H. ELEMENTOS DE PROTECCIÓN PERSONAL
I. ENFERMEDADES PROFESIONALES
For the implementation of the empirical work consisting of the model building and model use studies (sections 3.5.1 and 3.5.2 respectively), a case on the UK prison population has been chosen. In the model building sessions, involving Verbal Protocol Analysis (VPA), the case study is used as the stimulus for model building. Participants verbalise their thoughts as part of this process. In the second study the survey of model use, DES and SD simulation models are built based on the same case, which are then given to the participants to use. The case study provides model users with background information regarding the problem considered, while the simulation models are the control variable, each given to two different groups of users, who are asked to use the models and make management decisions. While in
Chapter 4
the first study the case is used as the stimulus for modellers to build a simulation model, in the second study it provides the necessary background information about the simulation models, which then participants use for decision making.
The same case is used for both studies implemented in the thesis. Due to the different aspects compared by the two studies, the integrity of the study is not considered to be affected by the use of the same case. The main criteria considered for the choice of an appropriate case are: simplicity, novelty and the potential to accommodate both DES and SD modelling. For the purposes of the model building study, a suitable case needs to be simple enough to enable the development of a simple model in the limited period of time available. On the other hand, for the model use study, using simple models ensures that users with little or no prior experience of using simulation models can understand and use them for decision- making. From a practical point of view, limited time availability was an issue for the sessions of each study. A time limit of 60 - 90 minutes was set for the modelling sessions involving busy expert modellers, to ensure their participation in the study. Whereas, the model use sessions needed to fit with the MBA course curriculum and requirements for a 1.5 hour long session.
Regarding novelty, the case used in the model building study needs to be equally familiar to both SD and DES modellers, to assure a common starting point for all modellers. For example, the beer game, which considers the bullwhip effect in a
Chapter 4
far more familiar with building such a model as compared to DES modellers. Using a novel case situation ties-up well with the VPA methodology, ensuring that the participants, without any prior preparations, are actively involved in the modelling exercise and thus providing live accounts of the model building process. With consideration to the survey of model use, the choice of a novel case study and its implementation is appropriate with the view to collecting participants’ unbiased opinions on the simulation models used.
Another important criterion is that the case needs to accommodate models from both simulation techniques, but on the other hand it should not lead to the choice of one or another approach. For example, some specific features of each modelling approach the presence of which is required are: randomness in DES vs.
deterministic models in SD, the aggregated presentation of entities in SD vs. the individual representation of entities in DES, etc. This is further discussed in section 4.4 regarding the suitability of the case chosen, that of the UK prison population.
4.2.1 Selection of the case
This section describes the process followed for the choice of the case. First, two previously considered problems are referred to, followed by the reasons why these were rejected. Then the selection of the prison population case is described.
Chapter 4
Supply chain management was originally chosen as a specific area of application for the empirical work. The supply chain is a topical area and moreover has been
extensively modelled in both the DES and SD literature. It was initially expected that a suitable case would be available among the already published case studies on supply chain management. For this reason, a number of case study databases were explored, such as: European Case Clearing House (ecch), Darden case collection, Richard Ivey and the Harvard Business School Cases. The case studies retrieved from these databases were considered to be general in scope, dealing with multiple issues and/or complicated issues which were not suitable for building simple simulation models.
Due to the difficulties encountered in finding a suitable pre-existing case study, the creation of a customised one, tailored to the needs of the research, was considered. An application in the supply chain sector was first attempted and a case of an imaginary doorstep dairy delivery company (Fresh Dairies) was chosen (Appendix A.1). The case study design process was an iterative process revolving between writing the case study and building the equivalent supply chain model based on it. The DES model was built in Simul8 and was found quite complicated, taking long hours of modelling. The complexities arose due to detailed information required for each echelon, i.e. demand, consumption, order processing, etc. Furthermore, as the design of the case progressed, it resembled a generic supply chain, similar to the beer game. As a result, the case of the milk delivery business was discontinued.
Chapter 4
In the meantime, the fishery model developed by Morecroft and Robinson (2005) was considered. Their model could serve the basis of a simple model, representing the fish stock in the sea, including the feedback effect of natural fish regeneration and the effect of fishing undertaken by human activity. The DES fishery model created attempts to match the structures and modelling aspects of the problem situation chosen and it is not deemed to be a representative DES model for the purposes of this study because it does not represent individuals. Hence the fishery model was not considered suitable for this study. Some models with a similar structure to the fishery model are generally those that incorporate birth/death
processes with inventory, such as: hospital occupancy, prison population, customers of a particular brand/product, workforce recruitment/retention, etc.
After thoughtful consideration, a case study on the UK prison population based on Grove et al. (1998) was chosen. The prison population is a topical subject both in the UK and elsewhere, e.g. (Korporaal et al., 2000). The inherent feedback that exists in the system, with prisoners entering and returning back to prison due to re- offending (recidivism), can be uniquely represented by each simulation approach, DES and SD. While simulation models of the prison population and the criminal justice sector are rare, DES and SD have both been used to model the prison population. DES models of the prison population have been developed by Kwak et al. (1984), Cox et al. (1978), Korporaal et al. (2000). Equivalent SD models of prison population have been developed by Bard (1978) and McKelvie, et al (2007), while the initial UK prison population model by Grove et al. (1998) is a flow model
Chapter 4
analogous to an SD model. Hence, the prison population case is deemed as an equally familiar application to both DES and SD modellers and also amenable to both modelling approaches. Therefore, the UK prison population case is considered suitable for use in comparing DES and SD model building and model use in the current research.