consistancy in fabrication of multiple units

huangyhg

consistancy in fabrication of multiple units
today the shop manager approached me and stated there was too much variation between each heat exchange unit being built. these composites are layed up by hand from an outside source, and are fabricated using faceplating by three different technicians. any advice or ideas on how to have three different technicians be able to cut and build the fabricated parts with no variation so as one technician's part is exactly the same as the next technician's part? it seems like an impossible task to me with just the mere fact that it is what it is: a fabricated part. although the units are all within tolerance they are not happy with the variation. please, does anybody have any experience and advice to lend on how to take a completely hand built part, built by multiple technicians and have the units turn out exactly the same???
find a job or post a job opening
i have little experience with composites, but some with technicians, and i think there is a way.
[ at my very first job ever, i made simple architectural drawings for a consulting engineer.  his stuff looked gorgeous, mostly because all the (freehand) lettering on every drawing appeared exactly the same, as if the same person had done every drawing in a set.  that was, of course, not true.  it was just a 'house rule' that everyone in the shop had to letter exactly like the chief draftsman.  and we did.  only the shop crew could tell who had done a particular drawing without looking at the title block.  there were individual differences, but they were subtle. ]
you can get your crew to buy into the idea of making the parts consistent, not by making it a 'quality' issue, but by making it a 'pride' issue.
"those clowns downstream think we're amateurs because of the cutting and layup variations they can see.  let's make it impossible for them to tell who did a particular layup, by developing a consistent style, and making every part appear to be exactly the same.   please agree among yourselves how you're going to do that, and make a record of it, with sketches, drawings, photographs, whatever it takes."  then give them time to do it.  make it an official project to the extent that you can.  in other words, give them some attention, and take advantage of the hawthorne effect.
"at the same time, i'd like to be able to tell at a glance whose work i'm looking at, so at the same time, please work out individual 'signatures' that you can work in without outside detection.  they have to be distinct from each other, but subtle enough that you need to be looking for them to find them."  
again, ask them to work together, and to fully document their efforts, and treat them as if their participation is valuable... because it truly is.
  
mike halloran
pembroke pines, fl, usa
" although the units are all within tolerance they are not happy with the variation. "
then the tolerances are wrong.
cheers
greg locock
i'd agree with greg ... if the parts are to spec (within tolereance) then they're good.  you own thought is .. is it the customer who's unhappy or the shop manager ?  if the parts work like they're supposed to then the customer should be ok with them.  possibly the customer wonders how these things that look different are going to perform with respect to the test articles ? (which could well be a valid concern)
what sort of "variation" are you talking about ... obviously nothing too physical, maybe just appearence ?
personally i don't think you're going to get the tech's to agree on one way to do the job, they each have their own experience and preferrence for doing things a particular way (for both rational and irrational reasons), but none of them are wrong.
i gues you could ask the shop manager what the specific problem (variation) is, and talk to the guys about it ?
after reading your post, i am curious as to exactly what type of variation that you are talking about. i'm going to make a couple of assumptions here. first, you are probably getting variation on the iml (bagged) side of the (out of autoclave) layup. i will assume that you have your ply orientation defined on your drawings, so that should not change between parts. possibly the variations is in the overlaps between pieces of kitted plies. one possible solution, if this is the case, is to cut your plies to templates. it's a fairly easy thing to create the template. once the template is made, each piece of the layup should be cut the same. the templates should reference your tool geometry (scribe lines), so you can consistently lay up each kitted ply. then each ply should be laid up from the ref geometry in your tool, outward beyond the edge of part. next you should define the maximum overlap between plies, or call for butt splices where you can.
this could get you much closer to the uniform part. keep in mind that you will never get a 100% identical part in an out of autoclave hand layup. but you can get pretty darned close.
obviously this statement makes a lot of assumptions, but if you post what the variation is maybe we can help some more.
wes c.
------------------------------
no trees were killed in the sending of this message, but a large number of electrons were terribly inconvenienced.
some pretty good general advice above, based on very little specific information.
we'll need more details to make more detailed suggestions, and perhaps you need more information as well.  here are a few things to consider:
greg noted
" although the units are all within tolerance they are not happy with the variation. "
then the tolerances are wrong.
and that very well may be true.  but consider this.  is it the shop manager's responsibility to define the tolerances or performance requirements for the heat exchanger?  is the shop having a specific problem that they believe is caused by part-to-part variation?
you seem gung-ho to go out an make the vendor make changes to his process.  you don't know what changes to ask for.  you don't know what the problem is.
this approach will add lots of cost to the part, very likely without adding any value.
step one in solving any problem is understand the problem.
thank you for all your replies so far. here is a bit more information on the situation. the horizontal and vertical points are measuring a thousandth of an inch off from corners to corners. the edges of the composite may not be sanded down consistantly enough from one tech to the next ie: the angle may look slightly different from the previous although as previously stated it all still falls within tolerance. i'm not aware of any tolerance issues pertaining to the finished edges of the composite. there are some tolerance problems but that has to do with the positioning of the nutplates. such as if we went with maximum tolerance in one direction with the nutplate row, it would result in the row of nutplates being out of tolerance. anyway, apparently what it's coming down to is the client wants a machined look to a handbuilt fabricated part by multiple technicians. being a hand built part we have to take advantages of the tolerances in order to make it work. maybe the client is being unreasonable in their request. i did think about building a jig to place the cut composite in and finish sanding to the edges to get a uniform part everytime that way.
ignore what i said earlier; i misread that to say they were _your_ techs.
a thou?  sanding?  yikes!  you won't get that even in a jig.
mike halloran
pembroke pines, fl, usa
as a practical note i hate sanding anywhere near finished composites. the danger of functional damage to a fibre is too great.
if the customer wants a machined appearance then you need to use a female mould in that area. i suggest you get the customer in, and see what exactly it is that they want. then give them the options. if you need to retool to meet their new requirements then that will cost them. if you can work out some acceptable compromise with your current process, great.
as usual most engineering problems are to do with cmmunication, not maths.
cheers
greg locock
quote:
the horizontal and vertical points are measuring a thousandth of an inch off from corners to corners.
what does that mean?  horizontal and vertical points of what?  off with respect to, or from what?  you stated earlier that everything fell within the allowed tolerance.  yes or no?
quote:
anyway, apparently what it's coming down to is the client wants a machined look to a handbuilt fabricated part
so it's an appearance issue only?  how about a nice pretty piece of glue-on trim?
otherwise, get the client to write down exactly what he wants you to change, and exactly where in the specification or contract it states that his request is a defined requirement.
if he can't do that, offer to price a change order.
at the risk of appearing to change the subject:  increasingly we are seeing fab/job shops being asked to fabricate to tolerances that are at or near machining tolerances.  some of this undoubtedly is due to the demand for more quality, but often absent from the effort is a realization (on customers, management, or even engineering's part) that such efforts require improved, perhaps special, methods, or the addition of a machining step, often without the ability to machine the assembly as a whole.   
some of the tolerances might be close to, or actually, impossible.  the one that comes to mind is a pump-skid production project (mild steel fabrication, square tubing and plate stock), where the skids are 46 feet long, 18 feet wide, and 18" high.  tir spec from end to end (along the 46' length) is 1/16 of an inch (~ 0.060"), as fabricated.  is this a reasonable spec?
bk
is there any written or other formalized requirement?  as others picked up on you say they meet tolerance, so presumably they are functionally ok?  in the original requirement did the customer require them to be visually 'indistinguishable'?
however, the above just means that contractually the client probably can't reject the parts.
if in the future though you've decided you need to try and please the customer then you'll want to update your product documentation to reflect this requirement.
is possible to tigten the tols to achieve what your client wants?  if you tighten the tolerance to meet the customers visual requirement are they still manufacturable with the same process/cost?
create a detailed spec or work instruction like mike h describes.  make sure and get input from the techs involved etc.  reference this on the drawing/mbd that the part shall comply with or be made in accordance with that spec.
if at present each techs parts are acceptably similar, it's just the variation between techs that's the problem, then this may be adequate (when backed up with qa).
one of our sister sites has 'cosmetic parts' where they have a number of requirements for the parts to basically look pretty.  cosmetic parts are identified on the drawing with a reference to the spec that gives the requirements.  can't recall the details, sorry.
kenat, probably the least qualified checker you'll ever meet...