A salon chain ordered 240 styling stools with circular footrest rings. The rings were welded onto chrome-plated columns, because the ring is fitted at final assembly and the columns come back from the plating shop already finished. The weld beads were smooth, uniform and completely convincing. Twenty-two rings were lying on the warehouse floor when the cartons were opened, and the other 218 had two millimetres of play in them.
Illustrative example This case is written from the patterns we see in furniture export orders, and the figures are representative rather than from a specific client file. It shows how a check works and what it finds. We will replace it with a real, anonymised case as soon as we can publish one with the buyer's permission.
Twenty-two rings on the warehouse floor
A salon chain opening twelve new outlets ordered 240 styling stools with circular footrest rings from a factory in Foshan. The unit price was $118 and the order was $28,320. Thirty per cent, $8,496, was paid on order. Seventy per cent, $19,824, was due against a pre-shipment inspection report, and the report had recommended release.
The stools reached the buyer's distribution centre 22 days after they left the factory. The first 24 cartons were opened in the goods-in area, and in six of them the footrest ring was lying in the bottom of the carton with the bracket still on the stool.
The buyer stopped the receipt and checked the rest. Of the 240 rings, 22 were off, 5 were cracked along the weld line, and 213 were still attached. Every one of the 213 moved when it was pushed. The play at the bracket, measured with a feeler gauge, ran from 1.0 to 4.3 mm. The brackets themselves were intact and the beads looked normal, which is what made the finding so strange.
This is an illustrative example built from the pattern behind plating and sequencing disputes on metal sub-assemblies: the readings, the order and the arithmetic are representative of the category rather than of any particular client file.
Where to go next
The part that carries the user and nobody measures
The footrest ring is not decoration. On a 700 mm styling stool it is where a person puts a foot, and it is where they push when they stand up. It sits 180 mm above the floor on three steel brackets, each a 5 mm plate 40 mm wide welded to the column housing, and it stands 20 mm proud of the base on every side.
That overhang is the fact the whole case turns on. The ring is 400 mm across and the base is 360 mm across, so the ring is the widest part of a packed stool. Everything else about the stool is protected by the base. The ring is not.
It is also the only part of the stool that no standard covers. EN 1728 tests the strength and durability of the seat and backrest. EN 1022 tests whether a seat overturns when you push it. EN 12520 covers domestic seating and the gas lifts in it. None of them puts a load on a footrest ring, because a footrest ring is an accessory and nobody has ever written a standard for one.
That means the only place a criterion for the ring can exist is the buyer's purchase order. On this order it did not exist, and in a salon furniture inspection the ring is exactly the item that gets looked at rather than measured.
Where to go next
The order the two sub-assemblies arrived in
The bracket had to be welded to the column housing, because the ring is fitted at final assembly, after the seat and the gas lift are on. The housing came back from the plating supplier first, chrome-plated and finished, in the same condition it would have been on the day it was made.
Nobody had written down that it had to be welded before plating, because the drawing did not say so. The drawing said the bracket was welded to the housing with a continuous fillet weld, 20 mm leg, all round the bracket, on both faces where the joint was accessible. It said nothing about the surface. On the bench, the bracket went onto the chrome.
The weld was laid over a decorative chrome plating 0.05 mm thick, in the usual nickel-underlayer-plus-chrome build. It is a very thin layer and a hard one: decorative chrome on a furniture tube runs around 800 to 1,000 HV on a mild steel substrate. The bead did not fuse with the steel. It fused with the plating.
| At the bracket | What the drawing called for | What was on the bench |
|---|---|---|
| Weld size | continuous fillet, 20 mm leg, all round the bracket, on both faces | one bead on the outside face only, leg measured 7.4 to 8.2 mm |
| Root gap | 0.5 mm maximum, no skip welding | not stated on the drawing and not checked on the bench |
| Surface under the bead | not stated | chrome-plated, 0.05 mm, not ground back |
| Order of operations | not stated | plated first, welded second, which is what the line happened to do |
| Inner face of the joint | welded where accessible | bare, and inaccessible once the ring was fitted |
The bead looked perfect, because it was
Nothing about the outside of that weld was wrong. It was uniform, it had a slightly convex face, there was no undercut, no porosity and no crack. A weld gauge gave a consistent leg between 7.4 and 8.2 mm. It was, as a weld, a good weld.
The trouble is that visual inspection grades what it can see. ISO 17637 asks for bead shape, undercut, overlap, porosity and cracks on the surface. A 0.05 mm chrome layer under the bead is not on the list, because on a surface like that no amount of looking helps. The defect is not in the weld. The defect is in the 0.05 mm between the weld and the part it was supposed to be holding.
The numbers are worth doing, because they are the whole case. The bead footprint on the column is 8 mm by 40 mm, which is 320 square millimetres of contact. The throat of that bead is 0.707 times 8 mm, or 5.7 mm, over the same 40 mm of length, which is 226 square millimetres of throat. If that bead had been welded to bare steel, EN 1993-1-1 would give the group a tension resistance across the throat of about 72 kN, taking 400 MPa for the weld metal and the usual partial factor of 1.25. Welded over chrome, the bond is not a fusion at all. It is the adhesion of a decorative plating to its substrate, and decorative plating adhesion on steel runs around 25 to 50 MPa.
So the same bead, in the same place, on the same stool, has four different numbers depending on the surface it was welded to and the direction you pull it. The weakest of them is the one in the direction the ring actually failed: 0.64 kN, which is 113 times less than the same weld on bare steel and 15 times less than the same weld manages in shear.
A 0.05 mm film has no stiffness in tension and almost none in peel, so the strength of the joint in that direction is the strength of the film itself, and the film is the weakest thing in the assembly by two orders of magnitude. A decorative film is porous and full of micro-cracks, so its effective strength is nowhere near the strength of solid chrome, and around 40 MPa is a reasonable planning figure.
| The same 8 mm bead, pulled four ways | What resists it | Capacity |
|---|---|---|
| Shear, on bare steel | fusion of weld metal to the column | 52.2 kN |
| Tension across the throat, on bare steel | the same fusion, loaded in peel | 72.4 kN |
| Shear, over chrome plating | adhesion of the plating to the steel, 30 MPa over 320 square millimetres | 9.6 kN |
| Peel, over chrome plating | the effective strength of a 0.05 mm film, about 40 MPa over the same area | 0.64 kN |
Where to go next
A check in the direction the joint was strong in
The factory did check the stool. At the end of the line every stool gets a push on the footring, and the ring is pushed by hand in the direction a foot would push it: in the plane of the ring, sideways. That is a shear load on the bracket, and the bracket's capacity in shear over chrome was 9.6 kN. The push was about 300 N. The check passed with a margin of 32 to one, and it would have passed a thousand times over.
The ring came off in a direction nobody pushed it in. A peel, perpendicular to the plane of the ring, at the rim, with the base held. In that direction the same joint had 0.64 kN, and a good part of that went the first time the ring moved and took the chrome with it.
There is a second problem with the factory's check that is worth separating out. The drawing called for a 20 mm leg. The bead measured 7.4 to 8.2 mm. A weld gauge run over every bracket would have found that in about a minute per stool. Nobody ran it, because a gauge check is not a checklist item and the bead looked fine to the eye. A bead of 8 mm is not a defect you can see. It is a defect you can only measure.
What the container did
The stools went into export cartons 620 by 620 by 900 mm, two to a carton, gross weight 21 kg. The internal width was 460 mm, which leaves 30 mm of clearance on each side of a 400 mm ring. The ring sits 180 mm above the carton floor and it is the only part of the packed stool that can reach the wall.
The 120 cartons went into a 40 foot container in two layers and the voyage was 28 days. Over that time the stack does what stacks do: the lashing relaxes, the cartons compress and settle, the vessel works, and the load moves. Cartons in the affected stacks shifted 30 to 50 mm, which is all it takes for a ring that is already 30 mm from the wall to reach it.
The cartons kept the evidence. Of the 24 cartons opened first, 18 had rub marks on the inside of the sidewall, 40 to 90 mm long, at 180 mm above the floor. The ring tubes themselves carried bright rub marks over 200 to 300 mm of arc. A stool that never moved does not mark its own carton.
Each of those shifts is a peel event on a joint with 0.64 kN of peel capacity. The ring was not being lifted by a great force; it was being levered against the carton wall a fraction of a millimetre at a time, in the one direction the joint could not take. The chrome let go along the footprint of the bead and the weld went with it. When the ring came away, the beads were still intact and still smooth, because the chrome layer had parted from the steel between them.
Where to go next
It was all 240, and no sample size would have helped
Nothing here was a welder's mistake and nothing was bad luck. The defect was in the order in which two sub-assemblies reached the same bench, and that order was the same on every stool in the lot. A systematic defect at a hundred per cent is caught by any sample at all, and in this case it was caught by all 240, at the buyer's warehouse, at the wrong end of the shipping schedule.
It is still worth doing the sampling arithmetic, because the instinct is to reach for a bigger sample. On ISO 2859-1 a lot of 240 falls under code letter G, and at general inspection level II that is a sample of 32 units, with an accept number of 2 and a reject number of 3 at AQL 2.5. Any one of those 32 stools would have shown it. More samples would not have found more, because there was nothing to find on the other 208. They had the same joint. The failure was not the sample size. It was the check list.
The check that did find it took ten seconds per stool. Push the ring sideways, listen for the rattle, and feel for movement at the bracket. A joint that has let go of its substrate has a rattle that a stool with a sound weld does not. The buyer ran that across all 240 in an afternoon, then pulled each ring perpendicular to its own plane with a spring scale: 22 came off, 5 cracked, and 213 held but showed 1.0 to 4.3 mm of play.
Where to go next
→ the AQL calculator on this site, with the code letter table
One cut tells you everything
The macro-section is what turns this from an opinion into a finding, and it takes one cut-off disc and forty minutes. The inspector sectioned a bracket, polished it, etched it and photographed it with a scale in frame.
A section is a destructive check, which is why it is not something you do on every stool. It is something you do once on the first article, and once per shift, and it is something you do before the balance payment leaves. Here it would have cost the price of a cutting disc and saved the whole lot.
The photograph shows the chrome as a bright hairline between the weld bead and the column. Along most of the 40 mm width the fusion line sits on the chrome. Where the ring had moved in the container, the chrome is gone and there is a gap between the bead and the steel, with the coating sitting loose in the bottom of it. The bead never touched the steel it was supposed to be holding, and the part of the joint that failed was the 0.05 mm.
| What the section showed | Reading |
|---|---|
| Chromium layer under the bead | present and continuous, about 0.05 mm |
| Fusion boundary | along the plating, not in the steel |
| Where the ring had moved | a gap between bead and substrate, 0.2 to 0.6 mm wide, with loose coating in it |
| Bead shape | sound, uniform, no porosity and no crack |
| Weld leg | 7.4 to 8.2 mm against 20 mm called for |
What it cost
The buyer had three options, and the numbers are the argument for checking early rather than checking carefully.
The third option is the one that worked, and it was only available because the buyer had checked all 240 and found the defect confined to the ring. The seat, the gas lift, the base and the finish were sound. A stool without a footring is still a stool, just a less comfortable one, and 240 new rings welded to bare metal before plating are a much smaller problem than 240 new stools.
So the difference between checking and not checking is $8,470 against $32,120, and four weeks against eight. The inspection that had already been done before the goods shipped cost $680, which is 2.4 per cent of the order value, and it had checked the ring with a push in the wrong direction.
The part of the loss that never appears in the ledger is the 213 stools that looked fine and were not. They were standing in a salon with a joint that had less than a hundredth of the strength it should have had in the direction it failed. The 22 that came off in the container were simply the ones that were used hard enough first.
| Option | What it costs | Time |
|---|---|---|
| Replace the lot | 240 at $118 is $28,320, plus $3,800 of freight | 8 weeks, and twelve openings without furniture |
| Send the stools back for a re-weld of all three brackets | strip the base, grind back, re-weld and re-plate at $61 a unit, so $14,640 | 6 weeks, and a re-plated area on every stool |
| Keep the stools and fit new rings | grind back the plating and re-weld 240 rings at $34 a unit, so $8,160, plus $310 of freight | 4 weeks |
Where to go next
What goes into the purchase order
None of this is a standard's fault, because no standard covers a footring. It is a specification's job, and the clauses that would have caught it are short.
Clause 8 is the one that would have saved this order, and it costs fifteen seconds a stool. Clause 1 is the one that would have saved it for nothing, and it is a sequencing decision taken once at the drawing stage rather than 240 times on a bench.
- All welding on the stool is completed before plating. No welding, cutting or grinding is permitted on a plated surface.
- Where an attachment must be added to a plated part, the plating is ground back to bare metal over the full length of the weld plus 10 mm each side, the edge of the prepared area is feathered to no steeper than 1 in 10, and the area is masked before plating.
- A floating electrode test to EN ISO 17637 is carried out on every prepared area before welding, and the result is recorded.
- Weld size: continuous fillet, minimum 20 mm leg, all round the bracket, on both faces where the joint is accessible. Stitch welding and skip welding are not permitted on a structural attachment.
- Root gap 0.5 mm maximum, measured on the fixture before the bracket is closed up.
- The welding procedure is qualified to EN ISO 15614 and the welders to EN ISO 9606-1. Weld quality is to ISO 5817 level C, with the leg and length verified by gauge on 100 per cent of brackets.
- First-article macro-section: one bracket per lot, cut, polished, etched and photographed with a scale in frame, showing the fusion line in bare steel with no plating between bead and column. Three sections where the lot spans more than one shift.
- Peel proof load on 100 per cent of units: the ring is loaded perpendicular to its own plane at the rim with 400 N, in the direction that opens the joint. Acceptance is no movement greater than 0.5 mm and no visible separation of the coating.
- Packaging and loading: the ring is removed and packed separately, or the cartons are marked so the ring is not the outermost feature. Cartons of stools are not stacked more than two high, and the lashing does not bear on any single point of a stool.
- The report states, for each item checked, the acceptance criterion and the measured value. For a weld it states the leg, the length and the method, and the first-article section photograph is included.
Where to go next
Ninety minutes
The whole thing, on 240 stools, fits into an afternoon. Twenty minutes to push every ring and listen for the rattle. Forty minutes to pull every ring perpendicular to its own plane with a spring scale and a strap, and to record the play. Thirty minutes to section three brackets, polish, etch and photograph them. The consumables are a cutting disc and some etchant. The labour is one inspector and a helper.
That is what a $680 pre-shipment inspection should have been spending its time on, instead of counting cartons.
None of it is exotic and none of it needs a laboratory. It needs the buyer to write down what the ring has to do, and the factory to be checked against that line rather than against a photograph of a good-looking bead.
The point of the case
The weld on this stool was strong. The joint was weak. The difference between the two was fifty microns of chrome and the order in which two sub-assemblies arrived at the same bench. Nobody lied, no measurement in the file was wrong, and the report recommended release because every criterion in it had been met. The criteria were the wrong ones.
That is the recurring shape of a salon furniture inspection problem: not a bad part, but an unwritten requirement. For a component that a standard does not cover, the purchase order is the only document that can carry the requirement, and the only place a report can fail is where it says nothing.
If you are about to place an order for stools, chairs or any metal-framed furniture that arrives plated, ask for the sequence: what is welded, what is plated, and in which order. The answer takes a minute to get and it is worth more than the goods.
Where to go next
→ send us the order details and we will tell you what a check can cover
Have an order you need verified? Send us your PI and factory address. Three levels, published prices, $149 per man-day from.