Professional Documents
Culture Documents
CrossMark
View Export
Online Citation
Depth profiling using the glancing‐incidence and glancing‐takeoff x‐ray fluorescence method
Rev Sci Instrum (October 1995)
Gang Liang (梁钢),1 He-Xia Huang (黄河峡),1,a) Hui-Jun Tan (谭慧俊),1,a) Zhong-Qi Luo (罗中岐),1
Xue-Bin Tang (唐学斌), Can-Min Li (李灿民), and Jia Cai (蔡佳)1,2
1 1
AFFILIATIONS
1
College of Energy and Power Engineering, Jiangsu Province Key Laboratory of Aerospace Power System,
Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
2
College of Aviation Engineering, Nanjing Vocational University of Industry Technology, Nanjing 210023, China
a)
Authors to whom correspondence should be addressed: huanghexia@nuaa.edu.cn and thj@263.net
ABSTRACT
curved isolator increases by 32% compared with the length predicted This paper is sequentially organized in the following ways. First,
using the Waltrup-Billig empirical formula,13 and a radial pressure the primary flow structure upstream of the shock train, i.e., the glanc-
gradient is found in the shock train region, which induces a secondary ing shock wave/boundary layer interaction (GSWBLI) is explored ade-
flow. On this basis, Huang14 studied the curved isolator with complex quately. Then, as the backpressure increases, the impact of the shock
background waves and found that the streamwise pressure gradient of train/glancing shock/boundary layer interaction (STGSBLI) on the
the shock train decreases, whereas the radial pressure gradient flow structure and migration of high- and low-momentum flows are
increases with the decrease in the duct curvature. Moreover, the stable investigated and revealed in detail. Subsequently, the effect of the cen-
operation range of the isolator decreases by about 20%, but appropri- terline curvature on the STGSBLI is analyzed. Ultimately, the quasi-
ate duct curvature can weaken the rapid forward motion phenomenon two-dimensional feature of the center part of the shock train in the
of the shock train because of the interaction of background waves. STGSBLI is discussed thoroughly.
He14 reported that the interaction between incident compression II. DESIGN METHOD OF THE CURVED ISOLATOR
waves and separation can establish a positive feedback mechanism, WITH SIDEWALL CONTRACTION
making it difficult for the shock train to stand still in simplified curved
isolators. First, the design method of such curved isolator with sidewall
In addition, the majority of research on high-speed isolators focus contraction is introduced. The key design principle is based on the
on cylindrical and rectangular ducts.15–17 Kawatsu et al.18 reported idea proposed in Ref. 36, wherein the arbitrary cross-sectional shape is
that the onset of boundary layer separation caused by the leading shock transformed to curvature distribution k(s) as the Cartesian coordinates
of the shock train occurs in diverging rectangular ducts differs from are correlated with the curvature uniqueness, as shown in Fig. 1(a).
that in the constant area duct. Billig et al.19 suggested that the shock Once the entrance and exit of the isolator are prescribed by calculating
train characteristics in the cylindrical and rectangular cross sections the curvature of each point in Fig. 1(b), the corresponding curvature
are similar, but the circular duct is more capable of resisting the com- distributions are also determined. Therefore, this method converts the
problem of shape transition between two sections into the problem of
bustion backpressure.20,21 As the isolator locates between the inlet and
curvature transition between two arbitrary functions with the same
combustor, the shape of the inlet exit is fairly complex, particularly for
defined domain.
inward-turning inlet19 and three-dimensional-compression inlet,22
In particular, the design process is illustrated in Fig. 2. The shape
whereas the shape of the combustor is relatively simple usually with a
of the entrance and exit of the isolator designed herein is shown in Fig.
circular or rectangular shape.23,24 Therefore, the isolator must fulfill a
2(a) (the offset of the isolator is not considered here to emphasize the
2 3 2 3 4
y x x x x x x x
¼ 0:323 þ3 18:450 C ¼6 8 þ3 ; (2)
Hiso Liso Liso Liso Liso Liso Liso Liso
4 5 2 3
x x x x x
þ 24:191 9:418 : (1) A ¼3 2 : (3)
Liso Liso Liso Liso Liso
In addition, the shape and area distribution functions of the isolator In this paper, the entrance shape is a ring-sector originated from a hyper-
are cubic polynomial shown in Eq. (2) and quartic polynomial shown sonic inlet in Ref. 22, of which the aspect ratio is 31.5 and that of the exit
in Eq. (3), respectively, is 3.6. The three-dimensional view of the curved isolator with sidewall
contraction from different perspectives are displayed in Fig. 3. The other cells are filled in the whole domain. The grids are refined near the wall,
specific design parameters of the isolator are shown in Table I. and the height of the first-layer near-wall grid is set to 0.01 mm to meet
the requirements of yþ 1 and zþ 1. As shown in Fig. 4(b), the
III. NUMERICAL SETUPS pressure far-field boundary is adopted at the entrance of the isolator,
A. Numerical methods and the pressure-outlet boundary is adopted at the exit of the isolator.
In the current study, the commercial computational fluid All solid surfaces are set as no-slip adiabatic walls, and symmetry
dynamic software ANSYS FLUENT, based on the finite volume boundary is adopted for the symmetry plane of the isolator. Since the
method, is used to solve the three-dimensional compressible curved isolator with sidewall contraction studied in this paper is
Reynolds-averaged Navier–Stokes equations. The j–x SST turbulence extracted from a hypersonic inlet/isolator, which is designed to cruse at
model37 is selected, and the molecular viscosity coefficient is modeled the Mach number of 7.0.22 The incoming airflow goes through a series
using the Sutherland formula. The implicit scheme is used for time- of shock and compression waves to the entrance of the isolator, the
marching to accelerate convergence. Roe’s scheme is used to calculate average Mach number at that cross section reduces to 3.46. Therefore,
FIG. 12. Pathlines originated from entrance in-state without backpressure: (a) upper surface and (b) bottom surface.
distributions are compared in Figs. 18 and 19. As can be seen, the shock-expansion waves within the shock train, a tiny discrepancy
center part of the shock train maintains an identical wave configu- of pressure can be noticed in the upper surface of the center part
ration and streamwise position in different spanwise slices. Since in Fig. 19(a), whereas the bottom flow in the center part is prelimi-
the surface pressure on the upper surface is dominated by the narily low-momentum flow, of which the pressure increases
FIG. 17. Surface pressure contours under different backpressures: (a) pexi/pent ¼ 5 and (b) pexi/pent ¼ 7.
monotonously, and there is almost no difference between these sli- different backpressures, which further indicates the quasi-two-dimen-
ces, as exhibited in Fig. 19(b). sional characteristics of the center part of the shock train. Affected by
the STGSBLI, the lateral pressure gradient caused by the glancing
shock is terminated at the intersection point N in Fig. 20(a), which is
C. Flow migration in the shock train/glancing the boundary of the center part of the shock train. Consequently,
shock/boundary layer interaction region when the backpressure is 5pent, the lateral migration of low-
Due to the aforementioned two parts of the shock train in the momentum flow from the sidewall (blue color) does not reach the
STGSBLI region, the migration characteristics of high- and low- symmetry plane, and it is terminated at the boundary of the center
momentum flows in the isolator show different tendencies and are part of the shock train, as shown in Figs. 20(a) and 20(b). In addition,
exhibited in Fig. 20 (the meanings of all colors and lines in Fig. 20 are when the backpressure increases to 7pent, the side part of the shock
consistent with those in Fig. 12). As shown in Fig. 20, the trajectories train is confined in a narrower region, leading to the shortening of the
of the main flow pathlines (red color) have no lateral migration under distance of the lateral migration of low-momentum flow from the
sidewall, as shown in Figs. 20(c) and 20(d). Therefore, the STGSBLI sidewall, the low-momentum flow principally accumulates in the cen-
contributes to the difference of flow structures in two regions. Near ter region, exhibiting quasi-two-dimensional feature. However, the
the symmetry plane, the flow structure is dominated by the center part supersonic high-speed flow shaded with red color lies near the side-
of the shock train with quasi-two-dimensional feature, in which the wall, which represents the aforementioned glancing shock, reflected
airflow is relatively uniform with no lateral migration. However, near shock and the side part of the shock train region. Since the side part of
the sidewall, the airflow is dominated by the multi-shock system and the shock train exhibits a hysteretic effect, compared with the center
the side part of the shock train with evident lateral migration. part of the shock train, the supersonic region extends longer along
Distinct flow-migration patterns in the two parts of the shock streamwise direction than that in the center part, terminating at the
train result in different distributions of high- and low-momentum exit of the isolator. When the backpressure increases to 7pent, the influ-
FIG. 20. Pathlines originated from entrance under different backpressures: (a) pexi/pent ¼ 5, upper surface; (b) pexi/pent ¼ 5, bottom surface; (c) pexi/pent ¼ 7, upper surface;
and (d) pexi/pent ¼ 7, bottom surface.
side part of the shock train is enhanced, contributing to the reduction dimensional characteristics of the center part. As the backpressure
in the velocity gradient between the high- and low-momentum flows, ascends from 5pent to 10pent, the total pressure recovery coefficient on
which ultimately results in a weaker mixing effect between them. the symmetry plane of the exit progressively increases approximately
Therefore, the high- and low-momentum flow distribution on by 37.5%, as shown in Fig. 23. Actually, this is principally due to the
the exit of isolator differs from that in the straight isolator. Figure 22 fact that the Mach number ahead of the shock train decrease from
compares the exit flow Mach number contours with different back- 3.62 to 3.46, which reduces the pressure loss of the shocks and the skin
pressures. As can be seen, the high-speed flows principally distribute friction behind of the shock train. Meanwhile, as the shock train
near the both sidewalls, especially when the backpressure is 5pent in moves upstream, the glancing shock as well as its reflected shock is
Fig. 22(a). Nevertheless, the low-speed flows with less spanwise gra- confined into a narrower region; thus, the related pressure loss
FIG. 22. Mach number contours on the exit (mirrored by the symmetry plane)
under different backpressures: (a) pexi/pent ¼ 5, (b) pexi/pent ¼ 7, (c) pexi/pent ¼ 9,
And (d) pexi/pent ¼ 10. FIG. 23. Total pressure distributions on the symmetry plane of the exit.
the rising backpressure contribute to the formation of the STGSBLI, side view of the curved sidewall-contraction isolators with different
which indicates that this phenomenon is the dominated flow in the centerline shapes
2 3
sidewall-contraction isolator. To demonstrate its universality, several y x x x
isolators with different centerline shapes are designed by the methods ¼ 0:323 þB þ ð4B 6:45Þ
Hiso Liso Liso Liso
introduced in Sec. II. The centerline shapes are mainly determined by 4 5
the parameter B as presented in Eq. (4). If B ¼ 3, Eq. (4) equals to x x
þ ð5B þ 9:19Þ þ ð2B 3:42Þ : (4)
Eq. (1), which corresponds to the baseline isolator analyzed above. If Liso Liso
B ¼ 3, the centerline shape is almost a straight line. As the value of B The flow structures in these isolators with different centerlines
decreases, the isolator bends more severely. Figure 24 compares the are exhibited in Fig. 25. It can be discovered that the flow structures of
FIG. 25. Wall pressure distributions and Mach number contours of isolators with different centerlines when the backpressure is 5pent.
(GSWBLI), resulting in the separation of the local boundary layer. The ACKNOWLEDGMENTS
GSWBLI rolls up vortices adjacent to the symmetry plane on upper
This work was funded by the National Natural Science Foundation
and bottom surfaces and drives them to migrate along the glancing
of China (Grant Nos. 51906104, 12272177, and U20A2070), the
shock. Once these vortices meet on the symmetry plane, they further
National Science and Technology Major Project (No. J2019-II-0014-
develop into two pairs of streamwise vortices, contributing to an evi-
0035), 1912 Project (Nos. 2019-JCJQ-DA-001-067 and 2019-JCJQ-DA-
dent accumulation of low-momentum flow near the symmetry plane.
001-164), the Foundation of National Key Laboratory of Transient
Then, the glancing shocks from both sides reflect on the symmetry
Physics (No. 6142604200212), and the priority academic program
plane, which induces GSWBLI again in the aft part of the isolator, roll-
development of Jiangsu higher education institutions.
8
Z. A. Wang, J. T. Chang, W. X. Hou, and D. R. Yu, “Low-frequency unsteadi- 25
C. P. Wang, X. A. Tian, and K. M. Cheng, “Numerical analysis of pseudo-
ness of shock-wave/boundary-layer interaction in an isolator with background shock flow diffusion phenomenon in variable cross-section ducts,” Proc. Inst.
waves,” Phys. Fluids 32(5), 056105 (2020). Mech. Eng., Part G 222, 1109–1121 (2008).
9
J. G. Tan, J. P. Wu, and Z. G. Wang, “Experimental and numerical investiga- 26
X. A. Tian, C. P. Wang, and K. M. Cheng, “Numerical investigation of pseudo-
tions on flow fields and performance of dual combustion ramjet,” Proc. Inst. shock waves in two kinds variable cross-section ducts,” in Second International
Mech. Eng., Part G 228, 920–929 (2014). Symposium on Physics of Fluids (NUAA, Nanjing, 2008), pp. 23–24.
10 27
A. V. Gnos, E. C. Watson, W. R. Seebaugb, R. J. Sanator, and J. P. Decarlo, Z. A. Wang, X. A. Xin, R. Z. Huang, C. Kong, C. K. Lv, and J. T. Chang,
“Investigation of flow fields within large scale hypersonic inlet models,” NASA “Mechanism of shock-train/boundary-layer interaction in spanwise concave
Technical Note No. D-7150 (NASA, 1973). isolator channels,” Acta Astronaut. 199, 232–248 (2022).
11
Z. A. Wang, J. T. Chang, N. Li, and K. J. Xu, “Experimental investigation of the 28
S. C. Xu, Y. Wang, Z. G. Wang, X. Q. Fan, and X. Y. Zhao, “Design and analy-
throttling process and oscillation mechanism in a supersonic cascade,” AIP sis of a hypersonic inlet with an integrated bump/forebody,” Chin. J. Aeronaut.
Adv. 9, 055103 (2019). 32, 2267–2274 (2019).
12
H. J. Tan and S. Sun, “Preliminary study of shock train in a curved variable- 29
S. C. Xu, Y. Wang, Z. G. Wang, X. Q. Fan, and B. Xiong, “Design method for
section diffuser,” J. Propul. Power 24, 245–252 (2008). hypersonic bump inlet based on transverse pressure gradient,” J. Zhejiang
13
P. J. Waltrup and F. S. Billig, “Structure of shock waves in cylindrical ducts,” Univ. Sci. A 23, 479–494 (2022).
AIAA J. 11, 1404–1408 (1973). 30
A. G. Panaras, “Review of the physics of swept-shock/boundary layer interac-
14
H. X. Huang, H. J. Tan, and S. Sun, “Behavior of shock train in curved isolators tions,” Prog. Aerosp. Sci. 32, 173–244 (1996).
with complex background waves,” AIAA J. 56, 329–341 (2018). 31
H. Kubota and J. L. Stollery, “An experimental study of the interaction between
15
L. H. He, H. Chen, L. J. Yue, Q. F. Zhang, and W. N. Wu, “Interaction of a a glancing shock wave and a turbulent boundary layer,” J. Fluid Mech. 116,
shock train with inherent isentropic waves in a curved isolator,” Phys. Fluids 431–458 (1982).
34(6), 066106 (2022). 32
F. S. Alvi and G. S. Settles, “Physical model of the swept shock wave/boundary-
16
K. R. Sekar et al., “On the unsteady throttling dynamics and scaling analy- layer interaction flowfield,” AIAA J. 30, 2252–2258 (1992).
sis in a typical hypersonic inlet–isolator flow,” Phys. Fluids 32, 126104 33
D. D. Knight, D. Badekast, C. C. Horstmant, and G. S. Settles, “Quasi-conical
(2020). flowfield structure of the three-dimensional single fin interaction,” AIAA J. 30,
17
Z. A. Wang, J. T. Chang, Y. M. Li, R. Y. Chen, W. X. Hou, J. F. Guo, and L. 2809–2816 (1992).
J. Yue, “Oscillation of the shock train under synchronous variation of 34
P. E. Rodi and D. S. Dolling, “Behavior of pressure and heat transfer in sharp
incoming Mach number and backpressure,” Phys. Fluids 34(4), 046104 fin-induced turbulent interactions,” AIAA J. 33, 2013–2019 (1995).
(2022). 35
J. Fang, Y. F. Yao, A. A. Zheltovodov, and L. P. Lu, “Investigation of three-
18
K. Kawatsu, S. Koike, T. Kumasaka, G. Masuya, and K. Takita, “Pseudo-shock dimensional shock wave/turbulent-boundary-layer interaction initiated by a
wave produced by backpressure in straight and diverging rectangular ducts,” single fin,” AIAA J. 55, 509–523 (2017).
AIAA Paper No. 2005-3285, 2005. 36
H. C. Zhou, H. J. Tan, and X. P. Li, “Unique design method of subsonic inlet with