Professional Documents
Culture Documents
In Eq. (32) F̂mn and Ĝmn denote the left- and right-running waves, J1 the bessel function of first kind and
first order, s10 = 1.8412 the 0-th root of the first order Bessel function derivative, R the diameter of the
considered cylinder (combustion chamber) and δ a phase angle with respect to the frame of reference. The
±
wave numbers kmn is given in [18] as
" r #
k s 2
± 10
kmn = −M ± 1 − (1 − M 2 ) , (33)
1 − M2 Rk
where k is given by
ω
k= . (34)
c
±
The parameters (ρ̄, c, kmn , R) are known for a given configuration. For the mode decomposition into p̂00 and
p̂10 , the set of unknown parameters ξ = (F̂00 , Ĝ00 , F̂10 , Ĝ10 , δ) is determined from the acoustic pressure field
p̂ for a specific frequency.
The problem of finding ξ is formulated in a nonlinear, least square optimization procedure of fitting the
given pressure field p̂ to the ansatz function Eq. (32), which yields
I X
J X
K
2 !
X
[p̂(xi , rj , φk ) − (p̂00 (xi , rj , φk , ξ) + p̂10 (xi , rj , φk , ξ))] = min. (35)
i=1 j=1 k=1
In Eq. (35) i, j, k indicate the spatial coordinates which are used for the fitting procedure. The optimization
problem is solved by utilizing the Newton-Raphson-Method in MATLAB.
A high number of points used for the fitting procedure in axial, radial and circumferential direction is
favourable. Fig. 1 shows a typical distribution of I, J, K = (10, 8, 10) in the cylindrical part as it is used in
this paper. Since the ansatz of Eq. (32) is only valid for cylindrical geometries the pressure field in the nozzle
is excluded for the fitting procedure.
Using the ansatz for p̂00 and p̂10 from Eq. (32) the nozzle admittance for the first longitudinal and first
transverse mode can be written as
F̂00 − Ĝ00
Y00 (ω) = (36)
F̂00 + Ĝ00
and
κ+ F̂10 + κ− Ĝ10
Y10 (ω) = (37)
F̂10 + Ĝ10
with
±
k10
κ± = ± . (38)
k − M k10
7
M. Schulze, J. Gikadi and T. Sattelmayer
ûi ni − Sf = 0, (40)
where S denotes the maximum velocity value. The small area ensures a low influence of the excitation on the
acoustic mode shapes but allows to excite the desired longitudinal and transverse modes.
Figure 2: Left: location of the excitation, further boundary conditions and computational mesh. Right:
shape of the excitation function.
8
Prediction of Acoustic Admittance of Frequency Space
By solving Eq. (12) - (13) and applying p̂ = c2 ρ̂, the acoustic pressure field is obtained. The finite ele-
ment discretization and the solution procedure is performed in COMSOL Multiphysics 4.3a using MUMPS
(Multifrontal Massively Parallel Sparse) direct solver. The three dimensional computational mesh consists
of approximately 7400 degrees of freedom using linear shape functions on tetrahedral elements in a unstruc-
tured grid for both considered configurations, see Fig. 2. In the region close to the excitation and within the
subsonic region slightly upstream of the nozzle throat, the mesh is finer to provide a higher resolution of the
acoustic waves. The time consumption for a single frequency calculation on a standard 2.6 GHz, 8 Gbyte
RAM machine is about 2 s, which allows for a very fine frequency resolution.
On the rigid wall free slip boundary conditions are applied according to
ûi ni = 0 (41)
The supersonic flow condition in the divergent part of the nozzle provides a natural anechoic end to the
computational domain, which allows to apply any kind of boundary condition at the outlet of the configura-
tion. For convenience free slip conditions are used.
At the inlet the air flow enters the domain with a homogeneously distributed axial velocity component
over the entire inlet area. Acoustically energy neutral boundary conditions have been proposed to ensure
physically consistent conditions [3]. However, in linear acoustics the inlet boundary condition may not influ-
ence the scattering properties of the nozzle which also allows to choose an arbitrary inlet boundary condition.
Again, for the sake of simplicity, free slip conditions are applied, see Eq. (41).
4 Nozzle designs
The determination and investigation of the nozzle admittance is performed on two nozzle configurations
under ambient temperature conditions using air as fluid. Nozzle A represents a geometrically downscaled
(”photo-scaled”) thrust nozzle, see Fig. 3 (left), with a combustion chamber diameter of 92 mm and a throat
diameter of 60 mm [3]. The convergent part of the nozzle features a length of lA,conv = 73 mm. Fig. 3 shows
the isentropic flow through the nozzle in form of the Mach number and clearly highlights the supersonic
condition in the divergent part.
Nozzle B is also derived from the original thrust nozzle geometry. While featuring the same diameter of
the combustion chamber part, Mach-scaling rules are applied to achieve a similar axial Mach number distri-
bution in the nozzle (”Mach-scaled”). Compared to nozzle A, Mach-scaling results in a significantly increased
length of the convergent nozzle part of lB,conv = 273 mm. Besides, the throat diameter is slightly smaller.
Please note that due to similar Mach numbers in the cylindrical part the cut-on frequency of the first
transverse mode is about 2100 Hz for both designs. Both the length of the cylindrical part and the location
of the excitation are the same for both designs. This provides a comparable influence of the excitation on
the acoustic field. Further information about both nozzle designs is given in Fig. 3, Tab. 1 and [5].
Figure 3: Geometry and Mach number distribution of nozzle A (left) and nozzle B (right).