# Heat transfer phenomena in frictional welding

1. Oct 9, 2007

### eks123

Hi,
I want to know the peak temperature attained and temperature distribution with time in a frictional welding process ( ultrasonic consolidation between an Al substrate and Al strip). The problem with normal thermocouples is that by the time it shows the temperature , the temperature has been decayed to much lower value. I think I can have two approaches:
1) To look for a thermocouple having very low response time (in microseconds)and to directly obtain the time distribution of temperature. (I cannot use non contact sensors because sonitrode blocks the welded surface)Also if you can suggest me the manufacturer of such thermocouples.
2) From the temperature profiles obtained from simple thermocouple only, using a mathematical theory of heat transfer,and getting the peak temperature ( reverse problem)
Please give me suggestions, specially for the second case.

2. Oct 9, 2007

### PerennialII

Hi eks123 - I certainly don't envy your problem, a very common one when working with welding & its heat transfer related problems.

1) Haven't come up with a thermocouple with the response time you're after -- if someone can might use one as well.

2) This is the common route have used several times. Have temperature - time information preferably from multiple locations, and calibrate a finite element model to reproduce the measured temperature - time results at those locations, and then use the model to predict temperature at locations where haven't done measurements or they are not feasible to do. First doing the analysis using FEM is relatively straightforward - commercial packages with a relatively steep learning curve can handle it quite well. The analysis, especially trying to figure out how good your results are, however, isn't really that simple. The problems, at least as I've encountered them, drill pretty deep into how complex of a FEM model you want to build and how good are your high temperature (near melt) material properties. The latter are scarce in general, and the former has to do with how do you introduce the actual welding process and the introduction of the weld metal to the problems (the actual creation of the weld)(actually this part goes to the material props quite a bit itself). The toughest way to approach would be to do a full fluid-structure interaction analysis taking into account what is going on in the weld pool itself (probably not the way to immediately unless you enjoy the challenge and/or master this stuff really well already), but in any case you may have to 'introduce new material' (new elements in FEM) as the welding progresses in order to get your heat transfer analysis to the right ballpark. The latter can be done with relative ease, but it's not the most simplest FE model to do, and requires "some" work.

One can simplify the situation further and work with a model where the weld is already there and then gain results from there by imposing initial and boundary conditions for weld filler temperature and so on, it pretty much has to do with how much time you want to invest and how accurate results you want to (sure can "build up" --- do the simplest first and then add complexity --- usually a good idea anyways).

Have you checked what sort of heat transfer modeling has been published for friction (stir?) welding .... I've seen some work, but don't know how well applicable those might be to your problem. Naturally some of those might help quite a bit at least seeing what sort of solutions have been attempted/proposed. There are then analytical solutions for the heat transfer PDEs for welding which can use always as a "first approximation", but those are really crude for most practical applications.

Any of this what you're after?

3. Oct 9, 2007

### eks123

Thanks PerrenialII for the reply. Actually I'am a student and just in process to learn FEM. What if I do the calculations myself by assuming this phenomena to follow a particular thermodynamic rule ( It should have transient conduction, convection applied ) Also it is ultrasonic consolidation friction welding, which not result in a weld pool but only an instantaneous plastic region, that plastic region then dissipates heat.. So the heat must transfer in the solid and convected out ( neglecting radiation loss) the temperature at the welded point thus decreases with time. So all I have to find is this temp-time distribution and the peak temperature that was attained but I'am not able to catch it because of slow response of thermocouple.
So can you give me suggestions how can I do it without FEM, by just applying thermodynamic formulations and also most importantly which thermodynamic conditions would apply for this situation, where can I get those equations, etc

4. Oct 9, 2007

### eks123

Sorry for writing your name wrong, it's PerennialII.

5. Oct 10, 2007

### PerennialII

Not a problem .... hardly remember what it is myself :rofl: .

The ultrasonic consolidation friction welding technique will probably ease the work a somewhat, a 'direct' conduction & convection approach might work, perhaps some of the existing analytical solutions. There are a couple of books about the subject which do a good job of collecting a number of usable solutions. I'll check my shelf and get back.

6. Oct 10, 2007

### PerennialII

There are a number of books out there about heat transfer in welding, one I've used and extracted analytical solutions of is by Oystein Grong - Metallurgical Modelling Of Welding - it has the "classical" analytical solutions for heat transfer during welding for one (the Rosenthal eq etc.)(taking a look at amazon with those keywords yields a number of hits, number of others are out there). Probably the good idea might be to search next door libraries first. There are lots of articles out there, but books are probably easier for getting going.

7. Oct 10, 2007

### eks123

I would just look for the book in Library, but do you know any site/link/ebook where some theory/helpful information is available?

8. Oct 10, 2007

### PerennialII

I've some links stored about modeling of welding which could be of use (pretty much deal with heat transfer solutions + numerical solution), I'll dig those up and paste here .... +I've a Dr. of the field in question a couple of rooms away, I'll ask him tomorrow over morning coffee if he knows whether anything good & "deep enough" is available online.

9. Oct 10, 2007

### eks123

thanks man, you've been very helpful.

10. Oct 12, 2007

### PerennialII

Some sources which are freely available online, quite a bit of them focus on applying the "classic" Rosenthal eq to differing problems (which might be a decent start):

http://happyinmotion.com/jez/coupling_paper.pdf [Broken]
http://www.scielo.br/scielo.php?script=sci_arttext&pid=S1516-14392004000400010
http://eagar.mit.edu/EagarPapers/Eagar032.pdf [Broken]
http://www.engr.psu.edu/ce/Divisions/Hydro/Reed/Education/CE 563 Projects/Kumar, Mishra.pdf
http://homepage3.nifty.com/yurioka/exp.html [Broken]
http://www.twi.co.uk/j32k/unprotected/band_1/sitg_fea_case6.html
http://www.scientific.net/0-87849-919-9/4099/
http://www.me.sc.edu/Research/cmmnde/Friction/papers/proc/thermal.htm
http://www.eng.upm.edu.my/~kaa/WEC/WECPapers-FinalVersion/AE16.doc

Last edited by a moderator: May 3, 2017
11. Oct 12, 2007

### eks123

thanks PerennialII, I'll just check out these links.