A historic bridge was one of the casualties of debris flows following abnormally heavy rain in eastern Scotland two years ago. Daryl Fossett, Maddie Clarke and Marion Duff explain how the damage was dealt with.
In early August 2004, abnormally heavy rainfall swamped Perth & Kinross and the east of Scotland.
By the evening of 11 August, some of the ground could take no more, triggering slips that continued into the following morning.
Several trunk roads were closed by debris slides, as was the C502, a minor road between Dunkeld to Rotmell. It had been part of the A9 until improvements in the 1970s introduced a bypass to Dunkeld; it was then retained as a local access and diversion route for the new A9 that runs parallel below it.
Damage to the C502 centred on two culverts, and an 18th century stone and lime mortar toll bridge was partially destroyed.
The bridge, originally 4m wide but widened to 7m in the 1970s, had been on one of the most important links in the military road network built by General George Wade, who was appointed Commander-in-Chief, North Britain by George I after the 1715 Jacobite rising.
Wade oversaw construction of barracks at key points in the Highlands. These were linked by a 400km road network, including 40 stone bridges, built between 1724 and 1740 to allow rapid movement of troops and artillery to crush any further uprisings. The Dunkeld to Inverness road was built between 1728 and 1730.
Following the slope failures offi als immediately closed the C502 as it was severely undermined in places. Perth & Kinross Council commissioned MouchelParkman to stabilise the area around two slides near the bridge, 4km north of Dunkeld, and design remedial solutions. The road reopened in June this year, but the historic bridge could not be saved.
The first failure ('Slide 6' in Figure 1) was 22m wide, centred on a culvert that carries a burn below the C502, and extended to the toe of the embankment, about 75m downhill.
At this point the burn was carried through another concrete culvert below the A9 to the River Tay, between 5m to 10m beneath the road.
The second failure ('Slide 7' in Figure 2), which centred on the toll bridge and its watercourse, was about 15m wide at the level of the C502, increasing to about 20m at its widest point. The slip was 50m long, with its toe 25m below road level. It featured near vertical slopes on its eastern and southern faces, made possible by the presence of semi-cemented dense fine sand.
Debris within the gulley it created included boulder size material with remnants of bridge, road surface and old gabion wall.
The amount of material and water ponding around the culvert access meant that during the initial inspection it was possible to see only the upper 500mm of the A9 culvert pipe at the base of the bridge slide.
Although debris initially washed through, the culvert eventually became blocked.
The area is forested with coniferous and deciduous trees, interspersed with patches of grass, low shrubs and gorse. The slip uprooted a large number of mature trees and deposited them in the bottom of the gully. Most of the displaced material ended up at the toe of the slope against the A9 embankment below.
The failure of the area around the toll bridge exposed a number of faces, allowing examination of the geological conditions.
Inspection revealed the sequence of material below the toll bridge as topsoil over glacial till - silty sand and gravel - over glacial outwash deposits comprised of fi ne sand and silt. The slide exposed two distinct layers of drift deposits, with general composition and distribution consistent with a glacial and fluvio-glacial origin. Bedrock was not uncovered.
Site inspections indicated a number of possible contributors to the failure at the toll bridge.
Passage of water
There had been a prolonged period of heavy rain in the area immediately before the failure. An excessive volume of water over the relatively short period meant the soil would have quickly reached saturation point.
It is believed water flowed around either end of the bridge and down its front facade. The extent of the bridge's structural failure suggests water scoured the spandrel wall foundations leading to its progressive collapse. The steep slopes of the gullies provided an easy route for water fl wing across the ground surface, which then passed into the gullies and the burn channels.
The drift deposits seen at the site - glacial silty sands and gravels and the uvial fine sand and silt - would be expected to behave in different ways in response to excessive volumes of groundwater. Generally, the granular nature of sand and gravel allows water to pass through relatively easily, but fine sand can liquefy quickly once wet.
It is plausible the fine sand liquefied because of the deluge's velocity head and continued flow, causing additional instability and continuing failure of the structure.
Loss of support to the sand, gravel and topsoil above the fine sand would have led to progressive collapse across the area.
It is thought water scoured the toll bridge at the western spandrel walls.
Progressive undermining of foundations and subsequent liquefaction of the formation layers are likely to have contributed to its failure.
The failure would have exposed fine sand below and to the west of the bridge. This in turn would allow water to erode freshly exposed surfaces and reduce support underneath the road and around the bridge.
The plucking failure of both the ground on the northern and southern faces and beneath the toll bridge is evident from its overstep and undermined condition. Ongoing spalling of material from the faces was undermining the slope and causing progressive failure.
Stabilisation and remedial solutions
3-D modelling software aided the remediation design and provided a check for rock fill volume calculations. It allowed comparison and appreciation of pre-site work and post-site work landscapes from different views. Figure 3 shows a 3-D model of the pre-works landscape.
A slope stability analysis software package calculated the factor of safety for the slopes as standing (pre-works) based on ground investigation data.
This software analysed the remnant debris slide pro les and the proposed slope reprofiling designs to ensure an adequate factor of safety.
For both debris slides, stabilisation and remedial solutions involved construction of two replacement culverts and placement of about 12,000t of stone to infill the scars and form new engineered channels. Gabion walls at the toe of each gulley supported these to minimise rock fill volumes.
After clearance of vegetation and reprofiing of the side slopes, workers laid a geotextile separation membrane on the surface of each slide, then placed rock on top of the geotextile and compacted it.
Engineered channels comprised gabion mattresses with box gabion toe cascades, made impermeable at the interface with the mass rock fill by a bentonite geocomposite.
The bentonite geocomposite was specified to induce water flow through and over the gabion channels and prevent water seeping onto the interface been the mass rock fill and fine sand formation. Figures 4 and 5 show the details of the culvert and cascade designs for Slide 6 and 7 respectively. Figures 6 and 7 show the emplacement of rock on top of the geotextile for Slide 6 and 7 respectively.
Figure 8 shows construction of the gabion toe wall for Slide 6. Mattresses were also placed within the existing channels to the culverts under the A9 to prevent scour on both slides.
On Slide 6 a precast concrete pipe was added to the existing culvert pipe.
This new pipe was skewed to aid the direction of water course towards the culvert at the toe of the slope.
On Slide 7 the original culvert was removed, as was the damaged toll bridge. This involved complete digging out of the existing culvert, placing a precast concrete pipe and surrounding it with compacted fill (Figure 9).
As well as the repairs to the two failures, a culvert (Culvert 8) about 93m west of the bridge failure on the C502 was found to have ruptured, allowing water to seep through its surrounding stonework (Figure 10).
This was repaired by inserting a 450mm internal diameter rigid pipe into it. A new headwall and gabion baskets were built to aid water flow through the new culvert (Figure 11).
At Culvert 8's out flow, workers used the same method as at the two failure sites, ie laying a geotextile with compacted rock placed on top. Gabion wall and mattresses were built and placed on top of the compacted rock to channel the water.
This method should have a greater ability to dissipate water energy at peak flow events.
The al stage of works for reinstatement of the C502 included some planing and resurfacing; and the installation of safety barriers and fencing. The fully reinstated Slide 6 is shown in Figure 12.
Hydroseeding will aid vegetation on the land adjacent to the slides and tree planting will begin in November.
Contractor George Leslie began main works on the £500,000 stabilisation and reinstatement scheme in April, nishing eight weeks later.
Daryl Fossett is principal engineer with MouchelParkman and Scotland Transerv. Maddie Clarke and Marion Duff are engineering geologists with MouchelParkman.