| dc.description.abstract | This research aims to determine if there is sufficient information encoded within construction workflow histories and document meta-data that may be exploited for the development of knowledge constructs such as diagnosis, thereby contributing to the body of knowledge of automation in construction, with a focus on advanced construction information systems. Electronic Product and Process Management [EPPM] systems provide the capability to establish and map information flow between different parties in a construction project as well as to model project processes. The wealth of information contained in an EPPM system can be exploited to extract knowledge that can provide significant benefits to construction companies. Much of the information relating to processes and their structure, the actors (people and machines) that operate them, and the data associated with each instance of a process is encapsulated within workflows. Workflows, therefore, provide an ideal medium for the capture of knowledge over the course of a project lifecycle.
Project managers have recognized that workflows provide greater visibility and help enforce stricter compliance standards for project processes. While workflows do facilitate process compliance by ensuring constituent tasks are executed as per ordered definitions, the compliance of these individual tasks and their impact on the compliance of the workflow has not been explored. A framework has been developed to address stricter quality control by capturing knowledge of the execution times of work-items, which was then used as a basis for filtering workflows that may violate compliance norms. This significantly reduces the number of workflow instances that would need to be analyzed in detail during an audit. The framework was applied to a case study of a construction project located in British Columbia and validated.
In an ideal EPPM system, the workflow engine would operate silently and seamlessly in the background, automating structured information exchange from the start to the end of a project. In reality, most workflows used in construction projects are of a semi-automated nature requiring manual involvement for tasks ranging from selection of participants to delegation of actors. An adaptive algorithm that is able to recognize and incorporate emergent patterns from prior executed workflow instances and also determine the relative availability of resources can greatly improve the performance of a workflow implementation by reducing its semi-automated nature. An algorithm was developed to demonstrate how a self-adapting workflow methodology could be applied to construction workflows, and two specific cases based on data from a construction project were analyzed showing promising results in terms of time savings.
During a construction project, it is important to ensure that accurate and pertinent knowledge is delivered on time to appropriate personnel. Determining the criticality of documents at different stages of the project can aid companies with managing the flow of information in an organized manner, while providing for the detection of potentially disruptive, erroneous material that could result in delays and costs. An algorithm was designed based upon the meta-data and access interaction logs associated with documents in an EPPM system to identify critical documents. A scenario based on a real event and real data was developed on an EPPM system implementation and a simulation was conducted to determine the applicability of the algorithm and demonstrate its effectiveness.
It is concluded that there is sufficient information encoded within construction workflow histories and document meta-data that may be exploited for the development of knowledge constructs such as diagnosis. Diagnosis based knowledge was used to discriminate between executed behavior and planned behavior to aid compliance checking. Analysis of workflow histories resulted in the development of patterns in workflows which demonstrated time savings if implemented as self-adapting workflows. | en |
