Session: 01-01/05-07 Joint Session: Aero-Engine Control & Diagnostics

Paper Number: 152478

Optimization of Variable Cycle Engine Throttle Schedules Using Gradient-Free Methods 

Although the Variable Cycle Engine (VCE) has been the subject of research for several decades, interest has sharply increased in recent years for both civil and military applications due to advances in material science and manufacturing mitigating reliability concerns plaguing previous designs. The key advantage of the VCE engine lies in its increased operational flexibility by introducing a set of variable geometries in the flowpath, such as compressor Inlet Guide Vanes (IGVs) and Variable Area Bypass Injectors (VABIs). These allow to dynamically adjust the engine operation to adapt for different flight regimes. For example, at subsonic speeds, the bypass ratio can be increased for improved efficiency, while at supersonic speeds, the bypass ratio is reduced to provide a higher specific thrust. In addition, VCEs allow for a so-called flow-holding capability when operating at reduced throttle, lowering the spillage and boattail drag.

The introduction of variable geometries in the flow path increases the operational degrees of freedom, making the generation of optimum throttle schedules at different flight Mach numbers, altitudes and thrust settings more complex and computationally expensive by necessitating an optimisation to be performed at each considered off-design point. Historically, the problem of computational cost is circumvented in two ways. One solution is to derive a throttle schedule from physical reasoning, and a second solution is to only perform the throttle schedule optimisation at a single set of flight conditions and extrapolate that knowledge to the whole flight envelope. Neither of these solutions yields the true optimum throttle schedule.

In this work, the authors aim to perform an optimisation of a three-stream VCE throttle schedule for the entire flight envelope. To achieve this engine performance is simulated using Modelon, a comprehensive object-oriented software platform that enables the detailed simulation of engine components and system-level performance modelling. Two methods to minimise the fuel consumption at a given flight condition and thrust output are benchmarked. The first is the gradient-free differential evolution algorithm that iteratively proposes new candidates based on the processes of mutation, recombination, and selection. The second method consists of a surrogate-based optimisation that builds a simplified meta-model approximating the true engine behaviour, which allows for more efficient exploration of the search space. Ultimately, this allows assessing the performance of existing approximate methods to the real optimum, both in terms of computational cost and accuracy.

Presenting Author: Karel Van Den Borre von Karman Institute for Fluid Dynamics

Presenting Author Biography: Karel Van den Borre received his Master of Science in Electromechanical Engineering at the Vrije Universiteit Brussels (VUB) in 2021. The year after, he graduated from the von Karman Institute for Fluid Dynamics (VKI), obtaining a Research Master in Fluid Dynamics. Currently, he is serving as a joint PhD candidate in both the Turbomachinery and Propulsion department at VKI and the FLOW research group at VUB. His primary research focuses on supersonic propulsion cycle modelling, reduced order modelling of combustion processes, and sustainable aviation fuels.

Authors:

Karel Van Den Borre von Karman Institute for Fluid Dynamics
Bayindir H. Saracoglu von Karman Institute for Fluid Dynamics