Tight coupling and complexity in modern process plants make it difficult to obtain
a clear and accurate assessment of the state of the process, especially when
disturbances occur. Highly interconnected plants are challenging to analyse as disturbances
propagate plant-wide over the process. Tasks such as troubleshooting,
maintenance planning and diagnosis of abnormalities become more diffcult in the
presence of unobserved dependencies between plant components. In order to support
engineers to better understand the connectivity among plant items, new methods
are required for modelling and visualising plant connectivity information.
The methods proposed in this thesis address physical connectivity (piping and instrumentation),
and also include the signals, the electrical sub-system that powers
the plant and the logical connections (cause-and-effect). The reason is that electrical
and utility systems are important propagation routes for the effects of process
disturbances. The research reported in this thesis includes the review of technologies
for data mapping and analysis, methods for modelling and integrating connectivity,
and novel algorithms for parsing, linking and visualising plant connectivity models.
The thesis presents the software prototype Topoviz, designed to achieve the research
objectives of this project. The development of Topoviz was driven by requirements
generated from site-visits and interviews with subject matter experts in the eld of
plant operations. Novel contributions to the research eld include the proposal of a
graph database using the property-graph model. Several algorithms were developed
and tested, including novel algorithms for creating plant connectivity networks from
engineering documents. These networks can be analysed to gain insights about
how the di erent sub-systems in the plant could a ect each other in case of disturbances.
For the rst time, the integration of connectivity information coming
from the process and electrical systems was achieved. The research in this thesis
shows successful results and visible advantages compared to previous methods for
analysing plant connectivity information.