The world's fastest air-breathing aircraft, the SR-71, cruises slightly above Mach 3. New air breathing aircraft capable of flying at hypersonic speeds (greater than Mach 5) are currently being developed, which will make use of an experimental engine called a SCRAMJET (supersonic combusting ramjet) [10]. A SCRAMJET starts operating at speeds over about Mach 7. SCRAMJETs use the speed of the aircraft to compress air, which then combusts with the fuel, usually hydrogen. An aircraft using this technology could be used as a single-stage-to-orbit (SSTO) space launch vehicle. Such an aircraft could take off at a conventional airport. Using its SCRAMJET, it could climb to a sufficient altitude and speed to engage a more conventional rocket which would take it into Earth orbit. Since there are no expendable stages and since they could use existing airports, vehicles such as this could greatly reduce the cost of access to space. A hypersonic aircraft could also be used as a fast, commercial intercontinental transport because of the speeds they attain.
A prototype of one such vehicle, which should start to undergo flight tests in 2000, is HyperX [4] (show here in Figure 1.1). HyperX is intended to be a demonstration of air-breathing hypersonic flight. It will be attached to a booster rocket which itself is launched from an aircraft. When the booster completes its burn the SCRAMJET will be started and the test flight of the hypersonic vehicle will begin.
Another example of a flight test of a SCRAMJET is the HyShot project [21], which is being undertaken by the University of Queensland. The purpose of the HyShot project is to compare results from a SCRAMJET flight test against those from shock tunnel experiments and those generated in computer simulations. The HyShot project launches a SCRAMJET engine as part of the payload of a rocket. When the trajectory of the rocket reaches Mach 8, which is actually when it is on the way down, the SCRAMJET experiment will begin.
Testing SCRAMJET designs is just one application of a shock tunnel. A shock tunnel is an aerodynamic testing facility whose test times are measured in fractions of a second. The short test times keep the energy cost of running the facility down as well as preserving the facility and test models from over heating. A diaphragm separating a high pressure driver gas from the test gas is ruptured causing a shock wave to be sent through the test gas. This shock wave travels down a tube and reflects at the end, creating a slug of high pressure and high temperature test gas which then expands through a nozzle into the test section. The reflected shock can interact with driver gas behind the interface sending disturbances back into the test gas. To avoid this a driver gas with the right speed of sound is chosen. The required speed of sound of the driver gas increases with the speed of the primary shock. Shock tunnels are discussed in greater detail in chapter 3.
Hypersonic flight experiments allow engineers and researchers to investigate all aspects of aerodynamics relating to the prototype vehicle. The problem is that they are costly and time consuming. To reduce future costs, one aim of these kind of flight experiments is to allow engineers to validate computational fluid dynamics (CFD) models of the engine designs. With working and reliable CFD simulationstions available, further development of SCRAMJETs and hypersonic aircraft can be validated at a greatly reduced cost.
For many subsonic flow problems, such as those encountered in conventional aircraft, the numerical methods available are simpler and more robust. In these cases CFD simulations can largely replace wind tunnel experiments.
Hypersonic flow problems, on the other hand, present greater problems for both wind tunnel experiments and CFD simulations. There are much higher speeds and temperatures involved. The flow is more complex with shock waves introducing their own aerodynamic effects. The increased energy of the flow itself can cause the molecules in air to dissociate and ionise, which further destabilises the flow. Aerodynamic heating of an aircraft at these velocities also presents further design problems.
Experiments in shock tunnels can present researchers with some difficulties, such as short test times and the presence of debris from a burst metal diaphragm. Various experiments have shown that the driver gas contaminates the test flow earlier than expected [31]. Thus, it is useful to validate shock tunnel test data against CFD simulations to rule out or explain unwanted effects that may be present in the results.
CFD calculations to simulate hypersonic flight are much more complex than their subsonic counterparts. The model equations being solved are hyperbolic and hence non-linear. Shock waves, which appear as discontinuities in flow solutions, present a range of problems to numerical researchers. Such problems include oscillation or instability in the solutions. To accurately simulate hypersonic flow, chemical reactions must also be taken into consideration.
The comparison of results obtained by numerical and experimental hypersonic researchers is a two way process. Each validates the other and increases the understanding of the problems involved.
Milthorpe [23] developed an algorithm for modelling unsteady flows containing discontinuities, such as those found in shock tunnel research. The algorithm he uses is a kind of direct numerical simulation and calculates flow values based on the conservation of mass, momentum and energy. His computer program has been used as a comparison for a number of shock tunnel experiments [11],[18],[26].
The current work is an extension of Milthorpe's method. The primary aim was to produce a computer program that was more flexible and extensible, allowing further modifications and extensions to be made in an efficient manner. To achieve this it was decided to implement the algorithm in C++, an object-oriented and modular computer language [33].
The functionality of Milthorpe's original computer program was extended to included parallel processing. One reason for this was to investigate the ease with which the new version of the computer program could be extended. Another reason is that a computer program that can make use of parallel processing takes much less time to complete when run on a set of networked workstations or on a multi-processor computer. The computer program was also made portable so that it could run on several types of computers and operating systems to make the parallel processing capabilities more flexible.
To contribute in another way to the field of hypersonics research, and to validate the computer code with a new problem, a particular shock tunnel simulation problem was tackled. The problem of driver gas contamination (discussed in more detail in Section 3.3) in the shock tunnel test section was modelled, in the hope that experimentalists might gain some useful insights to the problem. The driver gas contamination problem involves several interesting flow features such as shock waves, shock reflection, shock-boundary layer interaction and contact surface mixing, making this a good test for the code. To simulate this problem, the computer code had to deal with interactions between two species of gas - the driver gas and the test gas.
While tailoring the computer program to the driver gas contamination problem, it was discovered that a certain set of parameters needed to be tuned in order to model the problem more accurately. It was decided to automate the tuning process because there were a number of parameters. A simplified version of the problem had to be run to test the parameters as they were changed. Included in this work is a description of the tuning algorithm used and the results it produced.
