You are on page 1of 15

THE STRUCTURAL DESIGN OF TALL AND SPECIAL BUILDINGS Struct. Design Tall Spec. Build.

13, 89103 (2004) Published online 7 July 2004 in Wiley Interscience (www.interscience.wiley.com). DOI:10.1002/tal.237

A NEW MODELLING TECHNIQUE USING A SUBSYSTEM FOR STRUCTURAL SCHEMATIC DESIGN OF TALL BUILDINGS
HANSOO KIM1*, JONGHYUN JUNG2 AND SUKHEE CHO1
2 1 Hyundai Institute of Construction Technology, Korea Department of Architectural Engineering, Kyungnam University, Korea

SUMMARY In the schematic design of a tall building structure, a structural designer builds structural analysis models for many schemes. However, conventional modelling techniques force the designer to view the scheme as an assembly of many members related in a complicated manner. Therefore, the modelling scheme is laborious, prone to error and time-consuming. In this study, we present a new modelling technique that uses a subsystemthat is, the assembly of the members participated in a specic type of load transferas the primary modelling unit. We expect this modelling technique to allow simple modelling of a scheme. Copyright 2004 John Wiley & Sons, Ltd.

1. 1.1 Background and objectives

INTRODUCTION

In the structural schematic design of tall buildings, it is important to select a safe and economic structural system based on the results of evaluation of safety, serviceability and economy. This makes many structural designers try to generate and evaluate various structural schemes. The work of generating and evaluating the structural schemes of tall buildings commences with conceiving the schemes and building analytical models of the schemes. When conceiving schemes of a tall building, the designer treats each scheme as an assembly not of members but of subsystems, each of which is the conceptual assembly of members that participate in a specic type of load transfer. Thus, when modifying a scheme in mind, in most cases the designer does not add, delete or move each individual member. The designer, instead, adds, deletes and moves a subsystem. This means that the designer treats a subsystem as a single unit in the modication of a scheme in mind. This conceptual operation using a subsystem is so easy and simple. On the other hand, in modelling a scheme, that is, in preparing the input le or performing preprocessing for structural analysis for a scheme using conventional modelling techniques, the designer is forced to treat the scheme as an assembly of so many members, each of which is related in a complicated manner. The designer is then likely to be confused because there are so many gures and complicated relations between members. Consequently, the modelling of a scheme is laborious, timeconsuming and prone to error. In this study, in order to solve the above-mentioned problems, we proposed an intuitive and convenient modelling technique that enables the designer to build a structural analysis model of a scheme of tall building just as he or she conceives the scheme in mind. We then developed a computer program
* Correspondence to: Hansoo Kim, Hyundai Institute of Construction Technology #102-4, Mabuk-Ri, Goosung-Eup, YonginSi, Gyunggi-Do 442-912, Korea

Copyright 2004 John Wiley & Sons, Ltd.

Received February 2003 Accepted March 2003

90

H. KIM ET AL.

that implements the proposed modelling technique and apply the program to an example building to verify the effectiveness of the modelling technique. 1.2 Contents

First, we investigated the process of structural schematic design of tall buildings focused on modelling and then identied the properties of modelling. We reviewed conventional modelling techniques to identify the problems with and requirements for the modelling. We then proposed a new modelling technique that could satisfy the requirements. Finally, we developed a computer program and veried the effectiveness of the modelling technique through the application of the program to the example building. 1.3 Scope

In the structural design of tall buildings, the main issues are lateral loads due to wind and earthquake. Therefore structural systems that resist mainly lateral loads are included in the scope of this study. Structural systems that resist mainly gravity loads are excluded. Floor assemblies such as one-way and two-way are also, excluded, as are buildings with a curved shape. However, oor assemblies and buildings of curved shape will be included in future studies. 2. 2.1 REVIEW OF STRUCTURAL SCHEMATIC DESIGN AND CONVENTIONAL MODELLING TECHNIQUES OF TALL BUILDINGS Structural schematic design of tall buildings

2.1.1 System and subsystem Structural systems of tall buildings have been classied with respect to the way they resists lateral loads. Braced frame system, shear wall system, tube system and so on are typical examples of structural systems of tall buildings. In fact, however, there are hardly any tall buildings designed to make use of a single type of structural system. In a tall building, various types of structural system are combined as needed. An excellent example . . . is the Overseas Union Bank Center in Singapore. Here a braced steel frame was used . . . and combined with concrete shear walls (CTBUH, 1995). Therefore, it is reasonable to treat a structural system of a tall building as the combination of structural systems each of which is the assembly of the members participating in a specic type of load transfer.1 In other words, a structural system of a tall building is treated as a combination of subsystems. Consequently, the process of structural schematic design of tall buildings can be thought of as the process of combining subsystems. This means that subsystems play a key role in structural schematic design. 2.1.2 Conceiving schemes Before modelling a scheme, a structural designer conceives a structural system, i.e., a combination of subsystems. Here the designer conceives a structural system from scratch or adapts on existing one. In any case, when conceiving the system, the designer treats it as a combination of subsystems. And, although a subsystem consists of many members, the designer ignores the corresponding individual members and treats a subsystem as a single unit. Therefore, the designer adds, deletes and moves a

In this study, we will use the term subsystem to refer to the structural system combined to form a structural system.
Struct. Design Tall Spec. Build. 13, 89103 (2004)

Copyright 2004 John Wiley & Sons, Ltd.

SUBSYSTEM FOR STRUCTURAL SCHEMATIC DESIGN

91

subsystem in mind just as he or she adds, deletes and moves a member. For example, when moving a braced frame to another location, the designer does not move the corresponding individual members. Instead, the designer considers the braced frame as a single unit into which the corresponding members are abstracted and then moves a braced frame at once. Although the subsystem has been moved, the topology of the corresponding members within the subsystem is not modied. Only the location of the subsystem itself and the connection relationships between the boundary members of the subsystem and the adjacent subsystem are modied. 2.1.3 Grids In many cases, a designer recognizes the location and size of a member using grids with labels. That is, a designer thinks a member is attached to a grid and recognizes the location and size of the member using the location of the corresponding grid. Also, the designer can recognize the topology of members using the relation of the grids. Here, of course, labels are used to identify each grid. Following is a typical example that demonstrates the role of grids. In Figure 1, the designer would prefer to think that a member is located on the grid Y3 and the left end point is the cross-point of grid X3 and Y3 and the right point is that of grid X8 and Y3. And, because the distance between grid X3 and X8 is 10 m, the length of the member is 10 m. The designer, however, would not think as following; the coordinates of the origin is (0, 0) and the coordinates of the end points of the member are (4, 4) and (14, 4) respectively. The situation described above is equally applicable to editing such as resizing and moving of members. The designer thinks that to move a member is to attach the member to another grid and the location and size of the member are determined by the gird to which the member is attached. If a grid is moved, the designer thinks all the members attached to the grid are also moved automatically. Therefore, in the structural schematic design of tall buildings, grids play not only a subsidiary role in entering points efciently but also a key role in determining the location and size of a member and in guring out the overall geometrical shape of the system in an organized way. 2.2 Conventional modelling techniques

2.2.1 Modelling technique using elements Modelling using elements originated from the nite element analysis method (FEM). In FEM, a structure is divided into so many nite elements the relations of which are represented using nodes. A designer therefore generates nodes rst and then generates elements based on the nodes. Most FEM programs, such as SAP2000 (CSI, 2000) and ABAQUS (Hibbitt, Karlsson & Sorensen, Inc., 1994), use this modelling technique.

8@2m Y4 Y3 Y2 Y1 X1 X2 X3 X4 X5 X6 X7 X8 X9 3@2m

Figure 1. Example of grid and member


Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

92

H. KIM ET AL.

Because FEM is a general-purpose analysis method that can be applied to almost all kinds of structures, a modelling technique using elements can be used in modelling buildings, aircraft, machinery, ships, and so on. Therefore, no characteristics of tall buildings are reected in this modelling technique. Consequently, there are difculties, described below, in modelling the structural systems of tall buildings using elements. First, because there is no concept of subsystem in conventional modelling, it is not easy to make use of a subsystem. In other words, a designer must build the model using so many individual elements. For example, when conceiving in mind, a designer can move a subsystem as shown in Figure 2(a). That is, the designer can move all the elements included in the subsystem at once. On the other hand, when modelling, the designer should move one element at a time as shown in Figure 2(b). Second, because a designer should refer to numerous identication numbers assigned to each element and node, it takes time to build the model. Also, because identication numbers do not convey any useful meaning, it is easy for the designer to make errors. Third, because grids are used only to help input points, there is a need to add or delete a large number of nodes and elements to reect the modication made by inserting a new grid or moving an existing grid. If the identication number must be assigned again to each node and element, the amount of work to be done will be greater. Although some of the nite element analysis programs offer functions to select or group elements in order to move or delete them at one time, there are still many things to improve for the efcient modelling of structural schemes of tall buildings. 2.2.2 Modelling technique using members and frames Modelling technique using members and frames was rst introduced by ETABS (Habibullah, 1995) and was developed for static and dynamic analysis of three-dimensional multi-storey frame and shear wall buildings.

Movement of subsystem : 1 times

(a) When conceiving

Movement of member : 25 times

(b) When modeling using elements

Figure 2. Comparison of moving a subsystem and elements


Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

SUBSYSTEM FOR STRUCTURAL SCHEMATIC DESIGN

93

In this modelling technique, a building is treated as a group of vertical frames connected by oor at each storey level and a frame is regarded as a group of members, such as column, beam, brace and wall, each of which is placed on a column line or bay. A column line, here, is a virtual vertical line on which columns are located; a bay, similar to a column line, is a virtual horizontal line for a beam, wall and brace. Therefore, in order to use this modelling technique, rst a designer should divide a building into frames and dene storey levels, column lines and bays. Then, for each frame, the designer should dene column lines and bays and assign the corresponding members using storey levels, column lines and bays. Generally, oors are not modelled in the procedure because each oor is assumed to be a diaphragm that has innite in-plane stiffness and no out-of-plane stiffness. This modelling technique makes use of column lines, bays and storey levels instead of nodes. This modelling technique also makes use of members and frames instead of elements. Thus, although the applicable scope is limited to buildings, this modelling technique provides the concept of member and frame, which helps a designer take advantage of the characteristics of buildings as a modelling component. Consequently it is possible to build an analytical model of buildings with ease and decrease the amount of work. This modelling technique is more advanced and specialized than that using elements. However, because this modelling technique is not developed exclusively for modelling of tall buildings, it still has some problems in building a model. The rst problem is that there is a need to perform the same operation on some members repeatedly. When conceiving, on the other hand, only one performance of the operation on a subsystem is needed. This is because the member is a basic modelling component. Of course, the actual amount of work is not decreased much. The second problem is that, although a frame is similar to a subsystem, it is difcult to generate various types of subsystem and generating, deleting and moving the frame is not allowed. Therefore, it is not easy to build a model intuitively. The third problem is that grids are used only to input points for column lines and bays. Therefore, just as in the modelling technique using elements, there is a need to add or delete a large number of members again to reect the modication made by inserting a new grid or moving an existing grid.

2.3

Requirements

Up to now, we have reviewed the characteristics of the structural schematic design of tall buildings and conventional modelling techniques and their problems. Here, based on the review, we identied the requirements that a new modelling technique for tall buildings should satisfy. First, a subsystem, which is the conceptual assembly of members participating in a specic type of load transfer, should be treated as a single unit that can be generated, assigned, moved and deleted at once, including the corresponding members and their relations. Second, the operations listed above should be performed with a minimal amount of mandatory data such as the number of bays in each direction, member sections, and so on. This requirement must be satised especially in the case of generation of subsystems whose layout and relation of included members are predened, such as tube, core and outrigger. Third, the location and topology of systems, subsystems and members should be represented based on grids with labels. The relations between grids and the corresponding systems, subsystems and members should be kept for future use. Also, modications made to the grids should be automatically propagated to the corresponding subsystems and members.
Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

94

H. KIM ET AL.

Fourth, the amount of repetitive operations should be minimized through the effective use of repetitive characteristics of tall buildings. For example, for members and subsystems of the same shape, a modication made to one should be automatically propagated to the others. Finally, a designer should be able to organize some members into a group in various ways and maintain the group for future reference and modication of the model. 3. 3.1 MODELLING TECHNIQUE USING SUBSYSTEM

Modelling components

3.1.1 Global grid system A global grid system is a set of grids representing the location and size of the building within the site. In the modelling technique using subsystem, a surface, instead of a point or line, is used as the component of the grid system. Figure 3 shows the global grid system composed of surfaces. To use surfaces as a component of the global grid system, it is required to indicate four corner points or equations of the surface. In this study, however, considering that most buildings are perpendicular to a horizontal line, vertical grids (grid X1, grid Y1 in Figure 3) are dened by means of the end points ((X1, Y1)(X2, Y2), (X3, Y3)(X4, Y4) in Figure 3) of the intersection lines of the vertical grids and horizontal grid (grid X1 in Figure 3). For easy management of the vertical grids that are not parallel to the X- or Y-axis, the vertical grids in a grid set are not permitted to intersect each other. Considering the oors of most buildings are horizontal, all horizontal grids are dened by means of the horizontal surface and represented by height (Z1 in Figure 3). With the grid system composed of surfaces, therefore, a designer can dene points, lines and solids easily by combining grids and ignore a large number of points or lines. Also the designer can manage complicated connections of points to points, points to lines, and lines to lines conveniently.

grid Y 1 Z

Y (X4 ,Y4) X grid Z 1 (X1 ,Y1) z y x Z Y (Z1)

(X2, Y2)

X (X3, Y3)

grid X 1

grid X 1 : (X 1, Y1) (X2, Y 2) grid Y 1 : (X 3, Y3) (X4, Y 4) grid Z 1 : (Z 1)

Figure 3. Global grid system composed of surfaces


Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

SUBSYSTEM FOR STRUCTURAL SCHEMATIC DESIGN

95

This way of dening and using a grid system is similar to the way the designer denes and uses it. Therefore, this grid system will make it easy to dene the boundaries and locations of the members and subsystems. This grid system will also make it easy to dene and maintain the attachment between grids and members or subsystems. Because the shape of a tall building is so complicated, it is not unusual to skip some grids by mistake and to insert or modify the grids later. But even in this case, once the new order of the grids is determined by distance between the grid and the origin, the grid system composed of surfaces will make it easy to divide members optionally, to move subsystems and to modify the attachments between grids and members or subsystems. 3.1.2 Subsystem A subsystem is the conceptual assembly of members that participate in a specic type of load transfer. In this modelling technique, the subsystem is introduced as a modelling component and should play the role of a single unit in modelling operations such as generation, movement and deletion. That is, as shown in Figure 2(a), the designer does not need to generate, move and delete each and every member included in the subsystem. Instead, the designer can generate, move and delete the members included in the subsystem at once. In order to make it possible for a subsystem to play the role of a single unit in modelling tall buildings, the following three conditions should be satised: (1) a-part-of relationship between a subsystem and members should be managed; (2) a subsystem should be generated conveniently; and (3) the topology of members within a subsystem should be managed conveniently. Fortunately, the rst condition can be easily satised by means of giving the identier of the subsystem to the included members. However, the second and third conditions cannot be easily satised. In order to satisfy the second condition, a method to generate a subsystem with a small amount of input data is introduced. In this method, once the predened input data of small amount is given, a subsystem can be generated automatically. The predened input data is as follows: a type of subsystem, the number of bays, grouping of members, member sections and so on. This predened input data contains all the required information for generation of the typical subsystems as follows: frame, core, tube, outrigger, belt truss, braced frame, truss, super column and super beam. These types of subsystem are selected based on Hong et al. (1998). The details of this method of generating a subsystem can be described using the example of Figure 4.

x1 z2

x2

x3 j j 2

x4

x5

1 z1 member 1 j : x3, y1, z2 i i : x2, y1, z1

i i : x4, y1, z1 me mber 2 j : x3, y1, z2

Figure 4. Example of generation of truss subsystem


Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

96

H. KIM ET AL.

In this example, the following input data is required to generate the truss subsystem: and internal shape of subsystem: truss, shape of X; type number bays in x, y and z directions (4, 0, 1); member of grouping: all (all members are included in the same member group); member section (W10X12). Once the above-listed input data is prepared, the number and relative location of vertical, horizontal and diagonal members can be determined. Therefore, the following can be generated automatically: ve vertical members, four top chords, four bottom chords, and eight diagonal members. All the generated members are included in one member group and have member section W10X12 and are given the identier of the truss subsystem. As shown in this example, in the predened input data there is no data concerning individual members. This means that no additional operations for the generation of individual members are required and, consequently, no repeated operations for the generation of the members of the same member section are required. Therefore this method of generating a subsystem with predened input data is very efcient for the generation of a subsystem that has a typical and simple shape. However, in the case of an irregular or complicated subsystem in which some members do not exist or are arranged in a different pattern, some extra operations are required. In this case, the rst thing to do is to generate a typical and simple subsystem that is similar to the irregular or complicated subsystem. The second thing to do is to add new members to the subsystem or to modify or delete some members of the subsystem. Therefore, this method of generating a subsystem with predened input data can be efciently applied to various types of subsystems and, nally, satises the second condition. Lastly, in order to satisfy the third condition, a local grid system is introduced. This is described in the next subsection. 3.1.3 Local grid system The local grid system, just like the global grid system, uses the surface as a grid component. Unlike the global grid system, however, all grids in the local grid system are parallel to the x-, y-, and z-axes. There is one local grid system per subsystem, which is generated implicitly based on the number of bays of the subsystem. Local grids have no coordinates. Instead, each grid is represented by the order determined by the number of grids located before and after. This local grid system is very useful for representation of the topology of members within a subsystem. Figure 4 shows how to represent the topology of members in a subsystem using the local grid system. In Figure 4, because the number of bays in the x, y and z direction is 4, 0 and 1 respectively, the local grid of x1, x2, x3, x4, x5, y1, z1 and z2 is generated implicitly. The end point i of member 1 is then represented as the point (x2, y1, z1). The point (x2, y1, z1) means the point that is dened by intersection of the second, the rst and the rst local grids in the x, y and z direction respectively. Here the second local grid in the x direction (x2) means that there is one grid before the grid in the local x direction. The rst local grid in the y and z direction (y1 and z1) means that there is no grid before the grid in the local y and z direction respectively. The end point j of member 1 is represented as the point (x3, y1, z2), which means the points dened by intersection of the third, rst and second local grid in x, y and z direction respectively. Again, here the third local grid in the x direction (x3) means that there are two grids before the local grid in the x direction. The end point i and j of member 2 are represented as the point (x4, y1, z1) and (x3, y1, z2) in the same way that the end point of member 1 is represented. We can then see that the end point j of member 1 is connected to the end point j of member 2.
Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

SUBSYSTEM FOR STRUCTURAL SCHEMATIC DESIGN

97

From this example, although no coordinates of the local grid system are given, it is found that the topology of members within a subsystem can be represented and managed conveniently using the local grid system. This means the third condition is satised by the introduction of the local grid system. Now, all three conditions listed above can be satised. Therefore, in this modelling technique, a subsystem based on a local grid system can be generated conveniently and play the role of a single modelling unit. 3.1.4 Assembly An assembly is a group of subsystems. An assembly is useful for making some subsystems into a single modelling unit that is used to move or delete subsystems with one operation. For example, in an outrigger system, several outriggers are connected between the interior core and exterior frame or super column. In most cases, the number, shape and layout of the outriggers are so diverse that it is difcult to generate these outriggers as one subsystem using only predened input data such as type, internal shape, number of bays, member grouping, member section and so on. Therefore, rst it is required to generate individual outrigger subsystems one by one. The generated outrigger subsystems can then be selected and grouped into an assembly. Once the outrigger subsystems are grouped into an assembly, it is possible to move or delete all the outrigger subsystems at a time by moving or deleting the assembly. In this study, an assembly is given a unique name or identier and a-part-of relationship between the subsystems and assembly is maintained. Consequently, an assembly is not a temporary modelling component that is generated once and then deleted immediately, but instead a persistent modelling component that can be used later. 3.2 Using modelling components

3.2.1 Assignment In this modelling technique, a subsystem is generated without geometry, that is, the location and size. This is because the subsystem is generated based on a local grid system that has only order of local grids without coordinates. The geometry of the subsystem is determined on completion of the assignment. The assignment is to connect each local grid of a subsystem to the corresponding global grid. Because the global grid system has actual coordinates, the location of the local grids can be determined. In turn, the location and size of subsystem can be determined from the location of the local grids. For example, as shown in Figure 5(a), if local grids of x1, x2, x3, z1 and z2 are connected to the global grids of X1, X2, X3, Z1 and Z2 respectively, the coordinates of the end points of member 1 would be (0, 0, 0) and (5, 0, 0).2 The length of member 1 would be 5 m. On the other hand, as shown in Figure 5(b), if local grids of x1, x2, x3, z1 and z2 are connected to the global grids of X2, X3, X4, Z1 and Z2 respectively, the coordinates of the end points of member 1 would be (5, 0, 0) and (8, 0, 0), and the length of member 1 would be 3 m. However, the connections of members 1 and 2 would be maintained without any changes. This would be true for all the other connections of the members. As shown in the example, if changes are made to the assignment, the geometry of members is modied automatically. This is because the end points of members are dened with local grids. Thus, if the local grids are assigned to the other global grids, the location and size of all the members for
2

These and the following coordinates are based on the assumption that local grid of y1 is connected to the global grid Y1 to which the distance from the origin is 0 (zero).
Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

98
x1 z2 x2 x3 local grid

H. KIM ET AL.
x1 z2 x2 x3 local grid

z1

z1 X2 X3 X4 Z2 5m

X1

X1

X2 X3

X4 Z2 5m

global grid

Z1 4m

5m (a)

3m

global grid

Z1

5m

3m (b)

4m

Figure 5. Assignment

which the end points are dened with the local grids are modied automatically according to the actual coordinates of the newly connected global grids. However, whatever changes are made to the assignment, the topology of members is not modied. This means that if some members are moved to other locations then the neighbouring members are also moved automatically to the corresponding neighbouring locations. Therefore, there is no need to perform additional and repeated operations for each individual member. These can be equally applied to the initial assignment and cancellation of the assignment. Finally these make it possible for a subsystem to be treated as a single modelling unit just like a member. 3.2.2 Repetitive assignment Often some subsystems in a structural scheme have the same shape, that is, some subsystems have the same layout of members that have the same sections. In this modelling technique, all subsystems must be generated and then assigned to determine the location and size. Therefore, if there are two or more subsystems of the same shape, one subsystem can be generated and then assigned repeatedly as needed. There is no need to generate the same subsystem repeatedly. Using this method of generation and assignment, many subsystems of the same shape can be modelled conveniently and the same modications to many subsystems of the same shape can be done with just one modication to the generated subsystem. Also, if the generation and assignment are utilized skilfully, it is possible to model similar subsystems with minor differences easily with a minimum amount of work. 3.2.3 Overlap of members Because tall buildings are composed of many subsystems and the subsystems are composed of many members, there exist members that are included in many subsystems at the same time. Therefore, if a subsystem already exists and the other subsystem is assigned, overlap of members may occur. In this case, as a general rule, the overlapped or overlapping member is deleted. However, in this modelling technique, none of them are deleted. Depending on the designers decision, one of them is set to be valid and the others are set to be invalid. If the subsystem that includes the valid member is deleted or moved, the valid member is deleted or moved and one of the remaining invalid members is set to be valid. If the subsystem that includes an invalid member is deleted or moved, the very invalid member is deleted or moved and no other changes occur.
Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

SUBSYSTEM FOR STRUCTURAL SCHEMATIC DESIGN

99

(a) in case of deleting

(b) in case of not deleting

Figure 6. Results of deleting and keeping the overlapped or overlapping members

Figure 6 shows the difference in result between deleting and keeping the overlapped or overlapping members in an example that is composed of a frame subsystem and braced frame subsystem. In the case of deleting, when the braced frame subsystem that includes the valid member is deleted or moved, as shown in Figure 6(a), the frame subsystem becomes the defective subsystem. To restore the defective subsystem, it is required to enter the deleted members again one by one. On the other hand, in the case of keeping, although the braced frame subsystem is deleted or moved, the frame subsystem is restored to its original without any defect, as shown in Figure 6(b). Management of the overlapped or overlapping members in this way will help the designer to treat the subsystem as a single unit just as was conceived. 4. EXAMPLE

In this study, we developed a computer program that implements the modelling technique using a subsystem. The program, written in Microsoft Visual Basic 6.0, reads the input data prepared in accordance with the proposed modelling technique and builds the structural analysis model of tall buildings. The program stores the analysis model in a database and then generates an ETABS input le for structural analysis from the model stored in the database. We applied the program to an example building. The example is an ofce building with 42 storeys and has a total height of approximately 170 m above ground. The example building has a central concrete core with exterior steel braced frames. The plan is shown in Figure 7. Here we generated two structural schemes. In the rst scheme, we placed the X-braced frame on the bay 1 (circled in Figure 7). Then, based on the assumption that the rst scheme turned out not to resist wind load sufciently, we generated the second scheme. In the second scheme, we moved the X-braced frame of the rst scheme to bay 2 and added another X-braced frame to bay 3. In this section, we describe the details of the global grid system, subsystem, local grid system and assignment related to the above operations. 4.1 Generation of X-braced frame subsystem

The rst stage of modelling the rst scheme is to input the global grid system, material and member sections. The second stage is to model the central core and the exterior steel frames with no braces. The details of these operations are omitted. The next operation is to generate the X-braced frame subsystem and locate it on bay 1 (Figure 7). For this purpose the number of bays, number of storeys, details of member grouping and its section should be prepared. The X-braced frame subsystem that does not have location and size is then generated using this input data. Figure 8 shows the generated X-braced frame subsystem. The grids in Figure 8 are not global grids but local grids. The members in this subsystem are dened by means of
Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

100

H. KIM ET AL.
54 000 90 00 Y4 3 Y3 1 Y2 2 Y1 X1 X2 X3 X4 X5 X6 X7 11 750 33 000 95 00 11 750 90 00 90 00 90 00 90 00 90 00

Figure 7. Plan of the example building

z43

z42

z2

z1 x1 x2

Figure 8. Elevation of the X-braced frame subsystem

local grids. For example, the start point and end point of member 1 (circled in Figure 8) are dened as (x1, y1, z2) and (x2, y1, z2) respectively.3 The edited and simplied input data for the X-braced frame subsystem is represented in the second row of Table 1. The next operation is to assign the local grids of the X-braced frame subsystem to the corresponding global grids. The local grids x1, x2, and y1 are assigned to the global grids Y2, Y3, and X1 respectively; the local grids z1, z2, z3, . . . , z42, and z43 are assigned to the global grids Z1, Z2, Z3, . . . , Z42, and Z43 respectively. The input data for this assignment is represented in the third row of Table 1. The left side of Figure 9 shows the edited image that ETABSIN, a pre-processor of ETABS,

The plane on which the subsystem is placed is the local grid of y1.
Struct. Design Tall Spec. Build. 13, 89103 (2004)

Copyright 2004 John Wiley & Sons, Ltd.

SUBSYSTEM FOR STRUCTURAL SCHEMATIC DESIGN

101

Table 1. Input data for the X-braced frame subsystem Input value (notes) Frame Braced frame (type of subsystem), 1, 42 (number of bays, number of storeys), X (layout of braces), Girder, All, G1 (all girders in this subsystem are included in the member group G1), Column, All, C1 (all columns in this subsystem are included in the member group C1), Brace, All, Br1 (all braces in this subsystem are included in the member group Br1) Exterior braced frame 1 (name of subsystem assigned to bay 1), Y2, Y3, X1, X1 (global grids to which the vertical boundary local grids are assigned), 1, 42 (storeys to which the horizontal boundary local grids are assigned)

Assignment

produced using the ETABS input le generated by the program from the results of this modelling. In the left of Figure 9, because the local grids x1 and x2 of the subsystem are assigned to the global grids Y2 and Y3, the width of the subsystem is 9500 mm, which is the distance between the global grids Y2 and Y3. Consequently, the length of member 1 in Figure 8 is 9500 mm, because member 1 is located between the global grids Y2 and Y3. 4.2 Move and addition of the X-braced frame subsystem

Here we assume that, based on the review of the analysis results, the rst scheme is not sufcient to resist wind load. Therefore, we will generate the second scheme based on the rst scheme. The rst operation for the generation of the second scheme is to move the X-braced frame subsystem to bay 2 (Figure 7). Modication only to assignments will result in moving the subsystem. In Figure 7, bay 2 is between the global grids Y1 and Y2. Therefore, the local grids x1 and x2 of the X-braced frame subsystem are assigned to the global grids Y1 and Y2 respectively. The size of the subsystem is then automatically changed to 11,750 mm, which is the distance between the global grids Y1 and Y2. The last operation is to add a subsystem of the same type to bay 3. Because the type of new subsystem is the same as the subsystem on bay 2, there is no need to generate the X-braced frame subsystem again. Only additional assignments will add the subsystem. At this time, the local grids x1 and x2 of the X-braced frame subsystem are assigned to the global grids Y3 and Y4 respectively. The new subsystem is then added on bay 3. Again, the size of the subsystem is 11,750 mm, which is the distance between the global grids Y3 and Y4. The italic parts in the third and fourth rows of Table 2 represent the edited and simplied input data for modication and addition of assignments of the subsystem respectively. The right part of Figure 9 shows the edited image that ETABSIN produced using the ETABS input le generated from the results of moving and addition of the X-braced frame subsystem. 5. CONCLUSIONS

The purpose of this study is to propose a new modelling technique that allows the structural designer to model the structural scheme of a tall building in the same way he or she conceives the scheme in mind. In order to achieve this purpose, rst, we introduced a global grid system, subsystem and global grid system as modelling components. We then introduced an assignment as a way of using these modelling components. In the proposed modelling technique, a subsystem is the assembly of members that participate in a specic type of load transfer. The subsystem has a local grid system with no coordiCopyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

102

H. KIM ET AL.

Table 2. Input data for the X-braced frame subsystem Input value (notes) Frame Braced frame (type of subsystem), 1, 42 (the number of horizontal and vertical bays), X (layout of braces), Girder, All, G1 (all girders in this subsystem are included in the member group G1), Column, All, C1 (all columns in this subsystem are included in the member group C1), Brace, All, Br1 (all braces in this subsystem are included in the member group Br1) Exterior braced frame 1 (name of subsystem assigned to bay 2), Y1, Y2, X1, X1 (global grids to which the vertical boundary local grids are assigned), 1, 42 (storeys to which the horizontal boundary local grids are assigned) Exterior braced frame 2 (name of subsystem assigned to bay 3), Y3, Y4, X1, X1 (global grids to which the vertical boundary local grids are assigned), 1, 42 (storeys to which the horizontal boundary local grids are assigned)

Assignment

Assignment

Z42

Z42

Z38

Z38

Z34

Z34

Z8

Z8

Z4

Z4

Z0

Z0

Y4 Y3 Y2 Y1

Y4

Y2 Y1

Figure 9. Edited results of modelling in ETABSIN

nates. Each local grid is represented by the order determined by the number of grids located before and after. The end points of the members included in the subsystem are dened by means of the local grids. The local grid is assigned to the corresponding global grid that has actual coordinates. The location and size of the subsystem are determined using the global grids that can be accessed through these assignments.
Copyright 2004 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 13, 89103 (2004)

SUBSYSTEM FOR STRUCTURAL SCHEMATIC DESIGN

103

In conclusion, through the development of a computer program and its application to the example building, it has been demonstrated that (1) the proposed modelling technique provides a way to generate, move and delete a subsystem just like a single modelling unit and (2) modelling in terms of subsystem results in an intuitive and convenient modelling of structural schemes of tall buildings.
ACKNOWLEDGEMENT

Financial support provided by the Ministry of Construction and Transportation (Korea) under grant no. 99-Architecture-10 is gratefully acknowledged.
REFERENCES

CTBUH (Council on Tall Buildings and Urban Habitat) Committee 3. 1995. Structural Systems for Tall Buildings. Tall Buildings and Urban Environment Series. McGraw-Hill: New York. CSI (Computers and Structures Inc.). 2000. SAP2000 Analysis Reference, Vol. 2. Berkeley, CA. Hibbitt, Karlsson & Sorensen, Inc. 1994. ABAQUS/Standard Version 54 Users Manual Vols I, II. Pawtucket, RI. Habibullah A. 1995. ETABS Users Manual Version 6.1. CSI (Computers and Structures Inc.): Berkeley, CA. Hong S. 1998. A study on the investigation for structural analysis and design of hybrid system (in Korean). Report no. STRESS-97-10, Advanced Structure Research Station: Seoul.

Copyright 2004 John Wiley & Sons, Ltd.

Struct. Design Tall Spec. Build. 13, 89103 (2004)