The creator opinions a weld program on the educate pendant. Photos: Triple Crown Trailers
On Monday morning, the gates regarded straight. By Wednesday afternoon, they didn’t. Operators at our Ocala, Fla., store have been prying on members with bars, stacking shims in a rush, and calling for high quality management (QC) each different pallet. The rework cart by no means had time to chill. We’d invested in a brand new robotic cell for the principle gate frames that went on to our automotive and business trailer merchandise. Certain, the cycle time dropped, however geometry drift stayed cussed, like a bubble beneath wallpaper.
“Possibly we simply have to tweak this system,” somebody stated.
We tried. The bubble moved.
That week taught us one thing easy and painful: We hadn’t purchased a robotic drawback; we had a system drawback. The cell would by no means be predictable till half presentation, warmth, code, and behavior labored collectively. That is the story of how we turned the nook, and the checklists we constructed so we wouldn’t lose the thread.
Why Robotic Welding?
Our administration group at Triple Crown Trailers didn’t need to use robotic welding for its personal sake. Certain, advertising photographs of a robotic weld cell look good on our web site as they create our numerous car- and utility-trailer merchandise. But when they didn’t profit from our welding expertise and improve general throughput, what can be the purpose?
We noticed a possibility for robots to weld the frames for the principle trailer gates (see Figures 1 and 2). The repetitive welds took a variety of time and didn’t make greatest use of our expert welders.
For a similar motive, we additionally wished to keep away from counting on fixtures that required tack welds. Our welders wished to weld, not tack all day. And moreover, the cycle time that tack welding added would cut back general throughput. If robotic welding extended the general cycle time and elevated labor necessities, why not weld the frames manually?
Then again, if we might design our fixtures and weld processes in order that staff might merely load merchandise into the fixture, provoke this system, and offload welded elements, throughput would rise—which, in fact, is the whole level of course of automation.
The oblong tubes on this utility are fixtured within the as-received situation, with no in-house weld prep. Our tube provider saw-cuts them at 90 levels (no bevel) and sends them to us, which we then examine, stage, and fixture for robotic welding.
We relied on teach-pendant programming and didn’t get pleasure from offline weld course of simulation. On reflection, we would have benefited from offline simulation, however regardless, after some laborious classes discovered, we prevailed.
FIGURE 1. The primary trailer gate consisted of an oblong tube meeting. Every tube was saw-cut and delivered to us in particular lengths.
Set up the Baseline
As a primary step, we stopped guessing, tweaking the robotic program, and guessing once more, solely to have QC regularly ship rework again to the weld cell. As a substitute, we spent a full week measuring. Sheets and tubes have been tagged as they arrived. Joint gaps have been mapped with pictures and sticky notes. We saved monitor of the clamp-unclamp instances and logged each rework mode: incomplete fusion right here, an undercut there.
Two findings modified every little thing. First, the fixture was arguing with physics; it held elements so completely that there was no room for thermal development. Second, our weld sequence fought our personal geometry. We pulled lengthy members towards the recent aspect and by no means allow them to calm down.
The robotic wasn’t the issue; it was us, faithfully repeating our errors. To keep away from repeating previous errors, we developed a guidelines to undergo earlier than making any adjustments to the robotic program:
-
Document any incoming materials variation. This included the sq. tube’s straightness, edge prep (the noticed lower high quality),
in addition to the radii of the tube’s outdoors corners. Each tube obtained from the mill has a tolerance that we would have liked to account
for within the robotic cell. - Map fit-up tolerance stackup by joint household and marked the place individuals “made it match.”
- Categorized the weld defects, be they porosity, undercut, incomplete penetration, or anything.
- Time the non-arc duties, together with clamping, unclamping, and half repositioning.
- {Photograph} and tag all the recent spots, or areas the place visible weld examination revealed we had important warmth enter.
All this gave us a baseline. We now had the knowledge we would have liked to cease guessing and begin bettering.
Fixtures That Respect Warmth
We discovered that any fixture we used wanted to respect the truth of thermal distortion. Our first fixture “enchancment” was mainly a vise, which actually simply over-constrained the half. Our welded gate merchandise left the fixture wanting excellent, however they curled on the cart 10 minutes later. The fixture hid actuality.
Subsequent, we rebuilt round 3-2-1 datum logic. This basic fixturing precept includes finding elements on the first (1), secondary (2), and tertiary (3) planes (see Determine 3). The first finding aircraft (datum) is your baseline, and each part inside the fixture ought to have a direct relationship to it. A fixture can have dozens of elements, and the very last thing you need is for one part to turn into “trapped”—that’s, have a number of different elements blocking that direct path to the first datum. This makes it subsequent to inconceivable to uncover the basis trigger behind errors.
On our new fixture, locators pressured right presentation; clamps allowed a managed slide as warmth gathered. We designed go/no-go options into the fixture, so a mistaken load was apparent earlier than the arc began. For the whole gate household, we added swappable locators for simple changeovers between totally different variants within the product household.
The primary setup we examined advised us we have been shut—much less persuasion with bars, fewer “maintain right here whereas I tack that” moments. And from all this, we developed the next guidelines:
-
Name out major, secondary, and tertiary datums on the print. - Depart a level of freedom the place thermal development must go.
- Construct mistake-proofing (poka-yoke) and go/no-go factors into the fixture.
- Plan entry (torch angles, cable sweep, reamer attain).
- Standardize put on objects; put alternative intervals on the job traveler.
- Use modular locators for households to shorten changeover time.
Schedule for Warmth
With presentation beneath management, distortion shrank however didn’t vanish. The final mile was warmth self-discipline. We rewrote the weld sequence so the gate breathed. We alternated from one aspect to the opposite (see Determine 4). The welding gun moved from a degree of restraint (that’s, sure components needed to stay the place they have been) to factors of freedom, the place we accounted for motion from thermal development.
We additionally stopped treating an enormous fillet as an indication of expertise. As a substitute, we certified the smallest bead that met the weld necessities (see Determine 5) and added interpass home windows and dwell instances to our digital work directions. We didn’t need to depend on tribal reminiscence. All the pieces was on display screen.
FIGURE 2. The tube-to-tube connection on the trailer gate created a joint with an open root. Operators examine elements for consistency, together with the skin dimension, wall thickness, and the radius of every of the sq. tube’s 4 corners. Any change impacts the joint geometry.
The following eight-hour run felt unremarkable. The robotic welded and elements shipped. No fuss. No drama. That was the purpose. And from all this, we developed one other guidelines, this one for controlling distortion and discovering the optimum weld sequence:
- Stability shrinkage with alternating passes; plan to maneuver from restraint to freedom.
- Qualify the bead measurement per the weld necessities; cease oversizing welds.
- Set preheat and interpass limits for important joints as essential, and really measure them.
- Construct cool-down into the routine the place geometry calls for it.
- Confirm straightness on the cart, not simply within the fixture.
Be aware that we wrote this (and all of our checklists) to function a baseline for present and future robotic welding work. As an example, our gate half household didn’t require any preheating, however we knew that some merchandise might require it, so we documented it for future reference.
We additionally wrote a “sanity examine” record to make sure we have been inspecting and measuring the weather that mattered:
- Pull the half sizzling and examine it on the cart.
- Look ahead to sprung elements, indicators your fixture is masking development.
- Examine locator put on. A mushroomed pin is a brand new datum you didn’t plan for.
Program for the Actual World
Our first weld packages assumed excellent inputs. The actual weld cell wanted cushions to account for the actual world. We added fallback routines the place joints drifted most, so the welding gun might discover the seam when variation crept in. As an example, we’d know that thermal development from sure welds would drive the meeting to float in a single course by about one weld-wire diameter (0.045 in.). So at this level, as a substitute of centering the gun place, the programmer would transfer the gun barely over. Expert welders account for this drift naturally; the robotic simply must be taught what to do.
We modeled not simply clamps and locators, but in addition the welding gun costume pack that protects its cables and hoses; we would have liked the gun to keep away from not solely the half but in addition itself. And we tied program revisions to the half drawing and fixture IDs. All the pieces—the drawing, the robotic code, the fixture—“moved” collectively. From all this, we developed a guidelines for our robotic weld programming and verification course of:
- Add joint discovering/contact sequences at identified drift factors to account for thermal development.
- Embody clamps, locators, and cables when modeling for attain collisions.
- Validate angles/approaches throughout actual tolerance ranges.
- Standardize consumables; change by arc-time, not guesses.
- Lock revision controls throughout drawings, fixtures, packages, and digital work directions.
Procedures Folks Really Comply with
We wrote welding process specs (WPSs) and process qualification data (PQRs) in order that they aligned to widespread observe, after which educated to the process, to not behavior.
The primary-piece examine measured two or three important datums on the cell utilizing easy, rugged gauges.
Our digital work directions confirmed clamp order, security notes, and rev IDs on the display screen subsequent to the work. If a drawing modified, the job traveler modified.
No person requested, “Which model are we on?” From all this, we developed the next guidelines for course of management, to make sure our clean operation remained a clean operation:
- Create WPSs/PQRs by joint household; qualify on consultant coupons.
- Practice and re-qualify on a cadence.
- Do first-piece checks on the station; maintain gauges easy.
- Use digital work directions with clamp order, security notes, inspection factors, and rev ID.
- Tie invoice of supplies, routing, drawing, digital work directions, and welding robotic packages to 1 supply of reality.
Comply with the Physics
After eight weeks, the info advised us we have been on the correct path. Cycle effectivity climbed into the mid-80% vary—and stayed there. This degree gave us the effectivity we would have liked whereas nonetheless leaving time to account for warmth distortion. First-pass yield on important datums stabilized, and distortion-driven rework fell by greater than half.
FIGURE 3. This fixture holds axle elements, not the trailer gate body, nevertheless it illustrates the datum logic we use. The blue is the first datum, which by no means strikes. Each different clamp level (in grey) has a direct path to it. When you discover an issue, you relate it again to the first datum, which helps uncover the basis reason for positional error.
As soon as all of our gate product households shared fixture logic and visible work directions, changeover time dropped considerably. We saved tabs on our KPIs, together with cycle time, rework charges, and changeover time measured from the fixture swap to the primary good half.
What didn’t change? Metal nonetheless moved when heated. Folks nonetheless did what the system allowed.
And we couldn’t beat physics. We adopted physics and, therefore, made the method predictable. As a substitute of over-constraining the half, the brand new fixture design left one diploma of freedom and let the sequence pull towards the clamps designed to soak up it. As soon as we let the half speak, it advised us what it wanted.
Immediately, the operator who used to shim quietly to “make the fixture work” now’s the one who finds missed locator put on factors and different sources of variation that the engineering group missed. For security, upkeep now performs a weekly, 10-minute interlock examine. And with complete digital work directions, a brand new rent can run the cell after simply two days of coaching. The method particulars reside on display screen, not in somebody’s head.
The underside line: The welding robotic didn’t save us; engineering the system did. The robotic simply makes the selections we make repeatable.
