Publisher Summary Numerical predictions on parallel machines with distributed memory are applied to a gas-turbine-combustor model, where dilution jets are injected radially inwards into a swirling flow. The predicted axial velocity along the centerline region is found to be highly sensitive to the representation of the diffusive transport, and hence convection schemes adopted. On parallel algorithm developments within the finite-volume numerical framework, experiments are focusing on the distributions of dataset and the message-passing models. In addition, parallel programs were developed on an experimental parallel C++ programming environment. The distributed array and emulated shared memory operator relieve the programming burden in data partition and communication. The problem size has a significant effect on the speed up characteristic, especially at situations in which higher numbers of processors are adopted. Additionally, adoption of alternating-direction implicit time-division multiple access (ADI TDMA) matrix solver requires redistribution of datasets during different computational phases and this involves excessive movements of dataset.
[1]
W. Jones,et al.
The prediction of laminarization with a two-equation model of turbulence
,
1972
.
[2]
Seth Copen Goldstein,et al.
Active messages: a mechanism for integrating communication and computation
,
1998,
ISCA '98.
[3]
S. Patankar.
Numerical Heat Transfer and Fluid Flow
,
2018,
Lecture Notes in Mechanical Engineering.
[4]
Jenq Kuen Lee,et al.
Programming an astrophysics application in an object-oriented parallel language
,
1992,
Proceedings Scalable High Performance Computing Conference SHPCC-92..
[5]
Jenq-Kuen Lee.
Object-oriented parallel programming paradigms and environments for supercomputers
,
1992
.
[6]
B. P. Leonard,et al.
A stable and accurate convective modelling procedure based on quadratic upstream interpolation
,
1990
.
[7]
W. Jones,et al.
The calculation of low-Reynolds-number phenomena with a two-equation model of turbulence
,
1973
.