Towards self-recovering construction schedules: a new method for periodically updating project plans and optimizing recovery actions

It is common for a construction schedule to deviate from its original-planned baseline, as uncertainty is inherent in all construction activities. Accordingly, planners are required to perform periodic schedule updates that learn from retrospective progress to more accurately schedule remaining activities and draw optimum recovery plans. This research proposes a method that utilizes a neural network regression model to forecast upcoming productivity rates based on retrospective progress and accordingly updates the schedule on a regular time interval with the required resource adjustments to meet the planned end date of the project with optimal cost. The method was tested on brickwork activities at a residential complex construction project in the UAE, using retrospective progress data of 1487 working days for 132 masons, and was found to be 98% accurate in predicting labor productivity, which was thus used as a basis to draw schedule recovery plans according to the proposed framework. In essence, this research provides a platform toward an automated self-recovering scheduling system, which serves construction managers in proactively preventing potential schedule deficiencies.