Modelling structure is better performed with a solid modeller, if we wish to be as explicit as possible in our modelling. A structural model can be used as a spatial database. If we wish to ask questions about the physical properties of a structure, then we must use solid modellers. In addition to the engineering properties of structures, solid modellers can also be used to answer questions about the economics of construction. For historical reasons, the construction industry has preferred to use surface modellers, but I argue for the advantages of solid modelling in the archaeological study of construction.
The first benefit of a structural model is simply to generate awareness that one is making an assumption about how elements are related within the structure. For example, one must model how two walls join together, even though that junction will be invisible under most circumstances. One ought also to be concerned more with the physical properties of the structure, rather than simply its appearance, and if physical properties are used, they should in some way be attached to the element in drawing.
One example is the Pompeii Forum Project, which is carrying out a fresh standing buildings survey of the east side of the town's forum, and producing a surface model in AutoCAD, because 'the existing plan is simply descriptive and does not attempt to present evidence of chronological or constructional phasing, building materials, repairs, masonry seams, or other construction details upon which conclusions can rest' (Dobbins 1996).
There are strong arguments to be made here for the use of a solid modeller. First, there is no logical coherence to the structural elements in a surface model, which would be found in a solid model. In a surface model, an element, such as an ashlar block, would be represented as a collection of faces. However, there is no internal relationship between those faces, and it is incumbent upon something external to the model, the user, to remember that those faces are structurally intrinsically bound up with one another.
Second, the physical properties of an element are directly part of that element in a solid model. For a surface model, the data could often be attached to the drawing, but as the structural element is represented by a number of drawing elements, it is not clear which, or all, of these elements should be given that data. Again, it is incumbent upon the user to find a way of discovering which drawing element holds the data and, if all do, to ensure that the data are not changed for some elements without the rest. It is possible to link a block of entities together into a higher level of data entity, and link to that entity. It is intuitively easier, and gives greater integrity, if a solid modeller is used for this.
In addition to linking data to the entity itself, it is possible to link the model to an external relational database. This is of great benefit in archaeological recording, where there might be a great deal of information recorded for an object, of which the three-dimensional model is only one element. This facility has not been exploited in the construction of structural models in archaeology. Such data integration was recommended by a Royal Institute of Chartered Surveyors committee set up to research the use of CAD in quantity surveying, as it reduces data redundancy and inconsistency, and should lead to more accurate bills of quantities (Atkin et al. 1987, 58). It would add greater integrity to the database and to the model if a solid model were used rather than a surface one.
That the model might contain more information than simply its appearance has already been suggested by Harrison Eiteljorg II (1993). Taking advantage of the ability of some CAD packages to have a very large (sometimes effectively infinite) number of layers, which can be given names, Eiteljorg has suggested that the each layer contain a specific set of data, and that the nature of that data be declared in the name. The idea of layering conventions is already to be found in other areas of CAD use such as architecture, and gives the user much greater access to elements in the drawing, both because it is obvious what a layer will contain, and it is possible to group layers together using search criteria within the software. The layering convention was devised using a surface model, but it would add greater integrity to the model if a solid modeller was used.
This is an example of Eiteljorg's (1993)
convention, applied to a model of the Acropolis entrance. The
layer:
PTPCWFXA-0489-0478 would contain all elements which were:
A database is defined simply as 'a collection of related data' (Elmasri & Navathe 1994, 2). The CAD package can be seen as acting together as a database management system. The CAD package provides access to the data, while the entity data provides the data model (the graphical information comes from the drawn entity and the non-graphical information from the entity's layer name). The high-level conceptual data model to which this is closest is the Entity-Relationship (ER) model (Elmasri & Navathe 1994, 39), although the relationships between the entities are poorly defined. Each letter position in the layer name provides an 'attribute' which describes the 'entity'. Several drawing entities can share the same layer name, but they are further distinguished by their spatial location and extent. It is in principle possible to have two entities occupying the same space with the same layer name (i.e. which are described as identical by the model but which are distinct), which would violate the concept of normalization (Elmasri & Navathe 1994, 407), but this is not a necessary requirement for an ER model.
Because a structural model is attempting to represent a real structure rather than merely its appearance, it should be possible to interrogate the model as to some or all of its physical qualities, and this is possible only with solid models. Armed with information about the material from which an entity or group of entities are constructed, solid modellers are able to calculate such things as the volume, mass (using finite element analysis), centre of gravity, moments of inertia, radii of gyration and stresses. The model is thus able to provide us with information which previously was only possible with laborious calculation, such as questions about the strength of the building, and the quantities of materials needed to build it. I can find no evidence that much, if any, work of this nature has been carried out in buildings archaeology.
Although finite element analysis was developed to answer structural questions, using density values for given materials, it is trivial to change the density values to some other variable, such as the monetary or labour cost per cubic metre of a given building material. From this it is possible to move on to ask 'what if' questions about possible alternative methods of design and construction. It also becomes possible to ask questions about the economics of construction with much more precise figures than have previously been available.
I am not alone in arguing for the use of solid modelling in costing production. There is a definite trend in non-visualizing three-dimensional modelling towards the use of solid models (e.g. Geiger & Dilts 1996; Rappoport 1996), because of the physical properties which solids can be given. At present, this work is being done in engineering rather than construction, where it is easier to model the processes which a component undergoes in its production.
The three-dimensional models which are used in construction management are surface based, and work either with projects which have already been costed, either individually by a quantity surveyor, or by using standard construction elements, which have been already costed. This option is not available to archaeologists. 'The work of the extant recorder is diametrically opposed to that of the designer' (Nickerson 1996) and we must work with the buildings which were actually constructed, even in our reconstructions, rather than model using standard construction elements.
Solid modelling gave entities volume and mass, and similar basic properties, while object-oriented modelling goes further, giving entities all their physical properties. 'Objects' model real world entities by encapsulating both the data and the processes which relate to them. In other words, I define an object not only by properties inherent to itself, but also by the way in which it interacts with the world - for example how a given process will affect the object.
This takes structural modelling one stage further on from solid modelling, and is already being introduced into the construction industry, at the level of practice and research. Object-oriented modellers need not be solid-based.
Neil Bowman's research at ACT-Reading into the application of object-oriented models to construction is an example of how techniques developed in engineering are now being brought into construction. As with archaeology, the construction industry's three-dimensional modelling paradigm has developed out of the use of CAD. The next generation of applications will, almost certainly, be developed out of engineering packages.
In looking at perception models, I looked mainly at ways in which modelling has been used up to the present in buildings archaeology, with some indications of where future research may lead. The use of structural models is much more limited, and so I have looked at areas of potential which have seen little exploitation. In almost all cases, it would be possible to use a surface modeller, but I have argued that it is intuitively more acceptable, and adds greater integrity to the model, if a solid modeller is used. In the case of examining physical properties, the use of a solid modeller is necessary.
©
Internet Archaeology
Last updated: Thu May 1 1997