Early-age thermal and shrinkage cracking in restrained walls is predictable in principle and stubborn in practice. The mechanism is well understood and the codes give methods to estimate it, yet basement and retaining walls keep cracking in the same places for the same reasons. It is worth looking at why, and at how close the predictions actually land against what happens on site.
The mechanism
A freshly cast wall does two things in its first days. As cement hydrates it generates heat and the wall expands slightly; then it cools to ambient and tries to contract. Layered on top, over weeks, is drying shrinkage as the concrete loses moisture. Both want the wall to get shorter.
The problem is that the wall is not free to shrink. It is cast onto an older, already-hardened base slab (and often between previously cast sections), which restrains the movement. Restrained contraction is tension. As the wall tries to shorten and cannot, tensile stress builds; once it exceeds the developing tensile strength of the young concrete, the wall cracks - and it does so at roughly regular spacing along its length.
Reinforcement controls width, it does not prevent cracking
This is the single most important and most misunderstood point:
Reinforcement does not prevent cracking - it controls crack width, by forcing many fine cracks instead of a few wide ones.
The driving force here is imposed deformation (restraint), not applied load. You cannot make the wall strong enough to resist its own restrained shrinkage; the strain is going to happen regardless. What the steel does is share that strain across many closely spaced, narrow cracks - each fine enough to be acceptable - rather than letting it concentrate into a handful of wide, visible, leaking ones. More, well-distributed steel means finer cracks.
Where the code meets the field
Across three recent basement pours, EC2's crack-width predictions came within about 30% of the measured values. That is genuinely useful for design - it puts you in the right band and lets you size distribution steel with confidence. But the gap is a reminder of what the formula cannot fully capture:
- Pour sequence and size change the restraint dramatically. A long wall cast in one go is far more restrained than the same wall cast in shorter bays.
- Curing and the thermal peak set how much the wall heats and cools, and therefore how much it contracts.
- Ambient conditions at the time of the pour shift the whole picture.
The practical lesson is that the calculation tells you the band, but the pour sequence and curing dominate the real outcome. Two walls with identical reinforcement can perform very differently depending on how and when they were cast.
Designing against it
- Plan the pour as part of the design. Bay sizes, construction joints and the casting sequence control restraint, and they are a structural decision, not a site convenience.
- Detail distribution steel for crack control, not just for nominal minimums - this is the steel that sets the crack width.
- Control the thermal peak through mix design and curing to reduce the contraction that drives the cracking.
The water-retaining crack-control article takes this mechanism into the watertightness case, where the same restraint cracking has to be held to a far tighter limit. The EC2 provisions are in the standards reference.