How to Control Welding Distortion

Knowing nothing about this subject, I would preheat the piece in an oven before welding, and make my welds while it was still hot.

Hi sir,

yes i know the tolerances is kind of unrealistic but my customer insist to stick at this tolerance. Do you have any idea that to minimized the distortion? (betwwen 25 micron is almost equal to 0.001" not 0.0001")

thanks

I don't know about that.

It doesn't seem like those kind of tolerances should apply to this thingy, but I have a former client that has a machine shop, and the machines, to do stuff that boggles my mind. No room for error, and tolerances are extremely tiny.

Does your customer have lots of money to throw at this, because it seems to be an impossible task with values asked?

Fabricating it to the closest tolerance you can maintain then having it machined to tolerance would be the only way you might have a remote possibility of meeting the specifications.

Using a machined to tolerance jig with the application of pre-heat blankets and/or ovens will help.

Good luck, you are going to need it.

Are you sure you didn't receive the job from another planet far advanced to ours?

tcmtech,

I think your conversion is right (according to my conversion programm

Best regards,

John

The online calculator I used says that 25 microns equals .000984252 inches. Rounded off that gives me .001 or 1/1000th of an inch.

Either way the tolerances are nowhere near reality given the materials and structural application being given to the OP.

My old metals books say that 5/8" .063 wall square tubing can have a +- tolerance of .020" (~500 microns) in most typical lots which in itself put the tolerances of the material themselves at some 40 times outside of the tolerances expected by the customer.

Hi again tcmtech,

My work colleagues were rushing me out the door last night and an a saying of an old boss of mine "it pays to dither" comes to mind, your correct at 0.001" (give or take a thou. or so)

I would have expected an overwhelming "no can do" but there seems to be some who think it can. I wonder how many his customer wants, hope it's not a lot as you might get a "few" rejects.

Best regards,

John

Perpendicularity and parallelism are angular; microns are not. (That's just another thing that sounds odd about this.)

If you use geometric tolerancing perpendicularity & parallelism can be expressed in microns. The faces would be defined as falling between 2 planes, 25µm apart. (Still a stupidly tight tolerance though).

You cannot get the raw tube steel to this tolerance. At work we have machines fabricated to this tolerance. The granite blocks are mounted on level, ground flat, 30 mm thick steel plates that have been grouted in place to the cement floor. This frame appears to be designed to hold something and to rest on something. Does the tolerance specify loaded or unloaded and supported on what type of platform. Depending on the weight supported and load distribution this can easily sag more than 25 microns. Who ever is asking for this frame, ask them how they intend to measure your work longest length piece is between 700.0125 mm and 699.9875 mm long. You will also have to have nearly perfect right angle cuts. If my calculation is correct the 5/8" tubing must be cut to a perfect 90° ±0.09°.

That tolerance is IMO bs.

And one does not need to be a newbie engineer to put this out. I'm sure there is a reason. But I new of and engineer that put out drawings carried out to 4 decimal places. He would listen to me, but he listened to the owner.

I share concerns similar to those in the other comments about achieving the specified tolerance. It isn't impossible, but it would be a tedious undertaking.

.

Still, perhaps I can assist you in getting closer to the goal.

.

1. Things will be far more simple if the tubing the rack is to be assembled is very straight and flat and have sides with perpendicularity and parallelism within the tolerance specified for the completed piece.

Finding the best stock then precision grinding or hand scraping are possibilities.

If end caps are to be welded on, it may work out best to weld these on first then proceed to torque/bend straight and grind/scrape flat those pieces of tubing.

.

If might also work well build and then true sub-assemblies, then join sub-assemblies and true the workpiece.

.

2. The tubes and gussets need to be shaped/mitered to high precision for a very tight fit/butt. It is a mistake to assume that a piece of square tubing butting up to the side of a similar piece only needs to be cut off square (as this will leave gaps where the cut fails to touch where two of the sides approach a rounded corner. With thin wall tubing there is no need to bevel the edge or leave a 'root gap'; full penetration can easily be achieved without.

EDM or even hand filing (in the right hands of course) are two ways this can be achieved.

Cuts to miter the tubes or fit the gussets should be smooth and well cleaned.

3. A 'tacking jig' needs to be built. This jig will hold firmly in the proper orientation all the members (tubing, gussets, endcaps, etc) of the frame-to-be. This jig needs to hold the parts in precisely the right spot. It is for tacking everything together exactly where it should be. It should not be used to attempts to get a warped frame back into tolerance. It should also not be used to attempt to restrain the work while welding (this causes more distortion of the piece).

.

4. I am assuming you are TIG welding. MIG isn't going to work well in this situation. There are some other more exotic welding processes that are similar to TIG that could work, but until I hear otherwise, TIG it is.

.

Purge the tubes and maintain good cleanliness.

.

Definitely make use of a pulser. A pulser can allow you to have a very tiny or almost no pool for most of the cycle, but a regular high peak that reliably provides full penetration then refreezes, before there is much opportunity to shift/distort.

.

Welding too slowly, welding too hot, and using excess filler rod will all increase distortion.

.

The pattern of welding on the frame and how each joint is performed often needs a little tinkering to find what works best for a particular design.

Systems that often result in low distortion are:

Welding each joint in a non-stop single pass (occasionally a brief readjustment halfway) OR welding each joint by completing single passes on sections on opposing sides (square tubing lends itself well to four: front then back, left then right., while round mitered tubing is sometimes broken up into 6 opposing section).

Completing all the welds of flat sections moving clockwise (or counter) to the next adjacent joint and moving from one side to the other can work on box like frames.

The thing to keep in mind is avoiding following welding a joint that pulls in a particular way with another joint that pulls in that same way (i.e. for a flat rectangular frame, you don't want to follow welding one corner by welding the corner diagonal to it next.

.

.

.

After completing the work and measuring, tolerances can be improved by torquing the frame, further grinding, or hand scraping among other processes.

I can't find fault with much except, "A 'tacking jig' needs to be built"

To be any good, this tacking jig will need to be built from a granite block.

And assembly and service temperature will also be critical. Thermal expansion over any distance will take everything out of tolerance.

Granite block....

Normally when one build/welds high tolerance weld assembles, we had used, used machine table beds from large milling machines. And they can be rather massive.

I had two of these beds. One being 60" x 90" while the other one was 84" x 164". And by being a machine bed, it had T-slots for fastening your work.

I don't believe that anybody would try to "build/welds high tolerance weld assembles" with commercial tubing, to those tolerances.

I'd build it with over-sized tubes and grind it into tolerance, if that's my only choice.

T-slots and clamps don't get you that kind of accuracy.

If its light gauge tubing, no. The clamping is not for holding or keeping it from warping. It's for placement prep and tack.

This is the current way we are using now, but this is definately not the wise way as when it turn too mass flow, the number of process involve which increasing the lead time are killing us...now still seek for an alternative way to solve this problem.

the number of process involve which increasing the lead time are killing us...now still seek for an alternative way to solve this problem.

In project management you had a rule of thumb from a list of three criteria's to pick from to determine your project success...........

• Cost
• Schedule
• Quality

Pick two!

Those are good points. The jig will not be cheap (though I do think cast iron might also be an option, it still won't be anywhere near cheap). It certainly needs to be kept at a constant temperature and build to be shielded from temperature fluctuations from tacking, etc.

The tacking jig won't be easy or cheap, but building it would make a closer approach to the specified tolerances more straight forward.

A ridiculous spec. The structure will move more than that under it's own weight and also if one side warms more quickly than another.

Never in my 20+ years in designing and fabricating large welded assemblies have I ever spec'd tolerances like that for tube frame structures...ever! First because it is not possible and second because it makes no sense that "support structure" requires tight tolerances. It's only the functional parts or interface features attached to that structure that would require the tight tolerance. The functional part/feature is final machined to tolerance after stress relieving the entire welded assembly.

If those tubes need to be perpendicular and parallel within .001 because they locate something then it's a s$%t design. You don't locate anything with tube. You weld on plates or blocks and use the tubes only as support structure. After welding, machine the plates or blocks to final dimension. If the tubes have enough wall thickness you may be able to machine the tube after welding but in most cases there's always too much distortion to allow for that.

I think this is a case of "some desk jockey newbie" (credit to Tmctech in #4) not knowing how to design.

From a relatively lay perspective in Fabrication (I am a Process Engineer), I would think that your 0.2 mm tolerance attained via fabrication method of welding on its own is quite good - even impressive. I cannot see how one can attain tolerences of 0.001 inch with welding on its own. My laymans' guess is that you would need some additional fabrication procedure or some sort of finishing method in order to attain such rediculious tolerances.

Another idea is that the client may have left out a good couple of zeroes in his tolerance spec. You should perhaps enquire about the final use of this fabriacted item. What does he need this hectic tolerance for? I have watched our own boiler-makers & welders in our workshops. I have not checked, but I doubt that they can achieve tolerances of 0.5 mm consistently on such items. In our applications, it probably is not necessary. We were doing new product development, prototypes that eventually lead to commercial products with small markets (1-10 units).

I hope this does not turn out to be a new product development excercise where you may have to develop a brand new Fabrication Technique. In this case, the client must bear some of the R&D costs (I would go as far as to say upfront). If your company is not into R&D or development work then you may have to turn down this request.

Kudos to TerraMan,

He has the correct solution. Redesign the frame with the mounting plate interfaces as TerraMan described and present your solution to your client. These welded on plates are commonly known in the tooling industry as "nut plates" and then machine these interface surfaces to the correct tolerances. We do this on a regular basis in the design of automotive assembly/welding line tooling which needs to have robust frames to hold the weight of the chassis/car but have tight tolerances in order to accomplish toleranced welds for locating sheet metal panels.

UFG

Is it possible OP is overlooking something on the spec dwg where it says something like "machining after welding permitted".? We're all jumping on the "stupid design" band wagon (myself included) but we still don't know what exactly needs to be perp an parallel.

I've seen weld filets spec'ed to .005. Junior "want to be" engineers, but everyone has to learn so I don't hold it against them. But if you quote the job like you were going to try to hold those tolerances, (which you can't) the price will be so outrageous the head engineer will look at it. It'll be a lesson for the young person.

LOL, yes I was trying to be polite and not ruffle feathers. But I will say, head engineers are almost always some of the smartest, down to earth folks you could meet.And I've dealt with meany. It is important not to disparage the young one, just point out the problem and the head man will take care of the problem.

There are a few comments that were down your lines of thinking. It was my first thought. Some are still trying to split hairs.

I feel for you, us, Ky.

By what everyone has said is good. Could it very well be a typo and it should be 02.5mm? With the material and size that would be a very good/easy tolerance.

The OP should sit down have a talk about it.

In my opinion you need to educate your client. Even IF you could attain the requested tolerance, it is unlikely that it will be maintained at all temperatures. The dimensions will be subject to expansion and contraction due to the surrounding environment, which will most likely exceed the tolerance.

I'm learning a great deal here. I am a Process Engineer with some exposure to design & fabrication (4 years). This was in product develiopment. I spent more of the time dealing with the clients but also in the workshops & putting together NPD Equipment. We often had clients that wanted rediculous service and in a short space of time. Even big clients with massive plants were unreasonable sometimes. Indeed we got exploited by a large plant through their clever use of scrope creep. We also had the decimal point mistake (many times). Then we had the case where a client takes out contract with us for supply fo the reactor vessel (only). But then later he sneaks in process & reaction design obligations.

I, myself, made the decimal point mistake in an exam once (factor of 1000) only to repeat the mistake later in reverse so that the two errors cancelled out.

As for educating your client. Good idea, but do so diplomatically & subtely. Also learn how to have frequent communication and honest communication with your client. This is the way that even extremely tense situations can be resolved. In our development work, which was in it heyday up to 2009, we found that often both client & our own development dept were lacking.

This often doubled up the problem. I documented around a dozen NPD projects I was involved in. Recently, I used this in my application to Pr. Eng. The exercise was great as it gave me insight into the internal problems we had. I noted the names of good & bad clients. I must say that I learnt a great deal out of this.

Do see this client and the highly demanding request on tolerance as a learning experience. You are there to learn from it all. If it is that the client really needs a seriously high spec and it is possible that someone in the world can attain it then you have a unique difficult task on your hands. If so then this is indeed development work. You must personally meet with the client and establish whether the guy is geniune or whether he is a chancer. This is the deciding step. Also, the chap with them money & control may not be knowledgeable in the area we are talking about.

This is very easy to establish. Also, what type of client is this. Is it NASA? Is it a low level outfit in the back waters of one of the farming states? And do they even have the money & backing for development work?

Do note : I try to put into practioce what I say. E.g. when I interact with our own boiler makers & welders, I know them well as i visit them at their homes over barbecue. Great if you could do similar with the client. Not barbecue to start off with but at least go right into his offices & workshops to see what he is doing. A few of our own suppliers do this & they know us inside out. If you do manage to attain the rediculously hiogh stabndards for a client, you should be entitled to share some of the benefits. Indeed you would want a special cut of the priofit on each of the items fabricated.

I have a database of previous customers & queries. I grade thes a scale of 1 to 5. The lowest being bad ones and the highest being the top-notch & trouble-free clients. And a zero for the ones that are pure chancers. Quite a dubuious category for a client to fall in. We remain interested - to eventually find category your "high tolerance" category your client falls into as per my grading system.

Note, if your client does turn out to be genuine and the problem is real, I am very very interested in following up how (and why) one may provide a service with such infinitesimally small tolerances. Please do let us know.

Some lateral thinking combined with ample suspicion - All tempered by my experience of having burnt my fingers because of my excessive empathy with dubuious clients . . . . has lead me to the idea that this contraption has the exact sizes & dimensions of a metal frame for the common old Personal Computer ! A sort of anti-theft frame for a PC as it were. I could be wrong, and if so, please do rap me ove the knuckles so as to avoid the client-burnt sections of my fingers.

Material prep is important since such an acute tolerance is required. Your stainless steel should be cut by saw blade instead of abrasive wheel to give you cleaner and straighter edge joints. This will then give you good precise seam welds.

JIG and CLAMPS must be pre-fabricated to ensure close precision and ease of weld. Use copper plates and copper angles in JIG and CLAMP fabrication. Copper material will help dissipate heat from the welding zone and reduce distortion of the work piece. For example, in the seam welding of 0.2mm or 0.3mm stainless cylinders for the vacuum flask industry, copper finger clamps are always used to clamp the longitudinal weld seams in order to dissipate the heat from the welding heat zone. These copper clamps are further cooled by water circulating through the clamps via a water cooling unit.

TIG (Tungsten Inert Gas) Welding is recommended. Tack welding must be performed on all seams to hold form the basic shape of your frame. This basic structure must be in place before full seam welding is performed on each required seam of the frame. Weld a seam at a time, starting at one corner and then proceeding to the opposite corner. Do not weld all the seams in a particular corner in one go.

Use continuous manual 'TACK / SPOT' welding to complete each seam weld if you just have a normal TIG machine (without pulse control). If you have a more advance machine with 'PULSE' welding, you can select this. Both these techniques offer lower heat input into your stainless steel and can reduce distortion of your stainless steel. Avoid using a 'CONTINUOUS' seam welding technique.

A typical saw blade (I'm imagining a band saw) isn't going to cut to those tolerances.

.

Additionally, because the outside corners of square tubing usually don't come to a sharp edge and are instead rounded, cutting flat across would leave two walls of the tubing with gap that would introduce all kinds of stress and distortion.

.

The tubes need to be mitered to fit precisely. This can be done in a number of ways; EDM, experienced hand filing/fitting, grinding, or on a milling machine, for example.

.

.

Another important point to note is that many weldments do not respond well to attempts at restraint via clamps and jigs during welding. Restraining the weldment adds significant stress and tends to exacerbate distortion. This will become apparent when it springs from the jig.

.

I also have my doubts any changes in heat transfer that occur from copper clamps being beneficial more often than not for the purposes of minimizing distortion. Creating small quenched areas where the contact happens to be decent in the HAZ isn't going to create a uniform effect around the tube.

Yes, the tubes need to be precise whether it be through saw cutting, milling, etc. However, the greatest problem here is the end result, the acute tolerance. Most production of these sort of frames require some sort of JIG and CLAMP system. There is no running away from this fact. Weldments do not respond well because heat is not dissipated from the work material effectively. Distortion occurs because the HAZ is exacerbated by following silly weld procedures. The selection of weld process (tack or pulse welding) is just as important as material preparation. Pulse welding brings about a smaller HAZ and lower material distortion as compared to straight continuous weld. Welding a seam at a time rotating a corner at a time will reduce heat input at a particular corner thereby reducing distortion. Copper JIGS and Clamp fingers do immense wonders. Your doubt is unfounded. All auto seam welders of thin gauge stainless material use copper clamp fingers to control heat and weld distortion. We have welded 0.2mm or 0.3mm 304 Stainless with minimal distortion. So I don't see issues with 1.5mm stainless tubes. Look at the entire picture and not just at snippets.

I am a big proponent of pulsers, as you will see from my previous comments here. So you will find no argument from me there.

.

However you are incorrect both in your assumption about my concerns being unfounded and about your dismissal of restraint induced distortion. I have some experience building lightweight structures from thin wall tubing (chromoly, aermet, aluminum and even titanium) for which alignment was critical.... you see, may years ago, back in my welding career, I built custom road bike and mountain bike frames.

.

I can tell you unequivocably that restraint causes unnecessary stress and distortion. The dynamics are not that difficult to understand.

.

As far as clamps made of copper, I can think of a number of reasons other than heat dissipation, including copper being a good conductor (of electricity) and copper being fairly soft (not marring the workpiece so easily).

.

You are right in another aspect as well. HAZ is exacerbated by 'following silly weld procedure'....one of which would be attempting to restrain while welding a frame of thin wall tubing.

Your arguments of jigs and clamps are noted. Welding and work process may have to be lengthened to allow sufficient cooling, re-positioning and re-measurement of work piece. However, looking at the situation once more, the tolerances may be way too difficult to achieve.

agreed.

Another important point to note is that many weldments do not respond well to attempts at restraint via clamps and jigs during welding. Restraining the weldment adds significant stress and tends to exacerbate distortion. This will become apparent when it springs from the jig.
We lost a very expensive front foil on a Boeing PHM boat. The foil actually sprang when taking the assembly out of the oven and destroyed the fixture and the foil. I was sent to evaluate the foil and ended up scrapping an expensive piece of 17-4 ph hardware and we had to re design the HT fixture. Things move when normalized.

syeltang87

Are you still with us Mate?

Normal expansion/contraction in a day to day environment will distort this frame beyond stated tolerance.

Use these technique to control the welding distortion
1) Tack welding2) Back-to-back assembly3) Stiffening

Thanks TB, I will be able to discuss this with our workshops. We don't work anywhere near these levels of tolerance.

But the above discussion has given me good insight into the subject. Previously the discussion group agreed that a temperature variation of even 5C would destroy any tolerance in the range quoted above (thru expansion, etc). Do you think that this tolerance is even achievable in the first place (0.001 of an inch)? My layman's feel tells me it does not sound possible.

The tolerance is attainable if restricted to being measured in a temperature controlled environment.

It would not be cheap, easy, or easily adapted to mass production.