In his definitive work 'Properties of Concrete', Professor Adam Neville states: 'Calcium chloride should never be incorporated into reinforced concrete; in prestressed concrete the risks are even higher.'
Chloride ion-accelerated corrosion of embedded steel is the major risk; so in one sense the news that calcium chloride was one of the contributing factors to the collapse of a footbridge in Concord, North Carolina, comes as no surprise.
There was a time when shovelfuls of CaCl were routinely added to every batch of concrete produced in precast concrete works. No cheaper way of boosting the early rate of strength gain was available.
Units could be demoulded quicker, costs slashed and operative bonuses maximised.
As a result, the philosophy which states 'one shovelful good, two shovelfuls better' gained almost universal acceptance. Chloride levels soared, reinforcement corrosion became a serious problem and it was not long before virtually every concrete code banned the use of accelerators based on calcium chloride.
It is not suggested that the producers of the double 'T' units which failed in North Carolina added CaCl to the concrete used for the unit itself.
These days strength gain is most commonly enhanced by some form of heating coupled with much safer water reducing admixtures. Rather, it seems that CaCl was used to accelerate the grout which filled a midspan void stretching down from the top surface of the unit to the prestressing tendons themselves.
Unfortunately, among the other unwelcome side effects of too much CaCl are excessive heat evolution during hydration and an increase in drying shrinkage by up to 50%. The grout would have cracked, then shrunk, then crumbled. Without a waterproofing membrane or wearing course to protect the double 'T' units at Concord, there was nothing to stop rainwater percolating through the grout - picking up plenty of chloride ions on the way - and penetrating to the tendons themselves.
Why there was a void in such a critical location is another question. Precast prestressed units were traditionally produced by stretching tendons between stout anchorages at each end of long beds. The main drawback is that straight tendons are not the most efficient way of prestressing concrete.
Precasters eventually began to experiment with more complex systems, which would deflect the tendons to give maximum prestress at midspan and reduce the need for shear reinforcement.
US producers largely went for the 'harping' system, where the tendons are pushed down at midspan by forked steel rods.
Unfortunately, withdrawing these rods is a complex operation which leaves a large midspan void behind. This then has to be filled with hardened grout before the unit can be lifted off the production line.
Thankfully, UK producers mainly choose pull-down systems. This means that any voids left after the stressing operation are in the much less vulnerable soffit. UK experts assert that the risk of a similar failure here is much lower than in the US - and I tend to agree with them.
Dave Parker is technical editor of NCE