Funchal airport occupies just about the only substantial area of flat land on the Portuguese-owned volcanic island of Madeira, some 700km off the north west African coast. The airport sits on a raised plateau, nearly 60m above the adjacent sea. The existing 1,800m long runway stretches right across this plateau with the ground dropping away alarmingly at either end. By modern standards the runway is short, which restricts the size of aeroplanes that provide the main link with Portugal and the rest of Europe.
Construction is at last taking place for the runway extension which has been planned ever since a plane failed to stop in the 1970s.
In the absence of anywhere else to put it, the new 900m section of prestressed concrete runway is being formed on huge concrete piers, constructed on a strip of reclaimed land running parallel to the coastline. Work is being carried out in two phases, with piling operations presently drawing to a close.
The deck for the runway is supported on a series of 3m diameter columns, each designed to carry a load of 80,000kN. Variability in the underlying beach deposits and volcanic rocks has created an interesting foundation problem.
Conditions vary from a compact basalt (which is stronger than concrete) to pyroclastic ash materials and lavas. The lavas essentially consist of alternating layers of basalt and scoria, material formed on the upper and lower surfaces of the flowing lava, which can be particularly voided and so is significantly weaker and more compressible than the basalt.
Establishing a decent conceptual ground model was an important part of the design process. This meant inspecting available rock exposures on the island, to get a feel for 'how the rocks are arranged in the big picture' and to develop an appreciation of the volcanic environment when the rocks were formed.
The real difficulty comes not from the range of materials encountered, but from the uncertainty of their variation, and that very weak materials may underlie extremely competent rock.
'In the lava sequences the most competent material can quickly change from several metres to less than a metre. A further hazard is the possible presence of lava tunnels, formed when the outsides of a lava flow cools allowing still molten interior to flow out create a tunnel,' explains Peter Rutty of consultant Mott MacDonald. Additionally a relatively highly compressible ancient palaeosol layer has been found beneath the pyroclastics in places.
Because of these variations and the high level of uncertainty, column foundations are generally piled, except where strong basalt layers at least 2m thick are identified above sea level. In these cases columns will be supported on 12.5m diameter concrete foundation pads.
Elsewhere each column is supported by a group of eight, 1.5m diameter piles, which are up to 50m in length. Piles are designed to carry loads predominantly by skin friction, with only a very small contribution from end bearing (available end bearing is assumed to be that provided by the clinker- like scoria - the weakest material identified at site).
According to Rutty, the pile design philosophy was 'because we could never be sure what was below the pile, we had to be confident that we had sufficient skin friction'.
Given the paucity of published data on the performance of piles in pyroclastic materials, Mott undertook five pile tests on piles instrumented with strain gauges in an attempt to identify the skin friction in the various rock types. Unexpectedly much of the load was carried in the reclaimed fill material, which was simply end tipped over beach deposits to form a working platform.
At the test loads of up to one and a half times working load, there with relatively little mobilisation of shaft friction within the pyroclastic rocks, making it difficult to extract economic design parameters with confidence.
'It has been suggested that the use of permanent casing through the fill and beach deposits overlying the underlying rock could have improved the pile testing' comments Rutty.
In the end, Mott's design assumed skin friction based on a published correlation with unconfined compressive strength. 'The design method was undoubtedly conservative as there was only negligible recorded settlement. However given the lack of previous case history examples, the known geological hazards, and the significant consequences of failure it was important that all parties had confidence in the eventual performance of the foundations,' he concludes.