Skip navigation
All Places > Simulation > Blog > Author: Joe Galliera
1 2 Previous Next


24 Posts authored by: Joe Galliera Employee

I thought I'd just post some additions to inform you all of the Simulation software that the developers have snuck into SP's of 2012.  Thanks to my friends and colleagues, Delphine G. & Omar Z. for providing me with these details.


I think you'll agree that I always look forward to the What's New every year.  What's even cooler is when you get gifts like these after the initial release.  It's like Xmas in July!


Simulation 2012 SP2 (not verified yet by yours truly, um just simply not enough time in the day)

  1. In NL, plot reaction or contact forces vs time
  2. In NL, select a time value instead of a step to view results
  3. In NL, plot contact contours, not just vectors
  4. In NL, animate contact contour and vector plots


Flow Sim 2012 SP1 (same disclaimer as above)

  1. Improved solution-adaptive meshing technology (SP1). The improved refinement technology improves the quality of the computational mesh in the high-gradient flow regions in case solution-adaptive refinement is involved.


Flow Sim 2012 SP2 (ditto)

  1. Chinese UI and Help.
  2. Improved fan model. The new model offers better convergence and less oscillations at the beginning of the calculation.
  3. Record video. Now you can record screen activities as an animation file
  4. Improved definition of Electrical Condition. Now the specified electrical resistance is applied to the exact area of contacts between conductive solids.
  5. Export materials and connection to external databases. Now you can export desired material from Engineering Database to XML files.
  6. Tutorial for Tracer Study


Copyright © 2012 Dassault Systèmes SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without the written consent of Dassault Systèmes SolidWorks Corp.

I was working with a new Simulation Professional customer recently on getting his frequency analysis model working right.  He knew what the frequency should be because they had performed a physical test and he could hear the frequency it was generating; being a musician himself he knew that it was just above a D-sharp, or 311 Hz.  It ends up that his model had some rigid-body modes that we took care of and got the 1st natural frequency right where he thought.  (Note: A rigid-body mode, or RBM, occurs when a model is not fully constrained in one (or many) of the translational or rotational degrees-of-freedom.  When animated a RBM moves the body back and forth in that direction as a single rigid unit and has a frequency of zero.)


I found the following video, which is interesting because it shows the strings of a double bass vibrate, so you can literally see what I meant above.  The person who shot the video footage made these notes: “Frequency of the bass strings and high shutter speed of the camera lead to this surprising string-wobble footage.  There is no slow-motion applied to the take.  Sound is original.  Video was filmed with a Canon 5D MarkII , Nikon 50mm lens on 1,8f.”


Watch a WMV of the video here: (26.1 MB, WMV file)

Download a ZIP of the video here: (25.1 MB, ZIP file)



Another way to see frequencies is to throw sand on say a flat plate and see where the sand accumulates in the areas where the plate is not vibrating.

Watch the original video with sound here: (20.5 MB, MPEG file)

Download a ZIP of it here: (19.7 MB, ZIP file)


For comparison to a frequency analysis done with our Simulation Professional package, I recreated the flat plate model restrained at the center and interspersed the original video with images of the results from the simulation.  This is a good video to show how well Simulation can match real-world results.  I took out the sound because it is a little annoying to the ears at higher frequencies.  To create the images in Simulation, I changed the color legend to greyscale and inverted them so that the low spots would be white like the sand.

Watch the comparison video here: (4.51 MB, WMV file)

Download a ZIP of it here: (4.43 MB, ZIP file)


Copyright © 2011 Dassault Systèmes SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without the written consent of Dassault Systèmes SolidWorks Corp.

I thought it might be a good idea to provide some basic definitions of terms that are used in analysis that are good to know to be able to communicate intelligently about Simulation.  These are not meant to be definitive technical definitions but more fundamental knowledge of these terms (i.e. they are mostly coming off the top of my head as I'm typing).  If you believe that I am off on any of these, please let me know.



Now that we learned about different stresses in the previous 2 sets, I'm going to explain some ways that they can be used to determine failure.  There are many types of mechanical failure, such as: yielding, fatigue, buckling, thermal stress, impact and creep; although there are also some other types of mechanical failure that cannot be predicted well by computational methods, such as corrosion and wear, so these will not be discussed.  In addition to something fracturing or breaking, a failure could also have said to occur if the body deforms or displaces more than its tolerance.


19. fatigue - occurs when an object is subjected to repeated loading and unloading that cause microscopic cracks to form initially on the surface and after spending about 95% of its time as a very small crack it will reach a critical size in which it will suddenly and violently fracture with the ability to cut through very thick structures (this is because the stress concentration factor of the growing crack goes exponential).  Fatigue can also sometimes occur due to internal defects in the material crystalline structure.  Tell-tale signs of a fatigue fracture are both dark and light areas showing slow crack growth areas followed by areas of granular sudden fracture; other visual features are from the crack growth striations called "beach marks" that look like sand washing up onto a beach.


20. high-cycle fatigue - greater than 1000 loading cycles and upwards of 10 million.  Max stresses are typically less than yield.  Tensile and shear stresses are important for high-cycle fatigue so focus is placed on keeping them low, and the Fatigue module in SW Simulation Professional can help determine the amount of damage a part undergoes by comparing the alternating stresses to an S-N material curve.  Cumulative damage of greater than or equal to 1 or 100% predicts fatigue failure.


21. S-N curve - characteristic material curve for fatigue (S for stress, and N for number of cycles) which is obtained by cyclically loading a sample material coupon and counting the number of cycles to failure; many samples are needed to generate a good curve because cycles to failure is not consistent.  Surfaces of coupons are typically well polished to reduce the surface microcracks, so this should be taken into consideration when comparing to actual design surface.  It's frequently hard to obtain a S-N curve for a given material (because of the large number of samples needed and expensive time for testing), except for well studied materials such as metals used in the aerospace field that has a published handbook of material curves formerly called MIL-HDBK-5 and is now called MMPDS handbook (Metallic Material Properties Development and Standardization).  ASM International publishes an Atlas of Fatigue Curves that SW Simulation uses for its material library and are denoted by (SN) at the suffix of the material name.  A common practice is to find a similar material and scale the values based on elastic moduli.


22. low-cycle fatigue - cycles less than 1000 and typically stresses go beyond a material's yield strength into plasticity.  Strain-based methods are a used for low-cycle fatigue prediction, and SW Simulation does not have a good solution to handle this type of fatigue well.


23. yielding - typically for ductile materials where the stress goes beyond the yield strength of the material.  You can also look at where strain goes beyond 0.2% or 0.002.


24. buckling - typically a slender body under a compressive stress (look at 3rd principal stress) can lead to buckling, where the body's transverse direction stiffness goes to zero causing a sudden and dramatic fracture.  The causes of buckling can be compounded by off-axis loading, geometry or material non-uniformity, and/or elevated temperatures.  Linear buckling (based on characteristic mode shapes from resonances) is handled by SW Simulation Professional where the value calculated is the BLF, buckling load factor, which is a ratio of critical buckling load to the applied load; linear buckling is a very non-conservative method meaning that you may get BLF values much higher than what happens in reality.  A better test of buckling is to run a nonlinear analysis where stress softening and geometry deformation is taken into consideration.


25. thermal stress - stresses imparted on a body due to the expansion (heating up) or contraction (cooling down) of a material, especially where bodies of different material (and thus differing thermal expansion coefficients, symbolized by Greek alpha a) are in contact with one another.


26. impact - a collision of 2 or more bodies causes a high force or shock over a very short period of time.  To analyze the contact of bodies together, one should perform a Nonlinear Dynamic analysis; if one can determine the force over time imparted and the area of contact, then it is much more computationally efficient (i.e a lot less time) to solve this as a Time History type of Linear Dynamic analysis.


27. static analysis - How to know if it is a static or dynamic analysis?  Static analysis assumes that the load is constant (or applied over a very long period of time) and the force vector (both magnitude and direction) do not change.  If the first natural frequency of the load has a period of more than 3 times the first natural frequency of the body in question, then the analysis could be run as a static problem.  (Since the period, t, is one over the frequency, f, that is t=1/f, another way to figure out if it's static is to say if the frequency of the load is less than 1/3 of the first natural frequency of the body.)


28. creep - refers to a permanent deformation that body takes on under constant loading typically at an elevated temperature.  If the body gets to half of its material's melting temperature, one should definitely check for creep.  I have a personal story where a friend had a TV sitting on some plastic shelves that were near a window; over time because of the sunlight heating the plastic you could visibly see the deformation of the shelf down under the weight of the TV that it didn't originally have when he put it on the shelf.


29. deformation - when a body displaces more than an allowed tolerance it could said to have failed even if the stresses did not cause the failure.  The point to make here is that it is also very important to look at and understand the displacements in addition to the stresses.

Copyright © 2011 Dassault Systèmes SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without the written consent of Dassault Systèmes SolidWorks Corp.

I always found it both paradoxical and interesting how during the winter in modern times we heat our houses to stay warm but then have our refrigerators to keep our foods cold while it sits within the warm house.  What a waste of energy!  Over the weekend I found a method of bringing the cold from outside into your frig.  Fill up some discarded plastic bottles from your recycling bin, fill them with water and place them outside in the freezing cold overnight.  (Granted this assumes that you have freezing temps outside, so sorry to the people who live in warmer climates who can't try this out.  Actually it was above freezing over the weekend in New England where I live, so I had to wait until temps dropped again.)  Back to the freezing water, in the morning take these now frozen bottles and place them in your frig.  This helps you save energy from every time you open the frig during the day to eat because the frig doesn't have to do as much work to remove the heat that you let in.  At the end of the day the bottles should be mostly melted so then you put them outside again at night to refreeze... rinse and repeat.


What does this have to do with Simulation?  Well you could figure out how well this works by doing a transient Thermal analysis.  Just like the great "Beer Can Barbeque Conundrum" that has been solved by SolidWorks Flow Simulation; watch the video to learn how long before your beer gets undrinkably warm when BBQ'ing outside on a nice summer day.


Since we're thinking about beer, did you know that one theory on the origin of the phrase "rule of thumb" is from earlier times before thermometers where a brewer would judge when to add the yeast to ferment the wort (or young beer) by sticking his thumb in the kettle so that it wouldn't kill the essential micro-organisms by being too hot or too cold, it would have to be just right by a "rule of thumb."  Now it's probably just a myth about the origin of the phrase, because we know that humans are not accurate as thermometers (that are required for proper brewing/cooking).  We can only tell relatively if something is warm or cold.  Same thing goes for rules of thumb when applied to design; they can be used to roughly estimate say how much material should be used to make a design safe, but using upfront analysis tools like SolidWorks Simulation provides us a much more accurate measurement, like a thermometer to a modern brewer, in order to duly define (and even optimize) the amount of material to meet the design's factor of safety.



Copyright © 2011 Dassault Systèmes SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without the written consent of Dassault Systèmes SolidWorks Corp.

I thought it might be a good idea to provide some basic definitions of terms that  are used in analysis that are good to know to be able to communicate  intelligently about Simulation.  These are not meant to be  definitive technical definitions but more fundamental knowledge of these  terms (i.e. they are mostly coming off the top of my head as I'm typing).  If  you believe that I am off on any of these, please let me  know.

Last time two of  the terms that I gave were von Mises and shear stresses; there's many more  components of stress than that.  Here's 12 to 18:

12. normal  stress - if you were to take a look at an infinitesimally small cube  oriented in the global coordinate system, the normal stress is the stress  that develops normal to the face.  Normal stress is denoted by the Greek letter sigma, σ, and subscript is the face that it is normal to, eg. σx.  From last time, shear stress is defined as  the stress that is tangential to the face, usually denoted by a Greek tau, τ, and subscript is the face and then the direction of  shear, for example τxz is on the x-face in the  z-direction.  Thus for a given cube face, there is a single normal stress and  two shear stresses.  Both normal and shear stresses are vectors, having  direction and magnitude, and since the proposed cube we are looking at is  infinitesimally small, the stresses on opposing faces are equal in magnitude but  in the opposite direction.  Thus for a given cube, there are a grand total of 3  normal stresses and 6 shear stresses.  The below image can be helpful in  describing the stress vectors:

13. shear  modulus - primarily denoted as G, is the ratio of shear stress to shear  strain.  For a linear isotropic material, it is useful to know that any 2 moduli  can determine any other elastic moduli (a full matrix can be found at the bottom  of   G = (E/2)/(1+ν), where E is Young's modulus and ν is Poisson's ratio.  In SW SIM, if you define E and ν for a linear isotropic material, I recommend keeping shear  modulus blank and the software will compute it by itself (if you put in a value  it will use that one instead so again best to keep it  blank).

14.  principal stresses - at any point in a stressed body (in this case  let's take the same point as given above), there exists three planes where there  are no shear stresses and the normal stress vectors to these planes are in  "principal" directions, and are denoted σ1, σ2, and σ3, (or  in SW SIM we name them P1, P2 and P3).  The normal and shear stresses from the  stress cube above go through a matrix transformation (resulting in a matrix  where all off-axis terms are zero) to obtain the principal directions, which are  orthogonal to one another meaning that they are completely independent  directions.  It is always true that P1 ≥ P2 ≥ P3.  In an pure tensile load, all  three principal stresses are positive; in a pure compressive load, all 3 are  negative; and when the signs of the principal stresses are mixed (P1 is positive  and P3 is negative), then this usually means the body is in bending.  I  typically recommend to set your Simulation options to always show P1 and P3  stress components in addition to von Mises stress.

15. bending  stress - Thinking of shell elements, they tend to bend easily because  they are thin.  During bending, compressive and tensile stresses develop; the  maximum value of these stresses are on opposite sides of the outermost thickness  of the shell, and the stress varies linearly through its thickness.  The bending  stress is the slope of the line between compressive and tensile stresses through  the thickness of the shell.

16.  membrane stress - is the stress in the plane of the shell that develops  if the compressive and tensile stresses have different magnitudes.  If the  tensile stress is greater, then the membrane is pulled along the direction  of the shell plane; if the compressive stress is greater, then the material in  the membrane is compressed.
So in conclusion  for a shell, show P1 & P3 for both top and bottom of the shells and both the  membrane and bending stresses.

17. torsional stress - is a stress due  to a twisting of an object due to an applied torque.  It is a type of shear  stress with an specific equation Tr/J to calculate it by hand based on the  applied torque T, the radius r from the axis of torque, and the object  (typically a cylindrical shaft) with a polar moment of inertia J.  The max  torsional stress is on the outer max radius, or surface, of the  shaft.

18.  Hertzian contact stress - the compressive stress due to contact of two  curved parts under load.  The important thing to note for Hertzian stresses is  that the max stress occurs close to the area of contact in the subsurface, or  just below its surface, thus the mesh needs to be finest below the surface so  creating a mesh control just inside the sphere near the contact is important.   Another important thing to know is that it is very difficult to get an accurate  Hertzian contact stress result.

Copyright © 2010 Dassault Systèmes  SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without  the written consent of Dassault Systèmes SolidWorks Corp.
OK, so I took a bit of a hiatus from the blog. I'll try to  make this up somehow.

During this time away, I  thought it might be a good idea to provide some basic definitions of terms that  are used in analysis that are good to know to be able to communicate  intelligently about Simulation.  These are not meant to be  definitive technical definitions but more fundamental knowledge of these  terms (i.e. they are mostly coming off the top of my head as I'm typing).

I have a list of  about 150 good to know analysis terms that I want to (hopefully) get through all of them eventually, so here  we go to start 1-11:

1. FEA (or  finite element analysis) - sometimes referred to as FEM (M meaning  method).  The method used by the structural analysis modules of SW SIM by  breaking the CAD models into smaller pieces, called elements, on which the  physical properties, loads and restraints are applied and finally solved  collectively.

2.  Pre-processing - what is done in setting up the analysis model prior to  solving.  This may include: creating the geometry; simplifying it for analysis  purposes; applying material properties; applying loads, restraints, connectors  and contacts; and meshing the model.  This is basically formulating a question;  if this is not done appropriately, people may refer to this as GARBAGE IN, which  leads to GARBAGE OUT (in the post-processing phase).

3.  Solving - takes the input by the user from the pre-processing phase,  puts it into a form preferred by the solver, and calculates a solution for the  question.  The solution is typically very precise, but whether it is accurate is  left for interpretation in the post-processing phase.

4.  Post-processing - or viewing and evaluating the results from the  solver.  There are many methods available to view the results, such as contour  plots, section/cut plots, probing, tables or listing of values, and charts.   Evaluating results can be challenging and can be helped by experience and  judgement, but I think that answering the question "Does this move or react to  the loads as expected in reality?"  Thus determining whether the output is good  or not (GARBAGE OUT), we can say whether we need to go back to the  pre-processing step or not.  If accurate results are required then more than one  solution for finer mesh sizes will be required.  Accurate results may not be  needed if one is comparing different design configurations where consistency is  more important.

5. Modulus  of Elasticity - also known as Young's Modulus.  It is a material  property relating the stress in the material to how much it is strained, and is  typically obtained by pulling a sample of the material in a testing machine.  It  is a linear ratio of stress over strain, so it has the same units of stress  (psi, ksi, Pa, MPa) since strain is without units, and is valid up to the point  of yielding in the material.  Materials which are linear-elastic follow Hooke's  Law.

6. Hooke's  Law - Hooke's law of elasticity is an approximation that states that  the extension of a spring is in direct proportion with the load added to it as  long as this load does not exceed the elastic limit. Materials for which Hooke's  law is a useful approximation are known as linear-elastic or "Hookean"  materials. Hooke's law in simple terms says that strain is directly proportional  to stress. (Definition taken directly from Wikipedia.)

7. von  Mises stress - or equivalent (tensile) stress.  The von Mises stress  is meant as a way to try to fully describe the multiaxial stress state as a  positive scalar value, which also makes it nice to show as a contour plot.  It  has its downfalls in that: it doesn't tell the whole story in how a part is  being stressed, thus one should not rely on this quantity alone to get the full  picture, so show additional stress components such as principal, normal and  shear stresses; and secondly it's based on what's called 'distortion energy'  simply meaning that it's good for deformations that distort the geometry, like  pushing a small area on the outside of a sphere, and ignores uniform deformation  (or hydrostatic stress), like a uniform pressure on the entire outside of the  sphere.  Note that von Mises yield criterion surface circumscribes (fully  envelops) the Tresca max shear stress criterion surface, thus von Mises is less  conservative.
History buffs might  like to know that while it primarily carries von Mises' name, it was formulated  by Maxwell many years before and others, so the entire mouthful name for the  stress criterion is Maxwell-Huber-Hencky-von Mises theory (which I'm  guessing was just shortened to 'von Mises.')

8. Shear stress - is the stress  applied tangential to a face of a material, as opposed to normal to the  face.

9.  Poisson's ratio - symbolized by the Greek letter nu, ν, is the ratio of how much a material contracts in the  direction perpendicular to the direction pulled, or transverse direction.  It  describes similarly how it much it expands transversely when compressed.  The  Poisson's ratio of an isotropic, linear-elastic material must be -1<ν<0.5, but most materials are greater than 0.   Orthotropic materials can have Poisson's ratios outside of these limits.  Rubber  materials have a Poisson's ratio very close to 0.5, such as 0.4999, and cork has  a Poisson's ratio of nearly zero which is why it's used for sealing bottles so  that it can be inserted and removed while still holding the internal pressure.   Negative Poisson's ratio materials are called auxetic, and here is a cool animation  of an example.

10. CFD (or  computational fluid dynamics) - CFD is a general term applied to the  approach to solving the fluid dynamics equations numerically with a computer, as  opposed to experimental or analytical methods.  The method that SW Flow  Simulation uses is the finite volume (FV) method.  Our SW Flow Simulation  software is classified as a CFD program, although Flow Simulation (and FloWorks  before it) helped pioneer a subset of CFD called EFD.

11. EFD (or  Engineering Fluid Dynamics) - is an upfront approach to CFD that offers  a straight-forward easy-to-use interface that speaks the language of the design  engineer working with fluids.  Key technologies include: direct use of  SolidWorks CAD data; automatic detection of fluid volume; Wizard interface;  automatic meshing; automatic laminar-transitional-turbulent modeling; automated  control of analysis runs; robust convergence behavior; simulation of design  variants; and results reporting and presentation in MS  Office.

Copyright © 2010 Dassault Systèmes  SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without  the written consent of Dassault Systèmes SolidWorks Corp.

This was a post to a message on a Meshing Problem that I am reposting here.

When you add parts or place model into an assembly, the default mesh size that is calculated changes from before (when there were fewer parts).  Try using automatic trials to fix the problem by successively reducing the mesh size by a factor < 1.  If that still fails, try applying manual mesh controls to individual components or faces.  There is also a nice option in the mesh control for a component mesh control to mesh it as if the mesh size was dependent on itself only.

The other comments above are helpful  also.  Here are some meshing tips that I compiled before:

Here are  the order of tips to use to overcoming meshing  problems:
1) First if the  standard mesher fails for the default mesh size, then I suggest doing one of two  things: (a) try meshing using the auto looping technique, or  (b) look at the failure diagnostics to determine where it failed to give a clue  as to why.


2) If you did the  auto looping in step 1 as suggested and it still fails, then this time at least  check the failure diagnostics.


3) Did you check  interference?  If there is interference, then: (a) fix the geometry, (b) create  a cavity in one part so that they are not interfering but be wary of small  geometry made by this feature, or (c) make it an incompatible  mesh.


4) Does the  geometry have some other issues like sliver areas or super small faces that are  a a result of imported geometry?  Can it be repaired easily?  These are good  questions to ask at this point but wait until you try more before doing the  harder task of fixing the model.


5) Set a mesh  control for the part or parts that failed and use the sliding bar for "Component  significance" towards the right to High.  The left end of the slider corresponds  to using the default global element size of the assembly (f=0), and the right  end of the slider corresponds to using the default element size if the component  is meshed independently (f=1).  The program calculates the element size (Ei) for  component i from the equation: Ei = G - (G-Ci)*f.  G is the global element size; Ci is the component size for component i; and f is the slider factor as described above.


6) Set mesh controls  on individual features of the parts.  Just as a recommendation for what I use  for the ratio and number of layers, I think 1.25 and 4 does a better job and  looks a bit smoother than the default 1.5 and 3.




7) Open the failing  part or parts individually and try to mesh them by themselves.   If it automatically meshes with the default, then try to find the largest mesh  size that works.  If the default size does not work, then use auto looping to  find one that does.  You may have to use feature-based mesh controls here as  well.  Make note of the mesh size(s) when it does mesh successfully, so that you  can use this as the mesh size for a component (or feature) mesh control back in  the assembly.




8) After trying 1-7  and it still fails, maybe there is a problem with how the parts are touching.   Consider sliver faces or bad geometry that may be created by the mesher  automatically imprinting one surface onto another.  Get a better visualization  of this by creating a split line if possible.  You may be required now at this  point to create an Incompatible mesh and create bonded contact sets  semi-automatically by using the "Find Contact Sets" command.




9) If all else fails,  then with your incompatible mesh, use the "Alternate" or curvature-based  mesher.  The alternate mesher uses the Global Size value to define maximum  element size and the Tolerance value for the minimum element size. The minimum  element size is used for boundaries with the highest curvature. The maximum  element size is used for boundaries with lowest curvature.


Continuing  on with the meshing tips, one addition:

1a) You may find in  the failure diagnostics that it will tell you to either change the element size  and/or raise or lower the tolerance value.  Changing the element size has the  effect of also changing the tolerance which sometimes is the deciding factor.   You can also manually change the tolerance size to something other than the  default 5% of the global element size.


First, it is  important to understand how meshing is performed.  After preparing the geometry  by imprinting touching faces and breaking up faces into logical sub-surfaces, a  surface mesh is created for each face independently.


Next, the tolerance  value is used in knitting the surfaces together to create a water-tight solid,  so you want to have a reasonable size tolerance value to be able to knit  surfaces together (5% by default).  Do not increase the tolerance value too  large, but up to 25-30% of the global element size is fine.  The surface meshing technology  was developed in-house which is typically done for any FEA code.  If there is  only a Shell mesh, then the meshing stops at this point.


For a Solid mesh,  it continues by filling in the volume with solid tet elements and again uses the  tolerance value to determine whether elements should be collapsed or not.  Here  you want to have a reasonably small tolerance size so that elements are not  collapsed unnecessarily.  If the mesher fails in the volume filling phase, you  will want to decrease the tolerance down to about 1% or sometimes smaller.  The  volume mesher is done by a third-party meshing product called  TetMesh-GHS3D from a company in France called Distene; again this  is the typical scenario used by many FEA codes.


So you have two  opposing concepts that fight for raising and lowering the Tolerance value.


Here are tips on when and how to change the tolerance value:
(a) If the  tolerance is too large, it will collapse nodes and create bad element shapes  causing the mesh to fail.  If you have features like a fillet radius or wall  thickness that is smaller than the tolerance value, then decrease the tolerance  to something at least half the size of the smallest feature.


(b) Be careful that  you don't make it too too small or you will run into the problem of having the  surface mesh not able to knit itself together to create a water-tight solid for  volume meshing.  Look for mesh failures coming up during the final stage of  meshing a part where it is filling in the volume.


(c) The tolerance  is a global value and so you should also consider the smallest mesh control size  that was defined.  Also as mentioned before, consider the smallest geometry  feature.  Make use of the SolidWorks tool "Check" for information about short  edges, minimum radius of curvature and other min/max  features.


(d) If the solver  fails because there are not enough restraints or parts are ripping apart from  one another when they should be bonded by a global contact condition, typically  when working with a shells or a mixed mesh, then either your Tolerance is not  large enough or you should define a Local contact set as  bonded.


My favorite new values are: (0.5)^(1/4) and (0.5)^(1/3), which are approximately 0.8409 and 0.7937, respectively.  Why?


Well, when I set my automatic mesh trials options, these values bring me to half the original global mesh size in 4 steps (or 3 steps).  To quarter in 8 steps (or 6),  to 1/8 in 12 steps (or 9), to 1/16 in 16 steps (or 12), and so on.


See below image for settings:




Copyright © 2010 Dassault Systèmes  SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without  the written consent of Dassault Systèmes SolidWorks Corp.

A friend of mine  recently made me aware of this awesome website:  Wolfram|Alpha.
Wolfram has always  been known for its Mathematica software but now Stephen Wolfram wants to  make the world's knowledge computable.
Check out the  gallery of examples, like "#10 screw":
Check out the Intro  to Wolfram|Alpha by Stephen with many more examples:
But my favorite is  Examples by Topic:
and especially  under the Engineering heading is fluid mechanics:
Flow around a  cylinder:
Here's my own  example, Dynamic viscosity of water at 100°F, 1 atm:
Here's info from  their website:
Today's Wolfram|Alpha is the first step in an ambitious  long-term project to make all systematic knowledge immediately computable by  anyone.  You enter your question or calculation and Wolfram|Alpha uses its  built-in algorithms and growing collection of data to compute the answer.  It's  based on a new kind of knowledge-based computing.
We aim to collect  and curate all objective data; implement every known model, method, and  algorithm; and make it possible to compute whatever can be computed about  anything. Our goal is to build on the achievements of science and other  systematizations of knowledge to provide a single source that can be relied on  by everyone for definitive answers to factual queries.

Copyright © 2010 Dassault Systèmes  SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without  the written consent of Dassault Systèmes SolidWorks Corp.

For shell bodies in Simulation, you can color the surfaces by either material or  thickness after the mesh is created by creating a mesh plot (right-click on the  Mesh icon after a mesh is created).  Or even before a mesh is created,  from the drop down menu: Simulation > Shells > Show color by  thickness (or Show color by material).  But for the latter, the plot  does not stay on the screen after closing the dialog box.  You can also change the colors by double-clicking the color box; see  below for result.




Copyright © 2010 Dassault Systèmes  SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without  the written consent of Dassault Systèmes SolidWorks Corp.

Joe Galliera

Material tests

Posted by Joe Galliera Employee Mar 8, 2010
Last week, I had a healthy conversation with a customer and VAR AE regarding testing methods to determine the mechanical properties for different material models used in SolidWorks Simulation.  The customer had an Instron machine for doing testing, although many customers would not have this testing equipment readily available to them.  Even if you do not have to perform these tests yourself, it may be important to understand how they are done.

I found a few websites that can help out in understanding the different tests that are done for material characterizations:

From Axel Physical  Testing services site: The links under the heading "Testing Services" are really good.  Also you can download PDFs with more details of testing different materials from the links below the heading "Technical Downloads" (or from

Datapoint Labs offers various TestPaks specifically for SolidWorks Simulation material models at:

The previously mentioned customer conversation was specifically about defining Mooney-Rivlin constants for hyperelastic materials. In the Knowledge Base on the Customer Portal, there is an  article number that provides details about how to calculate the Mooney-Rivlin constants with the SolidWorks Simulation Nonlinear bundle and also gives an example problem.  The Solution ID number is S-034890.

Copyright © 2010 Dassault Systèmes  SolidWorks Corp. All rights reserved.
Do not distribute or reproduce without  the written consent of Dassault Systèmes SolidWorks Corp.

If you always wanted to know how to enter special  symbols, like 5.0±0.1 or or  176°F or µm, it's really very easy.  You have  to hold in the Alt key and use the number keypad to type in digits that  correspond to the special ASCII codes.  You can use this link as a reference: [HTML format] , or [PDF]


Copyright © 2010 Dassault Systèmes SolidWorks Corp. All rights  reserved.
Do not distribute or reproduce without the written consent of  Dassault Systèmes SolidWorks Corp.

Ever wish that you could easily read numbers off a plot, well there is a freeware solution out there for just this reason: Plot Digitizer.  I really like the  Zoom feature so that I can easily pinpoint my selections!

Download the  program from:   No need to install, it comes as an executable and includes a help file, info on  how it works and a sample JPEG image of a plot to try it out.  First, calibrate  by selecting any 3 points (manually typing in the plot location for each point),  and then click on the data points using the zoom preview to get it right on the  spot, and automatically it creates a nice spreadsheet of all your points which  you can save out to many formats including Excel.  Remember that  SolidWorks Simulation can read in a text file for material properties  and load time curves, so this can be very handy for those purposes.

Here are the features from the website:

Plot Digitizer: Plot Digitizer is a useful program for extracting  data from a linear, semi-log, or log-log plot.
Using an optical scanner, create a Bitmap or JPEG  image of the plot  and open the image file in Plot Digitizer.  Then, after calibration, you can  extract the data values by merely clicking on the data points.

    * A zoom window aides you in clicking on precisely  the point you are interested in.
    * Automatically corrects for the  rotation of the image you're analyzing as well as the non-orthogonality of the  axes.
    * Any 3 non-collinear points can be used for calibration.  (Calibration points do not need to be on the axes.)
    * Save your progress  at any time in a "project" file.
    * Label the calibration points you have  selected. (Labels can be toggled on and off.)
    * Label the points you're  digitizing on the calibrated image. (Labels can be toggled on and off.)
    *  Export the calibrated image (showing the point labels) to a bmp or jpeg  file.
    * Export data to an ASCII, an MS Excel or an MS Word  file

Copyright © 2010 Dassault Systèmes SolidWorks Corp. All rights  reserved.
Do not distribute or reproduce without the written consent of  Dassault Systèmes SolidWorks Corp.

Recently I helped a customer who wanted to make his images stand out, so he inquired how to get images like the ones seen in our marketing images.  Here I offer an alternative to using Photoshop by using an out-of-the-box approach.


First make sure that your Simulation > Options are set correctly under the Plot sub-category (see attached settings); the default setting is to show excluded and hidden bodies as transparent.  Now to exclude a part from the analysis, right-click a body and choose to exclude it from the analysis.  Set up the restraints/fixtures and loads as if the excluded bodies are not there; better yet-- isolate the remaining components and create a Display State.

To capture the image, my preference is to use the "Image Capture" tool that can be found under the Screen Capture tools while customizing the toolbar icons.



Copyright © 2010 Dassault Systèmes SolidWorks Corp. All rights  reserved.
Do not distribute or reproduce without the written consent of  Dassault Systèmes SolidWorks Corp.

Sometimes I have tips for software other than SolidWorks Simulation.  I tend to like to use animated GIFs in my PowerPoint slides, but with MS Office 2007 the animated GIFs were not working as it turns out in MS's effort to reduce the file size.  So they made it a little difficult for me until I learned the following tip that I'd like to share.

It turns out that PowerPoint 2007 compresses pictures (including  animated gifs) by default every time you save the presentation. You can turn that off for a single presentation or edit the registry to turn it off for all.

So here's how to turn off image compression:

To turn compression off (but only for the current  presentation):
Choose File, Save As.
In the Save As dialog box click  Tools.
Choose Compress Pictures.
On the Compress Pictures dialog box,  click Options.
On the Compression Settings dialog box that appears, remove  the check next to "Automatically perform basic compression on  save".
Optionally remove the check next to "Delete cropped areas of  pictures".
Click OK to dismiss the Compression Settings dialog box.
Click  OK to dismiss the Compress Pictures dialog box.
Back in the Save As dialog  box, you can either go ahead and save or cancel. The compression options you  just set will be preserved either way.

Turn  off compression for ALL presentations:
***  N.B. Before changing your Windows registry, make certain to make a  backup.***
Close  PowerPoint
Open registry editor (Start > Run and type Regedit)
Go to  [HKEY_CURRENT_USER\Software\Microsoft\Office\12.0\PowerPoint\Options]
Create  a new Dword value named AutomaticPictureCompressionDefault
Make sure that the  value of the new Dword is 0
Close registry editor

Copyright © 2009 Dassault Systèmes SolidWorks Corp. All rights  reserved.
Do not distribute or reproduce without the written consent of  Dassault Systèmes SolidWorks Corp.
Joe Galliera

Steps for Analysis

Posted by Joe Galliera Employee Dec 29, 2009
Back when I used to  do training, I would start off the course by going through the steps that I go  through for any and all analyses each and every time.  These steps are important  and just take some practice to get it used to doing it until it's like second  nature, so that's why I recommend at first writing them down until you do are doing them in your head unconsciously like I do  now.

Steps  for Analysis

1. What is the objective of the analysis?  This step here sets the tone for the  rest of the analysis steps so it is most important to identify all objectives  clearly.  Sometimes this is where your expectations have to come down to reality or feasibility within a given timeframe.
- Are you looking  for stress or displacement (or for another analysis any particular  quantity)?
- How accurate do the results need to be?
- Are you just comparing designs where accuracy takes second fiddle to  consistency?
2. How do I expect the model to move or what result is expected?  This may sound counter-intuituve such that you should already know what the answer is before you do the analysis, but in fact it is very important to be very familiar with the product design and how it behaves.  By knowing the answer to this question, it can help in both setting up the problem in later steps and as a check of the results.  The final results of the simulation are required for all the detailed numbers.
When someone asks me how long it takes to learn how to use SolidWorks Simulation, I typically tell them that: 1) by knowing SolidWorks you are already about 75% of the way there, since you're always working inside SW, and you can take training to learn the rest of the specific Simulation interface; and 2) how familiar you are with your products contributes a lot to the learning process and how to handle your particular analysis models.
3. What type of analysis do I need to perform?
- Can I make  certain assumptions to fit this problem into a linear static solution, or do I  need to do a transient, or dynamic, or nonlinear, or etc...
4. How can  I simplify the analysis?  I want to meet my objective as quickly as I  can, no matter how fast and how much RAM computers have these  days.
- Can I utilize symmetry?  Symmetry not only reduces the size of the geometry but also removes DOFs because of the need to define the natural boundary condisitons on the cut surfaces.
- Can I remove  extraneous geometry features, such as fillets or chamfers, which will not effect the result for what I am looking for?
- Can I use some type of connector to simplify the problem?
- How many parts in  the assembly do I need to include? The Rule of Thumb here is:
  • If Ea is the elasticity of part A (A  for that part you are doing the Analysis on) and  Eb is the elasticity of part B (the one that you could possibly replace with a Boundary condition), then if:
  • Ea ≈ Eb, you need to include part  B in the analysis model,
  • Ea » Eb, try using a remote load  condition for part B instead (as a displacement type  of remote load to allow rotations),
  • and finally Ea « Eb, you can replace part  B with a boundary condition because it is much stiffer than the part you are analyzing.
From here on you  follow the "recipe" set forth by the analysis tree:
5. What  material properties do I need, and where do I get them  from?
- Can I use the  properties that SW Simulation (SWS) provides or should I get them from somewhere  else?
- Remember that all  inputs should be evaluated so check the numbers that SWS reports because they  might be different from the actual material used.  For example from my  experience, the strengths that the SWS material library reports tends to be  on the conservative side.
(Note: I definitely  will need to expand on this topic with a later tip.  Besides boundary conditions, this is probably the next question that I get most often from  users.)
6. How do I  restrain and apply forces on the model (i.e. boundary  conditions)?
- The trick here is  that you don't want to impose a condition that would not have otherwise happened  without that boundary condition.
- Again I will have  to expand on this a little more in a separate tip, but I tend to try to stay away from fixed restraints because this condition does not exist in nature.
- Look for the Simulation Companion series for tips on applying loads & restraints, so for the time being I would refer you to those.  You can access this through the Customer Portal by clicking on the link for References > Webinars
7. How do  parts interact with one another (i.e.  contact/gaps)?
- Bonded? Free? No  penetration?
- If no penetration, then node-to-node, node-to-surface, or surface-to-surface?  Help (or the training manual) can aid in figuring out when one is better than others, so click on the ? to find out more.
8. What  element quality do I use? How do I mesh the model?
- Draft quality  is generally good for pure axial loads, or finding displacements, temperatures or frequencies, or lastly if you are just testing out your setup before  you launch a longer high-quality mesh run.
- If you have  meshing issues, have you tried meshing looping first; try manual mesh controls  next; mesh failure diagnostics can point you towards bad geometry that needs  a-fixin'; and lastly, go to the alternate (curvature-based) mesher after you switch to an incompatible mesh, and defined bonded contacts where  needed.
9. Which  solver do I use?
- Right now, all  I'll say about this is that the Automatic solver option for novice users will  choose the right solver for them.


10. View and  evaluate the results.  This is where the job of a design engineer comes in, because it's not just about viewing but also evaluating the results.
- Animate the results to see if it moves as one would expect it to move in real  life.  Going back to thinking about step 2, does it behave as you expected?
- Is von Mises  equivalent stress appropriate or do you also need to create other stress plots, like principal or shear stresses?  Cartesian or cylindrical coordinates? Is  strain more appropriate?
- Novice users can  use the Analysis Advisor for help in interpreting results.


11. Do I  need to go back to any previous step and rerun the  analysis?


I hope that you get  a lot out of these steps as guidelines in what you should think about each and every time when performing any analysis.  Please comment if there's something that I might have missed or if you want to share something that you do every time?
Copyright © 2009 Dassault Systèmes SolidWorks Corp. All rights  reserved.
Do not distribute or reproduce without the written consent of  Dassault Systèmes SolidWorks Corp.

Filter Blog

By date: By tag: