BIM-Enabled Conceptual Modelling and Representation of Building Circulation

2.1 Circulation within Buildings

In the AEC-FM field, the domain of building circulation modelling and its representation is of special interest to several participants. Architects and designers need to express their design intent using spatial planning programs; building owners and occupants need to foresee circulation patterns within buildings; and design evaluators need to assess building circulation according to intended circulation. In addition to some qualitative factors of circulation representation, we emphasize two aspects of circulation representation methods. The first is to explicitly describe the spatial topology of circulation between two spaces, and the second aspect is to denote measured walking distances in a precise manner. Obviously, many people involved in a building design project are familiar with conventional topological graphs on a building floor plan. For example, a bubble diagram for a space program denotes the adjacency of programmed spaces or zones in the early phase of design. It is a well-known topological graph, and one of the pragmatic, metric, graph-based approaches; its algorithm was developed by [7].

2.2 Circulation Research in a Spatial Morphology Area

Within the AEC-FM domain, access and the topology of spatial networks have their own fields of study, called ‘spatial morphology’ and ‘space syntax’. The general idea is that spaces can be broken down into components, analysed as networks of choices, and then represented as maps and graphs that describe the relative connectivity and integration of those spaces. This has since grown to become a tool used around the world in a variety of research projects and design applications in the fields of architecture, interior design, urban design, planning, transport, and so on. In general, the analysis uses one of the many software programs that allow researchers to analyse the graphs of one (or more) of the primary spatial components. In this paper, some of the interesting issues in space planning within buildings are tackled based on these related approaches and applications, but with the primary focus on the conceptual modelling and visualization of circulation. In addition, the ‘spatial component’ here means the space object defined in a given BIM model.

2.3 Graph and Information Visualization

The visualization of relational data is important, but more fundamental and critical in display efficiency is the way in which information is visually translated and encoded. The value of visualizing such data emanates from the fact that the perception of visual information and patterns is an enhanced intrinsic characteristic in people, as opposed to the exhaustive search for corresponding patterns in alphanumeric datasets. Graph theory-based visualization, as shown in Figure 2, is an example of the visualization of relational data that facilitates navigation through the visual representation of that data. It is an increasingly significant area of research that has been widely implemented in many applications in order to view large and complex networks.

OPEN IN VIEWER

Figure 2.

Example representation of circulation paths using a graph structure where each node represents a space object

Palmer and Rock [8] demonstrated a basic aspect of node-link diagrams through an experiment. They introduced the gestalt principle of connectedness as a key issue that supersedes proximity, size, colour and shape, perceptually. People tend to perceive nodes that are linked by line drawings as belonging together, much more than nodes that are closer in proximity or share the same colour or shape. The results from this experiment strongly support the value of the node-link diagram -where edges are drawn as lines between nodes - as an effective method of representation.

3.1 Design-oriented Representations

Werner et al. [11] classify navigational knowledge using route graphs in terms of various agents. The graph-based approach to circulation and navigational knowledge has been acknowledged by many researchers because of its simplicity and efficiency. Architects use graphs in the form of freehand sketches, such as bubble diagrams, to represent space-to-space relationships in the early phase of design. Pioneering researchers seeking to adopt graph theory in space planning include Eastman [12, 13], who developed a graph theory-based system for floor plan generation in the early design phase, and Roth, who researched a method to convert adjacency graphs to a floor plan [14].

The previous systems represent information based on the topological relations within spaces in early concept design phases. Topologically represented graphs provide spatial adjacency and connectivity information. Precise metric information, such as area, volume or distance, is unsuited for representation by such topological graphs. This detailed geometrical data could be represented by a graph structure, mainly controlled by the designer's intentions. Hashimshony described this process as three phases of architectural design [15]: a programmatic phase, a topological phase and a geometrical phase.

3.2 Analysis-oriented Representations

We focus on a given building model, where graphs are generated based on the predefined geometry of the model rather than the prediction or observation of agents' movements. Literature from the architecture, environment and behaviour, space syntax, and spatial cognition areas provides insights on issues of interest, such as reflecting human behaviour in graph-based representations, as well as providing a basis for enhanced perceptual and cognitive aspects of graph visualization. Bafna introduced the visibility-based isovist and its applications [16], while Turner devised an algorithmic definition of the axial map for an isovist [17]. An isovist denotes the area visible from a given location in a floor plan, and its line-based representation is an axial map. We observe that people generally move from one place to another following these axial lines because they are visible paths, and they tend to walk following an easy and efficient path. In other words, the most ‘human-like’ walking trace will follow the shortest, easiest and most visible path(s). Calculating walking distances based on these path lines would provide more accuracy as well as better visualization. Accordingly, [7] has researched and implemented just such a metric graph-based circulation representation method. This is shown in Figure 3, which also includes other types of graph representations. This is one of the BIM-enabled circulation visualizations.

OPEN IN VIEWER

Figure 3.

Various graph-based representations of circulation between two spaces. The bottom-right one shows the metric graph structure ([7], modified).

4.1 Fundamental Specification of a Circulation Graph

There are several useful aspects of building circulation reflecting various of the representation methods and approaches reviewed in the former section. Given the fact that BIM models provide rich spatial and technical data for generating circulation information, we utilize the previously presented issues in order to informally describe specifications of BIM-enabled circulation graphs, as follows:

4.2 Circulation Information derived from an IFC

Space information is required in order to handle a building's circulation issues. The space information can be represented by name, geometric shape, boundary walls and doors. A BIM-enabled design process allows architects to define this information explicitly. Lee [5] shows those objects and their properties according to four different criteria: basic (given) properties, relations, computed relations, and computed and derived properties. BIM tools utilize this data model to consider data sharing and the transferring of building information. Figure 4 provides an example of the description of the relationship between space and boundary design components in terms of the IFC schema. The IfcRelSpaceBoundary entity defines the physical or virtual delimiter of a space as its relationship to the surrounding building elements. The entity is defined as an objectified relationship that handles the element-to-space relationship by objectifying the relationship between an element (IfcElement) and the space (IfcSpace) it bounds.

OPEN IN VIEWER

Figure 4.

Simplified diagram illustrating how IFC data contains a spatial topology: The IfcRelSpaceBoundary entity defines relational information between spaces

As shown in a simple example model in Figure 5, a space called ‘Conference Room’ is enclosed by four walls and three doors. This example is represented as physical data in Table 1. It shows a physical file format based on ISO/STEP Part 21 [18, 19]. The physical data can provide information about space and boundary elements, and also spatial connectivity data - that is, how a space is adjacent to other spaces - in order to retrieve a spatial topology.

OPEN IN VIEWER

Table 1.

IFC schema and its instantiation for circulation-related IFC entities reflecting the example in Figure 5

#1906=IFCRELSPACEBOUNDARY(#1866,#1332,#1905,.PHYSICAL.,.EXTERNAL …); #1866=IFCSPACE(‘Conference Room2019,.ELEMENT.,.INTERNAL …); #1332=IFCDOOR(‘Single-Flush:36″ × 84″:36″ × 84″:1100′ …); #1961=IFCRELSPACEBOUNDARY(#1866,#1271,#1960,.PHYSICAL.,.EXTERNAL …); #1271=IFCDOOR(‘Single-Flush:36″ × 84″:36″ × 84″:1098′ …); #7463=IFCRELSPACEBOUNDARY(#1866,#4240,#1937,.PHYSICAL.,.EXTERNAL …); #4240=IFCDOOR(″Single-Flush:36″ × 84″:36″ × 84″:4988′ …); ……

OPEN IN VIEWER

Figure 5.

An example of the entity “IfcRelSpaceBoundary” and related entities

5.1 Definition of Circulation

Inter-space circulation simply means a state of movement from a particular space to another. All these spaces are in a building - that is, within a built environment - and they are defined by users utilizing space objects in the BIM-authoring tools. A route is a concatenation of directed Route Segments from one place to another [11]. A route segment consists of two spaces - a start and a target space - which are connected by a path. Mathematically, the number of traversal types in general for all paths between two nodes is infinite in number. In this paper, all paths P are reduced to closed sub-paths P′ and other iterated paths P″. A path has no repeated edges or vertices.

Using spatial topology data defined in a given BIM model (refer to Figure 4), circulation can be represented in a graph structure - as shown in the example in Figure 2 -and especially as a digraph, which has movement direction. Put simply, this kind of conceptual modelling concerns the relations between space objects, which are represented by the nodes in its graph structure, and a certain group of edges represents a circulation path. Based on the example in Figure 2, the circulation path and circulation elements could be defined as follows.

A circulation path P means a valid path:

P = Pall - Pinvalid

where Pinvalid denotes closed sub-paths, repeated and cycle paths.

Valid paths from A to G:

P1 = <A, h, B, i, C, l, F, o, G>

P2 = <A, h, B, j, D, m, G>

P3 = <A, h, B, i, C, k, E, n, F, o, G>

Circulation C between A and G consists of all valid paths:

C = <P1, P2, P3>

5.2 Circulation Rules and Rule-compatible Circulation

For architectural design, several rules regulate numerous paths between two spaces. They are called ‘circulation rules’. Well-known rules exist for certain types of buildings based on standard manuals and guides, such as a generic building design guide, a specific building design guide (e.g., hospitals, government office buildings), a safety-related code, a code for disabled people, general design guides suggested by architects and researchers, and so on. Our emphasis in building design is to have paths that not only meet all the rules but which are also efficient. ‘Efficient paths’ has a number of qualitative and quantitative interpretations. However, from the computational perspective, this means “the shortest path” between two spaces. There could be several more efficient paths, such as the simplest path, the path most recommended by a designer, the path which is the most visible to (or most appropriate for) a given agent, and so on.

Circulation rules can employ any definition of spatial connectivity, but we handle them as a base point of the conceptual modelling for the use of circulation representation. One of the most important applications of the circulation representation will be the analysis of building circulation. The following are some circulation rule examples from various references available on the Internet:

5.3 Circulation Agents

As shown in 5.2, circulation rules in natural language describe specific spatial relationships, and can be formalized in a certain way based on the relationship between space objects. Figure 6 shows this agent-oriented relationship using the Entity-Relation (ER) schema.

OPEN IN VIEWER

Figure 6.

Entity-Relation model representing relationships between a person and a space

The class ‘PERSON’ in Figure 6 can be defined with various properties. Those person entities are the agents traversing each space. Each person has their own access level in a building, and based on their organization and modifier, their additional conditions and roles for accessing spaces are determined and regulated. Natural language sentences contain some simple person-noun descriptors, but each person-noun should be translated into an exactly defined person-entity. As this paper's scope lies within the scheme of IFC, the person-entity type is not supported by the IFC as the data structure of BIM. Every person-subject rule should be translated into space-subject sentences based on a person's given attributes. For example, a person's access level determines all those spaces that he/she can access. According to pragmatic project developments, most agent-oriented properties could be migrated into space objects.

5.4 Role of Intermediate Spaces in Circulation

No person entities have been defined in the building information model scheme, as mentioned. For this reason, the spaces and related circulation rule-oriented model are shown in Figure 7. The basic unit of circulation consists of at least one each of start space (SS) and a target space (TS), performed by a person based on his/her specific task that required the person to change his/her location. The movement may require zero, one or more intermediate spaces (ISs). Logically, a circulation is a sequence of direct accesses of those two spaces, SS and TS. However, in most building environments, intermediate spaces and their various conditions determine a certain circulation rule. This is an elementary fact for representing a basic circulation rule in terms of an intermediate class; the different types of intermediate space connections give rise to the different general circulation types described in Figure 7.

OPEN IN VIEWER

Figure 7.

A circulation rule involves at least two space objects as a SS and TS, and has intermediate space objects in a many-to-many relationship

In terms of the relation between SS-IS-TS, at least two or more space objects will be involved in a single circulation path. The topological relationship of these spaces can be defined as in Figure 8. In circulation rules, most constraints and rules are generated by intermediate spaces and their properties that define their conditions, as well as the properties transferred from the person-entity (see Figure 6).

OPEN IN VIEWER

Figure 8.

Types of inter-space circulation based on the topological relationship: SS-IS-TS

Based on this kind of graph structure, architects usually analyse the spatial network in terms of the relationship of building spaces. The most well-known concept is ‘spatial depth’. The depth of one space from another can be directly measured by counting the intervening number of spaces between two spaces [16].

The number of ISs defines the topological spatial depth between the SS and the TS. In this structure, spatial depth D is easily acquired, as follows:

D = nIS + 1

In this paper, although our focus is on the conceptual modelling and representation of circulation in terms of BIM -as well as a topological spatial network graph - a metric spatial graph should be considered for property modelling. Contrary to the topological graph, if we need to calculate the metric distance from SS to TS, all the circulation paths should have a distance property based on their metric graph edge weight, which simply means its length. For instance, the metric distance M is acquired as follows [7]:

M = Σ ( dIS ) + dSS + dTS

where d is the distance (length) of each direct access path within spaces.

Additionally, the cardinality of the circulation-involved spaces is important to retrieve their relations and evaluation results. For example, as described in some circulation rule examples in Section 5.2, SS or IS can be one or many cardinalities. In all cases, a circulation model should be satisfied with a given number of space objects for the appropriate handling of the rules. Table 2 overviews this relationship.

OPEN IN VIEWER

Table 2.

Instance number of Start Space (SS) and Intermediate Space (IS) for accessing a Target Space (TS)

Num of SS	SS	IS	TS	Evaluated Result
At least one	1	0	1	TS is directly accessible from SS
At least one	1	1	1	TS is accessible only through an IS
At least one	1	*	1	TS is accessible from SS
Any (all)	*	1	1	TS is accessible only through an IS from all SS
Any (all)	*	*	1	TS is accessible from all SS

5.5 Generalization and Conceptualization of Building Circulation by Translating an Entity-Relation into Space Objects

In the ER modelling technique, a circulation rule involves the space classes SS and TS and potentially many ISs, as shown in Figure 9. A person traverses space objects and a person determines his/her own accessible security level based on his/her access level. Thus, a given person-based rule could be translated to a space object-based rule. For instance, a person class ‘Nurse’ may access all public areas and restricted areas in all spaces of a certain kind of healthcare unit. ‘Nurse’ has a role of ‘traverse’ within spaces, and it could be translated to the role ‘accessible from/to’ of all spaces he/she can access. From the perspective of agent-role relations, another modelling technique called ‘object-role modelling’ has been popular [20]. The main entities of circulation rules have their own rules, and could be represented as in Figure 9. After translating non-IFC entities in this ER model into the relation between associated space objects, it could be the conceptual and generic part of the building circulation representation. In Figure 10, the top layer shows this conceptual model layer that is derived from various types of circulation rules. In this layer, all issues should be translated into space objects without persons or other different types that are not defined in an IFC scheme. The bottom layer in Figure 10 is a BIM model instance layer, and it depicts an actual given model that can contains any type of space object translated in the top layer. The middle layer would be the implementation layer, which is in charge of mapping between the conceptual model and any given model instances.

OPEN IN VIEWER

Figure 9.

ER model for a building circulation rule: a generalized approach to building circulation and rules

OPEN IN VIEWER

Figure 10.

Three layers of the conceptual modelling of building circulation for its applications