Information Technology Reference
In-Depth Information
2 Related Work
One of the ways to estimate and optimize production is to create a mathemati-
cal formulation of the plant equipment eciency. Pongsakdi et al. [9] developed
a production planning model by taking into consideration the uncertainty of
product demand and price, and clarified that their model can sugget solutions
to reduce financial risks and improve profits. Gothe-Lundgren et al. [10] proposed
an optimization model in which the storage capacity minimizes the production
cost, and clarified that an increase in the storage capacity reduces the total
production cost. These models provide useful decision making information for
eciently operating plants. Additionally, there are methods that focus on con-
trolling the equipment in the plants [11]. The control methods, which are called
Model Predictive Controls (MPCs), calculate the eciency of the plant opera-
tions based on the input and output of each equipment, and provide feedback
to future control strategies. For example, Meum et al. [12] applied a nonlinear
MPC to reservoir production. The simulation results showed that the method
using a nonlinear MPC can increase the reservoir recovery performance. The
studies we explained above provide the methods for estimating the production
rate, amount, and eciency from the viewpoints of equipment specification and
equipment control. However, as mentioned in the previous section, the human
factor has to be considered in order to ensure safer production.
Agent-based simulation is one of the solutions to consider such human factors.
It is used to represent the emergent phenomena based on the individuals and for
analyzing complex systems. One of the complex systems is the human behavior
within organizations such as companies. Typical ABS usage is to measure the
organizational performance. For example, OrgAhead [13] focuses on the relation-
ship between the organizational structure and task processing performance, and
finds the optimal organizational structure to achieve a maximum performance
using machine learning techniques. The task represents product developments,
military operations, and so on. SimVision
1
[14] focuses on the relationship be-
tween the information flow among employees and the task structure represented
using the Program Evaluation and Review Technique [15], and estimates the
organizational performance. These simulations can estimate the organizational
performance for the given tasks. On the other hand, they are inadequate at rep-
resenting the human related risks causing accidents. The plant workers have to
tackle the problems in the plants, not the given tasks. Moreover, while the work-
ers represented in the above models do their work to achieve their goals as fast
as possible, the plant workers solve problems caused by daily production to keep
the plant running and safe. In other words, while the workers represented in the
above models are actively performing their duties, the plant workers passively
perform theirs. Therefore, for our purpose, we have to construct a new model
that has the following features, the human factors deciding the daily production
rate, and the plant workers passively solving their problems.
1
SimVision is a registered trademark of ePM, LLC in the U.S.A. and other countries.
