by AS ZELJIC · Cited by 13 — In designing the façade for Shanghai. Tower, a 124-level, 632-meter (2,074 feet) highrise, Gensler introduced a combination exterior and interior curtain wall

129 KB – 18 Pages

PAGE – 1 ============
| ˜˚In designing the façade for Shanghai Tower, a 124-level, 632-meter (2,074 feet) highrise, Gensler introduced a combination exterior and interior curtain wall system totaling 210,000 square meters (2.26 million square feet) of glazing area. This paper traces the development of the overall curtain wall system, focusing on exterior proposed design options and the issues associated with each of them, and discusses the underlying decision-making that led to the ˜nal documented option. Shanghai is located at ˚˛˜°˝ ˚™~ ˚˛˛° ˚˛™ east and ˙˜°ˆ˜™~˙ ˚°˝˙™ north, in the eastern part of Asia. It is on the west coast of the Paciˇc Ocean, the center-point of the north and south coast in the Peoples Republic of China, on the edge of the East China Sea. The prevailing climate is a subtropical monsoon climate, with weather that is hot and very humid during summer. It has four distinct but mild seasons with full sunshine and plentiful rain. Outside atmospheric temperature range from ˛˘F (Œ˛°C) to ˝F (˙˝°C), with an annual average temperature in the urban district oˆF ( ˚°C). Humidity levels vary daily but are constant through the year. Annual precipitation is more than ˚,ˆˆ˜ mm. Fifty percent of the annual precipitation falls in the ood season between May and September. There are many northwestern and southeastern winds throughout the year. The average annual amount of sunlight is ˚,˝ˆ˘ hours, with insulation varying from ˛.˝o ˝. ˚˝ kWh/m²/day. Now under construction, Shanghai Tower is the third and ˇnal planned super-high-rise building in Shanghai™s Pudong area that completes the development of the Lu Jia Zui Central Financial District. With a large program totaling about ˝ˆ˜,˜˜˜ m˛ (˝, ˚˝,˜˜˜ square feet) of built enclosed area, ˙˜,˜˜˜ m˛ (ˆ, ˚˜˜,˜˜˜ square feet) are above grade and ˚˜,˜˜˜ m ˛ (˚,˘ ˚˝,˜˜˜ square feet) are below grade. The tower has been designed as a soft vertical spiral rotating at about ˚˛˜ degrees and scaling at ˝˝% rate exponentially. The tower functions as a self-sustaining vertical city. It is a mixed-used building of unique, vertically interconnected neighborhoods that evolve as the tower slowly rises toward the sky. The building will comprise ˚˛oors plus four additional BY ALEKSANDAR SASHA ZELJIC, AIA, LEED AP This paper was delivered at the 2010 International Conference on Building Envelope Systems and Technologies (ICBEST 2010) held in Vancouver, Canada. Shanghai Tower Façade Design Process

PAGE – 2 ============
| ˜˛oors of equipment rooms and Tuned Mass Damper (TMD). The top of the structure is a˙˛ meters (˛,˜˘ˆ feet) height. The building is divided into nine zones with ˇve main functions: oce; boutique oce; luxury boutique hotel; themed retail, entertainment and cultural venues at the podium; and the observation experience at the tower™s pinnacle. Within each zone are atrium spaces that operate as activity centers and a gathering place for people within their fizonefl community. Additionally, each atrium is designed to accommodate access by the general public. The concept of the podium is to become activated with people, allowing uninterrupted public circulation between three adjacent fisuper-high-rises,fl and be open and interconnected with the neighboring community. Design Considerations Building Geometry The tower™s profound twist expression is the result of its geometry, which can be broken down into three key components that are controlled in total by four variables: ˚. Horizontal pro˜le (Figure ˙): The proˇle shape is based on an equilateral triangle. Two tangential curves oset at ˜ degrees were used to create a smooth shape. This Figure 1: Shanghai Tower Site, Lu Jia Zui Figure 2: Proposed vertical zone division Figure 3: Horizontal pro˜le geometry

PAGE – 3 ============
| ˜˙shape is driven by two variables: the radius of the large circle and its location relative to the center of the equilateral triangle (proˇle). It should be noted that the actual shape of the proˇle is independent of the remaining two key geometric drivers. As a result, Gensler had the ability to look at the eect of modifying the horizontal proˇle and the impact such changes had on the tower form at all stages of the design. ˛. Vertical pro˜le (Figure ˆ): The concept of the form is to take the horizontal proˇle and extrude it vertically and conform to the vertical proˇle. From a functional point of view, it was important to maintain a wide footprint for the lower third of the tower, with a slender footprint at the upper thirdŠa reduction of about ˝˝% overall. This proportional distribution allowed for large lease spans within the oce portion of the tower and smaller spans within the upper-level hotel/boutique oces. Early in the design, it was found that a basic exponential curve provided the desired result. This is the same basic formula used in the ˇnance industry for continuous compounding and/or discounting. Adjusting the two values in the horizontal proˇle and this third value in the vertical proˇle, we now have complete control of vertical ratio, grosoor area and building form. Figure 4: Vertical pro˜le geometry ˙. Rate of twist: This is a simple linear rotation from base to top. The fact that this ˇnal value can be changed independently allowed for greaexibility in the design stage, especially in selecting the best combined overall building performance. Wind Tunnel Testing Results Wind tunnel testing was essential for understanding building performance and was conducted at Rowan, Williams, Davies & Irwin Inc. (RWDI). The wind tunnel test procedures were based on requirements set out in Secti.f the ASCE Discounting formula based on exponential function adapted to suit geometric analysis for Shanghai Tower Figure 5: Wind tunnel study model, 1:500 scale Figure 6: Reynolds number study model, 1:85 scale (RWDI)

PAGE – 4 ============
| ˜ˆ˘-˜˝ Standard and the Load Code for the Design of Building Structures GB ˝˜˜˜-˛˜˜ ˚ for the P.R.C. Additionally, to predict the full-scale structural response and more detail pressure loads, the wind tunnel data were combined with a statistical model of the local wind climate. The wind climate model was based on local surface wind measurements taken at Hong Qiao International Airport and a computer simulation of typhoons provided by Applied Research Associates, Raleigh, North Carolina. All testing was conducted on a ˚:˝˜˜ model. Additionally, a ˚˝ scale model was tested for results of the Reynolds number correction factor that was used for more precise data on loading and the impact of wind vortex split on round exterior wall surfaces. The Gensler design team had anticipated that signiˇcant reduction in both tower structural wind loading and wind cladding pressures could be established if the building further improved its proposed geometry following the variables previously explained. To establish the best possible case for reducing these loads, several scenarios were proposed involving rotation a˜°, ˚˛˜°, ˚˝˜°, ˚˜° and ˛ ˚˜° and then scaling o˝%, ˆ˜% , ˝˝%, ˘˜% an˝%. All these scenarios were analyzed against each other and then compared to the base case scenario that was proposed, in the form of a tapered box. Results acquired through this process have shown that a scaling factor of about ˝˝% and rotation at ˚˛˜° can account for up to ˛ˆ% savings in structural wind loading and cladding pressure Figure 7: Wind tunnel study scaling models Figure 8: Wind tunnel study rotation models reduction as compared to base-case tapered box. This equates to about $˝˜ million (USD) in savings in the building structure alone. Additionally, it helped optimize and distribute maximum cladding loads on the building while maintaining desired aesthetics. Aesthetic concerns prevented the ˚˜° rotation from being pursued, even though it would reduce loading by an additiona% (Figure ˚˜). Ongoing testing procedures included Reynolds number testing conducted with a ˇnal model at ˚˝ scale. During this testing, constraints particular to the site were exempliˇed with Jin Mao and Shanghai World Financial Center, which combined generate a localized increase in lateral turbulence intensity between ˚ˆ% and ˆ˜%. During testing for the high Reynolds number, the following was concluded: Figure 9a: Main structure and building systems diagram

PAGE – 5 ============
| ˜˝fiWhile the positive pressures are unected by the Reynolds number, the negative pressures could be increased at a high Reynolds number. Approaching wind turbulence tends to reduce the Reynolds-number eects. To account for potential Reynolds-number eects for cladding design, it is recommended that the exterior peak negative pressures around the building corners determined from the ˚:˝˜˜ scale model tests should be increased by ˚˜%. This correction is applicable to the upper third of the building. For lower portions of the building, the Reynolds-number eects tend to be insigniˇcant due to high turbulence levels. Similar corrections should also be considered in the structural wind loads for the curtain wall support system.fl ˚ Final cladding loads testing results revealed that peak positive loads (pressure) are at about ˛.˜ to ˛.˝ kPa for abou˘% of the building, with ˛.˘˝ kPa maximum. Peak negative loads (suction), on the other hand, is at ˆ.˝ kPa for abou˝% of the building, wit.˝ kPa maximum. Peak negative loads are distributed considerably around corners and at the upper building half toward the top. At the same time, the highest instant dierential pressure between two subsequent curtain panels on exterior wall can reach +/Œ ˚.˝ kPa in either horizontal or vertical direction. Interior curtain wall cladding loads were at ˛. ˚ kPa. Here the Gensler team made an assumption in order to coordinate the erection sequence of Curtain Wall B, which will not see real wind-imposed pressures. The proposed curtain wall design strategy suggested addressin˝% of the negative ˆ.˝ kPa with a standardized segmented unitized system that is uniform throughout the building™s exterior glass wall. Peak loads of up t.˝ kPa corresponded to locations on the building where curtain panels were smaller in width due to the tower scaling factor on the same number of curtain panels peoor. This allowed that design glass thickness stayed uniform while vertical mullions were reinforced where needed to respond to high lateral stresses. Curtain Wall Support System Responding to an array of conditions in Shanghai, Gensler™s team proposed a building design that employs a curtain wall system designed as a symbiosis of two glazed wallsŠan exterior curtain wall (Curtain Wall A) and an interior curtain wall (Curtain Wall B)Šwith a tapering atrium in between. The main support for the exterior curtain wall is a horizontal ring beam consisting of a horizontal pipe ˙˝m in diameter laterally supported, at ˚˜ meters on-center in Zone ˛ and ˘ meters on-center in Zone , by a radial pipe strut support. This variation is a result of the geometry that included tapering and rotation of the tower. Figure 9b: Comparison models used for RWDI ˜nal studies Figure 10: Wind loading testing results comparison (RWDI)

PAGE – 6 ============
| ˜The horizontal radial pipe strut supports consist of a ˛ ˚-mm diameter pipe (with varied but mainly ˛˛ mm wall thickness) that transfers the exterior façade lateral load to the inner circular building slab edge. The radial strut pipes are rigidly connected with the horizontal girt while using a hinge connection on the other sideŠat the interior slab edge steel supportŠto allow the exterior façade to move up and down relative to the inner structure. To carry the gravity load of the façade and façade support structure, tw˜-to-˜-mm high-strength rods (depending on the zone) are hung from the mechanical room/refuge area above, with a robust steel structure designed within, to the horizontal ˙˝-mm ring pipe beams at ˆ.˝ meters (ˆ.˙ meters in Zones ˘ an) on-center vertically at every strut location, including an amenitoor that uses steel bushings instead of perpendicular struts to limit lateral movement. Steel bushings move in vertical direction to allow for expected combined closing and opening movements to be largest at Zone ˛, at ˚˚ ˆ mm. This is also where the curtain wall system vertical expansion joint is located. In the horizontal direction, there are total of eight expansion joints, each allowing typically about ˝m of combined open and closing movement. Special considerations were given to ˇre protective requirements of the Curtain Wall Support System (CWSS), and additional allowances had to be provided to total vertical and horizontal movements. ˛Figure 11: Diagram of positive and negative wind cladding loads (RWDI) Figure 12: Section perspectives with curtain wall systems description Figure 14: Typical atrium top and bottom Curtain Wall A connections Figure 13: Tower curtain wall support system (CWSS)

PAGE – 9 ============
| ˜Wall A was to be considered as an enclosure for conditioned space of the building, then makeup for it had to be an insulated glass unit. This created an additional challenge given the large size of the glass panelsŠvarying from ˛.˛ by ˆ.˝ meters to ˚.˛ by ˆ.˙ meters. The glazing unit would have to be not only insulated unit make-up, impacting with that desired visual transmittance ratio (targeted very highŠup to ˜.), but would also require individually thicker glass lights to respond to high wind-load peaks. In-plane glass deection had to be less than ˛˝ mm and with the insulated unit, there was a danger of two lights touching each other at high peak loads, thus creating the danger of possible peak incidental breakage. The idea of adding a spacer in the middle of the glass unitŠ although possibleŠwas not entertained. It has been calculated that if units were required to be insulated, then glass would have to be of a ˚˝ mm glass + ˚˜ mm air + ˚˝ mm glass make- up. This was a signiˇcant increase from the ˚˛ mm glass + SGP interlayer + ˚˛mm glass laminated make-up that was targeted. At current weight, betwee˜˜ to ˚,˜˜˜ kilograms (˛,˛˜˜ pounds) per glass unit (the largest units at Zones ˛ and ˙), this direction would result in an additional ˛˝% increase in exterior glass weight. Ultimately, this would impact the CWSS in its eective size and visual appearance in atrium spaces, as well as on individual member weight, which would also impact the total building weight expected to be approximatel˝˜,˜˜˜ metric tons, spiraling all the way to potential redesign of an already approved complex foundation system (including about ˙,˝˜˜ piles at ˚,˜˜˜ mm in radius and ,˜˜˜ mm high matt foundation) on a limited site area. It is common that the total exterior wall weight is within the ratio of up to ˛% of total building weight; however, the intent of the design team was to truly follow principles of China™s Three-Star Rating, based on implementing high-eciency standards with reduction and multiple usage of individual members where possibl eŠfiDo more with less.fl After going through an extensive and complex review process with various client, city and government expert panels, it was determined that Exterior Curtain Wall A was to be considered as a weather enclosure for ventilated and unconditioned atrium space, and that the true exterior wall is to be Interior Curtain Wall B. This allowed for Curtain Wall A to employ a laminated glass assembly and maintxterior wall-to-weight ratio, while maintaining desired transparency and glass- area ratio. Additionally, various strategies were employed to maintain atrium performance at a comfortable level (Figures ˚˝ and ˚Final Glass Selection As ˇnal glass selection will be contingent to series of mockups that are scheduled to be prepared in the next ˚˛ months, the Gensler team has proposed the following generic glass types for two major curtain wall Systems: Curtain Wall A: ted glass assembly: ˚˛ mm low-iron glass +˚.˝˛ mm SGP interlayer + low-e coating + ˚˛ mm low-iron glass. The upper ˛˝% of the panel will have dissolving frit pattern from ˘˝% down to ˚˝% Curtain Wall B: ˙˜ mm insulated glass assembly: ˚˜ mm low-iron glass with low-e coating + ˚˛ mm air space + tion of the panel between fil have dissolving frit pattern from ˚˝% down to ˘˝% and ˚˝% again Figure 17: Curtain Wall A and B standard panels

PAGE – 10 ============
| ˚˜Curtain Wall Design Options Curtain wall design development began as an exercise that at ˇrst attempted to resolve complex building geometry. The remainder of this paper will focus primarily on Curtain Wall A, analyzing ideas and various options, with an overview at the end of this paper on the current status of Curtain Wall B. Curtain Wall A Studies In developing resolutions for curtain wall geometry and ˇnal design, Gensler™s façade team used a variety of available software that involvxibility in analysis. Early digital tools were Revit and Generative Components; however later studies on exterior wall were conducted exclusively through Rhino with Grasshopper parametric mechanism as well as ˙D Max and AutoCAD. This allowed for a constant precise geometrical understanding of the various exterior wall schemes being proposed and their relationship to building form. Figures ˚associated discussion captures the results of these complex studies, highlighting the geometry involved. It should be noted that Revit was used as main software for tower documentation and consultant coordination. Figure 18: Curtain Wall Systems B and A Figure 19: Curtain Wall AŒCurtain panels sizes change vertically Figure 20: Typical zone plan view with superimposed curtain wall structural supporting system

PAGE – 11 ============
| ˚˚Figure 21a: Curtain Wall A: Pro˜le control points divisionŠpartial view Figure 21b: Curtain Wall A: The layout of two adjacent ˚oors Figure 22: Curtain Wall A: Control points interaction between subsequent ˚oors The starting point for all studies was a default proˇle and desire to have division along the curve est. In the competition phase, the Gensler team decided not to pursue surface diagrid or triangulation schemes, given that the design intent was to have two curtain wall skins with as minimal obstruction of the view out as possible. Going forward, this was a major design criterion for the client. Understanding glass-size limitation and desired scaling, the Gensler team decided that a single piece of glass should not be larger than ˛.˙ metommodate ting, coating and thermal glassŒprocessing capabilities. The default proˇle is divided along the curves into ˚ˆˆ control points; this ˚ˆˆ-point division resulted in ˚ˆˆ panels. The largest distance between control points was about ˛.˛˝ meters (~˘™-ˆfl) at Zone ˚ (ˇrst and about ˚.˛˝ meters (~ˆ™-˛fl) at Z Early studies suggested that the best location for a starting point for division was to be at fiV-strikefl area, with full panel size following. However, coordination of major structural elements behind the curtain wall required that a second point on the curve be moved ˙˙% of the panel size to allow for a clean connection of the strut, and sag rods to the perimeter girt, avoiding possible ct with the vertical glass ˇn and mullion assembly. This is why the ˇrst and last panel along the curve are f an actual panel. The logical starting point in resolving the curtain wall was to connect these control points directly and have a smooth appearance on the exterior of the wall. This involved angling the vertical mullions in two directions, which is what the Gensler team proposed as one of the early schemes. Due to a combination of rotation and scaling of t out of four points deˇning the panel will always be out of the glass plane (Figures ˛ ˚a and b), creating a warping of the glass and, in some situations, separation as lar This situation prompted a series of schemes that proposed vertical fishinglingfl of glass panels with a fithree-partfl mullion system (three individual components in a single structural line of vertical mullions) to deal with panel plane detion on the fourth point. There were three distinct shingled schemes proposed: shingling along vertical mullions, along horizontal mullions, and along both. As the process moved forward, the horizontal shingle remained the only option. To rtive vertical size of the aluminum proˇle, a glass ˇn was introduced along the vertical mullion, braced at top, bottom and mid-span, reducing the mullion™tive

129 KB – 18 Pages