Blast densification offers a solution to the problem of deep compaction of loose soils. Brian Wilson reports from Canada.
Loose soils have always presented a challenge to the geotechnical profession, particularly in regions of seismic activity where, in addition to settlement, liquefaction is a major concern.
A range of techniques has been developed to densify, cement or compact loose soils, but achieving this at depth is a heavy physical and financial burden.
A proposed tailings dam raise in Ontario, Canada complicated this situation with a highly variable silt content in hydraulically placed loose tailings.The solution was a combination of techniques, centred around blast densification.
The contract of just under C$5M (£2.24M) was carried out by ground engineering and environmental firm Gaia Contractors, part of Golder Associates, with a team of subcontractors including Geodensification, Rapid Impact Compactors, Nilex, ExRT, Rush Contracting, Orica, and Day Construction.
As part of the nickel mining process in Sudbury, Ontario, the crushed rock slurry (tailings) generated each day is discharged into large impoundments where the solids settle and water is recycled for use in metal extraction.
The tailings contain a significant amount of silt. Deposition in the ponds often results in local segregation of the tailings, giving rise to zones of slimes (with a very high silt content) as well as zones of very silty sands or sandy silts - soils previously considered impractical to densify.
After nearly a century of accumulation and numerous upgrades, the facility in Sudbury was approaching capacity.There was no more space, so it was decided to expand the impoundment facility by building new dams founded on the tailings.
There were two complicating factors. First, a portion of the perimeter dam would be founded on sand and silt tailings that had been hydraulically placed (and so were loose), and second, the latest design standards required the dam be designed to resist a magnitude 7 interplate earthquake, which could result in liquefaction of its foundation soils.
This could result in dam foundation failure, or in excessive movement, possibly causing a breach.
The solution involved densification of a zone of the foundation soils. This 'shear key' would increase the resistance to lateral movement during a large seismic event. The proposed treatment area was 3km along the dam axis and 30m wide.
Gaia was awarded the contract in March 2004 based on an approach combining three complementary ground improvement methods: blast densification, dynamic compaction, and surface treatment with a rapid impact compactor.
Dynamic compaction is a well-known and relatively inexpensive method of ground improvement, but, even with the weights and equipment readily available in Ontario, was likely only to be effective to depths of 8m to 10m in the tailings.
One section of the site was targeted for this technique because of the shallow depth of treatment required. Blast densification, which is effective to much greater depths, was reserved for the deepest sections of the dam.
The technique involves placing explosive charges at various levels or 'decks' The charges are contained in plastic tubes drilled into the ground in a grid pattern.
The depth to each deck depends on the total depth of the blast tube, the size of explosive charge and the desired vertical spacing for the soil type. Blast densification excels at treating sites with deep, loose deposits, but is not effective near the surface due to the lack of confining stress and, often, the depth to the water table.
The Rapid Impact Compaction (RIC) system, developed by the British Royal Engineers for rapid repair of bombed runways, was considered an ideal complement to the process because of its efficiency at compacting nearsurface deposits.
Gaia moved on site in May 2004, before main works began, to carry out ground profiling using cone penetration testing to supplement the tender data.
This indicated the soil was as variable horizontally along the dam as it was vertically, with sandier and siltier soils interbedded within the densification zone. The tailings in the area of the proposed dam foundation ranged from silty sands to almost pure slimes (80% to 90% silt).
The sandy tailings had a silt-sized and finer fraction generally in excess of 35%, while the silty zones contained more than 60% silt-sized (and finer) particles. Although guidelines for liquefaction assessment suggest any soil with more than 35% fines is resistant to liquefaction, tailings have been found to be an anomaly, so the soils within the foundation zone were assessed to be liquefiable even at silt contents of 60% to 70%.
Key issues for successful blast densification are the blast hole spacing, the charge weights and the timing and sequencing of detonations.
Augers pre-drilled holes through the dried dense crust, with plastic casing for the explosives installed in a single stroke using a modified rig.
The RIC was used over shallow bedrock, and also to follow through immediately after the explosive compaction, densifying the surface soils and, while pore pressures were still elevated, providing additional energy input into the soils at depth. Dynamic compaction was used in areas of the site where the depth of tailings were 4m to 9m.
There have been enough blast densification projects worldwide to allow first estimates of the likely requirements to be developed from past experience. Based on these, the method is optimised to actual soils at the site.
The goals of treatment are to use delays to extend the duration of explosive-induced shaking to at least several seconds, and with enough charges detonated simultaneously to get a resonant response in the site. This simulates a locally intense earthquake.
For the Sudbury project, fine tuning of the blast densification protocol was completed over 10 trial blasts carried out during the first three weeks. Detailed blast designs were developed by Mike Jefferies of Golder Associates, based in Vancouver.
The final procedure resulted in a nominal 8m spacing of the blast holes, with the duration of a standard pattern detonation averaging about nine seconds. This succeeded in inducing strong vibration of the site, continuing for as much as three seconds after the last detonation.
As an example, one treatment panel (for primary and secondary blasting) comprised 193 holes, with about 3.4km of pipe installed. The holes were detonated row by row, 'walking' the induced disturbance across the panel. About 7.4t of explosives was used to treat 65,650m3 of loose tailings.
Blast densification does not cause immediate settlement, as the induced ground response is the creation of large excess pore pressures through liquefaction.
Settlements arise as the excess pore pressures dissipate, with often spectacular quantities of water forced out of the ground. At Sudbury, a drainage ditch was used to remove the water, which flowed for several hours.
Given the gradation of the treated material, the times for decay of excess pore pressures were markedly longer than encountered in the types of sands usually densified with this or more conventional techniques.Measured excess pore pressure decay times exceeded 15 hours in most instances.
Typically, three to four days elapsed between primary and secondary blasts in a panel, with the RIC unit following shortly after the secondary shot to provide additional energy. The machine made a second pass a few days later to compact the loosened near-surface material.
Specification in sandy areas was easily met, despite the fines content. In the higher silt and slime zones, CPT resistances before blasting were much lower. However, blast densification still managed to double or triple the pre-compaction CPT resistances, and actually achieved the specification in many areas of the site.
Interestingly, the induced settlements were relatively large for all the tailings and corresponded to a vertical strain of 10% to 13%. This is twice the strain normally expected for blast densification of even very loose soils.
Once the blast densification procedures were established, production blasting work proceeded quickly. Full production work got under way in early June and the ground improvement was substantially completed by early September, four weeks ahead of schedule.
Brian Wilson is British Columbia regional leader of Gaia, based in Vancouver.