Scheduling Analysis of Stored Combat Load Retrieval Using Monte Carlo Sampling
![open_access](/assets/open_access_icon-bc813276d7282c52345af89ac81c71bae160e2ab623e35c5c41385a25c92c3b1.png)
Open Access
- Author:
- Smith, William Thomas
- Graduate Program:
- Industrial Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 12, 2018
- Committee Members:
- Guodong Pang, Dissertation Advisor/Co-Advisor
Guodong Pang, Committee Chair/Co-Chair
Soundar Kumara, Committee Member
Vittaldas V Prabhu, Committee Member
Minghui Zhu, Outside Member
Michael Andrew Yukish, Special Member - Keywords:
- Scheduling
Monte Carlo
Ammunition
Stored Combat Load
Parallel Machine Scheduling
Army
Military - Abstract:
- The United States has military units positioned abroad to support the defense of allied nations. These units must be ready to “fight tonight,” yet logistic requirements often hinder immediate combat readiness. This is especially true for ammunition where retrieval may require significant time. If the unit is not in an active combat zone, munition safety, security, and accountability requirements preclude the bulk of a unit’s ammunition combat load (ACL) from being stored locally and easily accessible. Instead, a portion of an Army unit’s ACL may remain under the custody of a servicing US ordnance company and is stored on a host nation ammunition depot or ammunition supply point (AD/ASP). When directed, customer units travel to the AD/ASP to secure their stored combat load (SCL) under the direction of their servicing ordnance company. US ordnance (ammunition) companies rarely have the necessary resources to serve all customers simultaneously. This shortage of resources is exacerbated when a majority of the support is provided under a Wartime Host Nation Support (WHNS) agreement. The support provided by the host nation in peacetime will be inadequate to support both US and host nation war efforts in a timely fashion. No known study exists of this time-consuming process to adequately estimate its impact on follow-on operations, nor have efforts been made to develop an alternate system which may reduce the makespan. The primary objective of this research is to provide military planners with 1) a means of evaluating the SCL retrieval process and 2) potential system modifications which may aid in reducing the time required for units to transition to full combat readiness. The analysis is accomplished employing Monte Carlo simulations populated with probability distributions representative of individual customer retrieval times and intra-AD/ASP travel times between magazines. While demonstrating system performance and various methods of job sequencing, the simulations use variable synthetic processing and travel times in place of sensitive operational requirements. The current system of SCL retrieval is modeled as a parallel machine scheduling problem - a traditional shop scheduling problem in industrial engineering. Presently, customer units (jobs) are assigned to ammunition company Soldiers (machines) on a first-come, first-served basis to retrieve their SCL. In the proposed alternate system, Soldiers (machines) process ammunition magazines (jobs), allowing customers to simultaneously retrieve their SCL. Both models are scaled across three principal components (numbers of Soldiers, customers and magazines) using several sequencing heuristics and genetic algorithms and evaluated using the resulting makespan. Analysis of variance is used to explore relative performance between systems. Additionally, customer prioritization via weighted completion times is demonstrated using both models as a means for planners to reduce the makespan while addressing operational requirements. Lastly, as the individual customer retrieval and travel times are variable, methodologies for finding a robust schedule while also mitigating the effects of uncertainty are explored and demonstrated. The methods within this dissertation can aid military planners in developing an SCL retrieval system and schedule, that will allow commanders to make better use of limited human resources by reducing the overall makespan of the process.