A column-and-cut generation algorithm for planning of Canadian armed forces tactical logistics distribution
Introduction
The Canadian military logistics system is a large-scale and complex network that involves many inter-related processes, decisions, and military bodies. The tactical part of the system involves operations in the field ranging from building of logistics networks to developing transportation strategies to supply deployed troops. A crucial component of the military tactical logistics is the transportation and distribution of supplies in the field. The Canadian Armed Forces (CAF) uses a variety of transportation modes including air/land/sea to distribute heterogeneous commodities (e.g., fuel, food, ammunition, etc.) in the field of operations.
Given the amount of supplies the CAF transports during each mission and the incurred cost, optimization of the logistics operating costs is becoming crucial to the effective and efficient support of the CAF. In the design of military logistics distribution strategies, as well as commercial logistics strategies, there are different trade-offs between the capital/operating costs and the achievable support performances [1], [2], [3]. Focusing solely on the logistics costs may result in strategies that are less flexible and not effective in meeting the requirements of deployed troops. For example, using the land transportation mode, e.g., trucks, to reduce transportation costs may increase the delivery time and the vulnerability of the transported supplies. Furthermore, a flexible supply strategy is the one that can be dynamically changed to respond to a new unexpected event with less disturbance of the whole logistics chain. Such a supply strategy is usually not necessarily highly optimized [4], [5], [6], [7].
In the military, the trade-off between efficiency and effectiveness appears at different levels during the deployment1 and sustainment2 phases. During the deployment phase, the operational support goals are to ensure deployment speed by reducing the time, as much as possible, while keeping its cost at its lowest value. Similarly, during the sustainment phase, the objective is to ensure that the deployed forces are able to achieve their objectives while minimizing the cost to provide the required support level. The CAF logistics planners are continually called to elaborate strategies and take decisions at different levels to balance between these two dual concepts. In tactical logistics, given the increasing cost of operations and the sensitivity of some missions, the trade-off between the efficiency and effectiveness of operations needs extensive analysis to find good balances for different logistics scenarios.
In this paper, we focus on the design problem of tactical logistics distribution strategies that achieve different balances between the support efficiency and its effectiveness. While efficiency measures how economic is the logistics strategy, we define the following metrics to measure the effectiveness of a tactical logistics distribution strategy: delivery time, loading and transportation safety, and security of transportation assets and commodities. We refer to these metrics as Quality-of-Support (QoS). Our contributions in this paper include: (i) an Integer-Linear Program (ILP) formulation to find the loading and routing of an optimized fleet mix and size of assets that minimize the operation cost, subject to constraints including: payload and bulk transportation capacities; transportation safety of commodities, suitability of the transportation assets to transport commodities of some classes, and priorities of the demands (delivery time). (ii) Following the Column Generation (CG) approach, we decompose our problem of building global logistics distribution strategies into building smaller loading and routing plans, referred in this paper by support plans. (iii) We analyze the integrality gap of the CG algorithm, and develop a strengthening cutting plane algorithm based on the Gomory–Chvátal cutting plane approach.
This paper is organized as follows. Section 2 gives an overview of our military tactical logistics network model. Section 3 reviews some existing work on logistics distribution. Section 4 provides mathematical optimization models and solution algorithms for our logistics distribution strategies design problem. Section 5 presents experimental results on the trade-off between efficiency and effectiveness. Section 6 concludes the paper.
Section snippets
Tactical logistics network topology
Fig. 1 illustrates a small example of a Canadian military tactical logistics distribution network. It is composed of a set of nodes including Air/Sea Port of Disembarkation ((A/S)PODs), Main Operating Bases (MOB),3 Forward Operating Bases (FOBs),4 interconnected by bidirectional links representing the routes. Plain and dash links are used to
Analysis of the problem and literature review
Our military logistics distribution problem consists in finding the loading and routing of an optimal fleet size and mix of transportation assets of two modes (air and land) to convey heterogeneous commodities to multiple destinations. The problem is modelled as a Fleet Size and Mix Heterogeneous VRP with Fixed Costs, and Vehicle Dependent Routing Costs with Time Windows (FSMHFDTW), a variant of the classical vehicle routing problem (VRP). Our FSMHFDTW model differs from the classical VRP in
Decomposition methodology
Given that the multiple visits are required to each destination at different times to deliver heterogeneous commodities, we develop a CG methodology that decomposes the whole distribution strategy into small deliveries of specific commodities to specific destinations during specific time slots. We develop a decomposition method that optimizes over a feasible set of support plans . A support plan is composed of a set of pairs of vehicle-link associations to deliver commodities to different
Computational results
In this section, we assess the performance of the proposed algorithm using some logistics scenarios, and analyze the trade-off between the efficiency and effectiveness of the designed tactical logistics strategies. The proposed algorithm was coded in C++, and the experiments were performed on a Linux computer equipped with a CPU of 2.53 GHz and 4 GB of RAM and running CPLEX 12.0 LP solver [29].
Conclusions and future work
In this paper, a tactical logistics planning method, that is to be integrated into a logistics decision support system, is developed. We developed an optimization algorithm based on column-and-cut generation techniques to find the fleet mix and size to minimize the logistics distribution operating cost subject to a set of constraints including transportation capacity of assets, reliability and suitability of transportation assets, security of commodities, safety of land routes, and delivery
References (30)
Supply chain design and analysismodels and methods
International Journal of Production Economics
(1998)Efficiency versus effectiveness in business networks
Journal of Business Research
(2006)- et al.
Adaptation of some vehicle fleet mix heuristics
Omega
(1992) - et al.
The fleet size and mix vehicle routing problem
Computers & Operations Research
(1984) - et al.
A cluster-based optimization approach for the multi-depot heterogeneous fleet vehicle routing problem with time windows
European Journal of Operational Research
(2007) - et al.
A record-to-record travel algorithm for solving the heterogeneous fleet vehicle routing problem
Computers & Operations Research
(2007) - et al.
A column generation approach to the heterogeneous fleet vehicle routing problem
Computers Operations Research
(2007) A deterministic tabu search algorithm for the fleet size and mix vehicle routing problem
European Journal of Operational Research
(2009)- et al.
A hybrid algorithm for the heterogeneous fleet vehicle routing problem
European Journal of Operational Research
(2012) The vehicle routing probleman overview of exact and approximate algorithms
European Journal of Operational Research
(1992)