the structural design of the hotel for the marques de riscal

Transcription

the structural design of the hotel for the marques de riscal
THE STRUCTURAL DESIGN OF THE HOTEL FOR THE
MARQUES DE RISCAL WINERY.
Miles Shephard (p), César Caicoya, Miguel Ángel Frías.
1.
INTRODUCTION.
The Marqués de Riscal Winery has commissioned Frank Gehry to design a hotel for
their facilities in Elciego, in La Rioja, Spain’s principal wine growing region. Gehry’s
design is for a highly innovative building with titanium and stone cladding. This paper
will describe how the engineers translated the Architect’s ideas into structural working
drawings. It will also describe the advanced computer techniques used in this process.
The project is now under construction, having commenced in May 2003.
2.
DESIGN ASPECTS.
2.1. Architectural Concepts.
The celebrated architect Frank Gehry was commissioned to design a new hotel building
at the Marqués de Riscal Winery in Elciego, in La Rioja region of Spain. This will be
the architect’s third project in Spain, the most famous being the Guggenheim Museum
in Bilbao. IDOM collaborated previously with Gehry on the Guggenheim and in this
project they are the Architect of Record. Their responsibilities include the detailed
architectural design, the structural design, the services design and the site supervision
and management.
The hotel for the Marqués de Riscal Winery will be an emblematic building clad in
stone, glass and titanium. With a built area of approximately 2000 m2, the building itself
is not large but its geometrical complexity makes it one of the most remarkable
buildings of this new millennium. The three floors and one mezzanine of the hotel are
supported above what appears to be a vineyard by three massive columns
(‘supercolumns’) and four inclined prisms that support the cantilevers of the end
elevations. The main structure is wrapped in a series of flowing coloured titanium
canopies supported directly off the main structure and from the ground on slender steel
sections. These provide shade from the intense Rioja sun.
The hotel reception area is a diaphanous space on the vineyard level. The next floor
incorporates 14 bedrooms, the floor above contains the restaurant and the top floor
houses the boardroom for Marques de Riscal. Although the hotel appears to be founded
in a vineyard this is, in fact, a false perception. The vineyard lies on top of a new wine
storage cellar and the building’s three ‘super columns’ pass through this slab and are
founded in rock 8 metres below the reception level.
295
Figure 1. Building Overview (model)
There are three floors, one mezzanine and the roof above the entrance level. Each floor
slab is highly irregular in shape with many peaks and troughs in their outlines. The slabs
gradually reduce in size successively giving the building a pyramidal shape. The
columns are not visible within the building and all forms of support are hidden within
the highly irregular internal walls. This has given rise to non-continuous and inclined
columns.
Figure 2. Canopy Detail with titanium cladding (model)
The architects made extensive use of physical models when designing the building.
Numerous scale models were made and changed in the process of fine tuning the
external appearance of the hotel. Each physical model was then translated into digital
form and the program CATIA was used for three-dimensional modelling. These CATIA
models were then translated into various different other programs for structural
calculations or in order to lay the design down on paper.
296
Figure 3. 3D Catia Models. Slabs and Canopies
2.2. Structural Concepts
Three ‘supercolumns’ with a total height of 16.5m start from the foundations below the
bottle store and pass through the second floor (the hotel entrance level) rising up to the
third floor. They are founded on large pad footings directly on rock below the bottle
store slab.
The third floor is a massive concrete slab that basically supports the rest of the building.
In addition to three concrete lift cores, a forest of small steel section columns hidden
within the internal partition walls of the building spring from this slab.
Figure 4. Structural Section.
297
The building’s internal spaces governed the positioning of these columns and therefore
they are highly irregular. In addition, the third floor supports the majority of the canopy
loads along its edges. The only supports for the slab are the three ‘supercolumns’
aligned along the central axis of the slab and four inclined columns, which support the
cantilevers at the ends. Transversely there are cantilevers of up to 10m in length. In
addition the slab is not centred about the longitudinal support axis and therefore the
‘supercolumns’ resist a permanent dead load moment.
The slab is of variable depth with a 1300mm section running longitudinally. The
transverse cantilevers have variable depth between 1300mm to 450mm at the edges
while the cantilevers supported by the inclined columns have a constant depth of
750mm.
This slab is very highly loaded and the transverse cantilevers are especially sensitive to
deflections. In order to counteract these deflections the slab is transversely post
tensioned. The irregularity of the slab edges, the amount of supports required at the slab
edges for the canopies and the fact that the tensioning must be done in phases in order to
prevent cracking under temporary conditions brings added complications to this
process.
Another complication for the slab as a whole is that it is supported by the very stiff
‘supercolumns’ and there are no movement joints. It is therefore designed to resist the
stresses induced by the stiff supports in shrinkage and temperature change. A concrete
pouring method is defined for the Contractor in order to reduce the shrinkage stresses to
a minimum.
The subsequent floors are designed as flat slabs supported on irregular steel columns.
As these columns are hidden in the internal partitions of the building they are highly
irregular. There are inclined columns and non continuous columns and this has led to
the slabs having to resist high point loads from columns being supported directly by the
slab at that level. In the majority of cases, the flat slabs are 450 mm thick with 200 mm
reduced depth areas in zones of less moment. The result appears like an irregular beam
and slab arrangement but functions like a flat slab.
Three cores rise from the ‘supercolumns’ in order to give the structure lateral stability.
In order to reduce negative effects produced by shrinkage and temperature difference
with three lateral supports, only the central core provides lateral support in both
directions. The end cores have walls transversely across the slab only, while the central
core is a box. Transversely all three cores provide horizontal support, while
longitudinally only the central core functions.
The canopies are attached to the edges of the main structure. Each canopy is designed as
a self-supporting structure with at least three points of support. The upper canopies are
supported directly off the slabs while the lower canopies are supported both from the
slabs and from columns rising off pad foundations sitting directly on the bottle store
roof slab.
298
2.3. Detailed Design of the Main Structure
Given the highly irregular nature of the supports, loads and shape of the slabs, each slab
was modelled in detail using finite element plates. The program chosen for modelling
was ANSYS due to its flexibility and high calculation capacity.
The most sensitive slab was the third floor. This was the slab with the highest loads as it
supports the rest of the structure and it also has the longest cantilevers, up to 10 metres
in places.
Figure 5. 3rd Floor Plan.
Extensive modelling was carried out on this slab in order to check the following load
cases:
•
•
•
•
•
Dead and live loads supported directly on the slab
Dead and live loads from the columns
Dead, live and wind loads from the canopies
Shrinkage loads caused by the three rigid supports
Loads from the post tensioning, including by phased tensioning
The deflections at the ends of the cantilevers were very carefully checked and the post
tensioning designed to reduce the deflection to within the building code limits. Phased
tensioning was required in order to remove the possibility of cracking in the soffit of the
slab. 55% tendons will be tensioned once the concrete has gained sufficient strength.
Once the forth floor has been cast and its formwork removed 25 % more of the cables
should be tensioned and the remaining 20% should be tensioned once formwork has
been removed from the roof slab.
299
Reinforcement was designed in both directions with the moment envelopes taken from
the computer models. Special attention was paid to the local detailing of reinforcement
in the irregular cantilever zones and over the ‘super column’ supports.
Figure 6. Model of 3rd Floor.
The remaining slabs were 450mm thick with 200mm thick sections in less stressed areas
with the exception of the mezzanine slab, which had a constant depth of 250mm. These
slabs were designed in a similar way to the third floor, with detailed models in ANSYS.
The irregular columns (non-continuous, sloping, etc) were modelled in order to give
accurate results for areas that will be locally very highly stressed by column point loads.
2.4. Detailed Design of the Secondary Structure
As much as anything, the design of the secondary structure supporting the canopies was
a geometrical problem. This was done in very close cooperation with the architects and
in order to achieve this an engineer was assigned to Gehry’s offices to carry out this
design. The whole structure will be visible and therefore aesthetic considerations were
as important as structural requirements.
The concept of each canopy is that it is a determinate structure supported at a minimum
of three points. The frames consist of two main beams (I sections), highly deformed
about their minor axis, supporting T sections, 90% of which were straight. These T
sections are supported at a variety of angles along the main beams thus producing a face
curved in two directions without the need for bending beams in two directions. The T
beams support hat channels that will be bent into shape during erection. Titanium plates
are then riveted to these channels.
300
Figure 7. Wood Frame Model / New York Mock Up.
The first problem encountered in the design of the canopies was that of estimating the
loads. The forms are so irregular that the wind codes could not be directly applied to the
structure. It was therefore decided that a wind tunnel test should be carried out. Using a
scale model produced by Gehry, RWDI tested the structure in their wind tunnel in
Canada. The data produced included patch loads of varying intensity for each canopy.
These were combined with dead loads, access maintenance loads, snow loads and
accumulated snow loads for the gullies, each multiplied by its given partial factor
depending on the load combination, to form a loading envelope for each canopy.
The design of the canopies was an iterative process. An initial design was carried out in
order to estimate possible section sizes. The structure was adjusted until the Architect
deemed the initial section size satisfactory.
In order to progress to the next stage of design a wire frame model of the canopies was
required. The architects were using CATIA for their three dimensional design. This
allowed all the building surfaces to be modelled in space. The canopy supporting
structure was added to the surface model and the result was then translated into the
Rhino 3D modelling programme so that it could be used for the basis of a structural
model.
The program RISA was used for the structural modelling of the canopies for its three
dimensional capabilities and its ease of use. The data from the wire frame model was
imported into RISA. The orientation of the members was checked and then a series of
load cases were input to complete the calculation. A large number of different load
cases were used in order to ensure that every possibility had been catered for.
301
Figure 8. Wire Frame Model.
A very sensitive part of the design was the connections to the slabs, especially on the
third floor. As previously described the third floor slab is post tensioned. In order to
allow access to the cables the highly irregular slab edge cannot be cast until tensioning
is complete. Therefore there are a number of important canopy supports that are
attached to sections of the slab that are not cast monolithically with the third floor.
Continuity of reinforcement to this newly cast edge section of slab will be vital in order
to transmit the forces from the canopies. This is achieved in two ways: threaded bars are
spliced onto the previous pour, thus ensuring continuity but leaving the area free for
cable tensioning and the canopy connection itself is formed from a fabricated I section
running along the edge of the slab so distributing the loads as much as possible. The I
section is attached to the concrete by way of stud shear connectors.
Figure 9. Canopy/Slab Connection
Canopy connections to the upper floors were simpler as they could be cast in with the
rest of the slab. For conformity with the third floor connections, Edge I beams were
used with shear studs to tie them to the slab.
One of the most innovative aspects of the canopy design was the way in which
computer programs were used in order to visually model the structure. This was
important for two reasons: firstly these models highlighted any points where structural
members or the cladding might clash with other canopies or the main structure.
Secondly, the visual models gave the architects the opportunity to review the connection
designs from an aesthetic angle before the design was translated into working drawings.
Full scale tests were carried out on the performance of the cladding for the canopies,
testing their reactions to wind pressure, suction, heat and freezing. The plates are
screwed into the structure. As the plates expand and shrink due to temperature
difference forces will be exerted on those screws. Tensions will be generated due to
wind forces. These stresses were checked with a full scale mock up of a section of a
canopy subjected to conditions in excess of those expected.
302
We had the opportunity to test the theory of the designs when three of the canopies were
constructed at 2/3 scale for an exhibition. The Guggenhiem Museum in New York held
a Gehry Retrospective Exhibition and scale models of the Marques de Riscal canopies
formed one of the pieces. This gave us the chance to see exactly how the structure fitted
together and it gave the architects the opportunity to see how the finished product would
look. The fabrication and erection of the structure was not unduly complicated and the
visual aspect of the finished canopies was very much as required by the architects.
Figure 10. Connection in CATIA / Reality in New York.
3.
CONSTRUCTION DOCUMENTATION
Given the complexity of the building and its structural calculations it was decided to
provide the Contractor with a high level of detail for each connection type and every
reinforcement detail. While the Contractor must geometrically detail the steelwork
connections, generic details provide the connection type and govern the aesthetic
qualities of the unions. The reinforcement details of complex areas such as the column
supports are drawn at scale 1:10 in order to reduce possible errors of interpretation. The
high level of detailing in the construction drawings is also coupled with a high level of
site supervision to ensure that the building is constructed in the exact way intended by
the design team.
The building itself is highly irregular and given the number of different drawings the
possibility of error within the AutoCAD plans is high. During the development of the
design, all drawings were taken from a CATIA masterplan where all the outlines of the
different layers (concrete, steel sections, cladding, etc.) were plotted in three
dimensions. It was therefore logical that the setting out of the structure should be based
on the CATIA drawings. Information from the CATIA file takes precedence over any
AutoCAD drawing. The Contractor has the full CATIA file loaded onto a PC on site,
which is dedicated to the setting out of the building.
4.
CONCLUSIONS
The design of the Marques de Riscal Winery Hotel building has been a great
opportunity for engineers and architects alike to push back the frontiers of what was
303
previously thought possible. The extensive use of computer software has been of prime
importance not only in structural calculations but also in aesthetic design and in
ensuring adequate fit between the different structures and the cladding.
The building is currently in progress on site. While it was never going to be an easy
structure to build due to its highly irregular shape, it is hoped that the extensive
planning and calculation during the design phase will assist the Contractors during the
construction phase.
Figure 11. Progress on Site. June 2003
ACKNOWLEDGEMENTS.
The author would like to acknowledge the assistance of the following people: Shyamala
Duraisingham and Karl Blette.
REFERENCES.
Duraisingham, S. and Barrett, R. and Blette, K. and Pérez, F. and Caicoya, C. The
Structural Design of the Hotel for the Marques de Riscal Winery, Spain. The Structural
Engineer. 2 July 2002. Volume 80 Number 13. Pages 23-29
CORRESPONDENCE.
Miles Shephard
Avda. Lehendakari Aguirre, 3
48014 BILBAO
94 479 76 00
miles@bilbao.idom.es
304