Jung-Ho Lewe

AN INTEGRATED DECISION-MAKING METHOD TO IDENTIFY DESIGN REQUIREMENTS THROUGH AGENT-BASED SIMULATION FOR PERSONAL AIR VEHICLE SYSTEM

Presented at AIAA Aircraft Technology, Integration, and Operation (ATIO) Technical Forum, October 1-3, 2002, Los Angeles, CA.

ABSTRACTION AND MODELING HYPOTHESIS FOR FUTURE TRANSPORTATION ARCHITECTURES

The goal of a future transportation architecture is an expansion in mobility, enabling new types of travel and commerce currently not affordable and thus pro- ducing induced societal benefit. From the design per- spective, the complexity, high dimensionality and di- verse nature of the design space make study of such architectures extremely difficult. An abstraction frame- work and modeling hypothesis are proposed, steps vital to the proper start of such an aggressive challenge. The core entities within a transportation architecture are abstracted: stakeholders (consumers, regulators, service providers, etc.), resources (vehicles, infrastructure, etc.) and networks (both explicit for resources and implicit for stakeholders). This abstraction leads to a general description for transportation that is useful from a con- ceptual modeling point of view - stakeholders employ particular resources, organized in networks, in order to achieve mobility objectives. The modeling hypothesis is created stemming from the description and focused upon the need to examine the architecture from a sys- tem-of-systems perspective, under the belief that the organization of transportation resources is just as im- portant as the nature and performance of those re- sources. Subsets of the methodologies are tested on three exploratory research thrusts and the findings are used to project a future path towards full validation of the modeling hypothesis. Ultimately, decision-makers at multiple levels can use the methodologies to quickly understand and visualize the relative merits of alterna- tive architectures.

Aviation Demand Forecasts through Validation of an Agent- Based Multimodal Transportation Model

The air transportation system is undergoing significant changes in order to be able to face the future challenges such as environmental concerns and increase in demand. Forecasts are essential to plan and make decisions on what improvements are crucial. While there are various forecast methodologies available, this study adopts the agent-based modeling and simulation technique as a way to generate transportation demand forecasts. Since any transportation mode is subject to competition and interactions with other modes, a multimodal perspective is emphasized. The existing agent-based model was originally calibrated for a single year due to data availability. This limited model validity is extended by a rigorous calibration process using a set of diverse databases for multiple years. The validated model produces results that closely match the available data for past years between 1995 and 2010. Future demand predictions under a set of different scenarios demonstrate a noticeably different trend from the FAA forecast after year 2020. 1. Introduction and Motivation The air transportation system is a complex system, with a lot of uncertainty under the influence of many elements beyond the scope of engineering and technology. It is likely to change significantly in the future, with improvements in Air Traffic Management (NextGen), more aggressive policy on aircraft emissions and a new airline business landscape due to market competition and unstable energy prices. It is not simple to predict how these changes unfold and it is even more difficult to understand their impacts on air transportation demand. Tools and models need to be developed to forecast such demand with high accuracy and confidence. There have been many works in the aviation domain to that end but it should be noted that the air transportation system is not an independent entity when demand for air transportation is assessed: it should be considered in the context of the whole national transportation system. Therefore, a multimodal approach is needed as aviation always remains in competition and/or complimentary with other modes, even if this practice will dramatically increase the complexity of the problem. An agent-based model has been developed and validated against the available long-distance multimodal database: the American Travel Survey in 1995 (1995 ATS). 1 As to the model credibility, the agentbased approach has an inherent advantage over approaches relying on a simple extrapolation and/or a structure of econometric parameters. The limited confidence, however, in the validity of the model due to the single year calibration needs to be addressed, which is the primary focus of the present study. 2. Multimodal Transportation Model and Databases

Multimodal Transportation Demand Forecast using System Dynamics and Agent Based Models

As the starting point of any transportation related study, an accurate analysis and forecast for transportation demand is important for appropriate planning of necessary technological improvements. Multiple modeling and simulation techniques are available, including bottom-up agent-based modeling and top-down system dynamics. The purpose of this research is to build a multi-modal transportation model using the System Dynamics approach (Ground and Air Modes Explorer) supported by an existing agent-based model Mi. These two complementary models were calibrated against each other to ensure equivalency of the outputs within a given range of input values. The system dynamics model matched historic data and generated future forecasts for various scenarios including different economic situations, capacity constraints on the air transportation system, and the

A Spotlight Search Method for Multi-criteria Optimization Problems

Presented at the 9th AIAA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, Atlanta, GA, September 4-6, 2002.

Feasibility Focused Design of Electric On-demand Aircraft Concepts

In this paper the feasibility of electric on-demand aircraft is discussed in a context that includes the performance characteristics of the aircraft but also its potential cost. Traditional vehicle sizing methods are utilized and weight regressions are updated to link the performance of the aircraft to its acquisition cost. Moreover, cost variables sensitive to performance are identified to enable the integration between the costs of operation and aircraft performance. This formulation is used to assess the feasibility of the aircraft by considering the coupling of its physical constraints and its cost. With this integration a design environment that enables design space exploration is constructed. The environment can aid in the identification of favorable design conditions that could make an electric ondemand vehicle feasible in the future.

Transportation System-of-Systems Simulator for Multimodal Demand and Emissions Forecasts

An accurate forecast for future transportation demand and emissions is important for all stakeholders in the National Airspace System. Due to uncertainties in future socio-economic variables and technologies, a scenario-based forecast and analysis is essential. In this sense an interactive simulation environment that provides capability to explore various future possibilities is desired. A Transportation System-of-Systems Simulator (TSS) was envisioned to answer the above challenges. This system-of-system level parametric tradeoff environment includes ground and aviation modes of transportation to capture dynamics in terms of mode choice, mode shift, and introduction of new modes for long distance travel within the continental United States. Parameters can be varied and their impacts on the time series of system level metrics can be interactively observed. Commonly used metrics Revenue Passenger Miles (RPM) and Vehicle Miles Traveled (VMT) are tracked. Resulting CO2 and NOx are quantified to measure environmental impact.

Scenario Exploration for Sustainability of the Multimodal Intercity Transportation System

As concern for environmental impact of various sectors grows, new methodologies and tools are needed to explore scenarios representing the wide variety of possible future socioeconomic environments and technologies. Sustainability of the transportation system is of paramount importance, as it is a driver of economic growth and a non negligible sector in terms of emissions. This study focuses on the two main modes of transportation for longdistance travel in the continental United States: commercial air transportation and automobiles. A parametric simulation environment, the Transportation Sytem-of-Systems simulator, has been developed and is used to assess the multimodal response of travelers to changes in prices and the impact on fuel use. The multimodal approach provides insight into potential mode shifts and tradeoffs in the transportation System-of-Systems. In high fuel price scenarios, both mode shifts and overall demand reduction occur. The introduction of more fuel efficient aircraft and automobiles is accelerated, which helps to maintain mobility while limiting CO2 emissions. With more fuel efficient vehicles, the effects of higher fuel prices is attenuated. Fuel price increase may be implemented separately on each mode at different times, and in most cases result in a decrease in CO2 emissions. The decrease in CO2 emissions resulting from increases in fuel price is limited (up to 2%) compared to the gain obtained through new technologies (of the order of 30%). Due to the multi-objective nature of this problem, multi-attribute decision making is used to identify the best scenario. High fuel price scenarios perform better if emphasis is on emissions reduction, but are penalized when mobility is weighted more strongly.

A Multi-Tier Evolution Model of Air Transportation Networks

This paper describes a model for the US air transportation network employing the multitier evolution concept. In an attempt to concisely represent the dynamics behind the complex network structure, a network is decomposed into multiple tiers. The primary tier is for a network of major hub airports whereas the secondary tier is for non-hub airports. A network evolution algorithm is utilized for modeling the primary tier. The main idea in primary tier is considering chronological and spatial network evolution. The secondary tier engages an access algorithm for tier switching and a shortest-path-finding algorithm for multi-stop routings. These algorithms find the shortest route combinations. The outcomes from these processes are combined together and the complete air transportation network topology is created. Given the simplicity of the algorithms, the overall result adequately agrees to the historical data. Finally, in order to figure out the impact on the network from other transportation technologies, ondemand aviation service is applied for a case study and how it affects established network properties is analyzed.

A Tradeoff Analysis of Future Small Aircraft Capacity from a Point-to-Point Operation Perspective

Determination of an appropriate aircraft capacity requirement is a fundamental, yet difficult decision making task in civil aircraft design. The difficulty arises from the fact that both market forces and the engineering of the aircraft itself must be considered. This paper examines the interacting effect of these two perspectives for the case of a small aircraft operating on a point-to-point network in the 2030-2035 timeframe. Existing demand estimation, airline cost estimation, aircraft conceptual design, and geographic information system tools are integrated with a clustering algorithm and an airport mapping process to create a viable point-to-point network for each aircraft capacity studied. Simulation results indicate that more markets can be served with a larger aircraft and despite more markets being served, the total number of operations remains approximately the same for aircraft with capacities between 24 and 28 passengers. Examination of the resulting point-to-point network reveals that the number of nodes (serviced airports) is independent of the aircraft size but is strongly correlated with the threshold load factor.