From Bridges across Great Belt and Øresund towards a Femern Belt Bridge

Niels J GIMSING Profesor Emeritus Techn. Univ. of Denmark DK-2800 Lyngby, Denmark njg@byg.dtu.dk

Professor Emeritus at the Department of Civil Engineering. Participated in the design of the Great Belt Bridge and the Øresund Bridge, and acted as specialist design adviser on numerous bridges around the World . Co-owner of Gimsing & Madsen, Consulting Engineers, Denmark.

Summary In Denmark the construction of three major bridges was initiated in the 1990.s: Storebælt Bridge (Great Belt Bridge), Øresund Bridge and Femern Belt Bridge. The first two were completed in 1998 and 2000, respectively, and the third is expected to be constructed during the second decade of the 21st century. In both design and construction procedures a number of innovative features have been introduced. Keywords: Box girder bridge, suspension bridge, cable-stayed bridge, wind tunnel test, composite action. 1. Introduction

Denmark consists of the peninsula Jutland and 406 islands, out of which 79 are inhabited. The second largest island, Funen, is separated from Jutland by the strait Little Belt and from the largest island, Zealand, by the strait Great Belt. In round figures 45% of the population lives in Jutland 10% on Funen and 45% on Zealand were the capital Copenhagen is situated. The substitution of ferry services by bridges across the straits separating the different parts of Denmark began in 1935 with the opening of the bridge across the Little Belt and in 1937 with the opening of the Storstrøm Bridge between Zealand and Falster. At the opening the 3.2 km long Storstrøm Bridge was the longest in Europe. After the completion during the 1930.s of more than ten bridges across a number of the smaller straits in Denmark the country was virtually joined in two infrastructure units separated by the 18 km wide Great Belt.

Fig. 1: Map of Denmark

Plans to build the final link - across the Great Belt - were worked out already in the late 1930.s but the outbreak of World War II and the subsequent post war period with lack of resources made it impossible to consider construction of a Great Belt Bridge for a period of more than twenty years.

In 1978 the Great Belt Bridge was close to reach the construction stage based on a design with either a 780 m cable stayed-bridge main span or a design with a 1416 m suspension bridge main span across the international navigation route between the Baltic Sea and the North Sea. The 1978 design never reached the construction stage as the government decided to give priority to a natural gas distribution net throughout the country.

Finally in 1987 the Danish Parliament approved the construction of a fixed traffic link across the Great Belt but now in a different configuration from that of 1978. In stead of carrying both the road and the railway across the East Channel on a bridge it was decided to let the railway cross in a bored tunnel and only carry the motorway across on a bridge. Across the West Channel both road and railway should be on a bridge (as assumed in 1978) but in a different alignment. 2. Great Belt Bridge The two bridges forming a part of the Great Belt crossing is the 6.6 km long West Bridge between Funen and Sprogø and the 6.8 km long East Bridge between Sprogø and Zealand. 2.1 West Bridge The West Bridge crosses domestic Danish waters without international restrictions regarding navigational requirements. The West Bridge could therefore be constructed as a low level bridge with a maximum vertical clearance of 18 m just to allow passage of coasters, fishing boats and leisure vessels [1].

The most unique feature of the West Bridge is the extensive use of precast elements for both the substructure and the superstructure. The elements were cast on a work site area close to the bridge abutment on Funen. Five production lines were used to cast the caissons and the pier shafts as well as the box girders for the road and the rail superstructures. The huge precast elements with a weight of up to 6000 tonnes (and slightly more) were moved onto a load-out pier where they could be picked up by the purpose built floating crane “Svanen” with a lifting

Fig. 2: Design from 1978 for a suspension bridge with a 1416 m main span for both road and railway

Fig. 3: Casting yard on Funen

capacity of 6000 tonnes. As some of the largest caissons had a weight in excess of the lifting capacity they had to have their lower part submerged during transportation to utilize buoyancy.

The superstructure of the West Bridge is divided into two parallel bridges with individual girders for the road and the railway, whereas the caissons in the substructure are common for both bridges. The precast box girder elements have the full length of the span (110 m) reaching from midspsn to midspan. That allowed the floating crane to pick up the girder elements at the center as the post-tensioned reinforcement both in the construction phase and in the final stage would have the right position at the top of the girders above the supports. From erection of the first caisson to installatiom of

the last box girder elements a period of only 25 months elapsed. 2.2 East Bridge The East Bridge crosses the international navigation route from the Baltic Sea to the North Sea. Therefore, the bridge had to be designed so that the largest existing ships could pass safely through the large main span across the navigation channel [2]. In 1978 it was believed that a 780 m span would be adequate to allow safe passage of the largest ships provided that it could be assured that the very large ships could be guided through the main span without encountering other ships in the immediate vicinity of the bridge. With a 1416 m span it was believed that ships could pass the main span without restrictions. Prior to tendering the East Bridge in 1990 a number of navigation studies were performed to ensure that the bridge would not affect the safe passage through the Great Belt of even the largest ships. In the meantime it also became clear that it would not be acceptable to impose one way passage of large ships through the main span. The final choice of the main span length of the East Bridge was based on manoeuvring simulations carried out at the Danish Maritime Institute. Here the navigational safety was studied for main spans of 900 m, 1200 m, 1400 m and 1600 m by having experienced pilots and ship captains to manoeuvre computer generated ships through the shipping lane under severe weather conditions. The result of the navigational studies was that only with a main span of 1600 m (and a vertical clearance of 65 m) could all the involved shipmasters agree that the bridge would have no decremental effect on safe passage. 2.2.1 Approach spans On either side of the 2700 m long suspension bridge approach bridges with lengths of 1600 m and 2500 m, respectively, have been constructed as continuous box girders in steel.

Fig. 4: The floating crane ”Svanen” with a precast element for the roadway superstructure

As for the West Bridge full span prefabricated elements were erected for the approach spans of the East Bridge. However, due to application of steel the weight of a 193 m long girder element was only 2500 tonnes in stead of the 5800 tonnes weight of the 110 m long precast elements of the West Bridge.

The steel girders for the East Bridge approach spans were fabricated in Italy and assembled into 193 m long erection units on a work site area at Aalborg in Denmark. From here they were transported on a barge to the bridge site and lifted onto the piers by two cranes positioned at either end of the erection unit. The extreme end of the erection unit was

initially supported 4 m above the final level of the bearing and in this position the continuity at the other end was completed by site welding. Subsequently, the girder was lowered onto the final bearing and by this procedure it was ensured that continuity would also be achieved for the dead load of the steel girders. During the detailed design of the East Bridge it was discovered that vortex excited oscillations of the slender approach span girders with a depth-to-span ratio around 1/25 could be experienced even for cross winds with a velocity of less than 25 m/sec. These vibrations would not endanger the structural integrity of the bridge but could result in discomfort by the users of the bridge. It was, therefore, decided to install large tuned mass dampers to eliminate all oscillations for wind speeds below 25 m/sec. For wind speeds above 25 m/sec the bridge would in any case be closed to all traffic so user comfort did not have to be considered. Besides the tuned mass dampers a number of dehumidification plants are also installed inside the box girders to exclude corrosion and make interior painting superfluous. This system was for the fist time used in the box girder of the Little Belt suspension bridge from 1970 and has since then been used intensively around the world as a most efficient way to exclude corrosion inside major box girders. 2.2.2 Anchor Blocks

The anchor blocks of the East Bridge are located far from the coasts at open sea and they are therefore very visible compared to anchor blocks on land where they can be partly hidden behind vegetation and buildings. It was, therefore, important to arrive at a less massive appearance than found in many existing anchor blocks. The anchor blocks are founded at a water depth of

approximately 10 m on 122 m long and 55 m wide precast concrete caissons. The upper visible part of the anchor blocks are cast in situ.

Fig, 5: Erection of a 193 m long box girder element for the East Bridge approach spans

Fig. 6: Anchor block of the East Bridge

2.2.3 Pylons

The two pylons reach a height of 254 m making them the highest concrete pylons in the world at the time of construction. The pylons are designed with a very simple and clean appearance characterised by just two cross beams between the quasi-vertical legs. In contrast to other large suspension bridges there is no cross beam immediately below the deck but half way to the top. In situ casting was chosen as slipforming was abandoned by the client and prefabrication of segments was unattractive due to weight problems. The batching plants were located on a work site area on Zealand and transported to the pylons by truck mixers on ferry boats. After arrival at the pier the mixed concrete was pumped to the top of the pylon and cast into the jump form. In each pylon the total quantity of concrete amounted to nearly 40,000 m3.

2.2.4 Cables Each main cable contains 18.648 galvanised wires with a minimum tensile stress of 1570 MPa and a diameter of 5.38 mm. Both the air-spinning method and the PPWS method was allowed by the client during bidding but only the air-spinning method proved to be competitive [3]. 2.2.5 Deck The deck of the suspension bridge consists of a slender box girder in steel with a depth of 4.3 m and a width of 31 m. It was fabricated in Italy in 48 m long segments and transported on a barge to Denmark.

With a span approximately 15% longer than that of the Humber Bridge there were special concerns about aerodynamic stability as past experience had indicated that the critical wind speeds for bridges with a mono-box deck would diminish as the span was increased. Initially the wind tunnel tests were performed on section models with different cross sectional layouts to determine a suitable configuration. The section model tests indicated that it would be possible to achieve the required critical flutter wind

speed of more than 60 m/sec with the mono-box but to give a further verification of the adequacy it was decided to test a full aeroelastic model in a scale of 1:200. As the continuous box forming the

Fig. 7: Pylon under construction

Fig. 8: Full bridge model in wind tunnel

deck of the suspension bridge is 2700 m long it was necessary to build a special 14 m wide wind tunnel at the Danish Maritime Institute. The result of the extensive wind tunnel tests was a critical flutter wind speed of 72 m/sec, comfortably higher than the limit of 60 m/sec. However, this was for a model with an open railing, but the client also wanted to know the influence of adding wind screens to improve the user comfort.

Adding wind screens proved to have a decisive effect as it reduced the critical flutter wind speed to 62 m/sec. In contrast to the majority of existing suspension bridges the deck of the Great Belt suspension bridge is continuous from one anchor block to the other - and with only lateral support at the pylons. The omission of a pylon cross beam below the deck and the continuity of the steel box clearly exhibits the fact that the vertical load of the deck is transferred to the main cables through the hanger cables.

3. Øresund Bridge The bridge constituting the eastern part of the Øresund Link has a total length of 7845 m and it consists of three main sections: the western approach bridge with a length of 3014 m; the main bridge (over the navigation channel) with a length of 1092 m; and the eastern approach bridge with a length of 3739 m.

Fig. 10: The Øresund Bridge

. Across the navigation channel the main span has a length of 490 m and a vertical clearance (air draught) of 56 m. The bridge is the longest cable-stayed bridge in the world with both road and railway traffic, and it is also one of the largest composite structures ever built [4]. 3.1 Approach bridges The two approach bridges have a combined length of 6753 m which constitutes approximately 85% of the total length of the bridge. It was, therefore, evident from the initial design phase that the emphasis should be laid on arriving at an optimum solution for these spans.

Fig. 9: Deck at pylon

A double deck configuration offered the advantage of separating the train and automobile traffic and at the same time the required stiffness would be achieved through the large structural depth imposed by the clearance requirements above the railway tracks. The double deck arrangement naturally leads to a structure with steel trusses between the two decks which in turn can be incorporated as chords to achieve maximum structural efficiency. With the train load constituting the major part of the live load it was important to position the two vertical main trusses just outside the clearance diagram of the railway so that the heavy load on the tracks could be carried as directly as possible through transverse girders to the lower nodes of the main trusses.

Fig. 11: Cross section and part elevation of approach spans

With the transverse position of the trusses determined by the clearance of the railway the wider roadway deck had to be made with overhangs, i.e. the deck slab should be cantilevered on either side of the trusses. This was in fact advantageous for the upper deck as it led to a favourable distribution between positive and negative moments in the slab and thereby made it possible to design the roadway deck as a solid concrete slab supported directly on the top chords of the main trusses. With a span of 12 m between the top chords the concrete slab had to be designed with a varying depth and with transverse prestressing. The design of the trusses was influenced by a number of considerations. The geometrical layout should be clean and simple not only for the overall appearance but also to give a harmonic rhythm when viewed from a passing train, the truss should be maintenance friendly, and the truss should be

so rigid that it could respect the strict requirements regarding deflections under passing trains. The solution was to use a Warren truss composed of unilateral triangles with relatively flat diagonals inclined approximately 45o in stead of the generally accepted value of 60o (a value arrived at when minimizing the quantity of structural

steel and neglecting the favourable cost implications of reducing the number of nodes). All truss members are made as closed boxes with stiffeners and diaphragms inside so that the exterior surfaces are clean and easy to maintain. The interior of all boxes is protected against corrosion by installation of dehumidification plants.

Fig. 12: Approach spans

The most efficient use of the different structural materials would clearly be achieved by establishing a composite action between the concrete slab and the top chords of the steel trusses. However, the composite action should not only cover longitudinal shear but also bending moments between the

top chords and the concrete slab so that a rigid closed frame could be established by the lower cross beams, the diagonals of the truss and the upper concrete deck. This required that the top plate of the chords was equipped with studs in two different sizes, normal studs in the central region of the top plate and long studs along the edges (above the webs of the box shaped chords).

The long studs would be efficiently anchored inside the concrete slab and thereby allow transfer of the moment as a couple consisting of two vertical forces of opposite sign. The short studs on the other hand were only to transfer shear. the assumption

For the design of the studs it was important to investigate the interaction between the long and the short studs to avoid that the long studs were overstressed before the short studs were efficiently activated. In 1989 when a double deck truss bridge was considered for the West Bridge of the Great Belt Link a full scale test was performed at the Department of Structural Engineering, Technical University of Denmark. From this test it could be concluded that it would be safe to base the design on the assumption that total shear was equally distributed over the long and the short studs whereas the moment transfer only affected the long studs. A great effort was made during the conceptual design phase to design the main trusses in such a way that they were not only structurally efficient but also pleasing in appearance. This resulted in gently curved gusset plates in a configuration that was favourable in relation to fatigue – a feature of special importance due to the train loading. Besides this it was also decided to keep the exterior surfaces plane so that variations in plate thickness were to be made by adding thickness inside the box sections. This implied a slight complication in the fabrication as the

interior diaphragms were to be geometrically adjusted as the side plate thickness was varied but this proved to be of minor importance.

Fig. 13. Joint between the top chord and the upper concrete slab

Fig. 14: Top chord with studs

Fig. 15: Test specimen in the structural laboratory at Techn. Univ. of Denmark

In the design of the trusses it was attempted to minimize the number of structural elements to give the structure a clear and logic layout. An important step was to reduce the number of nodes by having a diagonal inclination of approximately 450. The effort resulted in a truss with only 64 members within the span length of 140 m: 14 top chord members, 14 bottom chord members, 28 diagonals and 8 transverse girders. In a traditional truss with stringers, wind bracings and verticals in the main trusses the number of members can easily amount to more than 350. The trusses of the approach spans are made mainly of high strength steels S 460 N with a yield stress of 460 MPa. Only the lower cross beams and the top chord members between the gusset plates are made of S 355 N. The steel structure for the approach bridges totalling approximately 65,000 tons were manufactured in Spain on a large work site area in Cádiz. Here the steel trusses were assembled and painted in elements with the full span length

(generally 140 m). Also in Cádiz the upper roadway slab was cast, and for this purpose the subcontractor had built a huge covered workshop with a roof spanning 150 m so that the steel trusses could be moved in and out sideways. Furthermore, the workshop had large air condition plants installed to reduce the temperature rise during concrete hardening.

From Cádiz the approach span girders were transported in pairs to the work site area in Malmö on large barges each capable of carrying two prefabricated spans each weighing almost 5000 tons,

At the work site area in Malmö the precast concrete trough girders were added at the lower deck before the full span unit was transported to its final location. Every second week two 140 m girders arrived in Malmö and were subsequently erected by “Svanen”. Consequently, the bridge grew in average by 20 m per day throughout the erection period. 3.2 Main bridge

With the main bridge forming a relatively small part (approx. 15%) of the total bridge length it was obvious that a structural solution should be sought where the approach spans (comprising

Fig. 16: Node of the main truss

Fig. 17: Truss at the storage area in Malmö.

Fig. 18: Covered workshop at the work site

Fig. 19: “Svanen” with 140 m truss element

continuous trusses) could continue into the main bridge without a complete change of structural system and materials, and without an abrupt visual transition. Also the strict requirements regarding strength and stiffness imposed by the passage of both heavy freight trains and high speed passenger trains proved to have a strong influence on the design of the main bridge with its 490 m main span – the longest for any cable-stayed bridge carrying both road and railway traffic. All these requirements clearly pointed towards a cable-stayed main span with a girder composed of two vertical steel trusses and an upper concrete deck - as in the approach spans of the bridge. The demand for a high degree of rigidity led to a harp-shaped cable system with relatively steep cables and intermediate support in the side spans.

Fig. 19: Elevation of cable-stayed spans

To facilitate the anchoring of the stay cables at the deck level it was decided to change the truss geometry in the cable supported regions. Thus, the geometry of the approach spans where all diagonals have the same length is modified so that the two diagonals leading to each node have different inclinations and lengths. By this it was achieved that the long diagonals became parallel to the stay cables. With the vertical trusses positioned in the same transverse distance as in the approach spans special structural elements are required to transfer the load to the cable system. Furthermore, the selection of a harp shaped cable system and free-standing pylons above the bridge deck imposed a distance of 31.5 m between the cable planes.

With the chosen lateral position of the vertical cable planes (imposed by the demand to have uniform compression across the pylon cross sections from vertical load) the distance between the cable anchorages at the girder level became 7 m larger than the width of the roadway. The load transfer from the main trusses to the stay cables is, therefore, established by adding triangular brackets (‘outriggers’) in the plane of the long diagonals. At the end of each outrigger two anchor tubes are positioned to give support to the sockets of the double stay cable.

Fig. 20: Adjustment of truss geometry from approach span to main span

Besides the adjusted truss geometry and the addition of outriggers the main span girder also deviates from the approach spans by having a lower railway deck made as a shallow box girder in steel. This was chosen to reduce the dead load and improve the capacity to transmit the large positive moments at midspan. The triangular brackets (‘outriggers’) and the adjusted truss geometry give the main span a quite unique appearance and at the same time it exhibits an honest structure where nothing is done to hide the flow of forces from the two decks to the stay cables and further to the 203.5 m high pylons. The steel structure of the main span girder was fabricated in Sweden at a shipyard in Karlskrona. The steel grade applied was generally S 420 N with a yield stress of 420 MPa. After completion of erection units with a length of 140 m or 120 m the steel structure was transported to the work site area in Malmö where the upper concrete slab was cast.

During erection the girder of the main span was supported on the pylon cross beams, on the permanent piers in the side span and on temporary piers in the main span. The temporary piers were allowed due to the fact that the navigation channel were to be relocated after completion of the main span erection, and consequently no shipping took place through the main span until after completion of the superstructure.

After erecting the superstructure from the transition pier at the end of the side

spans to midspan the cable erection was initiated and subsequently the load was transferred from the temporary piers to the cable system. With the chosen procedure comprising temporary piers the 490 m long main span could be erected as four girder units with the concrete slab already in place as it was cast at the work site area. Only

Fig. 21: The main span girder with outriggers at the work site area

Fig. 22: Deck cross section in the cable-stayed spans

Fig. 23: Partly erected main span on temporary piers

relatively small portions of the deck slab at the joints between the erection units had to be cast on site.

The stay cables of the Øresund Bridge contains approximately 70 seven-wire strands inside a HDPE tube with an external diameter of 250 mm. To counteract rain-wind induced vibrations the surface of the tubes were made with a double helical spiral 2 mm thick. However, this modest measure did not prove sufficient to suppress the vibrations and especially in case of sleet large oscillations of the longest stays were observed. Further measures in the form of dampers at the stay anchorages and on the free length then had to be taken.

4. Femern Belt Bridge Only two years elapsed from the opening of the Great Belt Bridge to the opening of the Øresund Bridge but almost twenty years will pass before the third if the large bridges, Femern Belt Bridge, will be ready to carry the traffic from Sweden, Norway and Eastern Denmark on the most direct route to Central Europe. The distance from Copenhagen to Hamburg will be shortened by 150 km when travelling via Femern in stead of via Great Belt. The final treaty between Denmark and Germany to build a bridge across the Femern Belt was signed on 3 September 2008. It was agreed that the bridge from coast to coast should be designed, financed and supervised by a Danish owned bridge authority whereas the approaches on land should be handled by each of the two countries. The preferred solution was indicated to be a high level bridge with a multi-span cable-stayed bridge across the international navigation channel. However, it was also stated that an immersed tunnel should be further investigated for comparison. 4.1 Feasibility study 1999 A feasibility study of the Femern Link was performed at the end of the 1990.s. It comprised traffic forecast, environmental issues, geotechnical investigations for a number of technical solutions. In the technical studies both immersed and bored tunnels were investigated for rail only as well as for rail+road. Bridges were studied only for combined road and railway traffic with a dual-two lane motorway and a double track railway. The bridge proposals were in their main design features similar to the Øresund Bridge, i.e. with a double deck truss carrying the two railway tracks on the lower level and the motorway on the upper level. The depth of the truss was, however, increased to 15 m to allow the approach bridges to have spans of 240 m. Across the navigation channel two solutions were studied, a multi-span cable-stayed bridge and a suspension bridge.

Fig. 24: Polyurethane tubes used as casings around the bundles of seven-wire strands

The cable-stayed bridge was designed with three 724 m main spans to allow one-way ship traffic in two navigation channels each with a horizontal clearance of 700 m. The suspension bridge was designed with a 1752 m main span and 588 m side spans. The cost estimates clearly pointed towards the cable-stayed option and after the conclusion of the Feasibility Study it was decided that the primary solution for the Femern Belt Link should be the bridge option with cable-stayed main spans and as a secondary solution an immersed tunnel.

More recently it has been questioned by some navigational authorities whether the cable-stayed bridge will give sufficient safety as shipmasters without experience in navigating through the bridge might be confused when approaching it and then select the central span in stead of the assumed outer starboard span. Also, it is realised that the nearest anchor pier in the side spans of the cable-stayed bridge would carry a special risk in relation to ship collision as the distance from the center of the navigation span to the nearest anchor pier is only 640 m .

Fig. 26: Suspension bridge from the Femern Belt Link Feasibility Study 1999 It is, therefore, likely that the bridge consultants will reconsider the design of the navigation spans in the Femern Belt, and maybe also reinvestigate a suspension bridge main span. 4.2 Ongoing investigations To give the best basis for the designers of the reference design to be tendered a number of investigations have already been initiated. Extensive environmental investigations have started to form the basis for the Environmental Impact Assessment (Hydrography, Marine Biology, Approaches on land, etc.). Geotechnical investigations of soil conditions is expected to continue until the end of 2011 within a total budget of 40 million Euro. Investigations into navigational safety are about to be concluded and will form a decisive basis for the final choice of main span and approach span configurations. With the unexpected appearance of individual cable vibrations in the Øresund Bridge a thorough study of cable vibrations under wind, rain and frost conditions has been initiated at the Department

Fig. 25: Cable-stayed bridge from the Femern Belt Link Feasibility Study 1999

of Civil Engineering, technical University of Denmark As part of this study a Climatic Wind Tunnel will be installed at Force Technology in Denmark. The specifications for the Femern Climatic Wind Tunnel:

- 20m x 8m closed-circuit - 25 m/sec mean wind velocity (min) with turbulence grid and flaps at -5OC - 40 m/sec mean velocity (min) without cooling elements - Water spray capabilities (simulated rain) and water-proofing - 2m×2m section minimum to allow for a maximum of 10% blockage on a 200 mm wide

cable or bridge section - Consistency of flow characteristics for both smooth and turbulent flow - Full turbulence spectrum (within reason) achieved through some form of flaps

(turbulence generators) - Approx. 1 MW tunnel power consumption

The total expected cost for the Climatic Wind Tunnel is approximately 1.3 Million Euro. The actual construction of the Femern Belt Link is planned to start in 2012 with a completion in 2018. 5. References [1] GIMSING, NIELS J. (editor), “The Great Belt Publicartions: West Bridge”, A/S Great Belt,

Copenhagen, 1998 [2] GIMSING, NIELS J. (editor), “The Great Belt Publicartions: East Bridge”, A/S Great Belt,

Copenhagen, 1998 [3] GIMSING, NIELS J., “Cable Supported Bridges – Concept and Design”, Wiley,

Chichester, U.K., 1998.

[4] GIMSING, NIELS J. (editor), “The Øresund Technical Publications: THE BRIDGE”, Øresundsbro konsortiet, Copenhagen, 2000.

[5] Trafikministeriet, “Femer Bælt-Forbindelsen, Forundersøgelser – Resumerapport (Femern Belt Link. Feasibility Study – Summarising Report)”, Trafikministeriet, Copenhagen, 1999.