Towards data-driven project planning

Towards data-driven project planning: Exploring the possibilities of generation and implementation of throughput times based on data analysis

Boelens, Niels (TU Delft Civil Engineering and Geosciences)

Bakker, H.L.M. (graduation committee)
de Bruijne, M.L.C. (mentor)
Rikhtegarnezami, M. (graduation committee)
van Tuinen, Barthelemy (graduation committee)

Delft University of Technology

Civil Engineering | Construction Management and Engineering

2022-05-20

The goal of this research is to create a design for a method to enable the generation and implementation of evidence-based throughput times in the planning process of the construction industry. The current planning process and the generation and collection of throughput times relies heavily on experience-based information. Generating throughput times with data analysis could contribute to the creation of a more accurate and predictable project planning. The proposed method is necessary since the current planning process in the construction industry has a number of factors that obstruct a direct path to implementing data analysis. The current planning process is analyzed by a number of interviews with project planners and additional talks with a panel of experts in the planning process. A literature review is used to analyze how other industries use data analysis in their processes and what conditions are necessary to make data analysis possible in a planning process. Additional interviews are used to find the possibilities of data analysis in the planning process. A case study of one infrastructure construction project is used to find the missing aspects in the current planning process that would possibly enable data analysis. The case study is analyzed by following the steps of the Cross Industry Standard Procedure for Data Mining (CRISP-DM). The final product of this research is an alternative version of the CRISP-DM method. This method is focused on establishing a database filled with planning information, design information and combining those two datasets. It proposes steps to avoid undermining currently present intricate negotiation processes between contractor and subcontractor. The method makes sure the database is expanded and diversified and that proper evaluations steps are in place and that a benchmark is met in the accuracy of output values. The steps are divided in separate cycles to be applied on individual construction activities (e.g., sheet piles) since there is so little information available and it would not be feasible and justified to implement the steps on all construction activities. It is recommended that the proposed method is used to provide a proof of concept of generating detailed throughput times of construction activities in the planning process of the construction industry. This proof of concept is necessary to expand the data mining operation and try to fit evidence-based information in the workflow of project planners.

Project planning
Data analysis
AEC industry
Throughput times
Data driven
CRISP-DM

http://resolver.tudelft.nl/uuid:71ed08fb-5ca7-4d72-90d6-19ad9899fc59

Student theses

master thesis

© 2022 Niels Boelens