One engineer’s observation about the world of simulation engineering beyond the folk tales of “oh its really easy”
The FEA Quiz Revisited
The basic challenge that every simulation engineer faces is how do we tell the difference between a cartoon and that of a simulation. As a friend of mine has remarked, I can make you a cartoon very inexpensively but an accurate simulation will cost you. It is a tough game since engineering is a blend of getting the job done on schedule and on budget and how does one decide that a simulation is good enough? I have no answer since it is the responsibility and the reputation of the simulation engineer to decide whether or not that the answer is good enough and it is a weighty responsibility. Combine this responsibility with the at times tiresome nature of model construction and checkout, and the pool of engineers willing to become simulation engineers is not exactly large. Which leads to another comment by a friend: “Many engineers think they might want to be simulation engineers.” But of course, once they start slogging thru the modeling process and get their asses kicked via bad simulations, they realize that maybe it ain’t the dream job that they had envisioned!
How to Hire Simulation Engineers
Get lucky has been my path toward success. I get the feeling at times that gamblers in Las Vegas have better luck than I do in hiring simulation engineers, but so it goes. What I have learned is that a good simulation engineer should be curious and has a thirst to understand the underlying theory of the tool that they are using. One screening tool that we use is to ask a few FEA questions. Most of our candidates fail miserably in answering these questions but that is part of the fun. If their reaction is to dig in and learn more, then all good and if not, one quickly knows that maybe it won’t be a good fit. I’ll throw out the questions and provide the answers at the very end. BTW, I have a simpler FEA Quiz at the Predictive Engineering web site but I want this quiz to be for more of the full-time dedicated simulation engineer. This quiz is more for an advanced simulation engineer or at least someone who does or aspires to do nonlinear and explicit work.
What One Should Really Know:
Question #1- Where are stresses calculated in an isoparametric element?
At the integration points.
Question #2 - In a perfect world, what aspect ratio would all your isoparametric elements (shell and solid) have?
They would all be unity since a perfect square or solid provides the most accurate area or volume calculation and thereby stress calculation.
Question #3 - Okay – let’s have a softball question. If you have square beam under simply- supported bending, the von Mises stress will be negative on the compression side of the beam. True or False?
Question #4 - For linear elastic stress analysis using a single material, if the load is displacement based, will the stresses double if the elastic modulus is doubled? And its buddy question, if the load is force based, will the stresses double if the elastic modulus is doubled?
Yes and No
What Would be Nice to Know for Advanced Simulation Work:
Question #5 - Why is a complete constraint set (6 DOF) necessary for a basic, run-of-the-mill static analysis to work? And its corollary, why is a constraint set not necessary for a transient analysis?
If we start with the EOM as ma+cv+ku = F(t), then a static analysis is solving ku = F(t) and without constraints it is unstable where as with acceleration (transient) it is stable. One could phrase the answer a number of different ways but that is the gist.
Question #6 - For a nonlinear elastic-plastic stress analysis of aluminum and steel rolled products, to what approximate level of plastic strain would it be appropriate to use an engineering stress versus strain curves as compared to the full-blown true stress versus strain curve?
Roughly 5% there is very little difference in the two curves.
Question #7- For a large strain plasticity analysis of common low carbon steel with good toughness, why is stress usually not a good measure of the severity of the deformation process?
The stress vs strain curve (whether engr or true) for low carbon steel is rather flat and thus one will see little changes in stress as versus large changes in plastic strain as the material deforms. Hence, estimating how close you are to failure is best done contouring the plastic strain in the component.
Question #8 - For an explicit analysis using common engineering materials and nominal impact or crash velocities, what two fundamental items determines the explicit time step?
Speed of sound in the material (Elastic modulus and density) and the element shape or size.
Review more case studies, services and training options at