As usual there is a pdf file downloadable here
Stockport viaduct is magnificent. It was designed by G W Buck and built in 1839/40.
The town is in a valley and the viaduct strides over it on 22 spans, most of which are 63ft (19.2m) semi circles standing on piers up to 46ft (14m) high. Each pier is founded on an enlarged pad on the rock.
Figure 1 Drawing from Brees showing the south half of the viaduct
Note, in this drawing, that there is one thicker pier (12ft) near the centre of the run, founded in the River Mersey. For reference, it seems that the river span is No 14 with span 16, which becomes important, having the 10’0” wide label (which refers to the piers).
The viaduct was widened from 2 tracks to four in 1887 by taking the western spandrel wall down to track bed level and adding 22ft 4in (6.5m) of viaduct to the initial 31ft (9.45m). The new had apparently identical piers and spans. This time, though, there were no sleeper walls and no spandrel wall at all against the original. There was no connection to the old either, which may influence what happened later.
After a hot summer in 1915, when the extension was 28 years old, it was noticed that there was a hump in the track at span 16. The track was simply adjusted, but by 1917 there was a 12mm hump in the parapets. Sighting irons were fixed over the piers and at mid span. Using these, a further rise of 20mm was noted by 1920, which increased to 45mm by summer 1927. Compression damage in the parapet led to a 300mm slot being cut to release the stress.
On 28th July 1929 two bricks fell from the arch soffit into Heaton Lane some 20m below. This prompted observation with binoculars, revealing transverse cracks in the arch soffit at the haunches, near the point where the backing met the extrados. What is more, it became clear that the springer at the north of span 16 in the new arch was offset 25mm south of the original pier.
A fire escape ladder was extended to a few feet short of the crown and the railway company (LMS) steeplejack climbed to inspect both visually and by probing with an iron shod pole. He observed a crack 25mm wide at the north haunch and another 15mm wide at the south. Near the crown was an area of 6m square where brick on the intrados had crushed. The arch had risen 150mm at the crown and deflected visibly when trains passed over. (Note that this is not a severe test. All viaducts deflect visibly when observed from close by, especially if there is a stationary reference, in this case the old viaduct).
The treatment of the viaduct was written up in a paper by Edward Harold Morris in 1948. The details above were extracted from that paper. What follows, though, is my own development.
Some questions
The 6” rise and 1” shortening (150mm and 25mm) may seem odd but the geometry is correct. The cracks at the arch haunches show that rotation was centred there. Looking at the thrust in Archie we get a picture like this:
Before movement, with a thrust of 300kN/m
And like this after, with a thrust of 760kN/m
We can model the 150mm rise of the arch by moving the thrust up at the crown like this, which reduces the thrust to 593kN/m width.
Thus, if all the movement were generated north of span 16, the spans to the south would see only increased thrust. If the southern spans (1-15) were also expanding, there is no reason why the piers there should not tilt northwards, but they didn’t, so it seems we must seek an issue that affected only half the viaduct. In this situation the 12ft wide pier will present no extra resistance to movement. It is higher and therefore equally slender.
Visual Inspection
It is some years since I spent a sunny day in Stockport but going through these pictures with hindsight from reading the paper shows some interesting things.
The hump in the damaged span clearly attracted my attention and is very visible in Figure 2.
Figure 2 The damaged span is visibly humped
Figure 3 shows damaged bond between the rings and a slightly open horizontal joint in the spandrel, which would be expected as the crown rose so much.
Figure 3 Evident damage to bond in the rings and the spandrel wall at the haunch of span 16
There is much less apparent damage in the northern half of the span, where the crack was wider. The narrower crack to the south means there is some other influence. Possibilities are:
There was a rotation elsewhere with one or more further cracks.
The two rising arms were of different lengths, which means that the hinges formed in different places, released differently by the backing.
·There was crushing in the upper rings where the crack was narrower.
Figure 4 Interestingly, the damage is less evident in the Northern half of span 16.
The 6m square area of crushed brick is a strange description. That is almost the full width, but if it were the full width, surely the report would have said so. Which suggests that there was little or no crushing under the spandrel wall, or it was much more concentrated there.
The distribution of offsets in the piers today fits well with earlier observation. The fact that I was there on a sunny day means that the overlap at the north pier of span 16 was emphasised by shadow (Figure 5 and Figure 6).
Figure 5 A 25mm step in the string course. N pier span 16
At span 17 the step is reduced, at span 14, there is no step. Both these statements agree with the observations recorded in the report
Figure 6a Step tapering down pier
Figure 6b Taper to nothing at ground level
Figure 7 Span 17 with a reduced step
Figure 8 Span 14 with no step
The next question is: why did the older viaduct not respond similarly to the heat? In considering that, it is useful to refer to the drawing presented with the paper, and that presented above. These can be considered in connection with the photographs below.
Figure 9 Drawing from paper showing internal construction
Figure 9 shows details of the internal construction of the old and new viaducts. The sleeper walls were obviously not known about or at least not mapped over into the new. But the drawing in Brees doesn’t show them at all. The lines of original backing in the new viaduct are presumably a record taken during the reconstruction so a sound record. It shows the backing dipping away from a tangent near the crown to about the original level of the crown intrados over the pier. The dip means that water is run to the pier head, raising a question about how it was dealt with there.
But first, a check on those internal walls.
In several spans adjoining Viaduct Street the old viaduct is clearly saturated. There are substantial water stains down the spandrels and under the arch (Figure 10). The stain on the spandrel is higher than on the arch because the easiest flow lines (through the mortar beds) there are horizontal. It may also be that there is a crack at that level as a result of movements that are the subject of a current investigation.
Figure 11 shows more detail at span 5. There are clearly steps in the water line across the width. These correspond with the position of the sleeper walls showing the water has built up to higher levels in some of the bays.
The reason for the water being there is made clear by comparison of Figure 1 and Figure 12. Figure 1 shows the drains emerging four stone courses down the pier, just above ground level. Figure 12 on the other hand shows but 2 courses above the ground. It seems likely that the ends of the drains have been buried blocking the outflow.
Figure 10 Water stains at Spans 4 and 5
Figure 11 Span 5 detail showing steps in water under arch
Figure 12 View from StreetView of Viaduct Street. Spans 5, 4 and 3 from left to right.
That dropped the question of expansion control. These two pictures of the structures might help
In the old backing, the sleeper walls will be level with the top of the arch and so provide a direct load path from end to end.
The backing on the new section falls to a trough at the pier centre. The load path now waves up and down from span to span so that expansion would encourage the arches to burst upwards. Where and when those failures would occur depends on many things. One would be the position and condition of bed joints. A patch of dry, or not well compacted, concrete near the top in one span would encourage the arch to lift the backing and free itself locally. The chance of such a dry patch occurring at the arch and at the same level both sides is small so it is likely that any response would be unsymmetrical.
If the fracture level were slightly different on the two sides, the thrust pattern would tend to look like this.
The arrows indicate the length of the rising arms. The longer one would produce a smaller rotation for the same rise.
And finally
Figure 13 View along the viaduct from the South showing humps of various sizes in the string course
Figure 13 shows a view from the higher levels of a footpath rising up beside the South abutment. There are very obvious humps at four spans and less obvious ones at a further 5 of the 19 spans visible here. The first major hump, just beyond the third gantry from the right, appears to be at span 9, which places the biggest one at span 16. These humps suggest that there was indeed expansion in the southern half but that it distributed over several spans and possibly didn’t create any pier tilt.
Causes
I still have difficulty attributing this to temperature. There has been a permanent expansion and temperature doesn’t do that. It is surely more likely that this is growth of the bricks or mortar or both. Bricks grow in early life by absorbing moisture from the atmosphere. There are many chemical reactions that might cause mortar to swell. The 1880 mortar will be different from the 1840 and may have reacted differently to the acid environment of industrial Stockport.