Alasdair’s Engineering Pages © A. N. Beal 2018    

Alasdair’s Engineering Pages

The Structural Engineer, Volume 74, No. 1, 2 Jan. 1996

‘Skyhooks’ at 18 foot centres: strengthening a precast OPC/HAC concrete frame at Hattersley Heaton, Nottinghamshire

A. N. Beal BSc (Eng) CEng MIStructE MICE

Thomason Partnership


A two-storey precast concrete framed office block dating from the 1960s was found to have high alumina cement (HAC) concrete beams at roof and first-floor levels. Detailed investigation showed that the first-floor beams were satisfactory but those at roof level could not be proved. In addition, problems were found with first-floor corbel details and the precast frame beam-column joints. Because the ground floor area had to remain in use during repairs and also the foundation conditions were poor, the roof was strengthened from above using ‘skyhooks’ - steel beams placed over it and preloaded to limit stresses in the HA beams. The first-floor corbels were strengthened by drilled-in prestressed bars.


Hattersley Heaton’s offices in Lenton Lane, Nottingham, occupy a fairly ordinary two-storey precast concrete framed block, with first floor and roof spanning 10.8m clear between precast concrete frames built into the external walls. It was built in 1967-68 by Vie Hallam, with ‘T6’ precast concrete frames supplied by Trent Concrete. At one end there is a rooftop tank room. The site is made ground, and the columns are supported on piles founded at depth.

The roof structure comprises 230mm x 203mm Trent H9 X-joists in pairs at 715mm and 810mm centres, spanning 10.8m. At the end bay over the staircase, the span is divided by an intermediate beam and the roof joists support large water tanks in a rooftop tank room. The first-floor beams are 375mm deep inverted T-beams at 915mm centres, acting compositely with thin precast concrete planks and in situ concrete topping. At the front and rear of the building, the roof and floor beams rest on deep L-shaped fascia boot beams which span between precast concrete columns.

The Trent T6 frame has rolled steel section cuttings cast into the concrete columns at each floor level; steel seating cleats bolted to these with HSFG bolts provide support to the incoming beams. The beams have cast-in steel inserts at their ends, so that at each beam-column junction there is a bolted steel-to-steel connection.

Fig 1. General view (note rooftop tank room)


Tests proved that the roof level X-joists, the first-floor T-beams and the internal beams over the stairwell (including the beam supporting the joists under the rooftop tank room) were made of HAC concrete, but the remainder of the structure, including all external beams and columns, was of OPC. As expected in a structure of this age, the HAC was substantially converted, with five initial samples showing conversion between 34% and 72%.

For strength assessment, reference was made to the recommendations in the Stone Committee report (BRAC Report by subcommittee P (High Alumina Cement Concrete)) [1]. This gives guidance on assessment of existing HAC structures. The roof and floor beams in this case fell outside the limits which can be deemed to be acceptable without calculation. Tabulated ultimate moment capacities for the Trent H9 joists are given as 24.1-25.9kNm (depending on reinforcement). Combining these with the recommended ‘beam performance factor’ of 1.3 and an average load factor of 1.5 gave an allowable moment at working loads of 20.9kNm/joist; the calculated maximum applied moment at working loads was 30kNm, so there was clearly a risk of failure under full load.

The tabulated beam moment capacities are based on an assumed strength for converted HAC concrete of 21N/mm², so in order to be acceptable the concrete in the roof joists would have needed to exceed a cube strength of 30N/mm². According to the Building Research Establishment, HAC in existing structures commonly has a strength of 30-40N/mm², so this was a possibility. However, it is not feasible to extract core samples from 230mm x 203mm X-joists without destroying their load carrying ability, so testing cores from any realistic proportion of the roof joists was obviously out of the question. Non-destructive testing did not offer a solution either. The Schmidt rebound hammer does not give accurate results on HAC. The BRE pullout test is affected by prestress (present here), reinforcement (present in the joist soffits) and curved surfaces (such as the sides of the X-joists); in any case, the calibration curve for this test [2] shows a substantial error margin, so even if suitable locations for pullout tests had been available, results would have needed to be consistently in excess of 40N/mm² to give reasonable confidence that actual strengths were at least the 30N/mm² required. In the absence of any realistic method of proving that the roof joists had adequate strength, there was no alternative but to develop a scheme to strengthen them.

The first-floor inverted T-beams are larger than the roof joists, and they act compositely in bending with an in situ OPC concrete topping. Calculations showed that their moment resistance was satisfactory (because the OPC concrete does not suffer strength loss) but the calculated principal tensile stress generated by shear was 1.11N/mm², giving a theoretical safety factor against failure (based on the BRAC Appendix K recommendations) of 1.5, compared with the usual BS8110 global safety factor for this of 1.5 x.1.25 = 1.875. It was felt that, for peace of mind, a safety factor of at least 1.8 against shear failure was desirable, for which the equivalent cube strength of the concrete would need to be at least 30N/mm².

Core samples were taken (from the beam bottom flanges, at approximately quarter span points) and crushed to determine their strength. They were also tested (by X-ray diffraction) for HAC conversion. The results are listed in table 1 below.

Core test results from HAC beam samples

The two samples which fell significantly below 30N/mm² were carefully examined. One contained three reinforcing wires and, unlike all the others, it had failed in shear, indicating a local plane of weakness. The other was noted to contain a piece of sandstone coarse aggregate, unlike the gravel found elsewhere in the concrete. For prudence, a second core sample was taken from this beam; this gave a strength of 47N/mm², and samples taken from the two neighbouring beams gave strengths of 46N/mm² and 45N/mm², supporting the view that the initial low result reflected a local pocket of poor concrete rather than a general weakness. The uniformly high conversion values gave confidence that further significant loss of strength was unlikely, so the beams were judged to be acceptable without strengthening.

Other problems

Although the major structural problem initially identified was the presence of HAC concrete, in the course of inspection and investigation some other problems were also identified.

(1) The grout at the external beam-column joints had cracked, because of differential stresses between the concrete structure (which will have tended to shrink) and the brick infill wall panels (which would not). The Trent T6 frame relies on structural steel connections for its integrity at beam-column junctions and these rely on the grout as corrosion protection, so the cracking could have potentially serious consequences in the long term.

(2) Standard Trent beam details for the T6 system from the early 1970s show the projecting nibs on fascia boot beams as being reinforced with mild steel 8mm links, regardless of the span of the floor beams they supported. A section of nib in the structure was therefore opened up on site to check its reinforcement, and this was found to be reinforced with ¼in mild steel links - much less than would be necessary to justify the design of the nib to support the applied loads. Clearly, the nibs were relying on the tensile strength of the concrete to work.

(3) The roof asphalt had reached the end of its useful life.

Fig 2. Beam-column connection opened up (note rusted steel seating cleat)

Fig 3. General section through building, showing remedial works scheme

Fig 4.
Detail at edge of first floor

Remedial works - roof

The client’s requirements were that, although parts of the first floor could be evacuated during the remedial works, the ground floor offices were to remain in use, as must the water tanks in the rooftop tank room. Combining these restrictions with the known poor ground conditions, it was clear that the simplest structural solution, insertion of internal columns on new foundations to support the roof and floor, was ruled out. In theory, new supporting beams could have been inserted below roof level to span across between the existing columns but this would have been very disruptive, with difficult steel-concrete connections and the limited floor-ceiling height (less than 2.7m) left insufficient depth available for new beams designed to span 10.8m.

With these solutions ruled out, the only alternative was to strengthen the roof structure from above, erecting ‘skyhooks’ - 610 x 305 UB steel beams spanning across the roof, seated directly on to the column heads. Long bolts were passed down through the roof from these beams to pick up secondary beams positioned below, designed to support the roof joists at mid-span. The secondary beams were aligned to run down the first-floor central corridor, thus eliminating the need for any structural work in the offices on either side, and 254 x 254 UC sections were used to keep the loss of headroom to a minimum. The roof top beams were seated on SK sliding bearings and secured with bolts in slotted holes, so as to minimise stresses generated by differential thermal movement, and great care was taken over flashing details.

To renew the roof asphalt at minimum cost and disruption, it was decided to overlay it with a layer of ‘Amascoflex’ reinforced asphalt and chippings. This additional weight further increased the stresses in the existing roof joists and simply introducing the additional support at midspan to relieve live load moments was not sufficient to reduce the stress to an acceptable level. To reduce the moments in the roof joists further, the new supports had to be preloaded, to relieve some of the dead load as well as the imposed load. A variety of specialised means are available for doing this, such as hydraulic flatjacks but it was felt that a simpler method should be sought, in order to keep cost and potential complications to a minimum.

The solution was to prestress the structure by controlled tightening of the supporting rods which passed through the roof. The main steel beams themselves were used as load cells to monitor the applied load. Calculations showed that if the bolts were preloaded to the required tension, this would deflect the main beams by 5mm. Using this technique, the contractor was able to prestress the beams by simply tightening the nuts on the support rods, controlling the applied preload by using an ordinary engineer’s level and staff to monitor the deflection of the beam. No difficulties were experienced on site, and the contractor found the process simple to understand and control.

Fig 5. Steel beams being inserted to support water tanks

Fig 6. Controlled prestressing of ‘skyhook’ beams

The rooftop tank room presented a different set of problems. Again there were strong reasons for trying to carry out all the work above roof level, so as to avoid disruption to the office below, but the space inside the tank room was very restricted, with the water tanks occupying most of its volume. The structural scheme adopted was simple: the insertion of a grillage of galvanised 203 x 203 UC beams beneath the tanks and to support them and relieve the existing HAC X-joists of their weight. However, its practical execution called for ingenuity by the contractor, particularly in view of the fact that the client wanted the tanks to remain in use during the work, with minimal interruption to the water supply.

The solution adopted was firstly to demolish one wall of the tankroom above roof level (for access) and to freeze the pipes to the tanks so that they could be cut and temporarily reconnected with flexible pipes. With this done, flat trolley jacks were slid in under the tanks, between the tank supports, and these were used to lift the tanks slightly, allowing just enough clearance to manoeuvre the new grillage support beams into place and lower the tanks on to them.

Remedial works - fascia beams

Although the fascia beam nibs which support the main first-floor beams showed no visible signs of distress, their reinforcement was so light that we could give no assurances about their safety. The opening up which had been carried out revealed that, although the projecting nib bars were only ¼in mild steel links, there were substantial, high tensile reinforcing bars running longitudinally in the beams.

To strengthen the nibs, horizontal, high tensile rods were inserted through them from the outside and prestressed to compress the nib and beam together, reducing the theoretical principal tensile stress in the concrete to an acceptable level. Site measurements of the nib which had been opened up indicated that there should be just enough space for drilled holes to pass through between these bars without cutting any main reinforcement; the links could be located on site by covermeter and the holes positioned to avoid any of them being cut.

The final detail adopted comprised M16 grade 8.8 steel rods with grade 10 nuts (as in steelwork HSFG bolts) passed horizontally through the beams and nibs. These were anchored on the inside by rectangular steel plates and on the outside by standard circular cast-iron pattress plates. Resin was injected under pressure around the bolts to protect them and also provide some bonding. The resin was introduced through holes in the external anchor plates, with witness holes drilled in the internal plates to check that there was full penetration. It had been hoped to use stainless steel for the prestressing rods but galvanised steel had to be used instead because of the lack of availability of nuts in a suitable grade of stainless steel.

A series of pilot holes was drilled on site to confirm feasibility before proceeding with drilling the main series of holes. The required prestress was applied by torque wrench, with the torque/tension relationship being calibrated before proceeding on site. Bolts were tightened to 90% of the specified tension and then retorqued to full tension at least 24h later to check for relaxation.

At the external fascia beam-column junctions, where the original protective grout had cracked, a compromise had to be made between the need to fill cracks solidly (to prevent water ingress and thus corrosion) and the need to allow some of the inevitable thermal movement to continue to occur. The solution adopted was to inject cementitious polymer grout into the area locally around the structural steel ‘T6’ connector and then to chase out the remaining lengths of concrete/concrete and concrete/brick joints above and seal these with a suitable mastic.

Fig 7. Finished ‘skyhooks’


‘Skyhooks’ do have their uses and sometimes solutions to a tricky problem can be ‘high tech’ in principle but ‘low tech’ in execution. This may not push back the frontiers of engineering technology but it can be satisfying to the engineer, fun for the builder - and give the client what he wants.


Thanks are due to Hattersley Heaton for permission to publish the material in this paper. Thanks are also due to the Building Research Establishment (Dick Currie and Nora Cramond) for helpful advice on HAC.

Core testing: Labtest

Chemical analysis: UK Analytical, Leeds

Building contractor: Friargate Ltd, Derby

Concrete repairs: SCL Ltd, Doncaster

Reroofing: Briggs Amasco


1.  Building Regulations Advisory Committee Report by subcommittee P (High Alumina Cement Concrete), Department of the Environment, Welsh Office.

2.  Chabowski, A. J., Bryden-Smith A. W.: ‘Internal fracture testing of in situ concrete: a method of assessing compressive strength’, Information paper 22/28, Building Research Establishment, October 1980.

The original copy of this paper is available from




















Conv. %



















Table 1 — Strength and conversion for core samples