Mental Models of the Physical Process

As described earlier, the operator has stored in his or her memory a mental model of the physical surroundings. The process control room operator, therefore, has stored in his or her brain’s long-term memory a more or less effective and useful model of the process under supervision. The operator continuously updates this mental model by interaction between sensory inputs and the short-term memory. This pro­cess probably takes place at a subconscious/tacit level. It is also probable that things are brought to the operator’s attention by a perceived lack of agreement between the updated model and the actual state of the process (that is, when the operator per­ceives that ‘something is not quite right’). When this occurs, conscious processing takes over and the operator starts to observe the process in order to analyse this lack of synchronisation. As human beings have a limited ability to process large amounts of information simultaneously, the operator is dependent upon having some form of summary description of the process. The various possible methods for storing this mental model will be examined. The way in which the model of the physical reality is actually stored in the brain is not known for certain. Based on the various methods of describing the process, however, we can make suppositions on the most suitable way to structure the process in the long-term memory. Through having a better idea of the way and form in which the mental mode of the physical process is stored, it should be easier to specify the ways in which the various forms of display devices should present the true status of the process to the operator.

Two main types of model may be used for the graphic description of a process: some type of physical presentation may be made, such as a component flow dia­gram, or a functional presentation of the process can be given. Singleton (1966) used this basis for distinction. Ivergard (1972) developed the method further in order to describe a process using different functional flow diagrams. Using this method, the starting point is a general system goal. From this goal, different subgoals may be produced where each subgoal consists of a function. The goals may be broken down to different degrees, thereby obtaining functions at different levels of detail. However, these functions must not be broken down in too much detail, as this loses the degree of abstraction as well as the ability to see the functions as a whole. Also, working with too many functions may make the model unusable. If the functions become too generalised there will be too many physical functions and human tasks. When this happens, the model will become of less practical use as, for example, in fault analysis.

Rasmussen (1979a) produced an excellent review of the many different types of conceivable models of how an operator could store structures of the physical process being supervised. Rasmussen starts with a taxonomy of model descriptions and then describes the following models:

1. Model of physical shape

2. Model of physical function

3. Model of functional structure

4. Model of abstract function

5. Model of functional meaning and objectives

The first two are examples of what Ivergard (1972) refers to as physical models. The other three are examples of functionally-orientated models. The model based on functional structure is, in Ivergard’s view, an intermediate step between physical and functional description. The logical functional flow diagram described by Rasmussen under the heading of model of meaning and objectives is that closest to the function flow diagram described by Ivergard (1972).

Experience from Ergolab’s studies (Ivergard, Istance, and Gunther, 1980) on Computerisation in the Process Industries, however, showed that there is probably a very large variation in the way in which operators build up their mental model of the process. The more theoretically orientated operators, with experience of handling abstract information, have a tendency to form functionally orientated mental models. Examples of such operators are those who consider that they work most efficiently if they just stay in the control room and do not consider that there is any advantage in working in the plant or in being a supervisory engineer (that is, the sort of operator who at certain times goes out in the plant to control, monitor, or even to carry out cer­tain types of maintenance work). Theoretically orientated operators feel instead that they can build up their understanding of the process by being in the control room for as long as possible. On the other hand, there are operators—the majority in Ergolab’s studies—who state that they are wholly dependent on being out and doing practical work in the plant and it is in this way they build up their knowledge. This latter form of operator is probably completely dependent on a physically orientated model.

If it is correct that different operators use different forms of models—that is, functionally orientated or physically orientated—they must also use different degrees of synthesis and different degrees of breakdown of the various subprocesses. It is very difficult to use this type of knowledge in order to design display devices for operators.