Thesis (MScEng (Civil Engineering))--University of Stellenbosch, 2006. / In order to plan the engineering work of large construction projects efficiently, a model of the engineering
process is required. An engineering process can be modelled by sets of persons, tasks, datasets and tools,
as well as the relationships between the elements of these sets. Tasks are more often than not dependent
on other tasks in the engineering process. In large projects these dependencies are not easily recognised,
and if tasks are not executed in the correct sequence, costly delays may occur.
The homogeneous binary relation “has to be executed before” in the set of tasks can be used to
determine the logical sequence of tasks algebraically. The relation can be described by a directed graph
in the set of tasks, and the logical sequence of tasks can be determined by sorting the graph topologically,
if the graph is acyclic. However, in an engineering process, this graph is not necessarily acyclic since
certain tasks have to be executed in parallel, causing cycles in the graph. After generating the graph
in the set of tasks, it is important to fuse all the cycles. This is achieved by finding the strongly
connected components of the graph. The reduced graph, in which each strongly connected component
is represented by a vertex, is a directed acyclic graph. The strongly connected components may be
determined by different methods, including Kosaraju’s, Tarjan’s and Gabow’s methods.
Considering the “has to be executed before” graph in the set of tasks, elementary paths through the
graph, i.e. paths which do not contain any vertex more than once, are useful to investigate the influence
of tasks on other tasks. For example, the longest elementary path of the graph is the logical critical
path. The solution of such path problems in a network may be reduced to the solution of systems of
equations using path algebras. The solution of the system of equations may be determined directly, i.e.
through Gauss elimination, or iteratively, through Jacobi’s or Gauss-Seidel’s methods or the forward
and back substitution method. The vertex sequence of an acyclic graph can be assigned in such a way
that the coefficient matrix of the system of equations is reduced to staggered form, after which the
solution is found by a simple back substitution. Since an engineering process has a start and an end,
it is more acyclic than cyclic. Consequently we can usually reduce a substantial part of the coefficient
matrix to staggered form. Using this technique, modifications of the solution methods mentioned above
were implemented, and the efficiency of the technique is determined and compared between the various
methods.
Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:sun/oai:scholar.sun.ac.za:10019.1/2376 |
Date | 03 1900 |
Creators | Cronje, Mercia |
Contributors | Van Rooyen, G. C., University of Stellenbosch. Faculty of Engineering. Dept. of Civil Engineering. |
Publisher | Stellenbosch : University of Stellenbosch |
Source Sets | South African National ETD Portal |
Language | English |
Detected Language | English |
Type | Thesis |
Format | 1517685 bytes, application/pdf |
Rights | University of Stellenbosch |
Page generated in 0.0016 seconds