How to read a weld penetration report for a seat frame welded assembly

Somebody has sent you a PDF with a dark, glossy photograph of a cut weld on it, a few millimetre figures, and a stamp. It is called a weld penetration report, or a macro section report, or a fusion depth report, depending on who wrote it. You have to decide whether it tells you that the supplier's welds are sound. Short answer: on its own, it does not. It tells you what one weld, made at one moment, on one sample, looked like below the surface. Whether that is evidence about your production parts depends almost entirely on fields that sit around the photograph — when the sample was taken, from where, against which criterion, and under which fixture and programme. This guide is about reading those fields, because the photograph is the easy part.

We are a welding shop, not a laboratory: Wuhan Yaxin Auto Parts Co., Ltd. (武汉亚新汽车零部件有限公司) makes seat cushion and seat back wire frames, bent tube sub-assemblies and the laser-cut brackets that go into them, and our quality department issues these reports part number by part number for the parts we build. So this is written from the side that produces the report rather than the side that files it. That also means we will say where the document is weak, including where ours would be, which is not a thing a laboratory brochure will tell you.

Skip the theory and send us a drawing →

What a weld penetration report is, and what it is not

A weld penetration report is the written output of a destructive metallographic examination. A welded sample is cut across the joint, the cut face is ground and polished, the polished face is etched with an acid solution so that the melted zone becomes visible against the unmelted parent metal, and the resulting boundary is measured and photographed under a calibrated microscope. The sample is consumed. It does not go back in the box.

Three consequences follow, and all three are routinely misread:

• It is not an NDT result. Dye penetrant, magnetic particle and ultrasonic testing are non-destructive and can, in principle, be applied to shipped parts. A macro section cannot. Your production parts are never sectioned.
• It does not certify a lot. It certifies that the process — machine, wire, gas, programme, fixture, operator — was capable of producing that fusion profile when the sample was made. Extending that to the container on your dock is an inference, and the strength of that inference is exactly what the surrounding fields are for.
• It is not a strength test. Fusion depth correlates with joint strength but does not equal it. Pull-off, push-out, torque and fatigue testing measure strength. A frame that passes a section can still fail a fatigue test, and the reverse also happens.

Why the seat industry asks for it at all

Welding is the standard textbook example of a special process: a process whose output cannot be verified by subsequent monitoring or measurement of the finished product. You can look at a finished seat frame all day and not know whether the bead on a bracket fused into the tube beneath it or merely sat on top of it. A cosmetically perfect weld with almost no fusion — a cold lap — is a real and common failure mode, and it is invisible from outside.

On a seat structure the stakes are unusually concentrated. The welded joints on a cushion frame are in the load path between the occupant and the vehicle floor. They are qualified not by appearance but by the seat maker's own structural and fatigue programmes, which are expensive and late. Finding out at seat-level testing that a frame supplier's fusion was marginal is the worst possible place to find out. That is why the penetration report gets asked for at sourcing time rather than at launch: it is one of the few pieces of evidence a buyer can look at before committing, and it is an evidence type that is very hard to fake convincingly in front of somebody who knows how to read it.

How the sample gets from the welding line to the photograph

Six steps, and each one is a place where the answer can be shaped. Knowing the sequence is what lets you ask the right question about a report you did not watch being made.

1. Sampling. A welded part — or a representative coupon — is removed from production. Which part, from where in the run, and welded under which fixture and programme, is the single most important fact in the entire report.
2. Sectioning. The joint is cut, normally on a plane perpendicular to the weld axis and through the feature of interest. The cut must be made in a way that does not itself alter the metal — abrasive cut-off under flood coolant, not a grinder that heats the face it is about to reveal.
3. Mounting. The cut piece is set in resin so the small, awkward geometry of a wire-to-tube joint can be held flat and square during grinding. Skipping the mount is how you get a face that is polished out of plane, which quietly exaggerates or shrinks every dimension measured on it.
4. Grinding and polishing. Progressively finer abrasives until the face is a mirror. Scratches left at this stage are read as flaws at the next one.
5. Etching. A dilute acid is applied — for the plain carbon and low-alloy steels used in seat frames, an alcoholic nitric acid solution is the usual choice. It attacks the fusion zone, the heat-affected zone and the parent metal at different rates, so the boundary between what melted and what did not becomes visible to the eye.
6. Measuring and photographing. The fusion boundary is measured under a calibrated instrument and photographed with a scale in the frame. A macrograph without a scale bar or a stated magnification is decoration, not data.

Seven things to look for on the macrograph itself

Assume you have the photograph in front of you. In rough order of how often each one is the thing that is actually wrong:

1. The scale. Find the scale bar or the magnification statement first. If there is neither, stop reading the photograph; every millimetre figure elsewhere in the report is now unverifiable against the image that supposedly supports it.
2. The fusion line, not the bead outline. The eye is drawn to the shiny cap of the weld. The line that matters runs below the original joint surface, where the parent metal melted and re-solidified. On an etched face it is a distinct, usually curved boundary. The bead sitting proud of the surface tells you almost nothing.
3. Which member the depth is referenced to. Penetration is a depth measured from somewhere into something. On a joint between two parts of different thickness — a φ2–φ14 mm wire against a φ8–φ52 mm tube wall, say — depth into the wire and depth into the tube are two different numbers with two different meanings. A report that gives one figure and does not say which is ambiguous by construction.
4. Leg length versus throat. On fillet welds these get confused constantly. Leg length is measured along the two joined surfaces; throat is the shortest distance from the joint root through the weld to the face. Leg length is easy to grow by adding filler without adding a single millimetre of fusion. If your criterion is written in leg length, it can be satisfied by a weld that never fused.
5. Root fusion. Look at the deepest corner of the joint. Lack of fusion at the root is both the most damaging defect on a cyclically loaded frame and the easiest to miss, because it appears as a fine dark line that a low-resolution scan of the report will erase entirely. Ask for the original image file, not a photocopied page, when this matters.
6. Discontinuities in the fused zone. Porosity, inclusions, solidification cracking. A few small pores are normal on production welds; a chain of them along the root is not. Your criterion should say which of these are acceptable and at what size, and if it does not, that is a gap in your requirement rather than in the supplier's report.
7. Heat-affected zone width. The band around the fused metal that changed structure without melting. Its width is a rough proxy for heat input. A very wide HAZ on a thin-wall tube is worth a question, because heat that went into softening the tube did not go into fusing the joint.

Wire-to-tube joints are not the plate joints the standards assume

This is the part that gets skipped, and it is the reason two honest suppliers can section the same joint and report different numbers.

Essentially every convention for expressing weld acceptance was written with flat plate or sheet in mind: two surfaces meeting over an area, a constant gap, symmetric heat flow. A seat frame joint is usually round wire meeting round tube, or round wire meeting a formed bracket edge. Three things change:

• Line contact instead of area contact. Two cylinders touch along a line. A section taken through the middle of that contact and a section taken a few millimetres to one side are looking at physically different joints, because the gap that the arc had to bridge is different at the two places. This is not measurement noise; it is geometry.
• Fit-up is driven by bending tolerance, not by the welder. The gap at the joint is whatever the wire bender and the tube bender left there, held by whatever the welding fixture clamps. On a good day the gap is near zero and fusion is deep. On a part at the other end of the bending tolerance band the arc spends its energy bridging. In our experience the fixture and the incoming bend tolerance move penetration around more than the weld current setting does — which is why a report that does not name the fixture is a weaker document than it looks.
• Asymmetric heat sink. A φ52 mm tube conducts heat away from the joint far faster than a φ4 mm wire does. The fused zone therefore grows preferentially into the smaller member. That is normal and usually harmless, but it means "50% penetration" measured into the wire and the same phrase measured into the tube describe wildly different welds.

The practical consequence for a buyer is one sentence: on wire-and-tube structures, insist that the report states the section plane and the reference member, and treat any report that does not as unreadable rather than as passing.

Where the acceptance number comes from — and where it does not

There is no universal penetration figure for seat frames. The criterion comes from one of three places, in this order of precedence: a note on your drawing; the weld standard your drawing invokes; or the customer-specific requirements attached to the programme. Seat makers differ from one another on joints that look identical, and the same seat maker can apply different criteria to a cushion frame and a back frame in the same seat.

We deliberately do not publish a house penetration figure on this site. Printing one would imply that we weld to our own criterion rather than to yours, and a supplier quietly substituting its own acceptance limit for the customer's is a much bigger problem than a missing number on a web page. Where a drawing reaches us without a criterion, we ask before quoting rather than after. That is not procedural politeness: the criterion sets the weld schedule, the schedule sets the cycle time, and the cycle time is a line in the price. A quotation issued against an unknown acceptance limit is a quotation that will have to be reopened.

One warning about criteria expressed as a percentage. "Minimum 50% penetration" is only meaningful with the denominator attached — percent of wire diameter, of wall thickness, or of the thinner member. Ambiguity here is the most common genuine defect we see in incoming requirements, and it is worth fixing in your own documents regardless of who ends up welding the part.

The eleven fields a usable report carries

Print this and hold it against the next report that arrives. A report missing items 1, 2, 3 or 9 is not evidence about your parts, whatever the photograph shows.

#	Field	Why it decides whether the report means anything
1	Part number and revision	A report against a superseded revision may describe a joint that no longer exists.
2	Joint identifier and location	"Weld 4 of 11, bracket to left rear tube" — not "the weld".
3	Sample date, shift, and point in the run	First-off, mid-run and last-off carry different information. A sample with no date is untraceable to any production event.
4	Fixture identity	The fixture holds the fit-up. Change the fixture and you have changed the joint.
5	Machine and programme identity	Which robot, which programme revision. Programme edits are the quietest source of drift on a welding cell.
6	Section plane, sketched or photographed	On curved-to-curved joints this is not optional. See the section above.
7	Etchant and method	Tells a reader that the dark zone in the image is the fusion zone and not a polishing artefact.
8	Macrograph with scale bar or stated magnification	Without it the numbers cannot be checked against the image.
9	Measured values and the acceptance criterion, side by side	A measured value with no criterion next to it is a number, not a verdict. A criterion filled in after the measurement is worse.
10	Instrument identity and calibration status	A measured 1.4 mm from an uncalibrated instrument is an opinion.
11	Named inspector and disposition	Somebody has to own the conclusion, by name, with a pass or fail written down.

Five ways a penetration report can be entirely true and still worthless

None of these require anyone to lie. Every one of them we have either seen, or could have produced by accident.

1. The golden sample. The section was taken from a part welded off-line by the most experienced technician, on an unclamped bench, specifically to be sectioned. Everything in the report is accurate. It describes a part that has nothing to do with production. Defence: field 3, plus asking who welded it.
2. The flattering section plane. On a curved joint, the plane through the deepest fusion is a real plane and a report taken there is a true report. It is also the best possible answer rather than a representative one. Defence: field 6, plus asking for two planes on the first article.
3. The criterion written afterwards. The measurement is made, and then a limit comfortably below it is entered in the acceptance column. This is depressingly easy to do without anyone intending fraud. Defence: the criterion should be traceable to your drawing or your standard, not typed into the form.
4. The orphaned revision. An immaculate report, against a part revision that changed six months ago when a bracket moved 8 mm. Defence: field 1, and checking it rather than assuming it.
5. The unrepeatable fixture. The report is from a period when a temporary fixture was in use, or after a fixture repair that was never re-validated. The welds were genuinely good and are no longer reproducible. Defence: field 4, plus a control plan that treats fixture repair as a trigger for a fresh section.

How this works at our plant

Welding at Wuhan Yaxin runs on 7 welding robots on the arc and spot lines, alongside 20 TIG sets, 2 projection welders and 4 spot welders for the joints and volumes where a robot cell is the wrong tool. Wire forming covers φ2–φ14 mm wire and tube working covers φ8–φ52 mm tube, so the curved-to-curved joints described above are the ordinary case here rather than an exception. Our quality department holds weld penetration and metallographic destructive testing capability in-house, together with a tensile tester and a torque and fatigue test rig, and issues penetration reports part number by part number. We are certified to IATF 16949:2016 for the manufacture of seat frame assemblies, with design responsibility excluded — which in this context means the acceptance criterion is yours to set and ours to meet, not ours to choose.

Two honest limits, stated here rather than discovered later. First, we have not yet made a PPAP submission directly to an overseas customer; our export shipments to Mexico, Thailand and Malaysia have gone through the customers' own channels. We can build the package and we would rather you know that it would be our first direct one. Second, surface treatment — e-coat and zinc plating — is the one process we place with qualified outside processors, so a penetration report from us describes the joint as welded, before plating, and any question about coating in the weld area has to be answered jointly with the processor.

On a new programme the sequence we would normally propose is: sample parts in 10 working days for a welded assembly, which includes building the welding fixture the sample needs; sections taken from those sample parts against your criterion, in the report format agreed beforehand; and 3D-printed check fixtures in-house so your dimensional loop is not waiting on hard tooling while the weld work is being settled. Wire-formed parts with no welded joint run to a 3-working-day sample instead.

If you are the one writing the requirement

Most of the trouble in this area is not supplier misbehaviour; it is requirements that were never specific enough to be met or missed. If you are drafting the weld section of an RFQ for a wire-and-tube structure, four sentences will remove most of it:

• State the acceptance criterion with its denominator — depth in millimetres, or percentage of a named member.
• Name the joints to be sectioned by identifier, not by "critical welds".
• Require the section plane to be stated, and on curved-to-curved joints require two planes at first article.
• List the triggers for a fresh section in the control plan — fixture repair or replacement, robot programme edit, material lot change, torch or tip change, restart after extended stoppage — rather than relying on the scheduled frequency alone.

Any competent supplier can price that. Very few can price "welds to be of good quality".

Common questions

Send the drawing

If you are evaluating a wire-and-tube frame supplier and want to see what our reporting actually looks like rather than read about it, send the drawing as PDF, STEP, DWG or DXF with your annual volume and your SOP date. Within 3 working days you get back either a three-line quotation — part processing, tooling amortisation, outsourced surface treatment — or the precise list of what we still need to know, and we will include the penetration report format we would issue against that part number so you can check it against your own requirement before anything is committed.