Document Type

Dissertation

Degree

Doctor of Philosophy (PhD)

Major/Program

Civil Engineering

First Advisor's Name

Nipesh Pradhananga

First Advisor's Committee Title

Committee Chair

Second Advisor's Name

Arindam Gan Chowdhury

Second Advisor's Committee Title

Co-Committee Chair

Third Advisor's Name

David Garber

Third Advisor's Committee Title

Committee Member

Fourth Advisor's Name

Seung Jae Lee

Fourth Advisor's Committee Title

Committee Member

Keywords

Civil and Environmental Engineering

Date of Defense

10-27-2021

Abstract

Labor-intensive repetitive activities are common in civil construction projects. Construction workers are prone to developing musculoskeletal disorders-related injuries while performing such tasks. The government regulatory agency provides minimum safety requirement guidelines to the construction industry that might not be sufficient to prevent accidents and injuries in a construction site. Also, the regulations do not provide insight into what can be done beyond the mandatory requirements to maximize safety and underscore the level of safety that can be attained and sustained on a site. The research addresses the aforestated problem in three stages: (i) identification of theoretical maximum attainable level of safety, safety frontier, (ii) identification of underlying system inefficiencies and operational inefficiencies, and (iii) identification of achievable level of safety, sustainable safety.

The research proposes a novel approach to identify the safety frontier by kinetic analysis of the human body while performing labor-intensive repetitive tasks. The task is a combination of different unique actions, which further involve several movements. For identifying a safe working procedure, each movement frame needs to be analyzed to compute the joint stress. Multiple instances of repetitive tasks can then be analyzed to identify unique actions exerting minimum stress on joints. The safety frontier is a combination of such unique actions. For this, the research proposes to track the skeletal positional data of workers performing different repetitive tasks. Unique actions involved in all tasks were identified for each movement frame. For this, several machine learning techniques were implemented. Moreover, the inverse dynamics principle was used to compute the stress induced by essential joints. In addition to the inverse dynamics principle, several machine learning algorithms were implemented to predict lower back moments. Then, the safety frontier was computed, combining the unique actions exerting minimum stress to the joints. Furthermore, the research conducted a questionnaire survey with construction experts to identify the factors affecting system inefficiencies that are not under the control of the project management team and operational inefficiencies that are under control. Then, the sustainable safety was computed by adding system inefficiencies to the safety frontier and removing operational inefficiencies from observed safety.

The research validated the applicability of the proposed methodology in a real construction site. The application of random forest classifier, one-vs-rest classifier, and support vector machine approach were validated with high accuracy (>95%). Similarly, random forest regressor, lasso regression, gradient boosting evaluation, stacking regression, and deep neural network were explored to predict the lower back moment. Random forest regressor and deep neural network predicted the lower back moment with an explained variance of 0.582 and 0.700, respectively. The computed safety frontier and sustainable safety can potentially facilitate the construction sector to improve safety strategies by providing a higher safety benchmark for monitoring, including the ability to monitor postural safety in real-time. Moreover, different industrial sectors such as manufacturing and agriculture can implement the similar approach to identify safe working postures for any labor-intensive repetitive task.

Identifier

FIDC010417

Share

COinS
 

Rights Statement

Rights Statement

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/
This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).