Models, Material Properties, Loads and Constraints
I put this page together for the people who are new to FEA. My intention here is to give you a basic overview of what I will need from you to run your analysis. There are four inputs that need to be establish before you can run an FEA. These are the Model, Material properties, Constraints and loading. All of these need to be accurate to get accurate results from an analysis. This page covers the basic details of all four to help with any questions you might have.
The model is your part and can be any shape, size or configuration. There are several things that you can do to make the most of FEA. First, use the extremes of you manufacturing tolerances. Use the thinnest sections with the largest hole diameters and the most offset from true location. These would be your worst case scenarios and would show the true minimum strength of you part. Most people however, run their parts at the mean dimension and in most cases this is fine but it will not give you the true weakest part that you will be producing.
Second, if you are analyzing a part with multiple components, remove all gaps and over laps. The only time you should have a gap or overlap is when you are actually trying to model these things. Otherwise, these will cause problems with your analysis and I will send your model back to have these corrected.
Third, remove unnecessary features such as internal and external threads, raised or indented lettering and holes or bosses that do not directly effect the part. This will reduce the mesh complexity and reduce the time it take for the analysis to run.
Material properties include Young’s Modulus, Poisson’s Ratio, yield stress, ultimate stress, density and thermal expansion coefficient to name a few. Depending on what you are analyzing you could need all of these and more. However, most analysis will only need a few of these. A good place to find material properties is MatWeb. These are generic material properties and in most cases these will work. For analysis work where the maximum stress is expected to be below the yield point, a linear analysis is appropriate. This will use only the Young’s Modulus and Poisson’s Ratio and will assume that the linear stress - strain relationship will continue forever. This linear analysis will give false high stresses in locations of stress concentrations or areas that exceed the yield point.
The next type of material input is Bilinear. This will use Young’s modulus and Poisson’s Ratio along with the material yield strength, ultimate strength and ultimate strain. This information is usually given as Engineering stress - strain and I will convert it to True stress-strain. For more information on True stress-strain read this <article>. This material data is converted in to a curve with two line segments. The first part is the linear region up to the yield point and the second segment is from the yield point out to the true ultimate strength. This allows the software to account for the change in stiffness after yielding. For most analysis that involves local yielding this type of analysis will be accurate enough.
The last type of material input is multi-linear. This will be made up of a number of line segments that approximate the actual stress-strain curve generated from a tensile test. These points are entered as true stress, true strain points that are above the yield point of the material. This is obviously the most accurate of the material models however it is seldom used because it requires actual lab testing of your materials.
Material Density and the Coefficient of thermal expansion are used if you want gravity loads included or if you want a thermal stress or weld pre-stress on your part. Here again, use the minimum properties that you are likely to encounter.
Loads are what generate the stresses in you model. These are typically forces, pressures or torque applied to the surface at a point, edge or surface. Most of the time these can be used to accurately load and analysis you part. However, sometimes it is necessary to apply the loading through a second body using body contact such as a pin loading in a hole. I will advise you when I think this would be appropriate.
Forces must have a direction and magnitude associated with them. This will be in the XYZ coordinate frame that your model is constructed in. My software will use your Global Coordinate System for its reference directions so you can describe your load in terms of these directions. Both magnitudes and directions can be either positive or negative so a load of -500 lbs. in the -X direction will be the same as 500 lbs in the X direction when it is applied. As the model deforms, The direction and magnitude of the applied force remain constant. Loads can also be specified as a vector in an off axis direction using a combination of loads in the XYZ directions. Please specify units with your loads
Pressure loads will always act perpendicular to the surface they are acting on. These accurately simulate gas and liquid pressure on a model. As with loads, pressure can be either positive or negative but direction of load will always be perpendicular to the surface. Please specify units with your pressure loads.
Torque loads can be applied to the inside or outside of any cylindrical surface and can be either positive or negative. Torque is specified by an axis and magnitude such as 200 in-lb about the X-axis. Use the right hand rule to determine the sign of the magnitude. Like the Force, Torque can be specified as a combination or torques along the X, Y or Z axis.
Displacement can also be specified for any point on your model. If you know the maximum deflection of a part, this can be applied to that point to simulate the stresses at that know deflection.
Constraints are used to hold your model in place during an analysis. Without this, your part would just move around and would be impossible to analyze. The basic supports are Fixed, Displacement, Frictionless, Compression only and Cylindrical. A properly constrained model will converge faster and will produce more accurate results.
Fixed Supports constrain a surface from moving or rotating in any direction. These are usually applied where a component is mounted. Any point on that surface is not allowed to move. This is great for most situations.
Displacement can be used to limit the translation or rotation of a surface to a specific plane.
Frictionless supports act as a smooth, hard surface that allow a surface to slide and rotate in two directions but prevents the surface from separating from the plane or passing beyond it. This type of support can also simulate a plane of symmetry on a part. If you have a model that is symmetric about one or more planes then it is more economical to analyze only a portion of you part and use frictionless supports on all surfaces of symmetry.
Compression Only supports are used when you expect a portion of a surface to deflect away from its original position. This is used for modeling bolted connections where parts of a flange might separate from its mating surface. A Fixed support will give false results in this case.
Cylindrical Supports on a shaft act like a bushing and in a hole they act like a shaft. DOF for this constraint are radial, rotational and axial. Any of these can be fixed or free allowing movement in that direction.