Rapid, assured planning for safe operation of integrated power, propulsion, and thermal systems

Butler, Cary Laird

Permalink

https://hdl.handle.net/2142/124291 Copy

Description

Title

Rapid, assured planning for safe operation of integrated power, propulsion, and thermal systems

Author(s)

Butler, Cary Laird

Issue Date

2024-04-19

Director of Research (if dissertation) or Advisor (if thesis)

Alleyne, Andrew

Doctoral Committee Chair(s)

Alleyne, Andrew

Committee Member(s)

• Beck, Carolyn
• Haran, Kiruba
• Miljkovic, Nenad
• Parry, Adam

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Energy systems
• thermal systems
• sampling-based planning
• predictive control
• experimental validation
• energy management
• unmanned aerial vehicle

Abstract

Use of electrified vehicles is on the rise in aerospace and marine applications, in part due to their provision of increased capability and efficiency compared to traditional combustion-powered vehicles. Leveraging their increased capabilities, electrified vehicles can perform missions with more complex and energy-intensive constraints such as operation with reduced emissions or noise during segments of the mission. At the same time, electrified vehicles’ integrated power, propulsion, and thermal (IPPT) systems exhibit interactions between multiple energy domains (including electrical, thermal, and mechanical) that can lead to unsafe conditions if inadequately managed. To guarantee safe operation for these systems, energy and thermal management methods for these systems must address challenges including multi-domain, multi-timescale dynamics, high-dimensional, nonconvex planning problems, and mission specifications driven by uncertain, changing external conditions. This dissertation addresses these challenges by introducing a two-stage approach to manage the multi-timescale dynamics. The first stage of this approach uses sampling-based planning methods in a novel application for energy and thermal management of IPPT systems. The second stage of this approach uses tracking control methods to robustly track reference trajectories from the first stage. Both of these stages leverage a graph-based modeling approach which is briefly surveyed. The sampling-based methods manage slower dynamics and generate long-term mission plans. Notably, these methods can rapidly generate feasible mission plans in nonconvex feasible regions. A rapidly-exploring random trees (RRT) algorithm is presented which generates long-term mission plans using a reduced order model and a finite set of energy primitives, or predefined behaviors. Methods for re-planning online are presented to enable consideration of changing mission specifications. The presented planning algorithms ensure robustness to bounded error in the reduced order model due to imperfect tracking control. Robust model predictive control (RMPC) is used to manage faster dynamics over a short time horizon and track the reference trajectory. The RMPC formulation used in this work considers a linearized, full order model to predict future system dynamics and ensures robustness to linearization error using error reachable sets. RMPC solves an optimization problem formulated to minimize tracking error in the reduced order model while satisfying the planner’s error bound. Constraints applied to the optimization problem, including the error bound constraint, are tightened to account for the linearization error. The computation of reachable sets and numerical optimization are both performed online to determine optimal inputs to apply to the plant. Three case studies are presented to demonstrate the application of this two-stage method to IPPT systems in marine and aerospace applications. These case studies include simulation results for a shipboard power system (SPS) and hybrid unmanned aerial vehicle (UAV) power, propulsion, and thermal system (PPTS), as well as experimental results for a hybrid UAV powertrain. In all three case studies, the two-stage approach yields rapid planning and assures constraint satisfaction even in the face of uncertain, time-varying constraints. In the SPS case study, scalability of the approach is demonstrated as the RRT algorithm reliably solves 20-dimensional, nonconvex planning problems in well under 1/1000th of the time taken to perform the mission. The hybrid UAV PPTS case study demonstrates the flexibility of this method to perform simultaneous energy and thermal management. Finally, the hybrid UAV powertrain case study demonstrates the real-time applicability of this approach to consider uncertain, time-varying constraints by validating the methods through implementation on an experimental testbed.

Graduation Semester

2024-05

Type of Resource

Thesis

Copyright and License Information

Copyright 2024 Cary Laird Butler

Owning Collections