About the event
Aero-engines are astonishing engineering feats. They are tasked with the efficient delivery of thrust—some generating as much as 400 kN—whilst adhering to stringent emission and safety regulations. From the centrifugal force acting on each fan blade (equivalent to the weight of a double decker bus), to the high temperatures experienced in the turbine (well above the melting point of the blades themselves), there is no shortage of mystifying and intriguing facts about the engine. Following a brief overview of the aero-engine and its operation, I will talk about three research themes that lie at the intersection of turbomachinery aero-thermodynamics and data-centric engineering.
The first theme concerns the way one designs and manufactures blades. Here ideas from the field of subspace-based dimension reduction can play a powerful role in ensuring more efficient preliminary and detailed design workflows, while simultaneously cutting down the costs associated with manufacturing to certain tolerances. The second theme revolves around how one experimentally estimates uncertainty in engine sub-system- and system-level performance; an extremely important task as greater uncertainties warrant either more instrumentation, higher quality of instrumentation, or the launch of a research program. Given that performance is estimated from numerous temperature, pressure and bleed measurements—each with their own precision and range—aggregating them to compute efficiency (even for a single sub-system) is a challenging task. To this end, I will present some solutions using the Delta method from statistics. The third theme I will discuss is computational fluid dynamics—our workhorse for estimating aero-engine performance in computer simulations. More specifically, I will outline both the aleatory and epistemic uncertainties in Reynolds average Navier Stokes (RANS) computer models in turbomachinery and the numerous strategies tailored at their quantification. The former largely revolves around ideas within the remit of polynomial chaos while the latter is based on machine learning approaches. Broadly speaking, more accurate computer simulations can significantly reduce the costs associated with expensive rig and engine tests.
Speaker: Dr Pranay Seshadri (University of Cambridge, UK)
Pranay Seshadri is a postdoctoral fellow at the Department of Engineering, University of Cambridge. He is also a Group Leader within the Data-Centric Engineering Programme, largely focused on aerospace. He obtained his PhD in turbomachinery and computational engineering at the University of Cambridge.
Add comment