A Java-Based Direct Monte Carlo Simulation of a Nano- Scale Pulse Detonation Engine

Recently, the Aerospace community has focused much of its effort on the development of micro-air vehicle technology. The treatment of problems at the micro-scale and nano-scale levels has traditionally been difficult, and one of the most challenging aspects of the design of these aircraft is the development of compact propulsion systems for them. Here, the pulse detonation engine is proposed as a means of propulsion for micro-air vehicles and nano-air vehicles and its performance is simulated using a DSMC flow solver. The major problem arising when attempting to implement the pulse detonation engine at such small length scales is the dominance of the wall effects inside the detonation chamber. To alleviate the loss of thrust due to wall effects, an adiabatic wall boundary condition was developed, ultimately affecting a thermodynamic environment that is more favorable for the formation of a detonation wave, while maintaining real-wall viscous effects. The chemistry and physics of the pulse detonation processes are then simulated in order to determine the minimum allowable combustion chamber dimensions to facilitate ample thrust for the micro-air vehicle and nano-air vehicle problems. The code discussed in this paper is written in Java 2 and compiled using both GNU GCJ v.1.0 and Sun Java Development Kit 1.3.1. Traditionally, FORTRAN has been used for the solution of scientific problems, but more recently, C and C++ have replaced FORTRAN in some cases. With an increase in computer power, however comes the ability to develop larger software packages, thus spawning the need for a more robust computational tool. The object-oriented structure of Java lends itself very well to the Direct Simulation Monte Carlo flow solver and to particle based flow solvers in general. In particular, the relatively complex particle sorting routine used in G.A. Birdís original DSMC flow solver [1] can be replaced by a much simpler one in an object oriented program environment. Although a similar object oriented environment exits in C++, Javaís intrinsic multi-threading capability, which C++ lacks, makes the problem quite amenable to solution on multiple processors in a shared memory environment without the added complexity of MPI or OpenMP. Furthermore, Javaís graphical capabilities can be particularly useful in monitoring the solution as it is averaged over many ensembles. The trade-off, of course, is a mild performance degradation as compared to FORTRAN and C/C++ on some computer systems. However, the difference in performance is found to be surprisingly small as compared to FORTRAN and, in some cases, Java even outperformed the FORTRAN 90 solution to an identical problem.