Oil canning in sheet metal is why a cabinet door passes every dimensional check on the bench and still looks wrong the moment it is hung and lit. The part is in tolerance, welds are clean, coating is even, yet under raking light the face reads as a shallow wave. That is not a bending error but a flatness stability failure: a thin sheet carrying residual stress, spanning further than it can hold flat. This guide covers the levers that move it, how to measure flatness, and what to write on the drawing. The general tolerance framework sits in our sheet metal tolerance guide.
What is oil canning in sheet metal?
Oil canning is the visible buckling of a large, thin, flat area under residual stress. The sheet was not bent out of shape by a machine; it settled into a buckled shape because that stores less strain energy than a flat one. Press a finger into an affected panel and it often snaps through to the opposite curvature with a soft pop, then snaps back, which is where the name comes from: the flat bottom of a steel oil can does the same thing.
Two facts explain most of the behavior. The buckled panel is bistable and holds either the bulged-out or the bulged-in shape. And resistance to buckling is geometric rather than a material strength problem, since critical buckling stress scales with (t/b)², where t is thickness and b is the unsupported width. Both terms enter as squares, so halving the span is as effective as doubling the thickness, and usually much cheaper.
Why does oil canning happen on large, thin, flat panels?
Because it needs all three conditions at once. Change any one and the problem usually goes quiet.
- Large and thin. The penalty is quadratic in the span-to-thickness ratio, so a panel can cross from acceptable to wavy over a modest change in either dimension.
- Flat. A flat surface has no geometric stiffness. Any break in the plane, even a shallow one, raises the second moment of area of that section substantially, which is why a ribbed door can be thinner than a dead-flat one and still look better.
- Residual stress present. Usual sources are rolling stress left in the coil, the leveling done to remove it, cutting heat input, punching shear, bend springback, and weld shrinkage.
The stress is usually there long before the wave is, and it shows up when restraint is released or the panel is thermally cycled. A powder cure oven is enough to do it.
What makes oil canning more visible?
A panel can be mechanically unchanged and still look far worse depending on finish and lighting. Four amplifiers account for most of it.
- Gloss. A high-gloss coat behaves like a weak mirror: a small change in slope moves the reflected highlight a long way. Matte and textured coats scatter it.
- Dark colors. The highlight sits bright against a low-luminance base, so the contrast step across a wave is larger.
- Raking light. Light arriving at a shallow angle maximizes the effect of slope change. Grazing daylight is the worst case.
- Large unbroken area. With no seam, label or louver field, the eye has no reference and reads the whole face as one continuous distortion.
Finish selection is therefore a legitimate flatness tool: moving from high gloss to satin or texture is an engineering compromise rather than a cover-up, and often costs less than the stiffening a mirror finish would demand. The tradeoffs are in the powder coating versus liquid painting guide and the powder coating defect troubleshooting guide, and the gloss and colour numbers that go with them are set out in the powder coating colour match guide.
Six cause families and what to do about each
Sort by where the wave sits and when it appeared.
| Cause family | How it presents | First thing to try | If that is not enough |
|---|---|---|---|
| Mill or coil residual stress | Broad wave running in the rolling direction, already present in the flat blank | Buy flatter stock; run the blank through a leveler | Rotate the blank so the wave runs the short way, or hide it behind a break in the plane |
| Thermal cutting input | Edge-side bow; distortion grows with outline length and long thin features | Rework the cut path and lead-ins, balance the heat around the nest, or punch the part cold | Level after cutting, then re-check after coating |
| Bend springback and flange restraint | Shallow dish in the flat field next to a long flange | Add corner relief, change the bend sequence, or put a bead between flange and field | Deeper return, or shorter flange |
| Weld shrinkage | Localized pucker along a weld line, usually on the panel side of the joint | Reduce heat input, use a balanced or skip sequence, clamp to a chill bar | Move the joint, or add a stiffener that hides the heat line |
| Handling, stacking and fixturing | Random, not repeatable, and it changes when the panel is turned over | Change the rack, the clamp point or the lift point | Add a temporary spreader for coating and transport |
| Coating cure plus assembly preload | Appeared only after coating, or only after the latch was torqued | Lower peak metal temperature and dwell; check latch preload and hinge shimming | Accept a texture, or stiffen at the latch points |
Design levers: how do you keep a large panel flat?
In the order we would try them:
- Break the plane. A step, seam, louver field or perforated field adds section stiffness and gives the eye a reference. Louvers and perforations are specified for airflow first, but they earn their keep on flatness too: see the enclosure ventilation guide and the perforated vent pattern guide.
- Add a formed stiffener. Beads, ribs and hat channels are the standard answer. Their geometry, spacing and depth limits are worked through in the stiffener rib design guide.
- Reduce the unsupported span. Move a hinge line, add an intermediate support, or split one wide door into two. The layout trades are in the door and cover design guide.
- Set the return depth deliberately. A deeper return stiffens the edge, but it adds springback and can dish the field beside it. See the bending guide.
- Choose material with the mechanics in mind. Qualitative trends only, since the numbers depend on grade and temper. Aluminum has a lower elastic modulus than steel, so equal thickness buys less stiffness and it springs back more, and its higher expansion means more movement through a cure oven. Stainless has steel-like modulus but higher yield strength and stronger work hardening, so springback is larger. Full comparison in the material selection guide.
| Geometry | What the panel wants to do | First lever | Escalation |
|---|---|---|---|
| Small field, generous thickness | Holds flat unaided | Nothing needed | None |
| Moderate field, stiffened edges | Slight dish near the center | Deeper return, or one shallow bead | Texture in the coating |
| Wide field, thin sheet, no breaks | Classical oil canning under raking light | Break the plane or shorten the span | Hat channel or rib behind the face |
| Wide field with latch preload | Pops when pressed; changes with latch force | Stiffen locally at the latch, reduce preload | Thicker sheet or an assembled panel |
| Field next to a long weld | Pucker along the weld line | Move the joint, lower heat input | Weld sequence redesign, then level and fill |
What can the shop actually fix?
Leveling. A precision leveler passes the sheet through a series of small-offset rolls, bending it slightly past yield repeatedly in alternating directions, which averages residual stress out rather than pushing the wave back. It works on coil and flat blanks but not on a panel with formed features in the way, and it will not remove a wave a weld has locked in.
Cutting method. A laser puts heat into the outline; a turret press shears cold and leaves a different stress pattern. On a large thin panel with a long perimeter that is a process choice, not a scheduling detail. The economics are compared in the laser cutting versus stamping guide, and the punching side is detailed in the turret press guide.
Bend sequence. The order in which flanges go up decides where springback accumulates and where it cancels. Sequence is set by the person programming the brake, which is why the drawing should call out the appearance requirement instead.
Weld sequence and heat input. Balanced, skip and back-step sequences, the lowest heat input that still achieves penetration, and clamping until the joint cools. The mechanics are in the weld distortion control guide.
The limit of rework. A buckled panel is not bent, so straightening it does not unbend anything. Hand work moves the wave rather than removing it, and on a coated part any hammering destroys the finish. Our rule: if a panel needs more than one light leveling pass, we remake it rather than chase it.
How do you measure sheet metal flatness?
Most measurement disputes come from the method, not the part.
| Method | What it gives you | Where it falls short |
|---|---|---|
| Straightedge and feeler gauge | A local deviation over a stated length. Cheap, fast, and the most widely documented method | It only samples the line you lay the rule on. Rotate the rule and you get a different number |
| Surface plate with dial indicator or height gauge | A deviation map across the whole face, referenced to a datum plane | The part has to fit on the plate, and the result depends entirely on how the part is supported |
| Laser scan or structured light | A dense deviation map, and the only practical option for a large door | Needs a defined alignment to datums, and glossy surfaces can confuse the scanner |
Free state versus clamped state is the whole argument. Measured while clamped flat, you are measuring the fixture. Measured in free state, you are measuring the part, and you must also say how it rests: on three named points, or on a surface plate of a stated grade, with no clamping force and the latch disengaged. State the measuring force too, since a heavy dial probe presses a thin panel down.
One point has to be on the print: a flatness callout with no measurement method attached is not an acceptance criterion, because both sides will measure honestly, get different numbers, and both will be right. Numeric tolerance and cosmetic approval are separate axes, which is why appearance needs its own stated lighting and viewing distance, as set out in the cosmetic surface class guide, and why both belong in the inspection record covered by the quality control and inspection guide.
What should the drawing say?
The callout below is the one we ask for. Fill in the blanks and most flatness disputes disappear before the first article is made.
FLATNESS / OIL CANNING REQUIREMENT - PANEL FACE
Scope ..................... Part ___, faces ___
Form tolerance ............ Flatness ___ mm over any ___ mm
(per ISO 1101, applied to the
finished, coated part)
Measurement state ......... FREE STATE ONLY. Part resting on
3 supports at ___, ___, ___ (or on a
grade ___ surface plate). No clamping
force, latch disengaged, hinges not
preloaded.
Measuring method .......... Straightedge + feeler gauge over
___ mm / dial indicator on plate /
laser scan aligned to datums A|B|C
(delete all but one)
Lighting / cosmetic check.. ___ lux at ___ deg incidence,
viewed from ___ m for ___ s,
Class ___ face per cosmetic zones
Finish .................... Gloss ___ GU at 60 deg, color ___,
texture ___ (note: high gloss and
dark colors amplify visible waviness)
Stiffening ................ Bead / rib / hat channel per
detail ___; spacing ___
Records required with FA .. Flatness map, gloss reading,
retained reference panel
Not acceptable ............ Visible waviness on Class ___ faces
under the lighting stated aboveThe three lines that do the most work are the measurement state, the lighting description, and the gloss level.
Frequently asked questions
What is oil canning in sheet metal?
It is the visible buckling of a large, thin, flat panel under residual stress. The sheet has settled into a domed shape because that carries the internal stress with less strain energy. It is a flatness stability failure rather than a dimensional defect.
Why is it called oil canning?
Because of the behavior, not the industry. Press the flat bottom of a steel oil can and it snaps through to the opposite curvature with a soft pop, then pops back when pressed again. An affected panel does the same under thumb pressure.
How do you prevent oil canning on a large flat panel?
Give the panel geometric stiffness before reaching for thickness. Break the plane with a step, seam, bead or rib, shorten the unsupported span, and set the return depth deliberately. Halving the span works as well as doubling thickness, and costs less.
Can oil canning be fixed after the part is made?
Partly, and the window is narrow. A precision leveler on a flat blank before forming works well. Once the panel is formed, welded and coated, hand straightening moves the wave rather than removing it, and it damages the finish.
Does a thicker sheet always fix wavy sheet metal?
No, and it is often the wrong first move. Thicker sheet adds weight, cost and bend tonnage while leaving the residual stress in place. Breaking the plane or reducing the span addresses the mechanics directly.
Why does oil canning only show up after powder coating?
Two things happen in the coating line. The cure oven thermally cycles the panel, releasing locked-in stress so it settles into a buckled shape. The finish then makes the wave readable, because gloss and dark colors amplify slope change under raking light.
The short version
Oil canning is not a defect you inspect out at the end. It comes from a wide unsupported span, a thin sheet, and residual stress left by rolling, cutting, forming and welding, and it is amplified by gloss, dark color and raking light. Design the stiffness in and state how flatness will be measured, because a print that omits the method has no requirement at all.
Where we fit
We are a sheet metal OEM and ODM factory in Shenzhen building custom enclosures and cabinets for North American and European buyers. Cutting, forming, welding, surface preparation and powder coating all happen in our own building, so the people deciding the cut method, the stiffener layout and the cure profile are the same people who reviewed your drawing. Send us the model for a free CAD and DFM review, and we will flag where a large face is likely to move.
What we can control and record, we do: cutting method, bend sequence and stiffener placement, weld sequence and heat input, leveling passes before and after forming, and flatness records against the method written on the print. What we do not do is third-party certification work, and we do not issue certification certificates. Our quality system runs to ISO 9001 and ISO 14001 and our materials are RoHS and REACH compliant; we do not hold CE. For reference: a 5,000 m² facility, a team of more than 40, MOQ 1, and standard lead time of 7 to 15 working days, with 20 to 30 days for custom programs.
Related reading: the stiffening options in detail, in the stiffener rib design guide; how weld heat drives distortion, in the weld distortion control guide; what a finish hides and what it reveals, in the powder coating versus liquid painting guide; and how to define appearance zones, in the cosmetic surface class guide.
Written by the Sanjun sheet metal team, Shenzhen. Last reviewed 11 October 2026.