THE RAILWAY LINE through the city of Delft dates back to 1847. It ran along the city’s western border. Since then, the city has grown substantially to the west, making the railway line a great barrier through the centre.

Because of increasing rail traffic, a two-track rail fly-over was built in the 1960’s. With its construction, an historic canal, once bordering the city, disappeared. Since the 1960’s rail traffic has further increased. Some 350 trains pass through Delft every day, and in the rush hour this reaches one train every 3 minutes. This has created significant nuisance from noise and vibrations. Legal noise standards were exceeded for more than 700 buildings. The capacity of the two-track line is almost fully occupied, and the future growth of rail traffic demands expansion to four tracks.

PLAN DEVELOPMENT

After various feasibility and environmental impact studies by Railinfrasolutions and Benthem Crouwel Architects the decision was made to build a 2.3km long four-track tunnel next to the rail viaduct. This tunnel enables the development of 800 houses and 40.000m2 for offices. Upon completion, the tunnel structure will be sufficient for four tracks. In the first stage only two rail tracks are to be built. In the future, the expansion to four tracks will occur without further disturbance to the city. A new underground station has been situated next to the old station building, and contains a parking facility for 7,000 bicycles. Several underground car parks further add to the quality of the new developments.

Part of this project is a two-story parking for 450 cars, located underneath the re-constructed city canal. The total cost of the plan is approximately EUR 1bn (USD 1.13bn).

The project is divided in several contracts. The largest contract contains the tunnel structure, the tunnel installations, the underground station, an underground car park and the reconstruction of the public spaces. This project was tendered as a design and construct contract. In order to give the contractors the possibility to optimise, all specifications, apart from aesthetic specifications and the development plan, were functionally described. The client, ProRail (Netherlands Railway authority) awarded this contract to Combination Cromme Lijn, a JV of CFE, Mobilis/TBI and Dura Vermeer. These works will be completed in 2017. Full four-track operation requires expansion of adjacent tracks and is expected in 2023.

Underground station and city hall

The new underground station is catered for the transport of 39.000 passengers each day. It comprises two platforms, located at 8 m below surface and 340m long. A lot of effort is made to create a pleasant, secure environment for the passengers. Optimal transparency is created by avoiding columns at the platforms, making the roof span 2x20m. With its 9m height, the underground station is very spacious. Natural daylight accentuates the central staircases. A special artificial light design was made to further improve the traveler’s orientation. From the platforms, stairs lead to an intermediate level, the mezzanine. From here, passengers can either continue to the bicycle parking at the same level or go further up, to the station hall. The hall has no columns, spans over 40 m and is integrated with the new city council building. The six-story building is founded on the tunnel walls.

safety concept

The tunnel consists of four separated tubes, one for each rail track. In an emergency passengers flee to an adjacent `safe` tube. This tube is reached through emergency exit doors that are present every 75m. Longitudinally placed ventilators create an air overpressure in the adjacent tubes, preventing smoke from entering through the rescue doors. The rail traffic is stopped automatically. Passengers flee through the safe tunnel to the portal or to the station, which is regarded as a safe haven.

Since the station has no separating walls, a different concept in chosen to create a safe and smoke-free area. In case of a fire incident in the station, a smoke and heat discharge system, positioned at the top of the station walls, will create a smoke free situation for at least five minutes, which allows passengers to flee to one of the five regular or emergency exits. Glass smoke screens (downstands) protect the staircases from smoke entry. Several simulations with a three-dimensional Computational Fluid Dynamics (CFD) model are performed to check the proper functioning of smoke and heat discharge system.

Pressure waves and draught

About 10 per cent of the trains will not stop at the station, but pass the tunnel and platforms with a speed of 140km/h. The resulting pressure waves are a possible nuisance to train passengers. Also, draughts at the platforms and the stairs to the station hall should be limited. Both aspects were studied through numerical modeling and scale tests. From these studies, it followed that several measures are necessary to create an acceptable wind climate. These include air release points in the tunnel, wind guiding structures at the platforms and airtight revolving doors in the station hall. Now the train services have started, the measures prove to be effective for limiting pressure waves and wind speeds.

However, station doors are not able to withstand pressures in extreme situations. Currently measures are taken to strengthen these doors.

train induced vibrations and low frequency noise

As the tunnel ameliorates the environment with respect to airborne noise significantly, vibrations and low frequency (ground borne) noise are known to potentially cause a larger environmental impact. Extensive numerical and analytical analyses have therefore been performed to assess the noise and vibration impact of the new tunnel. The different components of a noise and vibration analyses, namely the source, the transmission paths, and the different receiver characteristics have been identified by several measurement campaigns and numerical studies.

The vibration source of the trains was back calculated from measurements at a nearby tunnel with a comparable structure and geological conditions. The transmission paths from tunnel floor to adjacent building (receivers) have been calculated by numerical models and compared with the results of several impact tests. A prediction of vibration and low frequency noise levels has been made and several types of mitigation measure have been studied. Based on the results of the different analyses, a ballast mat has been installed in the tunnel. This was expected to fulfill the required reduction of both vibrations and low frequency noise.

The different measurements and validation tests showed that the prediction of vibration and low frequency noise levels included different uncertainties. This resulted in a statistical approach of the prediction results for residences on both sides of the tunnel separately, using a lognormal distribution. In this way expected values as well as variances have been identified. After completion of the tunnel both vibration and low frequency noise measurements have been carried out in several residences. Preliminary results show good agreement with the predicted levels.

Geological and Geohydrological Conditions

The geological layers encountered in the Delft region are typical for the river delta in the western part of the Netherlands. The top layer is a rather thin layer of sand and recent debris. Under this top layer, the geological profile consists of an accumulation of almost 20m of soft soil of Holocene origin underlain by medium to dense sands of Pleistocene origin. The thickness of the Pleistocene sand layer is generally more than 15 to 20m (thus to a depth of approximately 35 to 40m), however at a few spots along the project, the underlying overconsolidated clay layer (Kedichem strata) is already found at a depth of 30 to 35m.

Three different water tables were detected within these soil strata. The phreatic water level is at about -0.4m NAP (NAP is the national reference water level in Amsterdam). In the deeper Holocene layers a hydraulic head of approximately -3.75m NAP was measured, in the Pleistocene sands this varies from -6 to -9m NAP.

This particular geohydrological situation is for the greater part caused by a factory at the northern end of the tunnel, which extracts large amounts of brackish groundwater from the Pleistocene sands. The extraction results in very low water tables in the Pleistocene sands thus forming an advantage during construction as it reduces the risk of uplift of the bottom of the excavation. As it is uncertain whether this situation will remain during the 100-year tunnel lifetime, the final structure is designed for both circumstances, with and without the present ground water extraction. Simulations with a regional geohydrological model predict that, without extraction, the head in the Pleistocene aquifer will rise to the level of approximately -2m NAP. As the very heterogeneous Holocene formation is a sequence of peat, sandy clay and clayey sand layers, a great number of piezometers are installed to measure the water pressures.

construction methods

For most tunnel sections, diaphragm walls form the retaining walls of the building pits. As the tunnel is often situated at a mere 3m from the surrounding historical buildings, this method was selected since it causes minimum noise and vibration disturbance. Moreover, as these diaphragm walls have a higher bending stiffness, the deformations due to excavation of the building pits are strongly reduced compared to traditional sheet pile retaining walls. Naturally, the diaphragm walls also serve as walls of the final structure, otherwise they would not be economically attractive.

The tunnel is constructed in two stages. As soon as the eastern tube could be carry train traffic, the second stage, starting with the removal of the old railway fly-over, was executed. Once the railway viaduct was removed, the western tube and an underground parking are realized. Due to the scarcity of the public space where roads, tramways and pedestrian roads should continuously be assured, a top-down building method was chosen. In this way, the roof could be used for storage of construction material and equipment.

The building sequences throughout the whole process strongly depend of the robustness of the nearby historical structures. The optimum width of the diaphragm walls is 7.3m. In such panel widths, two reinforcement cages with a maximum width of 3.1m each (necessitating transport by lorry in an urban environment) can be installed, resulting in a better ratio of reinforced width to total width of 85 per cent compared to the 81 per cent for 3.8m panels.

By doing this, the amount of panel joints, and therefore the risk of future leakage is reduced by half. However, during the stage when the excavation of the diaphragm wall is supported by bentonite mud, wide panels produce larger ground deformations. For particularly close passes to buildings, a maximum panel width of 3.8m was chosen.

At critical sections, the first excavation down to bottom level of the (future) tunnel roof could not be done without the installation of supplementary struts above the tunnel roof. Once the roof concrete is poured and hardened, excavation under the roof could take place to an intermediate level where struts, formed by steel tubes, were installed. Finally, excavation to the final level and the subsequent pouring of the tunnel floor were realised. During all excavation works, the water table inside the building pit was lowered to 0.5m below excavation level in order to reduce deformations of the retaining wall and nearby structures. Once the eastern tube was ready for trains, a similar building sequence could take place to realise the western tube and the underground parking.

The underground parking is executed simultaneously with the western tube. Therefore, the diaphragm wall between the western tube and parking space was replaced by an alternation of previously installed barrettes and afterwards poured in situ walls. These in situ walls only reach from floor to roof level, thus reducing the amount of concrete works. Moreover, as the underground parking was to be built in a bottom-up sequence (due to the greater width between the diaphragm walls, which would cause thick roof and intermediate floors), the excavation of the western tube could be done through the left open spaces between the yet installed barrettes. Leakage problems with diaphragm walls are mostly related to discontinuities in the concrete works, often related to a too dense reinforcement grid, especially in the coupling area between the diaphragm wall and connected horizontal slabs. Therefore, during execution of the eastern tube, a redesign of the western tube and parking lot was performed introducing hinged connections between slabs and diaphragm walls instead of the previously implemented rigid connections.

This resulted in much better rheological circumstances for the bentonite concrete exchange process of the diaphragm wall execution. After a thorough check whether this change would not affect the deformations of the adjacent buildings, this redesign was successfully implemented.

In the underground station area, concrete columns are foreseen at the middle of the cross section. Due to the large vertical forces, resulting from the office building on top of the station, foundations by means of diaphragm wall barrettes (points bearing in Pleistocene sand layers) are designed. As the demolition of the barrettes between floor and roof level of the station is very time and money consuming, the barrettes are filled with gravel over the height between these two horizontal slabs. To ensure the stability of the roof during excavation, steel H-beams are installed in the reinforcement cages of the barrettes. Once the total excavation was done, final concrete columns could be poured around the H-beams.

Settlement risk assessment and building damage control

At the northern part on the project area, the distance between the tunnel walls and adjacent buildings varies from 3 to 10m. The allowable deformations of the contiguities are small. The buildings are divided into four different quality classes (indicated by their colour) on the basis of the actual condition and the related allowable additional deformation. The criteria are set as per the graph in fig. 6, according to the Limiting Tensile Stress Method (Boscardin), where the angular distortion and the horizontal strain have to be under or to the left of the boundary line of the respective building class.

These parameters are determined by FEM-calculations. The model includes the tunnel and surrounding soil. The building is only modelled through its weight. No interaction between building and soil is taken into account: the assumption is that the building will completely follow the soil deformations. This is a conservative assumption.

The 2D-FEM calculation includes all stages of construction, excluding the construction of the D-wall. For this, a separate 3D analysis is performed. Calculations were carried out both for average and lower boundary values of soil and structural parameters. When using lower boundary values, more measures like an extra strut level above the tunnel roof and the introduction of pre-stress forces in the struts were required to fulfil the requirements. Therefore, the tunnel construction was prepared for these measures. The validation of these measures happened on basis of observation and analyses of the monitoring data during all construction stages. The angular distortion and horizontal strain were translated to front wall movements. These movements were the key day to day monitoring parameters. Robotic total stations continuously measured prisms, attached to the facades, in all phases.

For each stage of construction, limits were set. The maximum displacements depend on the classification and the exact location of the building in the settlement trough. Typically, limits are set to about 6-18 mm both vertically and horizontally. Apart from building measurements, also water pressures and soil movements were measured. Water pressures were measured in all three aquifers, horizontal soil movements were measured through inclinometers behind the D-walls. It can be concluded that, for the buildings adjacent to the top-down diaphragm wall tunnel, the measured vertical deformations mounted up to 60 per cent of the predicted values while the horizontal deformations were very often up to 100 per cent of the predicted values. However, for the buildings close to tunnel sections built inside temporary cofferdams comprising sheet pile walls two cases of significant excess of predicted values were encountered.

In one case a moderate settlement of 30mm led to some unforeseen damage and the observational method was successfully implemented as best way out strategy, while in the other case a significant deformation of 80mm was absorbed without any damage due to the rigid behavior of the building. Each execution stage was evaluated on beforehand on basis of a LTSM level III model (FEM including masonry and soil).

The rose windmill

A major challenge was the crossing of the windmill "the Rose". This monument dates from 1679 and is right in the middle of the tunnel trace and therefore a huge obstacle for tunnel construction. It is a highly sensitive building. Due to several reconstructions over the centuries and the poor soil conditions the building has tilted fivedegrees.

The presence of attached living quarters further adds to the complexity. Because of the monumental status, it was demanded to leave the entire structure intact. Below the mill and attached living quarters a reinforced concrete foundation slab was cast and after hardening of this slab the subjacent existing foundation was separated from the structure above, which – in the mean time – was taken over by a previously installed temporary pile foundation. Then, the monument was lifted over a height of 1m to make space for the construction of the tunnel underneath. Upon completion of the tunnel, the entire structure of the windmill and its attachments lowered back upon the tunnel roof while assuring a perfect load exchange without the slightest deformation of the foundation slab