's making something. Probably the best angle in that direction is chemical engineering? Fracture mechanics won't really help much because little in manufacturing should be fracturing by design as part of the manufacturing process. – J... Oct 19 '21 at 15:52
  • This looks like it would depend on what the company does. What can 4.0 do for your boss and/or company? – Yakk - Adam Nevraumont Oct 19 '21 at 19:32
  • I doubt but most of my coursemates from computational phys/math backgrounds ended up as software devs unrelated to CFD/modelling/simulation fields :( – 2Napasa Oct 19 '21 at 21:34

4 Answers4

16

The number of places where you can get paid a high salary for computation mechanics algorithm and software development can be counted in two fingers of the hand.

That includes a few commercial software companies (Ansys, Dassault, Siemens, Comsol, etc.) and a few research labs (LANL/LLNL/Sandia etc., EDF, ..).

The rest of computational mechanics software is developed by low-paid students and post-docs at universities. These people are concerned mostly with speed of completion rather than long-term usability of the software.

In my opinion, the best way to continue doing software development in the field is to identify problems in your industry that cannot be easily solved with computational tools and try to create in-house tools that solve those problems. It has to be a labor of love which may be appreciated after a long period of time (or may not be).

Biswajit Banerjee
  • 942
  • 5
  • 16
  • 3
    I'm pretty sure that there's a third theoretical possibility, but it's not one that's eventuated yet: people interested in building an AI capable of a hard-takeoff singularity. – nick012000 Oct 18 '21 at 08:55
  • 6
    "The number of places where you can get paid a high salary for computation mechanics algorithm and software development" -- that's just plain wrong. These jobs exist in every company with 1,000 or more employees that are actually producing physical objects, and there are a very large number of such companies. – Wolfgang Bangerth Oct 18 '21 at 20:24
  • 1
    @nick012000: Having tried to use machine learning in mechanics, I have a reasonably good feel for the limitations of known approaches. Since current AI is based on machine learning, I doubt that the singularity is anywhere near as far as computational mechanics is involved. However, computational mechanics is used extensively in visual effects and, more recently, in virtual/augmented reality. But these fields are dominated by CS graduates and a mechanics person will run into the "software development" issue the OP mentioned. – Biswajit Banerjee Oct 18 '21 at 20:29
  • @WolfgangBangerth: Not in my experience, and I've been in the field for more than 30 years in three different countries. They use commercial computational mechanics software almost exclusively today. Would you consider John Deere, most car manufacturers, Caterpillar large enough? – Biswajit Banerjee Oct 18 '21 at 20:33
  • 1
    A good way of finding out the trends in the industry is to look at Frost and Sullivan market reports (they are expensive but if your institution has a subscription, they are informative). Another way is to scrape the web (and web archives) for job listings with terms relating to computational mechanics. You can then run a few simple machine learning algorithms on those to extract trends. – Biswajit Banerjee Oct 18 '21 at 20:47
  • 1
    @BiswajitBanerjee Yes, John Deere and Caterpillar are certainly large enough. But using commercial software and code development are not mutually exclusive. You can use Abaqus but if you want to use a material that isn't in their database yet, you have to program a material law routine that describes, for example, the stress-strain relationship. And once you've done your simulations, you still have to do postprocessing and evaluation, and that often involves writing Python scripts or loops around the simulation. There is room for both commercial software and software development. – Wolfgang Bangerth Oct 18 '21 at 21:44
  • @WolfgangBangerth: Material model development and implementation are rare outside of academia and typically subcontracted out to academia is a large company needs it. I've been in the business and found that it's hard to earn a regular living from that sort of activity. I don't think scripting is what the OP was thinking of, but I may be wrong. – Biswajit Banerjee Oct 18 '21 at 23:54
  • 3
    Are you sure you don't mean "on the fingers of two hands" in your first sentence? – Federico Poloni Oct 19 '21 at 18:44
  • @FedericoPoloni: A quick search for "computational mechanics" in job search engines will give you an idea of how many fingers will be needed. Keep in mind that most of these jobs are actually analysis jobs. – Biswajit Banerjee Oct 19 '21 at 19:49
6

There are many many technical jobs in computational science and engineering that pay well and are fulfilling. Whether you call it Industry 4.0 or something else, the fact of the matter is that computational tools (simulation, optimization, characterization) are widely used in industry to design better products. For example, diaper companies use computational fluid dynamics to better understand how their products absorb fluids. The oil and gas industry simulates fluid flows in the subsurface, as well as in their "downstream" areas (after stuff comes out of a well). The aerospace industry uses CFD. Speaker companies simulate speaker frequency responses. Car companies simulate crashes. Tire companies simulate tire behavior. You can pretty much name the product and find someone who does computer simulations for it, and uses these simulations to optimize the product to have something that is better than what the competition does. There is literally no end to applications in computational science and engineering (CS&E). If you are interested, you might want to look through the program of the SIAM CSE conferences over the last few years; some of the keynote talks may be available to stream online.

How that relates to your specific predicament is a different question. What I read most from your question is that over the years you've been in management, you lost touch with the CS&E community doing computer simulations a bit, and are no longer fluid in what is being done in actual practice. There is of course also the question of what your company is doing and how you moving into a practitioner role would actually help your company. These are questions only you can answer.

Wolfgang Bangerth
  • 55,373
  • 59
  • 119
  • 2
    This. The way for the Dude to tick most of the stated boxes (... to include modeling and simulation and programming at the same time.) would be to find a place a research-heavy establishment (research lab, R&D of some company, etc.), which uses in the best case many open source or in-house tools. – Dohn Joe Oct 19 '21 at 11:45
  • @njuffa I tacitly subsumed "computational fluid dynamics" under "computational mechanics". There is substantial overlap in applications (fluid structure interactions, for example, but in general the techniques aren't so different). – Wolfgang Bangerth Oct 19 '21 at 21:16
2

I am a machine learning PhD student and only marginally had contact with the fields of your question. I understand industry 4.0 as an automation of the industrial process through digitalizing it and using the acquired data to make processes smart, where smart is for me quite equivalent to using some form of machine learning (e.g. predictive maintenance).

Therefore, I could see a computational mechanics guy programming simulation models in a digital twin of the industrial process. I can also picture a computational mechanics guy working on machine learning applications like predictive maintenance for some mechanical processes which require a lot of physical/mechanical knowledge and constraints. I can also see a computational mechanics guy working on automating mechanical simulations buy designing and training deep neural networks (to be aware of all the physical/mechanical constraints) to output accurate simulations. Deep learning models have the potential to output simulations multiple orders of magnitudes faster than actually simulating stuff in the traditional way through solving differential equations.

saper0
  • 129
  • 2
0

I just want to add one aspect using the jargon of the question.

If you are an in-house developer, then the software is not production in regards to the company. So e.g. if the company sells hammers, then the value of the software is based on how much it helps to sell hammers. E.g. to pay salary to people that would quit and not produce hammers if they don't get payed.

And in the same way with the 4th industrial revolution then I think it is beneficial to think about how you help a company with its core business with the software you produce, instead of a narrow focus on the software itself.

Thomas Koelle
  • 109
  • 2