Previously, we spoke about the matter of meshing when it comes to 3D printed parts. We will now tackle the next challenge: the conditioning of the part, or as fellow analysis consultants use for a less pejorative effect: “boundary conditions modeling.”
When we last spoke about FEA applied on 3D printed parts, we left at the point where our 3D CAD is a free floating meshed beauty with no constraints in the wonderful void of CAD software. But when it comes to modeling, the meshed part will need to be modeled after the most comprehensive conditions it faces throughout its working cycle.
What is the work environment of your part?
Remember when we spoke extensively about “getting your input data straight” and “knowing your part”? This advice will come in handy and will be stated here for the THIRD time (because I think it’s that important):
Know where your part is working, and how it works, and how it reacts with the whole environment.
Are you fixing your part to a shaking mechanism?
It’s going to vibrate.
Is your part close to a heat release source?
It’s going to pack heat.
Is your part moving or part of a motion generating system?
It will endure strong and/or repeated impacts.
There is just so much that your part can be going through and I’m not even going to list “earthquake” and “biohazard.” In short, virtually speaking, your 3D printed part’s environment and working conditions can include just about anything. But realistically speaking, you can’t protect your 3D printed part from everything. I always find it puzzling that I need to remind my customers that “the part will fail fatefully, from a known or unknown reason.” Corporate doesn’t really like it when I say that to customers, I should mention. Still, I don’t mind hurting someone’s feelings if they think their parts are indestructible. They are not.
So yes, know where your 3D printed part will be working, and how it will be working. And if it’s an assembly, even a part with some intricate flanges, latches, and tunnels, be aware of how flow takes place within it and how the parts connecting to it behave. This will help you narrow the models you will use and the studies you will pursue when you first start out.
Afterwards of course, according to results, maybe you will have to add a study case that wasn’t deemed necessary in the first place, or maybe you will find out that one wasn’t necessary. But all in all, you should never cross a modeling study case that proves relevant through technological reasoning unless you already have iterations of results that prove it’s obsolete.
This is how you play it safe with designs. Your design is fated to wither (we’re all going to die, including your design!). But then again, it’s no excuse to not take the work as thoroughly as possible, and to not use all the means at your disposal to ensure a hypothetical number with an error range that would make your part hypothetically (give or take) reliable.
What are the study cases to pursue?
The ease of answering this question will be proportionate to the efforts you’ve put in answering the first query. Granted, it might still depend on some other parameters, but those parameters are a matter of previous models, benchmarking, and R&D statistics. And they prove minor when you’re not working on aerospace or not having to go through a long process of certifying your part and meeting a security coefficient. It’s through experience and extensive coverage of data concerning older parts that your study cases can be “pointier.”
But meanwhile, you will achieve about 80% of accuracy when you know your part’s environment and can therefore state that it will require “static/dynamic” study, “vibration/noise/thermal/electric” one and so on…
In my personal case, my parts have gone through at least three types of analysis. The primary ones were static analysis, dynamic analysis, and thermal analysis.
Static analysis is placing your 3D printed under the hypothesis of its fixation while being subjected to the charges it will have to support: inner load, contacts, thermal data, charges, displacements…etc. This allows a first insight to see if your 3D printed part will even “stand to merely exist” under the weight of your expectations.
Oh, you thought the question mark feature had enough thickness to at least support itself? Boom! It’s subject to deformation.
What about the fact that your part will be in 100°C environment for at least 3 hours a day? You’ll have to dial it to half an hour, because after two hours, your part is starting to lose it.
You wanted your part to support this part here, that part here, and the other one through this latch? Tough luck, choose only one.
And what do you mean by placing a wiring hole in this spot? It’s making the part weak just by itself, I can’t imagine wiring going through it to top it off!
So first things first: go for static analysis. It will make you weep over your creativity, over your CAD model, and weep some more when you’ll realize you might have to mesh it over again. And if your outcome from the first iteration was sweet, well….take a picture with it and send it to me so that I know hope is still out there.
Dynamic analysis is for direct and indirect motion, although the first one requires particular modeling features (given a velocity equation or an excitation). It covers several purposes according to the working part and the needs of analysis. Amongst them, it helps in checking if the contacts of your 3D printed part will hold steadfast, and if your part will require vibration analysis (see Campbell diagrams). It gives a first insight as well as a model on how your part will actually behave in the physical reality and the order of failure (See Goodman modeling).
Finally, dynamic analysis is basically the pointer to other types of motion analysis and failure/wear ones. By this stage, you will understand how following one model of hypothesis over another on can give a different result and will therefore need to keep a tight memory and understanding on the type you’re using, its limitations and key-figures. Usually, the customer might have a specific standard of his/her own, but it’s always better to understand where you’re putting your foot in.
In case your part vibrates, a vibration analysis is compulsory. Yes, vibrating 3D printed parts can be funny up until a one micron crack spreads through your artwork in few days and you won’t understand the cause (or the noise pollution will drive you out of your workshop but at least then you will understand the cause). If your part belongs to a system, you’d better not let it have a common mode with the rest of the system. Resonance might be a goal for some practices, but in mechanical design, you avoid it like it’s the plague.
If your part is prone to high thermal stresses, it's kind of important to understand the virtual outcome BEFORE the actual outcome! Thermal analysis can be a broad study case, however when it comes to 3D printed parts it should be given very close attention. It’s partly because of the nature of materials used in 3D printing and partly because of how they can react in working conditions if the environment is a little bit endothermic.
There are so many other types of analyses, each implying a set of hypothesis and each requiring certain input data and/or certain compute power. When certification terms or customer requirements happen to specify some, you’d think it would make things easier. It might, but I myself have never faced such a scenario. It either results in me iterating, modifying, and reiterating to get the fixed results, or going around and performing other analysis because that one showed the need for another one and on and on.
For example, one time my static analysis showed a disagreeable flow of heat (where there wasn’t supposed to be one) and a certain concentration of stress in a regular surface as a result. It turned out I had to go through thermal analysis because the plastic my 3D printed part was made of can pack stress from heat. I just hope my customer gave my observations on materials to their materials guys. And I sure hope they’re not as grumpy as they sound in my head.
What about conditions?
This part is either the “HURRAY” part because my customer gives me the excel files or a 3D mockup/lookalike/similar/old version of the 3D printed part filled with the numbers that I need to inject to my model, or it's the “NOOOO” part where my customer gives me a pack of 2D pieces from which I’m supposed to extract data. In analysis, inputting data in is doubly unfun – it’s boring AND requires a lot of focus.
The more conditions and nodes, the more focus needed to not mess up numbers or screw up the modeling with double conditions somewhere. So, there you go: your part will have to support a 6 pound load, you will have to know the force it will deliver and chop it on the three-dimensional axis. You’ll also need to know where and how to exactly to apply that force. Your part will be fixed on a support, so you will have to know how and therefore what the best way to model it is. Rigid contacts or relative ones? Actual bolted assembly or modeled one with RBE lines?
Moreover, when it comes to thermal data, it’s so complex that it requires it’s own modelers. Those thermal designers will deliver a rough meshed model of your own with packed data on temperature at every step of your calculation and more, but you will have to perform the delicate task of applying that mesh to your own. And I assure you, when it comes to 3D on 3D thermal/mechanical meshes, I’ve had my fair share of losses and iterations. It helps to have your data neatly organized in excel and to go through them step by step – all the while saving at each step so you can retrace your steps in case your messed up.
The one thing that helps me not get something messed up is the fact that I have to give my customer an extensive report on the conditions, how I modeled them, why I modeled them that way, and screenshots showing that I did put the values written in files in my models. While building this report along with the model, I’m able to see when I messed up and go through it again. It takes time and repetition, but luckily enough, no one prepared me to a linear one-shot kind of work when I stepped into the mechanical design realm. Professors and senior engineers always depicted it as going back and forth, reviewing through working and being psychologically prepared to repeat it all again.
What about material?
You forgot to put the specific features of your materials in your part’s history! No need to go further! Even if you insert your material’s common numbers, you have to create the personalized features and specifics that 3D printing implies. With a 3D printed part, the heat, elasticity and void characterizations are very important and will most likely (due to the lack of demand for them or their uniqueness) not be included in the default menu. You will have to seek to add those options and quantify some of them with your customer, or materials provider.
When it comes to applying a material on a part, you will have to pay attention to hypothesis you endow every kind of elements with. 2D elements that serve as skin and are regarded as degree 2 elements, 2D elements that are regarded as degree 1 elements, quality of 3D elements…etc.
The meshing data needs hypothesis that the material’s own hypothesis can be applied to. Keep that in mind and be prepared to review your mesh because reviewing the material (at least at your level) won’t prove possible. Just be prepared once more to reiterate.
Be thorough, be brave and be psychologically prepared!
You will notice that none of my advice was technical. It’s easy to seek information from more senior engineers or to just know when you won’t be able to find one. It’s easy to learn to use a certain piece of software and to be aware of how to use your theoretical knowledge. Of course, it doesn’t seem that easy to my friends majoring in architecture and management, or to my dad who happens to be a brilliant accountant.
It’s easier, so much easier, to actually know how to manage the stress of knowing you will have to redo a lot of things all over again. It’s easier to just roll with the last iteration than to fight the appeal. It’s easier for me to just crack open Shigley’s than to have my stressed out project manager stress ME out because I’m going through my twenty seventh iteration (and my eighth one in the conditioning with only a few days left to run results, hope for results, and maybe get to dissect them).
So pay attention to your conditions and I promise that you will suffer less when you process your data. It’s also pretty great to earn the right to state very bluntly that you've injected the right numbers and if it doesn’t converge, it’s the solver’s fault.
I say you will suffer less but you should probably brace yourself anyway, you may have to figure out why why your convergence didn’t converge or why your model started deforming wildly when it wasn’t supposed to. Or was it supposed to? Maybe your customer doesn’t know either and he just wants good results and can’t bear that his printed part is at fault or maybe, just maybe…you forgot some set of conditions of hypothesis somewhere! Who said analysis was easy anyway?
We're making professional 3D printing easy and accessible
There's a new solution coming your way this summer. Click here or on the graphic to the left to request updates and beta access.