sume that one wants to determine through the use of process models how the two are interrelated, so that reasonable inferences can be drawn regarding the relationship between staffing system design and staffing system performance.
The specific approach we use to illustrate process modeling is known as task network modeling. In a task network model, system functions (e.g., performing all the required inspections in a region) are decomposed into a series of subfunctions, which are then decomposed into tasks. This is, in engineering terms, a task analysis. The sequence of tasks is defined by constructing a task network .
This concept is illustrated in Figure A-1, which shows a sample task network for a simple procedure—responding to a warning indication. The appropriate level of system decomposition and the portion of the system that is simulated depend on the particular problem. Staffing models have been developed that examine human behavior at the molecular level (e.g., detailed individual user interaction with the human-computer interface) and at a much more aggregated level (e.g., at which the task-level behaviors take hours or days).
In the ASI staffing context, the model might be at a gross level of granularity, since it need only represent inspectors or groups of inspectors as “busy” or “available.” The details of what they are doing are important only to the extent that they relate to factors that drive work demand and capacity.
The task network must also represent in some way the dynamics and dependencies of each task. Such factors include time to perform each task (possibly means and standard deviations), conditions that must be met for the task to start (e.g., available inspector resources, completion of a prerequisite task), and how performance of a given task interacts with other parts of the system.
Every time more than one path out of a task is defined in the network, a decision must be made by a human or other system element on what potential course of action should be followed, and the decision rules must be included in the model. The branching probabilities or decision logic can be represented by numbers, equations, algorithms, and logic of any complexity. In an ASI staffing model, the decision rules would reflect such factors as ASI task prioritization (e.g., required, planned, or demand work).
The paragraphs above describe the essential information that must be defined to adequately represent the relationships underlying a complex process such as the ASI staffing system. However, before a model can be useful in making predictions, it must be capable of accepting and processing changes in its inputs. In particular, the model must be able to accept and process a changing set of demands on the system (e.g., changing the