Skip to content
Research Fellows Directory

Stephen Jarvis

Professor Stephen Jarvis

Research Fellow


University of Warwick

Research summary

A Rolls Royce turbofan Trent engine (found on an Airbus A380 or a Boeing 777)

generates around 90,000 horsepower on take off, equivalent to the power of about

1,200 family cars. When the engine is running it sucks in more than a ton of air a

second at a speed of 350 miles an hour and by the time the air leaves the engine

it is traveling at over 1000 miles an hour.

How do Rolls Royce design such an engine? Today a large part of the design

process is supported by computer modelling. An abstract model of the engine

component is represented on the computer and physical properties, such as

aerodynamic performance, are simulated using sophisticated computer programs.

What types of computers are used and are they expensive? The computer

modelling needed is highly intricate and so supercomputers are used,

simultaneously executing millions of mathematical calculations per second.

Supercomputers can cost millions of pounds and just cooling the computer can

cost tens of thousands of pounds a year. Hiring staff to write the simulation

programs and run the computers is also expensive – thus these computers and

codes are a big investment for companies like Rolls Royce.

How does this project help? There are many different supercomputer designs;

choosing the right one can be difficult and making the wrong choice can mean

less simulation for your money. In this project we model the simulation codes and

the supercomputers, effectively simulating the simulation! Just as Rolls Royce

simulate the different designs of turbine blades, so we simulate different

supercomputers and simulation codes. We can therefore compare alternative

supercomputing designs before Rolls Royce purchase a new machine. We can

even simulate supercomputers which do not yet exist, and ensure that the code

development is compatible with future supercomputer designs. This work is not

limited to just engineering sciences; any industry that uses supercomputers can

benefit, from climate research to cryptanalysis.

Grants awarded

Assessing Future Computing Requirements using Application Performance Modelling

Scheme: Industry Fellowship

Dates: Oct 2009 - Sep 2013

Value: £125,237.78