| Περίληψη: | Objective of the present study is the development of methods for the assessment of
manufacturing systems complexity and the investigation of flexibility and complexity
relationship. Towards this target, a complete approach based on information theory
permitting the analytical, quantitative and systematic modeling and quantification of both
static and dynamic manufacturing complexity is proposed. Static complexity concerns the
structure of the manufacturing systems, namely the products, the processes, the resources that
constitute the systems as well as their interconnections. Static complexity is treated as the
information that is required for the description of a manufacturing system. Multi domain
matrices modeling the relationships between products, processes and resources are formalized
as networks following the notions of graph theory. The information content of each matrix is
assessed employing Shannon entropy measure and their aggregation yields the static
complexity. Dynamic complexity is related to the uncertainty in the behaviour of a
manufacturing system and in the present study is associated with the unpredictability of the
performance indicators timeseries. The unpredictability of the performance indicators
timeseries, which are provided by computer simulation, is captured employing the Lempel
Ziv algorithm that calculates the Kolmogorov complexity. The dynamic complexity is either
the unpredictability of a specific timeseries or the weighted mean of a series of performance
indicators timeseries produced under different product demand scenarios. The relationship
between flexibility and complexity is investigated for a group of 19 different configurations of
a manufacturing system. In particular, operation flexibility that refers to the system’s ability
to produce a set of products through different machines, materials, operations and sequences
of operations and total complexity, and both static and dynamic are examined employing a
utility function. As a case study, two assembly lines producing three car floor model types at
three different product mixes are investigated. The dynamic complexity of each assembly
line is assessed and the relationship between product mix and dynamic complexity is studied.
The evaluation of the case study revealed the efficiency of the suggested approach validated
its applicability to industrial environments.
|
|---|
