Archive for July, 2008

Projecting FEA to Management effectively

Friday, July 25th, 2008

I have always wondered about the perception of management towards purchase and implementation of CAE software. The benefits are seldom evident, and the success of the tool is almost always inevitably buried under the blanket. The timely rollout of the product line, minimal failures, and increased revenue are more direct indicators observed by the management. I guess the yeast is never appreciated for good bread. The dough and the bread makers are!!!

So it often becomes a big challenge when management needs to be convinced about the potential for a product like COSMOS. I found this interesting article written by Vince Adams, one of the Product Managers for the COSMOS FEA suite, where he addresses the resistance from the management and how to counter them.

http://designsmarter.typepad.com/vinceadams/

As you would perceive after reading this article, the issue lies in educating management sufficiently for them to perceive the benefits and liabilities associated with FEA. Setting good expectations would relieve a lot of stress downstream.

Vikram Vedantham

Simulation Product Manager 3DVision Technologies

Redundancies in COSMOSMotion 2008 – Part II

Wednesday, July 23rd, 2008

Continued from the previous article (Redundancies in COSMOSMotion 2008 – Part I)

http://www.3dvision.com/wordpress/2008/07/18/redundancies-in-cosmosmotion-2008-part-i/

THE CURRENT SOLUTIONS

The user has two choices when it comes to addressing the above problem:

  1. Put some thought into creating mates. Address the five degrees of freedom by distributing them at the three hinge locations without repeating them. (As you can imagine, this can become a tedious procedure where you have to re-think the process of creating mates). However, this will ensure that the loading is distributed evenly to the three hinges.
  2. If redundancies are already present in a pre-existing model, then a workaround is to somehow make the joints have finite stiffness, instead of infinite stiffness. This will ensure that the extra weight of the door gets distributed to the other hinges. COSMOSMotion 2008 permits the user to specify finite stiffness to the joints by specifying the values when defining the mates (Create a mate > Go to the Analysis tab in the mate property manager). However, this presents two challenges by itself:

 a) The user needs to account for a good approximation for the stiffness value and the damping effect so that the joint is neither over-stiff nor over-soft.

b) The user needs to figure out the right combination of stiffness values for the rotational and translational degrees of freedom for the mate.

THE FUTURE

The interface behind COSMOSMotion is now being rapidly revamped, and is guaranteed to grow by leaps and bounds in the years to come. If redundancies can be addressed during the mate-creation stage, rather than putting out fires later on, then running motion simulations would become a part and parcel of every design. One possible suggestion (if the developers are hearing) is to indicate graphically on the screen right during the mate creation stage that a particular part already carries constraints in specific directions, and has only so many degrees of freedom remaining. This would help the user make the best mate-choice while initially setting up the assembly in order to achieve solution 1 (from above) a modeling practice instead of a Simulation fix!!

BOTTOMLINE

The user has to spend a little time getting these issues addressed if there are redundancies in the model. Don’t get me wrong here – COSMOSMotion is a stellar product and does more than a great job in bringing virtual simulation to desktops!! I have seen a lot of success stories with it. One customer recently told me that using it in a project helped them finish a project on time (which was a company record!!). However, it seems to be one product that is facing an uphill battle in customers accepting the technology, learning it and using it in their day-to-day decisions.

The team that went back with the training from this week is now already on their first project in getting engineering data from using Motion. They used to generate prototypes that took forever to generate, test and gather valuable data, not to mention the resources, man-hours, time and money! COSMOSMotion 2008 has equipped them with the potential to revolutionize their product cycle and time to manufacturing!

Vikram Vedantham

Simulation Product Manager 3DVision Technologies

Redundancies in COSMOSMotion 2008 – Part I

Friday, July 18th, 2008

I taught Motion 2008 this week to a group of engineers from one of the larger accounts that we handle in Cincinnati, Ohio. As you might have gathered from my previous post, Motion 2008 has a new revamped interface that makes the tool easier to use. Yet, the ubiquitous problem with rigid body kinematics and dynamics – Redundancies, continues to haunt the user and make them think twice about integrating simulation directly into their conceptual stages.

BACKGROUND: What are redundancies?

Apart from a single-word definition (nuisance!!), redundancies can be understood easier by taking note of an example. Consider the door to your room that is mounted to the frame by three hinges. From Newton’s falling apple theory, the weight of the door because of gravity is opposed by equally distributed reaction forces in the opposite direction at all the three hinges. Now, if we remove the two lower hinges and let the door be mounted just with the top hinge, two things happen:

  1. The opposing reaction force is seen at the top hinge, and is equal and opposite to the weight of the door. This leads to the hinge material (aluminum/steel) bending, and ultimately deforming/breaking because of its inability to carry the moment acting about the center of mass of the door.
  2. The person mounting that hinge is rushed to the ER!!

The same event happening in COSMOSMotion would give the ultimately desired effect of the door standing upright without any problem! The reason for this is that COSMOSMotion considers treats every part in the assembly as rigid bodies, and every joint as a rigid joint. In other words, the hinge being used is not made of a material with a finite stiffness, but has infinite stiffness and can carry any load by itself without any deformation! What a pain – since the door motion can now be simulated by just one hinge, and the other two hinges become redundant!!!

Thus, the output in COSMOSMotion would show the opposing reaction force to be equal and opposite to the weight of the door on the top hinge, and show zero reaction force on the other two hinges!!

THE PROBLEM: What does a user do when his SolidWorks model has three hinges?

We know that any part held in 3-D space has six degrees of freedom – three rotations and three translations. When you create mates in an assembly, you are removing degrees of freedom from this part. In other words, when you mate the door to one hinge, you ensure that the only degree of freedom remaining for the door is the rotation about the axis of the hinge. The other five degrees of freedom are constrained by the mates created in SolidWorks.

Redundancies are created if the same five degrees of freedom for the door are repeatedly constrained along the same axis (even if at different locations, such as the other two co-axial hinges). In other words, if you have constrained all five degrees of freedom at one hinge, then you cannot re-account for the same at the other locations.

(To be Continued…Part II : The Solution)

Vikram Vedantham

Simulation Product Manager 3DVision Technologies

Mass transfer – the detour

Sunday, July 13th, 2008

Friday I left my customer in New Philadelphia to look at his 5000 files (see part 2) for another customer in Cincinnati. They had just bought PDMWorks Enterprise and have asked me to convert their legacy data from Activault.

As you would expect, no Activault manuals could be found. It took me a while, but I was finally able to figure their Activault’s database schema. Once this was done I was pretty much home free to write code to find the old file, match it with the custom properties they wanted to keep then add it to the PDMWorks Enterprise vault.

I actually wrote the code in the Activault’s .mbd file. I am beginning to like witting code like this in MS Access. I find I get two great advantages: I don’t have to install Visual Studio on the customer’s machine, and the VBA in Access gives pretty quick ‘n’ easy access to the database.

….plus I got to enjoy the irony of using Activault’s database to contain code to cause its demise. (That being said Activault’s database structure is quite clean and appears to provide a snappy interface.)

The customer’s new PDMWorks Enterprise server is a Dell PowerEdge 2970. I cannot sing enough praises over this machine. I first ran a test conversion on my Dell M90 laptop and it took almost eight hours to convert their 19,752 files. Their PowerEdge did it in about 50 minutes.

Whooowee working with a database like found in PDMWorks Enterprise is a lot more fun than working with PDMWorks Workgroup! …Back to New Philadelphia next week to finish that project.

Jeff Sweeney

CSWE Engineering Data Specialist 3DVision Technologies

Checking Deformation from Thermal Loading

Friday, July 11th, 2008

Can I check if two parts will come apart if a thermal load is applied?

Well, it depends upon how they are put together in the first place. Two parts can either be snap fit/press fit together, or injection molded together. Let’s see how to work with both these cases:

SNAP FIT SCENARIO:  If the two components are snap-fit, then they already experience a pre-load that needs to be captured. In such an event, the analysis becomes a two-step process, and has to be solved as a non-linear analysis using the SolidWorks Simulation Premium version (COSMOSWorks Advanced Professional).

STEP 1:  Setup the snap-fit problem and simulate the two parts being fit together. This can be done by putting a load/prescribed displacement to move the components together, while capturing the no-penetration contact as they move into place. Make sure that the time interval is defined correctly to make this the first event in the loading curve. The end result at this stage would be the stresses after the snap-fit process (pre-load for step 2).

STEP2: Re-start the analysis (under the properties of the study) for the next time interval while using the temperature load at this instant. This will show the stresses and deformations caused because of the temperature load. However, this would have captured the effect of the snap fit and carried the analysis forward from there.NOTE: This can also be setup in one stage if you stagger the time-intervals for the two loading conditions accordingly.

INJECTION MOLDED FIT:  This becomes a little trickier because the interface between the two connected components is as good as glued. The possibility of separation becomes a function of the strength of the bond at the contact interface. Usually, one of the two components is plastic, and the molten surface of the plastic stick to the adjoining component. Capturing this property becomes more tedious, and the alternative is to search for an in-direct approach to setup the case.

The problem can be set up as linear or non-linear based upon the material properties of the two components. If the components are linear materials, the linear module is sufficient. The problem is setup exactly as any other linear analysis, except for the contact condition and the solver type. Make sure you set the contact to Bonded, Incompatible, and the solver to Direct Sparse.

If the materials are non-linear, the material properties need to be appropriately modified as Von-Mises plastic or Hyper-Elastic in order to capture their subsequent effects.

In either case, the results of interest would be the stress plot. If the stresses at the contact interface exceed the yield strength of the material, the possibility of separation escalates. Thus, capturing the stresses at the interface is an indirect indication of failure at the contact areas.

Vikram Vedantham

Simulation Product Manager 3DVision Technologies

Working with COSMOSMotion 2008

Thursday, July 10th, 2008

When SolidWorks came out with the new revamped interface for Motion Simulation in 2008, I was both eager and skeptical at the same time. The interface looks a lot like Animator, and timelines dictate the simulation setup. However, as I tried my hand at it, I was a little disappointed at some of the functionality that seemed to have been stripped off. I was concerned about three main areas:

  1. Joint definition: You can no longer define joints manually.
  2. Contacts: 3D Contacts are the only choice in 2008 (as against both 2D and 3D contacts in 2007).
  3. Markers: This was probably what caught me by surprise since 2008 does not have markers anymore. Markers are ways to setup complex expressions in Motion in order to make a motion driver dependent on the results of another moving component (result-dependent properties). Difficult to understand at first, they become extremely useful once you understand their purpose and usage.

So I proceeded with the notion that 2008 Motion is fairly weak in these areas. However, over the past few months, I have had an opportunity to use it a little more, and I have come to realize that these three areas are addressed in the new release (just buried deep enough!!).

  1. Joint Definition: The SW mates have been enhanced to include the entire library of joints. I like this feature, but the onus is upon the user to ensure that he sets up mates correctly.
  2. Contacts: 2D contacts did not generate much data about contact forces and so forth. The only direct benefit of 2D contacts was displacement and velocity information. 3D Contacts are a much more realistic simulation of the problem. SolidWorks is right in terms of wanting to move contacts into the 3D Domain. Here, once again, the onus is upon the user to ensure that the correct impact properties are defined.  As problems get bigger, the solver might take more time to solve though.
  3. Markers: These tools are, in fact, much simpler to use now. Markers have been replaced with the ability to call a previously generated result directly into the expression property window, and define relationships to these result quantities in one step. The user does not have to look up all the FORTRAN based functions anymore to setup an expression.

FINAL VERDICT: Motion 2008 is not as bad as I originally thought it to be. It grows on you as you use it more and more. I have been pleasantly surprised with its ease of use, and the integration provided with SolidWorks mates.

Once again, way to go – SolidWorks!!

Vikram Vedantham

Simulation Product Manager 3DVision Technologies

Mass transfer – Part 2

Thursday, July 10th, 2008

Okay, wow…the bulk load tool took nearly eight hours to check the files into the vault. The log reports that just under 11,000 files went into the vault. The remaining files appear to be duplicates of files already in the vault. A nice side benefit of the bulk load tool. We’ll have to spend some more time analyzing the 5,000 files that did not go into the vault to confirm this…

Looks like we just dodged a bullet. The legacy files are on the same server as the Workgroup vault. Essentially we duplicated 11,000 files and their server is dangerously full! (Need to confirm this is why 5,000 files didn’t make the trip!)

So looks like I need to write a routine to delete the legacy files. This means I need to do two things I don’t like to do:

  1. Write code quickly with people looking over my shoulder
  2. Write destructive code…I hate writing code to remove data. Too much stress! :) [A moment of prayer to the backup gods.]

The bulk load tool doesn’t store the full file path, so I need to search for each file one at a time. This code is going to run slower than the bulk load tool. Looks like another routine to run overnight. At least this can run as a parallel process to the rest of the import.

Customer is looking at the files that didn’t make the trip, if all looks good, it will be time for my programming to shine.

Jeff Sweeney

CSWE Engineering Data Specialist 3DVision Technologies

Mass transfer – Part 1

Wednesday, July 9th, 2008

What does an Engineering Data Specialist do? Mostly I help people create engineering data [DriveWorks, SolidWorks, API, best practices] and maintain their existing data. [PDM, API].

I have just received an assignment to help one of our customers in New Philadelphia, Ohio (just south of Canton) move their legacy data into PDMWorks Workgroup. However there are a few twists:

  • They have been using Workgroup for a while now, current assemblies can reference files in or out of the vault.
  • The legacy files use a different naming convention than they use now. Most notably the file name contains the revision letter. They want all files that go into the vault to meet their current naming convention.
  • The files need to go into the proper Workgroup project….this project is calculated by the file name.

So the battle plan is:

  • use the bulk load tool to put the files into a temporary project
  • use PDMWorks’ API to rename the files
  • use API to delete all but the latest version of the files (first ensuring the files aren’t referenced somewhere)
  • use the API to move the files into the proper project

Putting the files in early and using the bulk load tool, gives us a couple of advantages. I don’t have to write code to add the files, I can use Workgroup’s the API to handle the moving and renaming -the file references for the SolidWorks files should remanin intact.

We are looking at adding just slightly over 16,000 files. It took nearly four hours just for the bulk load tool to dig through their existing directories – looking for SolidWorks files to add. (We’ll do their AutoCAD files later.)

The files are being checked in now, it looks like it is going to be a long process. I’m going to let it run overnight and we’ll look at the report in the morning.

Jeff Sweeney

CSWE Engineering Data Specialist 3DVision Technologies

Follow
Get every new post delivered to your inbox
Join other followers
Powered By WPFruits.com
Bear