Stage 1: Develop the model of CAD complexity

The first stage of the project delineates the criteria to be used in the evaluation of the complexity of a CAD model. This includes:

identifying the criteria used in determining complexity, e.g., numbers, repetition, differentiation, symmetry and hierarchy, as described earlier.
identifying the sources of CAD complexity;
identifying subsystems of CAD organisation.

We provide here an informal discussion of the issues we have identified to date, with the task in this stage of the project to incorporate these criteria and elements into a formal model.

Sources of CAD Complexity

Three possible sources of CAD complexity can be distinguished: CAD data, or the information content of the CAD model; CAD structure, associated with the CAD model file organisation; and properties associated with application software functionality. These are described in greater detail below.

CAD data

Items in this category include the actual components of the CAD model, consisting of shapes and annotations:

Shapes or figural information includes vectors and inserted raster images. Graphic symbols, such as North points and hatch blocks may also be considered as belonging to this category.
Annotations include dimensions, other numeric data, and drawing notations, such as headings, notes or title blocks. Also included is documentation to facilitate communication, but not necessarily part of the model.

Possible measures of CAD data include quantities such as numbers of objects, numbers of different objects, shape complexity and properties other than shape, including end points and scale factors. In measuring the complexity of notation, fonts, scales and other properties to meet different presentation requirements are relevant. Further distinction can be made between data that is part of the design and data that is part of the presentation of the design, although this is not crucial to the main ideas expressed in this proposal.

CAD structure

Items in this category include file variables and inter-file variables:

File variables include functions that support differentiation and organisation of data within individual files, such as colours, line types, layers and complex objects. Measurement of complexity may be similar in principle to that employed to measure CAD data.
Inter-file variables include functions that support organisation and referencing of data from file to file. Typically files are of the same type. Different types of files normally require interfacing and sometimes translation of data from one form to another.

CAD model structural elements, of the kinds briefly referred to above, are highly capable of analysis and decomposition by reference to CAD functions. Many support hierarchical organisation. This renders them suitable for use in developing a description of CAD complexity based on organisational subsystems that follows the next section.

Application software

CAD software products vary significantly, in extent and manner of supporting CAD production, editing and display functions. This influences not only how things are done, but also what is done. Two CAD packages used extensively in teaching in the Faculty of Architecture are AutoCAD and ArchiCAD, which use significantly different approaches to modelling. The potential exists to compare these two different approaches in terms of the CAD models produced.

Modelling approaches can also vary within a single piece of software. For example, one may choose to produce a 3D representation of an object using wireframe, surface or solid modelling techniques, all of which generate models of different purpose and complexity.

Subsystems of CAD organisation

It is possible from this point to identify and describe various CAD model hierarchies. We have categorised five subsystems of CAD organisation on the basis of organisational processes. There are four categories listed in order from low to high level of organisation. This order broadly coincides with least to most technically advanced, and least to greatest potential to provide high levels of model realisation at high levels of efficiency. A fifth category (pre-sets) is included to incorporate peripheral factors which are integral to the communication of a model, but are not part of the model.

Object Differentiation

Differentiation is an elementary, non-hierarchical structuring that implies grouping within a model file. It may be achieved by varying one or more appearance characteristics, enabling a person to visually identify different classes of element, or particular parts of a model. The main characteristics include:

Object differentiation:
varying shape, colour, line type

Shape: Different shapes can represent different objects. Proximity and orientation of shapes to each other or to some ordered form, such as a grid, may also be employed to assist recognition.
Colour: Variation in hue, saturation and other object colour properties enable different classes or conditions of objects to be visually distinguished regardless of shape. Some CAD systems (e.g., AutoCAD) allow association of colours with pens, enabling printed line thickness to be modulated by means of colour.
Line type: Particular repeating patterns of shorter lines, dots, symbols and spaces are enable the visual distinction of linear objects such as centrelines, especially when these overlay other objects.

Object Grouping

Grouping is a method of relating objects within a model file, to act as one, by means of one or more of the following functions:

Informal grouping		Polyline
Object class grouping		Layer hierarchies: Admin_bldg.level1.architectural Admin_bldg.level1.structural Admin_bldg.level1.mechanical ... Admin_bldg.level2.architectural ... Physics_bldg.level1.architectural ...
Named object grouping		"Piping assembly": valve1, pump1, elbow1, pipe1, ...

Informal or unnamed grouping: Most CAD software supports informal groupings, such as AutoCAD polylines (a series of end-to-end objects combined into one linear object), created initially as a continuous string of lines, or converted after initial placement.
Formal object-class grouping: Many CAD applications support open ended class groupings by permitting designers to assign and reassign elements to named layers. Use of layers enables designers to control display properties such as colour and line type, and selection properties of different classes of design elements. Layer hierarchies, of two or more levels, can be created by the use of layer name conventions.
Formal or named-object grouping: The ability to group objects and build complex hierarchies of nested objects has long been associated with drawing applications. Use of named 'blocks' in AutoCAD to modularise a design enables designers to achieve significant economies of file size and to simplify the production and editing of a CAD model. Different blocks need have nothing in common. However, it is possible to build up complex hierarchies based on consistent subcomponents.

File Grouping

CAD models requiring more than a few hours to produce, or more than one author, generally warrant organisation of objects and their representation into multiple files. ArchiCAD supports file grouping by means of a class of files called Library files, which are an alternative to the AutoCAD use of blocks. External file referencing (e.g., in AutoCAD) allows any model file to be 'attached' to any other. On large projects external references may be used instead of blocks in order to eliminate duplication. Hierarchies of attachments can be developed with validation procedures to prevent circular referencing. The use of external references versus blocks is one of the key issues that we plan to address in this study.

Application Grouping

There is a general trend in the evolution of computer applications toward increasing complexity. Indicators of this trend include the number of additional functions with each new version, their occupation of progressively more disk space and ever increasing demands on memory and processing. Associated with the general growth is a growing mobility of data between associated applications.

Linking CAD software and data base or spreadsheet applications is now a common technique. A recent trend toward the development of more sophisticated dynamic models, capable of simulating complex facilities, will further stimulate multiple file and application structuring, making this a very fertile source of new hierarchies.

Pre-Sets

Selection, simply as an activity, is not a factor in CAD complexity. However predefined settings to display or suppress the display of objects facilitate communication of a model of substantial complexity. While not part of the model, pre-sets act as agents for the expression of the organisational hierarchy contained within model files. Being both separate and relevant to CAD complexity we therefore consider them as part of an overall strategy to manage complexity. Examples of pre-sets in CAD packages include the following:

Filter lists: Filter options permit designers to limit the display of different classes of object, based on colour, line type and other more abstract properties, such as location, current in-use status, or their association with externally referenced files.
Layer Sets: Combinations of layers files may be defined, named and saved, as named layer sets.
Named Views: Some software packages allow designers to save combinations of viewpoint coordinates, orientation and zoom settings in the form of named views.
Other Display settings: Other important pre-sets include page layout functions, e.g. ArchiCAD Plotmaker and AutoCAD Paperspace.