2.2 List the Specific Issues to be Addressed
Often the simulationist is not the engineer who is familiar with the design and operation of the system. By listing all issues the
study should address and discussing them with all concerned parties, the simulationist will often gain further insight into the
operation of the system. In addition, it is not uncommon for various groups within the client's organization to have different
understandings of how the system functions. An early walk-through meeting with all interested parties to discuss the flow logic of
the operations of the system will clarify the specific issues that the study should address. At this step, the client should specify the
minimum and maximum values of the variables to be considered.
The itemized list of issues to be addressed by the study should be documented together with the objectives of the study as part
of the Project Functional Specifications. For example, in a study to find the best number of AGVs in a material handling system,
the issues to be addressed may include the AGV routing criteria, AGV speeds, idle AGV logic, AGV battery recharge policies.
etc.
2.3 Determine the Boundary or Domain of the Study
The objectives of the study together with the specific issues to be addressed by the study identify the information required from
the simulation model as well as the inputs and components needed by the model in order to generate the required output
information. The task of determining which components of the real system to include and exclude from the simulation model
requires both insight on how the real system operates and experience in simulation modeling. A good method is to make a list of
all components in the real system and identify those needed for the simulation model. For example, for the AGV material handling
study one may include the following components of the real system: stations to be visited by the AGVs in order to pick up or drop
parts, AGV path, and AGV battery recharge stations.
Each component of the real system to be included as part of the model is identified in a manner such that the modeler and the
project team agree that the component may have a significant direct or indirect effect on the simulation model output. Each
component's effect on the other components of the system should be discussed before finalizing the list of components to be
included in the model. Model boundary should be kept at a minimum in the beginning of the study and it should be extended later
in the study only if crucial system behavior can not be exhibited by the model.
2.4 Determine the Level of Detail or Proper Abstraction Level
The model should include enough information to get confident answers for the specific questions asked from the study. In many
cases, the availability of data and time, experience of the modeler, animation requirements, and expectations of the client are more
dominant factors in determining the level of detail than the specific issues to be addressed by the study. The objective of the
modeler should be to build the highest-level macro model that will satisfy the objectives of the study. For example, for the AGV
material handling study, the manufacturing cells that request the AGVs to drop and pick up parts may be modeled as black boxes.
Each manufacturing cell can be described with one interarrival distribution for AGV pickup requests and another interarrival
distribution for AGV retrieval requests. On the other hand, in a more complex AGV study where there are synchronization issues
among the manufacturing cells, one may need to describe in detail the schedule of operations at each cell.
The broad or macro-level assumptions of the model decided at this step of the process are influenced equally by the objectives
of the study and by the availability of resources (time, personnel, and funds) for the project. In some cases, lack of time may force
the modeler to build a macro level model that satisfies only a subset of the original objectives of the study.
2.5 Determine If a Simulation Model is Actually Needed; Will an Analytical Method Work?
The experience of the modeler with analytical techniques is very crucial for this step. In many cases, spreadsheet analysis,
mathematical programming and optimization approaches such as linear programming and branch and bound technique, or
statistical modeling techniques such as regression modeling may be more appropriate to use than simulation. For example, if the
objective of the AGV study is to determine the number of AGVs in the system, a number of analytical models are available for this
purpose and they work well for relatively simple systems (Wilhelm and Evans (1988), Maxwell and Muckstadt (1982), Ulgen and
Kedia (1991)).
The appropriateness of an analytical model for the study may not be easy to identify at the first phase of the study but may
become evident as late as at the fifth phase of the study while the simulation m