Piled embankment design, construction,and monitoring for A1/N1 dual-roadway
By Wyatt Orsmond
The A1/N1 section of highway between Newry and Dundalk forms part of a strategic transportation link between Belfast and Dublin. This section includes 14.5km (9mi.) of high-quality, dual-carriageway (divided-highway) construction and was challenged by a tight route corridor threaded between steep mountains and peat bogs.
Crossing the Flurry Bog — a wetland almost 1km (0.62mi.) long — was a significant engineering challenge, with peat and very soft silts to 9m (29.5ft) deep. The solution included a combination of excavate-and-replace with a 400m (438yd) section of piled embankment. The piled section was preloaded prior to construction of the road pavement to induce initial settlements and reduce long-term creep.
This article covers the design process undertaken, the challenge of constructing the piled embankment with more than 2,700 piles, monitoring of the basal reinforcement deflections and soil loading, and comparison of these with design values. The monitoring included horizontal inclinometers along the basal reinforcement and pile interface, settlement plates at pile-cap level and on the surface, and load cells placed on the pile caps and on the basal reinforcement between pile caps.
This 14.5km (9mi.) section of dual carriageway included 13 structures and required 1.9 million m3 (2.5 million yd3) of cut-to-fill, of which 540,000m3 (705,780yd3) was rock and required blasting. Excavate-and-replace of unsuitable soil totalled 160,000m3 (209,120yd3).
The Flurry Bog is a cutaway bog where the levels had been reduced to groundwater level with poor drainage exacerbated by an adjacent salmonoid-designated section of an adjacent river that is prone to flooding.
The area, therefore, resembled more a wetland than bog and the peat had limited fiber strength making surface access exceptionally difficult, even on foot. The difficulties of access were evident in the limited ground investigation information required for detailed design. This was achieved with the aid of an all-terrain lightweight tracked utility vehicle from which probing and window sampling was undertaken. However, even this specialized vehicle had its limitations. Figure 1 demonstrates the ground conditions anticipated.
The main area of peat and soft silt extended for about 900m (985yd) and up to 9m (29.5ft) deep, thinning out to a nominal layer of peat over firm glacial soil. To provide further clarification on the extent and depth of peat, a “soft soil” model was generated using MapInfo with a typical output presented in Figure 2.
Although other construction options were considered including excavate-and-replace within a cofferdam, the piled embankment proved the most cost-effective with the least impact on the adjacent river.
Design of the load-transfer platform was based primarily on BS80061. However, consideration was given to the variation in fill height both along length and width (crossfall) of the embankment.
Variations in span distance between piles was also considered, as was lateral movement and the transition zone from piling to excavate and replace sections. Due to alignment constraints, the embankment height was about 3.0m (9.8ft) over the piles. Because of this, it was decided to preload the embankment in order to pre-strain the basal reinforcement thereby reducing the effects of long-term strain deformation.
The final design incorporated a woven polyester geotextile to act as the basal reinforcements accommodating the design loads of 540kN (longitudinal) and 660kN (transverse) for an embankment height of ~3.4m and 700kN (longitudinal) and 780kN (transverse) for an embankment height of ~2.6m.
A number of issues had to be addressed to construct this embankment. Gaining access to the surface for piling rigs and deliveries was the most important. Other issues included the procurement of precast piles (2700No.), piling tolerances and pile cap construction.
To gain access, a working platform was constructed. Considering the very weak surface soils and the need to keep the platform as thin as possible to limit immediate settlement, a number of geogrid and geocell options were proposed. The trial of a separation geotextile and two layers of geogrid spaced within 600mm (23.6in.) rockfill proved sufficient for the intended loads but weak enough for the piles to be driven through it. The platform performed well and typically settled about 400mm (16in.) but in places settled about 800mm (31.5in.) leaving parts of the working platform submerged.
To validate the design and assess the effects of the preload, the embankment was instrumented and monitored at two locations, A and B (Figure 2), with fill heights of 3.4m (11ft) and 2.6m (8.5ft). The monitoring included the installation of:
- settlement plates at base, positioned on top of pile caps and at midspan above the basal reinforcement.
- horizontal inclinometers positioned directly over pile row, midway between piles and 1m (3.25ft) above these positions.
- pressure cells on pile caps, midspan between piles (above and below the basal reinforcement).
- fiber-optic strain gauges over pile caps, through perpendicular and oblique span.
- surface monitoring settlement pins.
The settlement points and inclinometer readings were undertaken weekly for 16 weeks and monthly thereafter. The pressure cells were connected to a data logger with hourly readings. Unfortunately, due to damage sustained apparently after installation and partial filling, the fiber-optic cables were broken and no strain measurements were recorded using this system.
Base settlement plates
One of the simplest ways to measure base settlement is by using plates positioned at the base of the embankment with a sleeved riser pipe to the surface. These are, however, prone to damage by filling operations and construction plant.
These plates were positioned at two points at each location (four points in total). One plate was positioned on top of the pile cap and the second on top of the basal reinforcement between adjacent piles. The data is presented in Figure 4 and is further examined in the Discussion section.
To monitor the pattern and magnitude of deflection of the basal reinforcement, horizontal inclinometers were positioned directly over a line of piles, parallel to this between two rows of piles and approximately 1m (3.25ft) vertically above each of these positions.
In general, this approach worked well, although as deflections reached 70-100mm (2.8-4in.), the tubes directly over the piles kinked and the inclinometer could not go through.
The data is presented in Figures 5–7 for two locations, labeled A and B,as follows:
- Figure 5 — Location A, midspan displacements between two rows of piles directly on top of basal reinforcement. Included in this graph is a typical displacement of the inclinometer directly over the pile caps. (Note: Data is not presented for a short section due to baseline line error. Convergence of lines at ~27m is due to linked pile caps.)
- Figure 6 — Location B, midspan displacements between two rows of piles directly on top of the basal reinforcement. Included in this graph is a typical displacement of the inclinometer directly over the pile caps (Note: Convergence of lines at ~6.5m is due to linked pile caps.)
- Figure 7 — Location A, displacement ~1m above pile row. Included in this graph is a typical displacement profile directly over the pile caps. (Note: Datum has been altered so that the line plots within the same range.)
Pressure cells were placed at both monitoring positions, A and B. The cells were positioned on top of the pile caps (A5, A10, B7, B9), midway diagonally between pile caps (A1, B2), directly below this underneath the basal reinforcement (B3), midway between two parallel pile caps (A6), and directly below this underneath the basal reinforcement (B4).
As the data indicates, there was no significant movement of the surface of the embankment. This would have been expected as the preload fill of 1m put the surface well above the empirical internal soil arch height.
The base plates gave some erratic readings indicating “lift” although considering that this is limited to about 10mm (0.4in.) it could be contributed to surveying errors, particularly in light of the difficult working conditions. The monitoring did, however, indicate that at midspan between two pile caps at location A, displacement was 130mm (5in.) and at location B displacement was close to 50mm (2in.). Considering that location A had an almost 0.8m (2.6ft) higher fill height, a difference could have been anticipated, but not to this magnitude considering that the basal reinforcement at B was significantly stronger.
The inclinometers directly over the pile caps produced good initial results, however it would appear that as the underlying peat and working platform settled, the profile changed to be more abrupt on the edge of the pile caps thereby kinking the tubes. The results included in Figures 5 and 6 do, however, show a consistent pattern of deflection although the magnitude varied. For both locations A and B, the average displacement was 78mm (3in.), with a typical maximum displacement of 105mm (4in.) although isolated values of up to 140mm (5.5in.) were recorded.
The inclinometer tubes placed parallel to the line of piles were not subject to kinking over the rigid pile caps and provided readings throughout the construction period. The data shown in Figures 5 and 6 show a difference in deflection between the two locations. Location A basal reinforcement typically deflected about 75mm (2.9in.) but up to 140mm (5.5in.) in places. For location B, displacements were typically about 100mm (3.9in.) with a peak of just over 150mm (5.9in.).
What is apparent is that the displacements measured along the midspan (i.e., controlled by the longitudinal basal reinforcement) are more variable. As the inclinometer tube is perpendicular to the longitudinal reinforcement, any variation in strain between parallel strips of reinforcement particularly relating to laying the material, could cause this.
In contrast, the inclinometer tube running over the pile caps will primarily be measuring the behavior of the transverse basal reinforcement, although the effects of the interaction of the pile caps will help control large variations in displacement (strain).
A review of the displacement of the basal reinforcement measured between pile caps has been undertaken in order to make a comparison with the BS8006 design method used. The results presented in Figure 9 show the average deflection measured, the typical maximum, and the absolute maximum. Note that the results for locations A and B were similar and the information presented is a summary of both locations. Using these average displacements, the strain of the basal reinforcement was calculated at between 0.5% to 1%. This is significantly lower than the design strain of 5%. However, when the material and load factors are accounted for, the expected design strain is about 2%.
The design method (BS8006) highlights the need for additional strength and lower strains for shallow embankments. This effect is highlighted by these results where the shallower embankment (location B) appears to have deflected similar to that overlain by higher fill (location A) even though its basal reinforcement was about 23% stronger.
When considering arching effects, the assumed arch profile and arching behaviour appears to vary from that recorded in this instance. As demonstrated in Figure 7, it appears that the soil overlying the piles has also displaced, partially filling the void created by the deforming basal reinforcement. Some mirroring of the reinforcement displacement is also evident. Much of this displacement would have occurred during filling and preload, which could have “disturbed” the arching effect. Due to the lack of displacement on the surface (Figure 3), it is apparent that the arching effect is occurring.
The results from the pressure cells were highly variable and offered limited insight to the distribution of loads onto the piles and basal reinforcement. Both pressure cells below the basal reinforcement and one above gave slightly negative readings although appeared to be recording correctly.
The pressures recorded on top of the basal reinforcement at midspans was in the order of 20kPa. However, on removal of the surcharge and replacement of the pavement layers, loads increased to about 50kPa. This would indicate that the reduction of applied loads has caused a reduction in the arching effects and an increase in the load on the basal reinforcement.
The pressure cells positioned on top of the pile caps (above the basal reinforcement) responded well to the loading and pre-load removal. However, the values recorded varied from 80kPa to 580kPa. On removal of the surcharge, the reduction of loads in all cases is rapid and distinct. On reapplication of part of the load through pavement construction, there is evidence of an increase in pressure to about 20% lower than that recorded under preload conditions. For cell A10, the pressure after preload removal and pavement construction exceeded that recoded prior to removal.
Monitoring of this type of geotechnical structure has its own particular challenges. The wide range of large degrees of localised differential displacements and small surface movements makes the selection of instrumentation critical.
Even with double or triple systems, between mechanical damage and equipment failure, very limited data can be forthcoming. In this particular case, there are some large variations in readings and, as yet, some unexplained results that are probably related to equipment and reading errors. However, the data that was collected provides an insight into how the basal reinforcement of a pile embankment behaves.
The settlement plates and horizontal inclinometers indicated that the basal reinforcement typically deflected downward in the order of 75mm to 100mm although displacements of up to 150mm were recorded.
When compared to the design strain values of 5%, (2% when design factors are excluded) it appears that actual strains are significantly lower at between 0.5% and 1%.
Pressure cells showed highly variable results and although this could be attributed to the variation in arching effects, installation conditions may also be influencing this. The readings did, however, highlight the level of pressure (stress) change that can occur with a variation of load conditions.
Wyatt Orsmond, NHD (Civ), MSc (Geotech), CEng MIEI, is technical director at RPS Consulting Engineers in Dublin, Ireland, (firstname.lastname@example.org).
British Standards Institution 1995. BS8006: 1995. Code of Practice of Strengthened/Reinforced Soils and Other Fills, London, England.