In the field of shipbuilding, there is a growing demand for faster and more efficient production processes, along with a need for swift adoption of new technologies. Modular production emerges as a potential solution, involving the development of a product family with a base platform and various modules. Instead of designing and producing each product individually, modular production allows for the combination of modules to create diverse products. Despite the recognized potential of this approach, there is a lack of quantitative results, and scheduling challenges in modular shipbuilding need to be addressed for its successful implementation.
This dissertation focuses on identifying and resolving three key challenges related to scheduling in modular production. The first challenge revolves around the definition and utilization of modules. Factors such as resource requirements, project sequencing influenced by module size, and project-specific variations in module usage are crucial considerations. The second challenge pertains to inventory management, where reduced production time increases the impact of long lead times, and standardized components spread inventory costs across multiple projects. The third challenge involves stochastic scheduling, leveraging the structural similarities among products in a modular production system to optimize schedules for future projects.
To address these challenges, the dissertation explores the Resource Constrained Project Scheduling Problem with a flexible Project Structure (RCPSP-PS). It introduces a Mixed Integer Linear Programming (MILP) model and a solution method, demonstrating its superiority over existing methods. Given the NP-hardness of the problem, heuristic methods, including group graphs, hybrid differential evolution, and ant colony optimization algorithms, are proposed to quickly find feasible solutions.
The scope expands to the production of a product family through the Resource Constrained Project Scheduling Problem with Modular construction and new Project arrivals (RCPSPMP). This extended problem incorporates stochastic project arrivals and inventory allocation, modeling the pre-assembly of modules. A Progressive Hedging (PH) algorithm is introduced to consider future project arrivals, ultimately aiming to create a profitable product family rather than individual products.
Finally, stochastic project arrivals are considered for the standard Resource Constrained Project Scheduling Problem (RCPSP). Simulation optimization is initially employed, but a data-assisted method using neural networks is introduced to significantly reduce computational costs while maintaining solution quality.
In conclusion, this dissertation presents comprehensive methods for scheduling in modular shipbuilding, addressing challenges related to flexible project structures, nonrenewable resources, resource allocation, and stochastic project arrivals. The versatility of these methods extends their applicability beyond shipbuilding to various industries.
|Doctor of Philosophy
- Delft University of Technology
- Aardal, K.I., Supervisor
- van Essen, J.T., Advisor
- Pruyn, J.F.J., Advisor
|12 Dec 2023
|Published - 2023
Horizon 2020 research and innovation programme (Contract No.: 769419).
- Modular shipbuilding
- Project scheduling
- Resource constrained project scheduling problem