ZACM4921 HYPERSONICS

CFD++ TECHNICAL REPORT

Introduction The aim of this report is to describe and discuss the CFD analysis of the convergentdivergent conical nozzle. This analysis was conducted to estimate the region in which the flow within the nozzle transforms from laminar to turbulent flow given known experimental measurements of the inflow. The outflow conditions will be taken after computation using CFD++ analysis software and compared to other hypersonic nozzle experiments. The expectation of the project is to provide a specified region of the nozzle in which the flow undergoes transformation from laminar to turbulent flow with a reasonable resolution of the nozzles length. Task This task to be performed began with the supplied dimension of a convergent-divergent conical nozzle. This model was constructed in Gridgen with inflow and outflow boundaries, as well as the nozzle wall and axis of symmetry. The inflow conditions have been provided as part of the task and are input into the analysis software in a single form. The input conditions used contained the concentrations and fluid properties of the gas as it enters the conical nozzle. The flow to be analysed contained varying concentrations of nitrogen and oxygen in their molecular and atomic forms and is representative of the inflow conditions for the T-ADFA shock tunnel. This data was provided by Russell Boyce as a single file for manipulation in the analysis software. The outflow conditions will be measured from CFD analysis. The data files from the successful simulations will be exported for manipulation, plotting and analysis of the data. The output data contains much information, of particular importance will be the pressure, density and temperature values in the nozzle which can be used to find a value for the gas constant R by:
p = RT

This value for the gas constant will then be used with the already known values for the velocity of the flow and the flow temperature in the equation 2 to give a value for the ratio of specific heats gamma.

M=

v v = a RT

Finally these values will be combined to give the normalised Pitot pressure values using the Raleighs Pitot tube formula as a function of the gamma number and the Mach number. The equation used in this is shown below.

p = p oo

1 1 2 1 M 1 + 2

Alternatively the group has an option of using another equation for an approximation f the expected results. This equation shown below has a resolution of up to 3 % and as such judged accurate enough for this reports approximations.

p pitot 0.93v 2
It is through these values for the normalised Pitot pressures that we expect to be able to compare our results to the experimental results actually measured. As such the most similar plot produced by us to the experimental plots will be indicative of the inactive turbulence region in the nozzle.

The mesh Several attempts were made to create a suitable mesh in order to adequately analyse the flow conditions of the conical nozzle. A rotated view of the nozzle design with its dimensions is displayed below in figure 1.

Figure 1: Dimensions of the rotated conical nozzle

Initially the mesh was to course in analysing the flow through the small throat region and the boundary layer growth on the side walls. This unstructured mesh was modified by increasing the number of cells to adequately analyse the hypersonic flow conditions in the conical nozzle. Increasing the number of segments was vital to producing a more accurate model of the flow conditions. This was particularly important for the throat and nozzle wall areas where a fine resolution is required to accurately map the results. In all, the final mesh generation utilised approximately 40,000 cells in an unstructured mesh to model the flow. The boundary conditions of the mesh are described in table 1.

Table 1: Boundary conditions of the mesh

Boundary Condition Number 1 Inflow

Details Inflow conditions supplied P = 10.6 MPa T = 3100 K MO2 = 0.2019 MO = 0.0037 MN = 0 MN2 = 0.0547 MN = 0.7397 Nozzle exit along the x and y directions. Nozzle wall from inflow to throat and exit Centreline of nozzle

2 3 4

Outflow Wall Axis of symmetry

These inflow conditions were supplied as the input conditions for a high speed test on the T-ADFA shock tunnel for low enthalpy testing. Through the use of the flow analysis software, several runs were conducted for varying turbulence inactive regions. These regions areas set areas in the mesh in which turbulent calculations are ignored and not allowed to develop. Such inactive regions were set initially very early in the model, increasing in size for each successive run. These conditions were imposed over six run until the majority of the conical nozzle was encompassed within a turbulent free zone. As described, initial tests allowed turbulent conditions to develop early while successive flow analysis tests imposed a longer restriction on the laminar region along the nozzle. The length of the turbulent inactive regions as set for each of the six CFD models is given in table 1.
Table 1: Dimensions of the turbulent inactive regions

Simulation Number 1 2 3 4 5 6

Distance from Nozzle Origin (mm) 57.3 82.7 304.7 526.7 748.7 970.7

Figure 2 highlights the turbulent inactive region in a graphical sense on an overlay of the mesh geometry.

Figure 2: The turbulent inactive regions set for test runs

Results and discussion Six software analysis runs of varying turbulent inactive regions were conducted. Figures 3 through 7 are contour plots of the flow Mach number within the nozzle. The contour range of each plot varies from the initial condition to Mach 9.5.

Figure 3: Mach number contour plot for simulation number 1

Figure 4: Mach number contour plot for simulation number 2

Figure 5: Mach number contour plot for simulation number 3

Figure 6: Mach number contour plot for simulation number 5

Figure 7: Mach number contour plot for simulation number 7

As can be viewed by the previous Mach number contour plots, there is very little difference in our runs for varying the turbulence inactive region. For each visible plot, the boundary layer on the conical nozzle wall is seen to grow slightly along its length. This said, it was expected that some form of separation was to occur and the boundary layer to

increase in size. This was not evident in any of our simulations, even with varying the turbulence inactive region from the throat to the nozzle exit. Figure 8 below is a plot of the approximated normalised Pitot pressure along the vertical outflow of the model, located at 200mm from the nozzle exit. This normalised Pitot pressure is plotted against y, as the distance, in meters, from the centreline of the model.
Figure 8: Approximated normalised Pitot pressure along vertical nozzle outflow
Normalised Pitot Pressure 200mm from nozzle exit (vertical)
3.50E+04 3.00E+04 2.50E+04 P/Ppitot 2.00E+04 1.50E+04 1.00E+04 5.00E+03 0.00E+00 0 0.02 0.04 0.06 0.08 y (m) 0.1 0.12 0.14 0.16 Run 1 Run 2 Run 3 Run 4 Run 5 Run 6

This plot indicates two points. Firstly, this seems to correlate with normalised Pitot pressure results obtained by Sirka [1] from experimental testing on the T-ADFA shock tunnel, for which this analysis is based. In the thesis by Sirka, normalised Pitot pressures are obtained for distances of 55 mm and 265 mm from the nozzle exit, of which ours lies between at the 200 mm case. The trend compares with that expected, the Pitot pressure being lower for further distances from the wall, as boundary layer effects prevail. The magnitude of the normalised pressure is also comparable, with our case being approximately 2.7e4, being in between Sirkas reported results from experimental data. The second point that can be drawn from this plot is the alarming similarity of successive runs, a trend evident previously in the contour plots of Mach number. There are two possible causes for these similarities. Either the turbulent effects and its onset has no effect on the flow conditions within the nozzle, or the turbulence initialisation effect in the analysis software has not been performed correctly, resulting in the runs being essentially identical. The first event is unlikely and not considered to be the cause. There should be a marked difference in the flow characteristics for fully laminar to fully turbulent allowable cases, as trend we do not see over our six simulations. It therefore seems the most likely cause is a failure to properly initialised the analysis software to correctly alter the turbulent region over the course of each simulation. So it appears these simulations have produced, at most, two independent trial runs, with run two through six being essentially unaltered. That is the most logical conclusion we can draw from the results obtained.

This similarity of runs is also supported for a plot of the normalised Pitot pressure along the centreline of the nozzle model (the axis of symmetry, being boundary condition 4). This plot is shown below in figure 9. Figure 9: Normalised Pitot pressure plot along the centreline
Normalised Pitot Pressure For All Runs
1.00E+00

1.00E-01

P / P pitot

1.00E-02

1.00E-03

RUN 6 RUN 5 RUN 4 RUN 3 RUN 2 RUN 1

1.00E-04

1.00E-05 0 0.09 0.18 0.27 0.36 0.45 0.54 0.63 0.71 0.8 0.89 0.98 1.07 1.16 X (m)

As can be viewed from this plot, the initial run one is also slightly different from the successive simulations. This test run was performed with the turbulence inactive region set to before the throat. With these two plots showing a small, but clear difference between the initial and the later runs, it is possible the turbulence inactive region was set correctly for this, but not for later, adjusted, simulations. Conclusion This report set out to compare results obtained from previous experimental data on the outflow conditions from a high speed shock tunnel. Its aim was, through the adjustment of the region within which turbulent effects would be allowed to propagate, make use of fluid analysis software to predict the general location within a conical nozzle where the laminar to turbulent transition takes place. This report was unable to deduce a reasonable region for which turbulent effects begin, most probably due to an inability to adequately develop simulations that altered the turbulence inactive reaction over each run. Although the limited findings obtained have shown a likeness to data present by Sirka in their experimental analysis of a high speed shock tunnel, further analysis is unavailable due to the little spread of data obtained.

References [1] Sirka, K. Thesis supplied by Russell Boyce Anderson, JD, 2006, Hypersonic and high-temperature gas dynamics, 2nd edn, AIAA, Virginia