An automated approach for developing integrated model-based construction project histories to support estimation of activity production rates.

摘要

When construction cost estimators estimate the production rates of construction activities in a project, they frequently refer to past production rates achieved in a similar project. Problems in the estimations based on such previous data are that the data stored in historical documents and databases do not provide detailed information depicting the conditions under which the activities were executed (i.e., contextual information) in a given project. In addition, these databases do not help estimators to interact with stored production data, at multiple levels of details. This situation results in having cost estimators estimating the same production rate differently, and consequently causes cost overruns or missing project opportunities due to overbids.; To improve decision making of estimators in construction companies, I developed a formalism for creating integrated model-based project histories. Creating an integrated model-based project history involved (1) identifying information requirements of estimators from past projects, (2) formalizing customization of contextual information requirements of estimators, and (3) formalizing the development of integrated model-based project histories, such that it supports storing activity-specific contextual and production information together, and aggregating this information to a level of detail desired.; The utilized research methods can be characterized as descriptive, qualitative and interpretive. It involved observations on construction sites and estimating departments, formal case studies, ethnographic observations and interviews. In addition, previous research studies and standardization efforts were synthesized to develop a data model for model-based project histories.; Validation of the research included validating (a) the comprehensiveness of the identified estimators' information requirements from past projects over different project types and estimators' experience levels; (b) the extensibility of the representation schema, and the precision, recall and efficiency of the retrieval mechanisms developed for enabling estimators to define their contextual information requirements in a customized way; (c) the generality of the reasoning mechanisms in storing contextual information in integrated model-based project histories, and in generating views for estimators; (d) the impact of having project histories on novice and expert decision makers' information use during cost estimating. Such metrics of validation were measured in user tests with utilization of prototype systems (i.e., ContextGen, i-ModelBasedHistoryGen) and a card-game.