Thermal shock: what steam cleaning does to a floor

A floor that has survived years of forklift traffic can fail within months of a change in cleaning method. The usual cause is temperature. Steam lances, hot water wash down and the discharge of hot process water impose a cycle that no mechanical test describes, and a resin system that is otherwise well specified can be destroyed by it.

The mechanism

Resin and concrete expand at different rates. Concrete has a coefficient of thermal expansion in the region of 10 to 12 microstrain per degree Celsius. An unfilled epoxy expands several times faster. When hot water at 80 degrees Celsius meets a floor at 15 degrees, the resin layer tries to grow while the concrete beneath it, which heats slowly, does not. The difference is taken up as shear stress at the bond line.

One cycle does nothing visible. Repeated cycles fatigue the bond, particularly at edges, at drains, at coving and at any point where the layer is already restrained. The failure appears as lifted edges, then as a hollow sound under a hammer, then as a sheet of resin that separates cleanly from the concrete. The resin itself is usually undamaged, which is what makes the diagnosis confusing: the material looks sound, so the cleaning is not suspected.

Thickness changes the picture. A thin coating heats through quickly and reaches the substrate temperature sooner, so the gradient is short lived but the coating has little mass to resist the stress. A thick unfilled screed holds the gradient longer. There is also a chemical limit: every resin has a glass transition temperature above which it softens, loses hardness and becomes vulnerable to marking and chemical attack. Hot water can exceed that temperature at the surface even when the room is cool.

Why PU cement behaves differently

Polyurethane cement is heavily filled with cement and graded aggregate. That filler brings its coefficient of thermal expansion close to that of concrete, so the differential movement that drives the failure is largely removed. Laid at sufficient thickness, commonly 6 to 9 mm where steam cleaning or thermal cycling is expected, it also carries enough mass to buffer the shock rather than transmitting it straight to the bond line.

Thickness is part of the specification, not a detail. A PU cement laid thin for cosmetic reasons in a room that is steam cleaned gives away much of the advantage it was chosen for. The same applies to detailing: coving, drain edges and thresholds see the hottest and most direct exposure and should be formed in the same material, keyed properly, rather than in a thinner patch system.

Setting the specification against the real regime

The relevant question is not whether a system resists heat but what temperature reaches the floor, how quickly, how often, and over what area. A once weekly deep clean is a different load from a twice daily wash down at shift change. Where a process discharges hot water or hot product onto a fixed spot, that spot is a local specification, not an average one.

• Record the actual water or steam temperature at the nozzle and at the floor.
• Record the frequency and duration of hot cleaning, and the resting temperature of the room.
• Identify fixed hot spots from equipment discharge, ovens, fryers and sterilisers.
• Check the system thickness proposed against the thermal load, not only against the mechanical load.
• Confirm that coving, drains and thresholds are formed in the same thermally matched material.
• Confirm the maximum service temperature and the cure time before hot cleaning may begin.