Large scale model order reduction for drag reduction of the Ahmed body
Publication Date
December 13, 2022
Creator
Abstract
Flow control of ground vehicles is a key area of research with the aim of reducing the drag that cars and trucks experience. There are many approaches being researched to achieve this, but a common approach is to use Synthetic Jet Actuators (SJA) located at the rear of the model. The SJAs are used to delay the flow separation at the rear of the model and reduce the size or control the behaviour of the wake. A closed loop control scheme monitors the external flow for changes and allows the SJAs to adapt, improving their efficiency over a larger range of conditions. The most common closed loop controller is the Extremum Seeking Controller (ESC), however, the ESC scheme can take a long time to converge as it requires no prior information about the model. To provide model information to the ESC scheme, a reduced order model (ROM) needs to be created as a CFD model is too slow to run in real time. There are many model order reduction schemes, but this research uses the Input-Output Dynamic Mode Decomposition (IODMD) scheme to reduce the size of the CFD model. These schemes have been shown to be successful on smaller scale systems, but never for the reduction of a system on the scale used in this research.
This research uses CFD simulations of an Ahmed body, coupled with SJAs located at the rear of the model. Unlike most SJA setups, the SJAs in this research are directed into the top of the wake, in the hope of reducing the size of the wake and increasing the static pressure on the rear face of the Ahmed body. An adaptation to the ESC is proposed with a reduced order model (ROM) created to provide information about the flow around the Ahmed body to the ESC scheme. By providing this information to the scheme, the convergence of the ESC scheme can be sped up. This would improve the potential for the closed loop control scheme to be used in real time control.
The 2D CFD simulations of the Ahmed body provide a good drag reduction, similar to results presented in the literature, of up to 20% for the Ahmed body with a 35 degree rear slant. The 3D CFD simulations also showed promising signs with the SJA setup providing a average reduction in the height of the wake by 3%, but detrimental effects in the wake cause the drag to increase. A ROM was produced which reduced the 3D CFD simulation from 4,000,000 cells in the mesh, to just a single mode to model the effect of the SJAs on the drag coefficient. The adapted ESC scheme used the 2D CFD simulation and showed a remarkable reduction to an almost instant convergence time by using the ROM to predict the optimal SJA velocity before the scheme took over. The adapted ESC scheme was also able to save 33% more power over a 30 second simulation.
Item Type
ethesis
Thesis Type
PhD
Supervisors
Subjects (LC)
Associated Schools / Departments
Department of Mechanical, Materials and Manufacturing Engineering (UK)
eprints ID
69533
UoN Repository URI
Except where otherwise noted, this item's license is described as
File(s)![Thumbnail Image]()
Name
Rhodes_Peter_Final_Thesis.pdf
Type
Full-text
Description
Examined
Size
14.29 MB
Format
Adobe PDF
Checksum (MD5)
53f6f400282fdce66d4aebae7829cda0