Why the edge takes the damage
A dining table lives in a hostile envelope. Plates land on it, chairs are dragged into it, trolleys pass at knee height, and the last 25 mm carries all three because that is where the geometry runs out. On a 25 mm top, the arris is a 90 degree corner with roughly 0.5 mm of coating wrapped over it. Any impact arriving at that corner acts on a contact patch small enough to concentrate stress well past the compressive strength of the core underneath.
In our own claims history across restaurant projects, between 60 and 70 percent of reported table damage starts at an edge or a corner rather than on the field of the top. The pattern is consistent: a chip at the arris in month three, a dark stain spreading along the core in month nine, a top replaced at year two that was structurally sound everywhere except the outline.
That matters because edge damage is not cosmetic in the way a scratch is. A scratched top is a finish problem. A chipped edge exposes particleboard or MDF, and exposed core absorbs liquid at a rate of 8 to 14 percent of its own weight in the first hour of contact. Once the core swells, the laminate lifts locally and the repair becomes a replacement. The profile is the cheapest insurance in the whole specification.
Four profiles and what each one actually does
A straight arris is the cheapest edge and the first to fail. The full 90 degree corner is left sharp, the coating wraps over it at a thickness that is thinner than anywhere else on the component, and the first knock takes a chip out of both the coating and the substrate behind it. It is acceptable on a side table against a wall that nothing touches. In a circulation route it is a two-year component.
An eased edge, typically a 2 to 3 mm radius, is the working default for commercial work. It removes the sharp arris without changing the visual thickness of the top, so a 25 mm table still reads as 25 mm. The machining pass costs 40 to 60 seconds on a CNC and adds nothing to material. Practically, going from a straight arris to a 2.5 mm radius removes the stress concentration at the corner and cuts the rate of edge chipping by roughly half in our drop tests, because the impact is distributed over a curved face rather than a point.
A chamfer is a cut, usually 45 degrees, producing a flat band around the top. An 8 mm chamfer on a 30 mm top gives a generous, architectural outline and hides the substrate behind a wider face, which is why it appears on so many of the restaurant dining tables in open-plan rooms where the table is visible from standing height. The trade is that the chamfer face becomes a wear surface of its own: it is where forearms rest, where cutlery is dropped, and where finish loss shows first because the film has been cut back at the same angle.
A banded edge, where a 6 to 10 mm solid timber lipping is glued to the core before machining, is the most durable option and the only one that can be re-sanded later without exposing the core. It adds 1.5 to 2.5 mm to the visual thickness, costs 3 to 5 percent of the table price, and turns an edge repair from a replacement into a 20 minute workshop job. On a plywood-top table with pyramid plate base, the band also protects the plywood plies, which delaminate at an exposed edge within one wet season if the lipping is missing.
Board thickness sets the radius ceiling
There is a hard relationship between how thick the top is and how much edge you can remove. The working rule on our drawings is that a chamfer depth should not exceed a quarter of the board thickness, and a radius should not exceed a third. On an 18 mm core that means a 4 mm chamfer or a 3 mm radius. On 25 mm, a 6 mm chamfer or a 5 mm radius. On 30 mm, 8 mm. On 40 mm, 10 mm or a 10 mm radius.
Push past that and the component fails a point load test at the corner even though it looks correct on a sample board. A 6 mm chamfer on an 18 mm top leaves a 12 mm land behind the chamfer, and a 60 kg point load applied 15 mm in from the corner puts the remaining section into tension across a very small area. We test at 60 kg for 24 hours on a 40 mm disc and the 18 mm board with an oversized chamfer creeps visibly.
Surface material constrains the radius from the other direction. A high pressure laminate needs a minimum wrap radius: standard 0.7 mm HPL wants 8 to 10 mm before it cracks at the bend, postforming grade will take 6 mm, and compact laminate at 12 mm thickness needs 3 mm minimum and is machined rather than wrapped. That is why a 3 mm eased edge on a laminate top is a machined detail, not a bent one, and why a 10 mm radius on a thin laminate top has to be banded instead.
Sintered slab goes further again. A pandora slab table with steel pedestal has a top that is 12 mm of pressed mineral material with no core to expose and no laminate to lift, so the edge is machined and then polished through four grits to a 2 mm radius. It will not chip the way a faced board chips, but it will spall if it is hit on a radius under 1 mm, which is why the polished radius is a functional minimum rather than a style choice.
Thicker is not automatically better. A 40 mm top on a 60 mm base plate looks heavy and restricts the legroom envelope at the knee, and it raises weight by 60 percent against a 25 mm top, which then drives base specification and freight cost. Choose thickness for the design and the span, then let thickness set the maximum profile.
Corners are a separate decision from the edge
Buyers routinely specify an edge radius and forget the corner, and the corner is where the injury and the worst damage live. On a 700 by 700 mm top, a 90 degree corner is a single point of contact. A radius of 40 mm reduces the exposed angle to about 70 percent of a right angle and removes the sharpest 5 mm. A radius of 60 mm removes closer to 45 percent and is the figure we recommend anywhere a child could run past the table.
The area cost is smaller than clients expect. Going from a square corner to a 60 mm radius removes 0.011 square metres per corner on a 700 mm top, which is 2.6 percent of the surface. Four corners at that radius cost a little over 10 percent of the top area, and a four-cover table still sets four covers because the setting positions sit well inside the corner.
There is a specification reason beyond safety. A sharp corner is the first thing to be struck by a chair being pulled out, because the chair back swings through exactly that arc. Corners with a 40 to 60 mm radius survive that repeated contact; 90 degree corners chip, and once chipped they cannot be made safe again without replacing the top. Insurers auditing a dining room for public liability tend to ask about exposed corners and glass, in that order.
On square restaurant tables all four corners face the room, so all four carry the same exposure and there is no hidden side to save money on. That is not an argument against the shape, only a reason to specify the radius once and apply it to every corner rather than rounding two and squaring two.
Corner radius should be listed separately on the drawing, per corner, because the two corners at the banquette end are often squared off to fit against a fixed seat while the two at the aisle end are rounded. Writing R40 all round when the seating plan needs a squared end produces a top that does not fit, and writing nothing produces whichever the operator had set up that morning.
The cleaning dead zone nobody specifies
A groove machined into the edge as a design reveal, usually 4 to 5 mm wide and 4 mm deep, collects debris at a rate several times that of a continuous profile. Food sits in the channel, the cleaner does not reach it because the cloth passes over the opening, and within a season the line is visibly darker than the rest of the edge. The same channel holds liquid, which is the mechanism that starts core swelling at an edge that was otherwise sound.
Laminated edges have a related weakness at the seam. Where the laminate is wrapped or joined at the underside, the joint line faces downward on a good top and sideways on a poor one. A sideways seam on a table that gets wiped 4 times a service will take moisture. Standard practice on commercial work is to keep the joint on the underside, seal it with a bead of PUR adhesive, and specify a 0.5 mm minimum overlap rather than a butt join.
Nightly cleaning chemistry matters here too. A pH 11 to 12 degreaser diluted 1 to 20 and left to dwell 3 to 5 minutes attacks any unsealed timber band at the edge faster than it attacks the top, because the band presents end grain. A 6 mm oak band with a 90 micron PU film and a sealed end-grain face will hold up for years. The same band with an unsealed end will darken and lift inside 18 months.
The practical rule is simple: an edge should be continuous, closed and cleanable with a single pass of a cloth. Any profile that creates an upward-facing groove, a step or an open joint is a maintenance liability dressed as a design detail.
Writing the edge onto the drawing
An edge specification needs six fields to be producible without a phone call: profile type, the numeric radius or chamfer, the tolerance, the banding species and thickness if banded, the adhesive class, and the finish build on the edge face. Without the tolerance the workshop will work to whatever the machine does, and the difference between 2 mm and 4 mm on a radius is visible across a room.
Set the tolerance at plus or minus 0.5 mm on the radius and 0.2 mm on a banding joint gap. Specify the adhesive as PUR or E0 class rather than generic glue, because edge banding is where a low-grade adhesive fails first in a humid room. State the finish build on the edge separately from the top, since a 90 micron film on the field of the top can be 60 microns on the machined edge if nobody asks, and the edge is the part that gets wiped.
Ask for a physical sample, not a photograph. Three pieces at least 300 mm long, in the actual core, the actual banding and the actual finish, with the corner radius machined on one end. A sample you can drop on a concrete floor tells you more than any specification sheet, and it costs a supplier 20 minutes. Anyone running custom dining tables as a regular line will have these on the shelf.
On restaurant tables in custom sizes, add a dimensioned detail drawing at 1 to 1 for the edge section. Scaling a 25 mm section down to fit an A3 sheet hides exactly the 2 mm that decides whether the profile is machinable on the core you have chosen.
Matching the profile to the zone
One profile across a whole order is the usual shortcut and it is usually wrong. Tables on a circulation route take repeated chair contact and trolley strikes, so they want an eased 3 mm radius over a banded core, or an 8 mm chamfer if the design needs the lighter look. Tables against a banquette take almost no side impact on the wall side, so the two hidden corners can be squared and only the aisle corners rounded.
Service stations and any top that holds water jugs, cutlery trays and condiment refills should have a bullnose of 10 mm or more and a fully sealed edge. The exposure there is continuous liquid rather than occasional impact, and a sharp arris on a wet top wicks moisture into the core at the first small failure of the film.
Communal and large-format tables deserve their own note because the span changes the failure mode. A 2400 mm top on two trestles deflects under load, and deflection concentrates stress at the edge where the top leaves the support. On those, a banded edge does structural work as well as cosmetic, and a plain chamfer on a thin core will show a hairline crack along the chamfer face within two years of daily use.
Children areas and family dining push the corner radius up to 60 mm and the edge radius to 6 mm minimum. That combination costs about 4 percent more than a square edge on the same top and removes the majority of the injury and damage vectors in one decision.
The same logic applies across commercial dining tables in a mixed room: the profile follows the traffic, not the product line. Two tables from the same range can legitimately ship with different edge treatments because one sits in an aisle and one sits against a wall.
Inspection before the container ships
Edge work is the easiest thing to inspect badly because it photographs well. Four checks on a sample of ten units take under an hour and catch nearly everything. First, measure the radius with a gauge set at 1 mm increments, on all four edges and both ends; a radius that reads under the drawing by more than 0.5 mm is a pass, anything measuring zero on a specified radius is a machining setup error.
Second, check every banding joint. Gap under 0.2 mm, no adhesive squeeze-out left proud, no step you can feel with a fingernail across the joint. Third, tap along the band with a small hammer; a hollow note means a void behind the banding, and a void holds moisture and lifts within a season. Fourth, the drop test: a 1 kg steel ball from 300 mm onto the corner, on a sacrificial unit. A 3 mm eased edge will bruise. A 60 mm radius corner will not chip. A straight arris will chip, which is the point of testing it on a sample before you have 300 of them.
Put the results in the same document as the load test and the material reports, and specify in writing that the sample approved at order stage must be kept and compared at pre-shipment. Profiles drift between the sample and mass production more than any other furniture detail, because a worn cutter changes a 3 mm radius into a 1.5 mm arris across a few hundred metres of cutting and nobody notices until the crates are open.
A supplier who can send the radius gauge readings and the joint photographs with the inspection report is a supplier who has the process, not just the machine. Ask for that on the first order and again on the third, and if you are still comparing options, request a restaurant furniture quote with the top sizes and the corner radius you need, and we will come back with the profile options and what each does to the unit price.