Why temperature differentials crack concrete

Hydration is exothermic

Cement hydration releases heat. In a thin section that heat dissipates about as fast as it is generated. In a thick section it does not: the core has nowhere to lose heat to, so its temperature rises — commonly up to 50 °C above placing temperature, depending on binder content and section thickness — while the surface stays close to ambient.

Concrete expands as it heats and contracts as it cools. A section at 45 °C in the core and 20 °C at the face is therefore not at uniform volume: an expanded core is bonded to a cooler, less expanded surface zone. That incompatibility is what generates tensile strain.

Two mechanisms, two crack patterns

The two are controlled by different quantities. Internal restraint is governed by the differential across the section. External restraint is governed by the temperature drop from peak and the degree of restraint. A pour can satisfy its differential limit and still crack through on cooling.

What the mix design controls

Most of what determines the temperature rise is fixed before the concrete is batched.

• Portland cement content. The dominant variable. Heat output broadly tracks the mass of Portland cement, so every additional kilogram is heat that has to be managed on site.
• Cement strength class and rate class. This is the cement's own class under BS EN 197-1 — 32.5, 42.5 or 52.5, with N or R for rate of strength gain — not the concrete's. A higher-class cement is more finely ground and chemically more reactive, so at the same cementitious content it hydrates faster and produces a higher, earlier peak. Fineness here means the specific surface area of the ground clinker, not a property of the concrete. Where early strength is not on the critical path, a lower class or the N rate is the better thermal choice.
• Cement class indicates the direction, not the magnitude. BS EN 197-1 sets minimum requirements only, so heat output varies within a single cement type — C766 reports a spread of roughly ±10 % across UK CEM I cements. For critical sections, obtain heat data for the actual cement or combination rather than assuming from the designation. BS EN 197-1 designates low heat (LH) cements; very low heat (VLH) special cements are covered separately by BS EN 14216. LH caps heat of hydration at 270 kJ/kg and VLH at 220 kJ/kg, both normally achieved with GGBS or fly ash combinations.
• GGBS and fly ash. The most effective lever on most projects. GGBS is used from 6 % to 95 % of the binder under BS EN 197-1, with most structural work in the 36–80 % range of CEM III/A and III/B. It reduces and delays the peak, which typically arrives as a plateau lasting much of a day rather than a sharp maximum.
• Placing temperature. Peak temperature tracks placing temperature close to one-for-one. Shaded stockpiles, chilled batch water, ice substitution and night pours all reduce it directly.
• Aggregate. The coefficient of thermal expansion is largely an aggregate property. A flint gravel concrete expands appreciably more per degree than a limestone one, so the same thermal history produces more strain — and a lower limiting differential.
• Concrete strength class. A different lever, and the more important one. A higher concrete class raises heat principally because it requires a higher total cementitious content to reach the characteristic strength — the class itself contributes nothing thermally. Specifying above what the design needs therefore buys heat for no structural benefit. It does raise tensile strain capacity slightly, and with it the limiting differential, but the extra binder normally outweighs that.

Compliance age and binder content

These interact, and the distinction matters. At high replacement levels a GGBS mix may not reach the specified strength class at 28 days, so the binder content has to be increased or the water-binder ratio reduced to compensate. Increasing binder content partially cancels the heat benefit the GGBS was specified for.

Allowing compliance at 56 or 90 days avoids that. It is not that later compliance permits a higher replacement level in isolation — it is that it lets you hold a high replacement level without raising total binder content to hit a 28-day figure. C766 makes the same point the other way round: specifying strength at a later age is effectively equivalent to specifying a lower strength class.

The limits, and their basis

A thermal control plan normally specifies two.

Limiting temperature differential

There is no single correct value. The limit is a function of the coefficient of thermal expansion and the tensile strain capacity — both largely aggregate-dependent — together with the restraint factor and strength class. Applying the C766 method, a gravel concrete reaches its limiting differential 15 °C below a limestone one — 20 °C against 35 °C. C766 tabulates values for common aggregate types in Table 7.1.

A round 20 °C remains widely quoted, and it is a real C766 figure — but only for gravel. It is the gravel column of the limiting-differential row, which reads 28 °C for granite and 35 °C for limestone. Table A5.2 restates the same differentials from earlier guidance; the two tables diverge in the limiting temperature change rows rather than the differential, by around 1 °C for gravel and more for limestone, with the older figures the less conservative. Where cracking is critical, C766 recommends measuring the coefficient of thermal expansion directly and deriving strain capacity from measured tensile strength and elastic modulus rather than adopting defaults.

Limiting peak temperature

The National Structural Concrete Specification sets 65 °C, as a single figure applied regardless of binder. This is not a cracking limit. Above the threshold, sulfate that would normally be bound into ettringite during setting can redissolve and re-form in hardened concrete, expanding as it does — delayed ettringite formation, which manifests years later and cannot be remediated on site.

The 65 °C figure is a conservative default rather than a fixed physical threshold. C766 Appendix A8.3 reproduces risk bands from Quillin (2001) that apply specifically to Portland cement concretes: no risk below 60 °C, very low risk below 70 °C, low risk below 80 °C. Replace enough of the cement — fly ash above 20 % or GGBS above 40 % — and the same source holds that DEF-induced expansion is prevented at peak temperatures up to 100 °C, because there is far less reactive sulfate and aluminate left to cause it.

NSCS does not build that distinction into its clause, so 65 °C is what applies unless the project agrees otherwise. In practice, a case for a higher project-specific limit — backed by the mix's SCM content and the DEF evidence behind it — is a normal thing to put to the design engineer on a GGBS or fly ash mix, not an exception to argue for from nothing.

Measurement

The differential is taken between a sensor at the hottest point in the core and one in the surface zone, typically at cover depth. Two things determine whether the value is meaningful: sensor position, and which face is instrumented. The critical face is normally the one losing heat fastest — the exposed top surface overnight, the windward face, or the first to be struck.

Limits apply through cooling as well as heating. The highest-risk point is often striking: removing formwork exposes a warm face to ambient and the differential can rise sharply within minutes. A section that has performed for three days can crack within an hour of being struck. Insulation retained after striking controls the rate of surface cooling.

Compliance record

Continuous temperature records demonstrating that the limits held, at what margin and for how long. That closes out the thermal control plan, and it informs whether the next pour's controls can be relaxed or need tightening.