Desktop Weld Optimization Software
  for automated welding

FAQ general



SmartWeld SLIDES

Sensitivity Parameters


ETE Table




SmartWeld HOME

Technical Discussion-2

I am an R&D metallurgist and am working with welding engineers on a new type of reactor. We are developing new GTAW and EB weld procedures. I recently obtained a copy of the SmartWeld software and am quite impressed with its capabilities. Unfortunately, zirconium is not one of the materials that can be selected. I tried using titanium which appeared to have the closest properties. My question is what would it take to get zirconium in the materials selection of the program? I can probably obtain a variety of state properties from other codes we have access to. Any information or comment would be appreciated.

Unfortunately, welding simulation and optimization are not commonly employed during weld process development. Because weld thermal modeling is not widely practiced, the necessary property data for most engineering alloys has not been measured and is simply not available. To facilitate weld modeling of the materials listed in SmartWeld, the measurement of necessary weld thermal properties was undertaken in the SNL welding lab. A small set of only 6 GTA welds with different currents and travel speeds on thin plate samples is all that is required for each alloy. It is not particularly difficult or expensive to do, and the procedure is outlined in a 2002 paper. If you can make the welds and metallographic measurements in your facility, the property calculations are straightforward. Then we can incorporate those values into the SmartWeld source code. We encourage SmartWeld users to do this and contribute to this open source project.

Can you provide little bit more detail on the weld sensitivity module in SmartWeld?

Sensitivity analysis is commonly used to characterize the effects of parameter perturbations on model output. Typically for welding these partial derivatives are calculated from the results of response surface experiments (see WJ paper). For many apps in SmartWeld these partial derivatives (sensitivity parameters) can be calculated directly from the conduction heat flow solution to the user input weld conditions ( e.g. material, joint type, travel speed). SmartWeld calculates weld dimension sensitivities for process parameters that might be difficult to control such as base metal temperature, part thickness, or transferred energy. There are a few examples of the sensitivity determinations in the slides on the website.

What's the best way to determine thermal contours around a small standing edge geometry weld? The application involves electron beam welding a 0.010" wall thickness Kovar tube coaxially into a similar wall thickness tube of stainless steel to achieve a hermetic seal. Power and travel speed are provided (40W @ 10mm/s), and the issue is whether or not the thermal gradients will generate appreciable stresses in a near-by ceramic-to-metal seal.

Normally, using ISOEDGE would be the correct way to do this. However, in this case the radius of the fusion zone calculated by ISOEDGE is small compared with the top width of the weld, thus predicting a fusion zone whose length in the direction of travel and depth are extremely small compared with its width, which is not very reasonable. A better way is to use ISO-2.5 D, and insert the Kovar wall thickness as the "plate" thickness. The symmetry of the heat flow allows you to "cut and fold" the plate into a geometry identical to a standing edge geometry, but because it is a point source, the shape of the fusion zone is now much more reasonable. Furthermore, by using a thicker than actual plate thickness, and/or by changing the material from Kovar to Al or Cu, the effect of a heat sink can also be approximated. This assumption is equivalent to assuming that the heat sink is in perfect contact with the material to be welded, and is flush with the top surface, which is not quite usual practice. Nevertheless, an upper bound approximation to the effect of heat-sinking is possible.

Is SmartWeld sensitive to beam quality effects or intensity distribution within the spot and also temporal profile.

Laser mode and beam quality is very laser specific, primarily it drives penetration and energy transfer efficiency. Energy transfer efficiency can be determined with SmartWeld using several of the applications and weld size measurements. There is an example of this approach in the slides.

For the weld penetration, width, and HAZ measurements, do you plan to include any microstructure prediction algorithms in future.

That is a good idea, but at present SmartWeld is a tool not an expert system. SmartWeld requires a minimum degree of knowledge and judgment on the part of the user to be applied successfully.

Looks like a very interesting piece of software. A quick question regarding the YAG application. The temperature of the weld spot - when does this temperature occur? Is it as soon as the YAG pulse ceases?

The YAG application is based on an empirical study so it cannot be queried in the same way as the other heat flow applications in SmartWeld. But in all cases the maximum temperature at a specific location is what is determined. The experimental measurements and the location of the thermocouples are shown in the Welding Journal paper.

I am currently a student and Cal Poly State University in San Luis Obispo, CA. I am currently working on a project for one of my engineering classes that is similar to your SmartWeld MATLAB suites, though on a much more limited scope. I am just supposed to use MATLAB to model the ISO-3D Rosenthal weld model. To that end, I was wondering if there was any way I could get a look at that part of your actual suite and compare it to what I'm doing. Thanks for any help you can give me.

The source code M files for ISO-3D are available for download on the SourceForge page.

Our interest in SmartWeld is only for the laser welding module, in particular application of key hole welding for a specific application for tailor welded blank welding. Can OSLW be used for dissimilar materials weld prediction?

There is no feature for dissimilar metals. The best way to handle dissimilar materials is to run the conditions separately for each material and bracket the problem with the two results.

Can it be used for dissimilar thickness butt joint configuration?

Again, the problem must be bracketed with separate solutions. OSLW provides results for partial penetration welds. Butt joints can be analyzed with Weld2D or Weld2.5D.

I believe the majority of our welds are done using a new fiber laser system and it is what I would be interested in modeling. It is different in that the fiber actually contains the lasing material in itself. It is very reliable and seems to be nicely linear with respect to weld penetration/width to power and defocus. I am wondering what conduction models you used? Does SOAR take into account laser reflections and such to first predict the keyhole shape, then use the energy balance to find the final shape of the weld, or is it a pure conduction model with a heat source of known distribution, i.e. Gaussian?

Yes, the laser type, beam quality, and focusing optics/position will all affect the depth of weld penetration. As a result, it can be difficult to predict what the penetration depth will be for a specific laser or even electron beam welder for that matter. The better way to predict penetration is to use and know power density, but that requires you to measure focused spot size. Most people don't really know what spotsize is at the workpiece. Kapton film works pretty well to easily measure spotsize, see our 2004 paper in Welding Journal. We have two laser apps in SmartWeld that were developed based on experimental spotsize data obtained from a large set of laser welds: OSLW and YAG. Without that type of data there is no good way to predict weld penetration with any computer model. You could build a model for your fiber laser if you make enough welds. Since most people don't really know their spotsize, we can't predict penetration. But that doesn't mean we can't understand/model the welding process. Most laser welds are made without a deep keyhole and conduction heat flow dominates. That is why we added all the conduction heat flow models in SmartWeld. If you make some welds with your new laser, then measure power, speed, and weld size, - you can then use SmartWeld to calculate what your energy transfer efficiency is. That will be useful in predicting what other welds will be like. SmartWeld is especially useful for minimizing heat input and the temperatures near the weld. We also have developed some sensitivity parameters that will provide a more robust weld procedure. There really is a lot you can learn about your process in this way.

I have just found and read your paper "Weld Procedure Development with OSLW - Optimization Software for Laser Welding" I am excited about the potential of using this type of software for our applications. My company is in the business of contract manufacturing photonic components. We primarily assemble telecommunication packages. We are currently using Nd:YAG laser welders for spot welding Kovar parts together. Can you recommend a source for literature and software to address modeling/predicting necessary weld parameters for our type of system?

We have recently published a paper investigating post weld shift in pulsed laser welding of photonic packages which can be downloaded from the related articles page on this website. SmartWeld has a spotweld application that is particularly useful for active alignment of fiber optic devices.

We are trying to model a pulsed YAG laser weld on both aluminum and stainless steel. I have supplied the Joules per pulse, rep. rate, travel speed etc. to our FEA analyst who is doing the modeling. One of the problems that we are having, is determining what percent of the energy is actually going into the part. Do you have any data on laser beam coupling efficiency, or anything in SmartWeld that might help in our modeling?

SmartWeld is a great way to verify an FEA model. It can be used to quickly determine coupling efficiency, parameter values, and temperature fields. Most SmartWeld solutions are based on the conduction heat flow equations. The outputs are of course limited by those equations. I believe that in many, many cases, SmartWeld can provide a very quick and sufficient approximation to the answer desired. But there are also mechanical stress problems where only FEA will do and using SmartWeld to check, calibrate, and even input thermal data into an FEA model is an ideal approach.

I plan to use OSLW with the following lasers 9kW CO2 from MLi, JK704 from GSIlumonics and D035 from Rofin. presently the DC035 is new and we plan to use this for all future welding. although we can get near gaussian output in this we also have a donut beam switch with this laser. What difference does it make if I use the donut mode and gauss mode. I believe the model takes only the Gauss mode. How accurately can this work for donut mode.

9 kW is significantly higher power than the laser used to develop OSLW. OSLW development was limited to 1600W. The other applications are not empirically based and will go up to 30KW, but they will not produce the lens and beam irradiance information that you are asking about. Laser mode and beam quality is very laser specific, primarily it determines energy transfer efficiency. Energy transfer efficiency can be determined with SmartWeld using several of the applications and weld size measurements. There is an example of this approach in the slides.

SmartWeld Copyright Sandia Corporation 1997- 2013.
Questions about this web page to:

Get SmartWeld at Fast, secure and Free Open Source software downloads