On a clear day the view from the top of the 100m high prill tower at Terra Nitrogen's Severnside works is breathtaking. To the west lies the Severn Estuary, with the dramatic S-curve of the Second Severn Crossing on the one hand and the elegant silhouette of the Severn Bridge on the other, and the Welsh hills beyond. But visitors making the trip to the top in the last few years might have been more concerned by an unusual sight much closer to hand - right beneath their feet, in fact.
There they would have seen the alarming signs of major concrete decay. The tower's 180mm thick roof was blistering and spalling, with exposed reinforcement corroding away. After more than 30 years producing up to 1,500t of fertiliser a day, the prill tower was in a bad state.
It was all down to the hot, concentrated solution of ammonium nitrate, which is constantly pumped to the top of the tower and sprayed down the inside.
Few textbooks cover the reaction between ammonium nitrate and concrete (see box). When Taywood Engineering was commissioned in 1996 to investigate the tower and come up with a viable refurbishment strategy some serious headscratching took place.
'A literature search yielded little specific information on ammonium nitrate attack and how to deal with it, ' reports Taywood senior concrete specialist Dave Cullen. 'And such papers as there were tended to be opinionated and contradictory. We concluded that it was a very complex mechanism, but not completely different to sulphate attack.'
Surveys showed levels of ammonium sulphate in the saturated concrete as high as 8% by weight. But there was a baffling apparent randomness in the distribution of high spots and the subsequent concrete deterioration. Taywood project manager Matt Gascoigne says that the worst deterioration had taken place in the most obvious place, the tower's roof beams, immediately above the sprayhead 'This wasn't thought to be too serious, a steel strengthening structure had been added to the roof some time ago, ' Gascoigne explains. 'But then a finite element study showed the original beams were needed for stability.'
More puzzling were a number of crumbling areas around the crucial interface between the tower proper and the five floor plant room which perches alongside its upper levels. Taywood eventually concluded the apparently random nature of the pattern of damage could be explained by intermittent discharges from the maze of pipework close to the common 180mm thick concrete wall.
Much depended on the degree of saturation at a given point, or the frequency of wetting and drying.
Terra even considered closure and demolition, such was the severity of the corrosion and the uncertainty over possible repair techniques. 'Our nightmare was the tower decaying so far it would be too dangerous either to repair or demolish it, ' says Terra maintenance manager Chris Blackman. However, Taywood's repair strategy allowed the tower to keep operating throughout. In late 1999 contactor Bierrum was awarded a £1.1M, 24 month contract to give the tower another 50 years of safe working life.
There were three key elements to Taywood's solution.
First, up to 10% of the cement in the repair concrete was replaced by microsilica - ultrafine spherical particles of reactive amorphous silica, which reduce concrete permeability and react preferentially with the vulnerable calcium hydroxide. Spray application was chosen for the wall sections to be replaced.
'Can you imagine trying to fix conventional shuttering behind all those pipes and brackets?'
The real key, however, was to use sophisticated finite element analysis to control the repair sequence and maintain adequate loadpaths as patches of concrete were cut away from the walls.
There was a particular problem at the eight level floor in the plantroom, from which hung the two 60t evaporators. Before floor repairs could begin new support steelwork had to be manoeuvred through the pipework to take the load from the evaporators down to the existing tower roof steelwork.
Jacking the evaporators off their existing concrete supports and transferring the loads to the new steelwork is the trickiest operation of all, says Gascoigne.
'To work properly, the evaporators have to be absolutely vertical. The maximum distance we could lift one side when were jacking was 1.5mm. On the first evaporator we completed the load transfer without raising it more than 0.3mm.'
In a few weeks the load from this first evaporator will be transferred back to the reconstructed concrete and the entire steel support structure dismantled and reassembled around the second unit. Spraying subcontractor Duracrete is well advanced with the wall repairs, using a proprietary Redaspraycrete 40 mix with added microsilica, which has very low rebound.
Bierrum expects to complete the bulk of the works by the end of this year. Repairs to the tower roof beams will take longer, however, as they are only accessible during the factory's annual two week shutdown.
In all, around 15t of temporary steelwork and 120m 3of microsilica repair concrete will be needed.