Paper name: “Bio – Inspired Heuristic Network Configuration in Air Transportation System-of-
Systems Design Optimization” (#37)
Authors: Ibrahim M. Chamseddine, Michael Kokkolaras Review by: Askarbek Pazylbekov
This research paper focuses on design optimization of Air Transportation System-of-Systems
(ATSoS), specifically on fleet allocation, aircraft sizing, and formulation of route-network using methods, which are inspired by biological transfer principles. The conventional methods of ATSoS design include decomposition-based approach (sampling to account for demand uncertainty and minimizing expected value of operating cost), semidefinite programming approach (analysis of profit and risk in mean-variance optimization problem), a simultaneous design of aircraft and its optimal trajectory and nested optimization formulation. This work compares adopted bio-inspired network configuration with nested configuration and shows advantages in cost reduction. The first step is reducing the complexity of ATSoS optimization formulation. Instead of concurrent usage of parameters of aircraft sizing (nested ATSoS optimization method) in network and fleet allocation design, the bio-inspired formulation uses step-by-step design analysis (feet allocation – aircraft sizing – aircraft design), which reduces the complexity of calculations. The limitation of this bio-inspired formulation includes unavailability to apply it to large-scale problems, however, it works for the case study presented in the paper – 15 – 20 cities. One important limitation of this research work is that demand is assumed to be symmetric, while in real life it varies day to day, depending on different factors. Bio-inspired network configurations are useful since biological transportation networks, such as bronchi or blood vessels obey fractal configurations, which are “hierarchical structures formed of branches that bifurcate successfully” This feature is very beneficial due to availability to construct multilevel trees of air transportation network that have smaller and increasingly distributed routes. As an example, assume the busiest airport as the hub O, set of airports with highest traffic ℍ and with lowest traffic 𝕃. Therefore, methods that derive the combinations of how elements of L and H should be connected to O are based on rules of bio-inspired networks. Once O, H, and L are determined it is needed to find how elements of L should be connected to elements of H. These computations could be highly-demanding and thus to avoid them, the virtual branch is introduced, which has a demand equal to arithmetic mean of all demands in the set L This allows us to calculate the optimal branching angle between the airports. After completion of route-network design, the problems of fleet allocation and aircraft sizing should be resolved. The solution of fleet allocation problem is concluded in allocation aircraft on generated network routes, such that direct operational cost will be minimized and demand will be also satisfied. Since the number of aircraft is an integer, the problem could be solved using linear programming (e.g. function “intlinprog” by MATLAB Optimization Toolbox). To simplify calculations of direct operational cost (DOC) and since aircraft can fly longer depending on loading, it was assumed that aircrafts’ maximal loading and seat utilization “as an approximation to the optimal aircraft operation due to the presence of fixed operational costs such as navigation fees and landing fees”. Another issue involves aircraft sizing and design. According to the paper, aircraft sizing is “the process of choosing the optimal range and capacity of new aircraft.” To estimate the potential effect of the updated design of aircraft on the whole fleet the Design of Experiment is conducted. The design variables of wing characteristics are usually optimized by FLOPS – FLight OPtimization System, which, additionally, can calculate direct optimization costs associated with new design. The case study is based on 15 Canadian airports. As a busiest airport or hub, the airport of Toronto was chosen. Conventional methods also suggest considering four hubs instead of one, however, in current design problem of ATSoS the cost of operating hubs is not considered but it is obvious that considering more hubs leads to higher cost. Comparative analysis of bio-inspired rule-based and network of nested formulation shoes that first is 3.7 % less in operational costs, while the number of routes is almost the same. One important difference includes missing two routes (Montreal -Vancouver, and Montreal- Calgary). It is also shown that introducing new aircraft increases cost savings from 42% to 60% percent in the bio-inspired rule-based network, while in nested formulation new aircraft increases cost saving by 15%. The application of this method on 20 cities revealed shocking results: proposed method builds the 75 active routes, while nested formulation networks build 190 routes to connect all the airports. Therefore, it is possible to conclude that using bio-inspired rule-based networks reduces costs, however, other different aspects of cost optimization are not included in this research paper.