Tunnel vision

Design software is indispensable in predicting the effects of digging a train tunnel directly beneath the heart of old Amsterdam. Jon Excell reports.

City planners in Amsterdam have taken the plunge and decided to ease burgeoning congestion problems with the construction of a new underground rail line. Due to be completed in 2011, it’s estimated that the north/south Metroline will ultimately transport around 200,000 passengers a day.

Six and a half miles of this tunnel will stretch beneath the narrow streets, tram lines, bridges, canals and historic buildings of the city’s picturesque centre.

While building an underground line is always a challenge, Amsterdam’s soft subsoil, the precarious wooden foundations of its oldest buildings and a desire to avoid the mistakes of the past have combined to create one of the biggest tunnelling challenges of recent times.

To accurately predict the consequences of creating what is effectively a small fault line through the middle of the city, the planners have called on the services of computer modelling and simulation tools more usually associated with product design.

Chief among these is Diana, a finite element analysis (FEA) package from single-software company TNO Diana BV. Specially tailored for civil applications, Diana has been used by Witteveen & Bos, the Dutch engineering consultant in charge of the project, to predict soil-structure interaction and provide information on how buildings are likely to respond to the tunnelling.

The software has also helped contribute to the suggested design of the tunnelling equipment.

Many of the factors the team has had to consider in the planning of thiscomplicated job are interdependent. For instance, the law requires tunnels to avoid passing beneath buildings by following, as far as possible, the present street pattern.

However, Amsterdam’s streets are narrow, meaning that not only must the tunnels be closely spaced, but along one section they must also actually be ‘stacked’ vertically.

Another important consideration is the presence of groundwater. From about 1m below the surface, the tunnel walls will be subjected to water pressures of around 3.5bar. While this clearly makes tunnelling even more difficult, it’s vitally important that the groundwater is undisturbed. Not only would disturbance cause serious ground deformation (or settlement), it would also affect the wooden stilts (or piles) that support many of the city’s older buildings.

Although the wet ground has caused many of these supports to start rotting, it’s thought that exposure to the elements by de-watering would cause them to decay completely.

Because of the risks involved in this project, the Witteveen & Bos team, led by Frank Kaalberg, also decided to validate its 3D models by carrying out a full-scale pile trial at the site of the the second Heinenoord tunnel near Rotterdam.

It soon became apparent that existing tunnel boring machine (TBM) technology was simply not up to the job. In the soft soil of Amsterdam it would lead to excessive surface deformations and building damage.

It was clear that the team would have to focus on improving the quality of the whole TBM process.

To understand the reason for this surface deformation one has to look at the principle of volume loss.

This is a familiar term in the world of tunnelling, and is essentially an expression of over-excavation. As the tunnelling machine cuts though the earth, the exposed tunnel face relaxes and begins to collapse inwards. If this isn’t prevented from happening the machine ends up removing more earth than it needs to.

Thus, if you excavate three per cent more earth than you actually need to (a three per cent volume loss), the ground around the tunnel has to reconsolidate, producing a settlement effect that carries right up to the surface. Using FEA, volume loss was used by the team to model the way relaxation translates into surface settlement.

Clearly the ultimate in tunnelling is to dig tunnels that have zero effect on the ground around them. For example, if you were to take out 100 per cent of the excavated volume, and then immediately secure the tunnel face, there would be no settlement. But as soon as the tunnel face is allowed to relax, settlement is inevitable.

Most current TBM technology produces average volume losses of between one and three per cent. Compared with the dramatic collapses (or ‘huge volume losses’) encountered by tunnellers of yesteryear, these figures are negligible; but Amsterdam’s architecture demands even greater precision than this, and Kaalberg’s team used FEA in the development of a TBM design that he claimed will keep volume loss below 0.5 per cent.

There are essentially two tunnelling machine options that would be suitable for a project of this kind – a slurry machine or an earth pressure balance unit (EPB). Slurry machines generally pump water into the tunnel face, mix this water with spoil, then pump the resulting slurry down the side of the machine. Pressurised slurry supplied to the front of the bulkhead keeps an excavation face stable.

The EPB method works by using the excavated material to balance the external pressure of the ground. As the machine advances, excavated earth ismixed with a foam material and maintained under a controlled pressure in the cutterhead.

In theory, both methods should allow only the removal of a precise amount of material and reduce the potential for settlement; in practice, each technique has its own advantages depending on the soil conditions.

Thus, the system design proposed by Witteveen & Bos relies on what Kaalberg calls a ‘variable shield system’ – a tunnelling machine that is able to alternate between slurry and EPB modes depending on the condition of the soil around it. While a change of mode involving separate machines would usually take several weeks, Kaalberg said that he thinks the new design will make it possible to effect the switchover in a few days.

Although the contract for the tunnelling machines has not yet been awarded, Kaalberg said it is highly likely that it will go to German expert Herrenknecht which, he said, has the best appreciation of the problem. Tunnelling is due to begin at the end of 2005.