Shock Boundary Layer Interactions – A Multiphysics

Approach

A thesis submitted to the

Graduate School of the University of Cincinnati

in partial fulfillment of the

requirements for the degree of

Master of Science

in the Department of Mechanical and Materials Engineering

of the College of Engineering and Applied Science
by

Kalyani R. Bhide

B.E., Don Bosco Institute of Technology, University of Mumbai, India, 2014

Committee:
Dr. Shaaban Abdallah, Chair
Dr. Milind Jog
Dr. Kiran Siddappaji
Abstract

Shock waves are a major source of drag in supersonic flows as they affect the aero-thermodynamic
performance and impose pressure and temperature loads on the structure under consideration.
The aim of this work is to address such multiphysics phenomenon and mitigate the losses due
to shock waves in a series of low to high aspect ratio rectangular supersonic nozzles with sharp
and smooth throats.
3D steady RANS CFD is performed on a baseline converging-diverging nozzle (exit aspect
ratio 2 and design Mach number 1.5) with sharp throat and the results are validated with the
experimental data. Turbulence model study shows that k-omega SST compares better with the
experimental data. A shock wave is present in the baseline nozzle due to the sharp throat. In
order to eliminate the sharp throat, a general-purpose curve generator - Gencurve is developed
and implemented using Python 3.5. Equivalent smooth nozzle geometries are created using
this tool. These show improvements in various performance parameters such as discharge
coefficients, thrust coefficients, etc., due to the mitigated shocks and reduced boundary layer
thicknesses.
The baseline and the equivalent smooth nozzle geometries are further analyzed at design
(NPR 3.67) and off-design conditions using multi-physics simulations, which are, for the first
time, performed on rectangular supersonic nozzles. This work highlights the significance of
Fluid-Thermal-Structural-Interaction (FTSI) simulations as a diagnosis of existing designs (exit
aspect ratio 2 with sharp throat) and as a means of preliminary investigation to ensure feasibility
of new designs before conducting experimental and field tests. Structural deformation in the
baseline design is far less than the boundary layer thickness as the impact of Shock Boundary
Layer Interactions (SBLI) is not as severe. FTSI demonstrates that the discharge coefficient of

i
the improved design is 0.99 and its structural integrity remains intact at off-design conditions.
This proves the feasibility of the improved design. Although the influence of FTSI is shown
for a nozzle, the approach is general and essential in any product design cycle or as a prelude
to building prototypes.
Investigating low (1, 2, 3) and high (8, 12) aspect ratio nozzles with sharp throats is also a
focus of this work. Although the throat shocks are present in these configurations, they become
weaker and the shock cell size reduces as the aspect ratio increases. Hence, the magnitude
of wall shear stress at the nozzle throat decreases. The boundary layer thickness on nozzle
exit plane along the minor axis decreases as the aspect ratio increases. This improves the
discharge coefficient due to the less blockage offered by the boundary layer. All jets spread
more in the minor axis than in the major axis and eventually become circular a few diameters
downstream the nozzle exit. The velocity decay in high aspect ratio nozzles is faster than in
the low aspect ratio nozzles. An issue of modeling quarter section of the flow domain using
symmetry boundary conditions is also addressed.

ii
Acknowledgments

Several people made this journey possible.

I’m very thankful to my family - my grandparents, my parents and my brother for ev-
erything. Their love, support and encouragement is invaluable. I dedicate this work to my
grandparents.
Special thanks to Dr. Abdallah for being my advisor who always encouraged me to become
an independent researcher. He also gave me opportunities to work on several other projects
where I could improve my knowledge and gain more experience.
I’m very fortunate to have Dr. Siddappaji as my mentor from day zero of my research. His
guidance, vision and ideas are invaluable. It would have been difficult to reach here without his
guidance.
Thanks to Dr. Jog for being on my thesis committee.
Thanks to Dr. Gutmark and Florian for sharing the experimental data and pointing me to
useful information.
Thanks to Rob for his assistance with the remote cluster.
Thanks to everyone who asked me some good questions during the course of this work. It
triggered my thought process and resulted into a better outcome.
Special thanks to Dr. Prasanna Nambiar for being an influential professor of CFD, fluid
mechanics and turbomachinery during my bachelor’s degree.
Thanks to my friends back at home and from UC for always being there for me.
Lastly thanks to my well wishers.

iv
Contents

Abstract i

Acknowledgments iv

Contents vii

List of Figures x

List of Tables xi

Nomenclature xii

1 Background 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2.1 Supersonic nozzle flow . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.2 Performance coefficients . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.3 Free jets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Literature Review 6

3 Methodology 9
3.1 3D steady RANS CFD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.1.1 Geometry, mesh and boundary conditions . . . . . . . . . . . . . . . . 10
3.1.2 Convergence metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

v
 3.1.3 Experimental validation . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1.4 Grid resolution study . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.5 Turbulence model study . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2 Multiphysics modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2.1 Governing equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2.2 Challenges in FTSI . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2.3 FTSI workflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

4 Gencurve – Python based smooth curve generator 25

4.1 Wall curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2 Parametric geometry with domain . . . . . . . . . . . . . . . . . . . . . . . . 28

5 Isentropic nozzle flow and oblique shock calculator 30

6 Results 33
6.1 3D steady RANS CFD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
6.1.1 Performance at various operating conditions – sharp vs smooth nozzles 33
6.1.2 Wall curve and shock boundary layer interactions . . . . . . . . . . . . 35
6.1.3 Symmetry boundary conditions and asymmetric flow . . . . . . . . . . 38
6.1.4 Aspect ratio effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.1.4.1 Low aspect ratio nozzles . . . . . . . . . . . . . . . . . . . . 41
6.1.4.2 High aspect ratio nozzles . . . . . . . . . . . . . . . . . . . 47
6.1.4.3 Wake profiles . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.2 Multiphysics modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.2.1 Structural deformation and boundary layer thickness . . . . . . . . . . 54
6.2.2 Existing (sharp AR 2) and improved (smooth AR 2) design . . . . . . . 55
6.2.3 Multiphysics modeling as a prelude to prototype . . . . . . . . . . . . 60

7 Conclusions 62
7.1 Gencurve - Python based curve generator . . . . . . . . . . . . . . . . . . . . 62
7.2 Aspect ratio and wall curvature effects . . . . . . . . . . . . . . . . . . . . . . 63

vi
 7.3 Fluid-thermal-structural interaction simulations . . . . . . . . . . . . . . . . . 63
7.4 Symmetry boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . 64

8 Discussion & future work 65

9 Contributions 67

Bibliography 69

A Jet spread 74

vii
List of Figures

1.1 Nozzle operation – (a) Over-expanded (b) Design (c) Under-expanded regime
[7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Illustration of displacement thickness as a boundary layer blockage [3]. . . . . 3
1.3 Free jet [6]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.1 F22 raptor [12]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.2 Sharp AR 2 rectangular nozzle - Baseline geometry [19]. . . . . . . . . . . . . 8

3.1 Physics models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.2 Baseline case – AR 2 rectangular nozzle taken from Viswanath et al. [13] (Di-
mensions in mm). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.3 Mesh and boundary conditions (Domain extends further downstream). . . . . . 11
3.4 Prism layers on nozzle walls. . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.5 Contours of y+ on nozzle walls. . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.6 Residuals from CFD computation. . . . . . . . . . . . . . . . . . . . . . . . . 13
3.7 Experimental validation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.8 Conical nozzle validation with JPL experiment [33]. . . . . . . . . . . . . . . . 15
3.9 3D perspective of grid shown on XY and XZ symmetry plane (quarter view). . 16
3.10 Grid resolution study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.11 Turbulence model study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.12 One-way coupled FTSI flowchart. . . . . . . . . . . . . . . . . . . . . . . . . 22
3.13 FTSI methodology with the sequence of steps numbered as shown in the image. 24

4.1 Improved nozzle design with contoured walls shown in green. . . . . . . . . . 26

viii
4.2 Nozzle geometry created using cubic polynomial. . . . . . . . . . . . . . . . . 26
4.3 Fifth order polynomial wall curve. . . . . . . . . . . . . . . . . . . . . . . . . 27
4.4 Parametric domain as a function of equivalent diameter. . . . . . . . . . . . . 29

5.1 Area ratio vs Mach [1]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

5.2 Oblique shock wave [45]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

6.1 NPR vs discharge coefficient – sharp vs smooth. . . . . . . . . . . . . . . . . . 34

6.2 NPR vs pressure recovery – sharp vs smooth. . . . . . . . . . . . . . . . . . . 34
6.3 NPR vs thrust coefficient – sharp vs smooth. . . . . . . . . . . . . . . . . . . . 35
6.4 Mach contours on XY and XZ symmetry planes. . . . . . . . . . . . . . . . . 36
6.5 Wall shear stress – sharp vs smooth. . . . . . . . . . . . . . . . . . . . . . . . 37
6.6 Velocity vectors with total pressure contours – sharp vs smooth. . . . . . . . . 38
6.7 Contours of streamwise vorticity on exit plane. . . . . . . . . . . . . . . . . . 40
6.8 Shadowgraph – XY symmetry plane. . . . . . . . . . . . . . . . . . . . . . . . 41
6.9 Contours of wall shear stress. . . . . . . . . . . . . . . . . . . . . . . . . . . 42
6.10 AR vs boundary layer profiles on exit plane along minor axis. . . . . . . . . . . 43
6.11 AR vs performance coefficients. . . . . . . . . . . . . . . . . . . . . . . . . . 43
6.12 AR vs PC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
6.13 Iso-surfaces of Mach 0.3 shown in green with Mach contours from x/De = 3 to
x/De = 20. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
6.14 Jet spread. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
6.15 Comparison of jet spread – AR 1, 2, 3.. . . . . . . . . . . . . . . . . . . . . . 47
6.16 Mach contours on XY and XZ symmetry plane (quarter view) – AR 1. . . . . . 48
6.17 Mach contours on XY and XZ symmetry plane (quarter view) – AR 2. . . . . . 48
6.18 Mach contours on XY and XZ symmetry plane (quarter view) – AR 3. . . . . . 49
6.19 Mach contours on XY and XZ symmetry plane (quarter view) – AR 8. . . . . . 49
6.20 Mach contours on XY and XZ symmetry plane (quarter view) – AR 12. . . . . 50
6.21 Contours of wall shear stress. . . . . . . . . . . . . . . . . . . . . . . . . . . 50
6.22 AR vs discharge coefficient. . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

ix
6.23 AR vs PC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
6.24 Velocity profile at x/De = 5 – minor axis symmetry plane. . . . . . . . . . . . . 52
6.25 Velocity profile at x/De = 5 – major axis symmetry plane. . . . . . . . . . . . . 53
6.26 Velocity profile at x/De = 10 – minor axis symmetry plane. . . . . . . . . . . . 53
6.27 Velocity profile at x/De = 10 – major axis symmetry plane. . . . . . . . . . . . 53
6.28 Velocity profile at x/De = 20 – minor axis symmetry plane. . . . . . . . . . . . 53
6.29 Velocity profile at x/De = 20 – major axis symmetry plane. . . . . . . . . . . . 54
6.30 Mach contours on symmetry planes in existing nozzle shown on a quarter section. 55
6.31 Contours of wall shear stress – sharp vs smooth AR 2. . . . . . . . . . . . . . . 56
6.32 Contours of Pt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.33 Total deformation (exaggerated 3x for better visualization) and thermal strain
in improved design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6.34 Off-design analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.35 Imported pressure loads on nozzle walls. . . . . . . . . . . . . . . . . . . . . . 60
6.36 Product development cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

A.1 Jet spread – AR 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

A.2 Jet spread – AR 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
A.3 Jet spread – AR 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
A.4 Jet spread – AR 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
A.5 Jet spread – AR 12. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

x
List of Tables

3.1 Grid details. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4.1 Input parameters to Gencurve – cubic polynomial. . . . . . . . . . . . . . . . 27

6.1 CFD calculated boundary layer thickness on exit plane. . . . . . . . . . . . . 42

xi
Nomenclature

ṁ Mass Flow Rate

Ae /A∗ Nozzle exit to throat area ratio

De Equivalent diameter

Pt Total pressure

Tt Total temperature

AR Aspect Ratio

CAD Computer Aided Design

CFD Computational Fluid Dynamics

FEA Finite Element Analysis

FTSI Fluid Thermal Structural Interaction

LES Large Eddy Simulation

NPR Nozzle Pressure Ratio

PC Potential Core

RANS Reynolds Averaged Navier Stokes

SST Shear Stress Transport

TR Temperature Ratio

xii
Chapter 1

Background

1.1 Motivation

Faster transportation has always been mankind’s interest. With the advent of jet engines it is
possible to achieve supersonic speeds. Converging-diverging nozzles aid in this process due
to their varying cross-sectional area by accelerating the turbine exhaust to supersonic speeds
which produces the thrust. Nozzles have a wide variety of applications ranging from spray
nozzles to rocket engine exhaust nozzles, etc. Supersonic nozzle flow is one of the most com-
plicated applications. The shape of the nozzle affects its performance and the properties down-
stream. Consistent efforts have been made to understand and improve this phenomenon either
by performing experiments and/or numerical simulations which could only be possible in a last
few years due to the latest developments in the simulation technology.
This work is one such contribution where a series of nozzles with different aspect ratios and
wall curves are analyzed using numerical simulations to understand the multiphysics effects
associated with it.

1.2 Introduction

Some important concepts with respect to supersonic nozzle flow are briefly discussed in this
section. For further information see references [1–6].

1
1.2.1 Supersonic nozzle flow

As the flow accelerates through a converging-diverging nozzle, the static pressure and temper-
ature decrease. When the back pressure Pb starts increasing, an oblique shock wave forms at
the nozzle exit due to the pressure difference. This condition is shown in Figure 1.1a and is
called over-expanded flow. When the back pressure equals the exit pressure, the flow is ideally
expanded and is represented in Figure 1.1b which is also called the design condition. If the
back pressure starts decreasing, expansion waves form at the nozzle exit as shown in Figure
1.1c and is called under-expanded flow. Addressing the flow physics associated with shock
wave formation will be a focus of this work.

Figure 1.1: Nozzle operation – (a) Over-expanded (b) Design (c) Under-expanded regime [7].

1.2.2 Performance coefficients

Performance coefficients [2] used as a metrics of nozzle performance are listed here.

2
 Figure 1.2: Illustration of displacement thickness as a boundary layer blockage [3].

There is boundary layer formation on nozzle walls as the flow develops through a nozzle.
The displacement thickness δ∗ is represented by the presence of boundary layer, which acts as
a blockage to the geometric flow area. This is shown in Figure 1.2. There is a loss of total
pressure due to the presence of boundary layer on the nozzle wall. A discharge coefficient
represents their combined effect on the mass flow rate and is given below.

ṁ actual
Cdg =
ṁideal

where, Cdg is the discharge coefficient, ṁ actual is the actual mass flow rate through the nozzle
and ṁideal is the ideal mass flow rate calculated with isentropic assumptions.
Shock wave formation affects the total pressure recovery which is quantified by total pres-
sure recovery coefficient as given below.

Ptactual
Ptrecovery =
Ptideal

Similarly, the thrust coefficient is represented by the following equation, where, C f g is the
thrust coefficient, Fgactual is the actual thrust and Fgideal is the ideal thrust calculated using Quasi-
1D assumptions.

3
 Fgactual
Cf g =
Fgideal

These coefficients will be used as a measure of nozzle performance for the nozzles under
consideration.

1.2.3 Free jets

Figure 1.3 illustrates the jet issuing from a nozzle. It shows various zones of the jet development
as the jet enters quiescent atmosphere. Typically there are four zones, namely, convergent zone
(potential core), transition zone, self-similar zone and termination zone [6]. For further details
see the reference [6].

Figure 1.3: Free jet [6].

Axis switching is an important factor generally present in asymmetric jets. It is a phe-

nomenon in which the cross-section of an asymmetric jet evolves in such a manner that, after a
certain distance from the nozzle, the major and the minor axes are interchanged [8].
There is a wide variety of literature which talks about jet spreading, mixing process and
the noise characteristics of asymmetric nozzles [8–10]. Various factors such as the nozzle
geometry, inlet velocity profile, inlet temperature affect the jet spread [6]. The nozzle exit
boundary layer thickness affects the jet noise [11]. One effect of altering the nozzle shape is to

4
alter the boundary layer thickness which affects the properties downstream the nozzle exit and
changes the overall nozzle performance. This will be discussed in detail.

1.3 Thesis outline

The effects of shock formation on the nozzle performance and the flow physics associated with
changing the nozzle aspect ratio and wall curve are addressed in this work. Shock mitigation is
also a focus of this work. A series of low to high aspect ratio nozzles with sharp and smooth
throats are considered. The nozzle flow properties largely depend on the exit to throat area
ratio which corresponds to a particular design pressure ratio of the nozzle. Hence, to have a
better reference of comparison, the area ratio and the length of all nozzle configurations are
kept constant.
A brief discussion on the benefits and capabilities of the in-house tool that models smooth
wall curves is presented. Nozzle performance at design and off-design conditions (over-expanded
and under-expanded) is evaluated. The multiphysics effects, particularly, the effect of shock
formation on the structural integrity, boundary layer thickness and the structural deformation
are also discussed.

5
Chapter 2

Literature Review

Rectangular nozzles have been a topic of interest for many researchers over the last fifty years
or so. They are currently being used in military aircrafts such as F-22 raptor (Figure 2.1), YF-
23 and B-2 [12]. Low noise emissions, enhanced mixing and thrust vectoring are some of the
desired features of rectangular nozzles [9, 13]. Considerable work has been done to study the
axis switching phenomenon, vorticity and acoustic emissions from low and high aspect ratio
rectangular nozzles [9, 14, 15]. High aspect ratio nozzles are particularly of interest and have
been investigated due to their applications in the distributed propulsion systems [16].

(a) Rectangular exhaust nozzles [12]. (b) Shock diamonds.

Figure 2.1: F22 raptor [12].

Noise characteristics of rectangular nozzles have been investigated both experimentally and
numerically (RANS/LES) [10, 17, 18]. Shock formation, turbulent mixing and screech tones
affect the noise generation [13, 14, 19, 20]. Double diamond shock pattern is present in rect-
angular nozzles with sharp throats [21] at design and off-design operating conditions. Throat

6
shock formation causes wave drag [4] and side loads due to the shock reflections and affects
the structural integrity and overall nozzle performance. This needs to be addressed further to
evaluate the stresses and strains induced in the nozzle. Although Finite Element Analysis was
performed by Frate and Bridges [22] during the design phase of rectangular nozzles, coupled
effects of multi-physics modeling are not addressed previously. The influence of shock wave
boundary layer interaction on the structure is crucial and must be accounted for as shown by
Riley et al. [23].
FTSI modeling gives an insight into the coupled effects of flow physics and structural in-
tegrity at high operating pressures and temperatures. Considering these factors, multi-physics
simulations are performed. Two nozzle designs are considered – 1) An existing experimentally
tested design which has in-nozzle shocks due to the sharp throat and 2) Improved design with
smoothly contoured nozzle walls which has higher discharge coefficient due to less blockage
offered by the boundary layer as shown by the author [24]. This design is further investigated
at off-design conditions to simulate the maximum operating pressure ratio. If the design can
withstand these loads, then it is safe at lower operating points.
Baier et al. [14] and Heeb et al. [19] conducted the experiments on sharp AR 2 nozzle at
University of Cincinnati’s Hot Jet Nozzle Rig (HJNR). Figure 2.2a shows the experimental set
up where the nozzle is mounted on a test stand and exhausts vertically upwards. Figure 2.2b
shows the cross-sectional view of the experimentally tested nozzle i.e. the baseline geometry.
The multiphysics simulations are performed such that they replicate the experimental set up.

7
 (a) Experimental set up [19]. (b) Cross sectional view of the nozzle [19].

Figure 2.2: Sharp AR 2 rectangular nozzle - Baseline geometry [19].

One way to mitigate the shock formation is to smoothly contour the nozzle walls as shown
by the author [24]. This is because the wall curve sets up the pressure gradient and affects the
shock formation [4]. According to previous findings, a variety of methods to smoothly contour
the nozzle walls have been developed by using the Method of Characteristics (MOC) [1] and
many other design codes [25,26]. This work presents a general-purpose smooth curve generator
for nozzles which can also be used for modeling other geometries without using CAD.
Another goal is to compare low vs high aspect ratio nozzles using various metrics of per-
formance. Boundary layer thicknesses at the nozzle exit planes and the jet development down-
stream the nozzle exit are also investigated for these nozzles.
Previous numerical studies [16, 17] used symmetry boundary conditions by taking advan-
tage of symmetry in the nozzle geometry about major and minor axis. Hence, this work also
presents a brief discussion on modeling quarter domain vs modeling the full flow domain.

8
Chapter 3

Methodology

The methodology for 3D steady RANS CFD and multi-physics modeling i.e. Fluid-thermal-
structural interaction (FTSI) is presented here. Solidworks [27] and Python 3.5 [28] are used
for geometry modeling. STAR-CCM+ v11.02 [29] is used for RANS CFD and ANSYS Work-
bench v14.0 & v18.1 [30] are used for multi-physics modeling. Tecplot 360 [31] is used for
post-processing the results of CFD and multiphysics simulations.

3.1 3D steady RANS CFD

In addition to the baseline geometry, simulations for a series of low to high aspect ratio equiv-
alent nozzles obtained by modifying the baseline nozzle configuration shown in Figure 3.2 at
over (NPR 3), ideally (NPR 3.67) and under expanded (NPR 4) flow conditions are presented.
All nozzle configurations are designed such that they have the same design NPR i.e. 3.67. This
is achieved by keeping De , overall length and area ratio of exit to throat the same for better
reference of comparison. Boundary conditions and computational domain are the same for all
simulations.
Standard RANS equations for mass, momentum and energy transport are solved on a Carte-
sian coordinate system with Menter’s k-omega SST turbulence model (refer 3.2.1). The physics
models are listed in Figure 3.1.

9
 Figure 3.1: Physics models.

3.1.1 Geometry, mesh and boundary conditions

Baseline geometry is shown in Figure 3.2. The computational domain is based on the work done
by Viswanath et. al [13] and extends 100 De downstream, 10 De upstream from the nozzle exit
and 15 De from major and minor axis planes. A Python based smooth wall curve generator -
Gencurve is developed to smoothly contour the nozzle walls by solving polynomial equations.
Refer Chapter 4 for further details. Polyhedral prism layer mesh is used with volumetric mesh
refinements for the nozzle and the jet. Polyhedral mesh offers a number of benefits in terms of
computational time as mentioned in reference [32].

Figure 3.2: Baseline case – AR 2 rectangular nozzle taken from Viswanath et al. [13] (Dimen-
sions in mm).

10
 Figure 3.3 shows the mesh, boundary conditions and prism layers. Since shocks are present
inside the nozzle, resolving the viscous sublayer is important to predict the boundary layer
accurately.

Figure 3.3: Mesh and boundary conditions (Domain extends further downstream).

Hence, y+ values are less than 3 on the nozzle walls (Figure 3.5) for all simulations. This
is achieved by using 15 prism layers, total prism layer thickness of 1 mm with stretching factor
of 1.5 as shown in Figure 3.4. NPR is 3.67 with Tt = 300 K and atmospheric pressure &
temperature are used as free stream.

11
 Figure 3.4: Prism layers on nozzle walls.

Figure 3.5: Contours of y+ on nozzle walls.

3.1.2 Convergence metrics

3D RANS steady CFD with k-omega SST turbulence model is performed with all y+ wall
treatment and grid sequencing. Along with the residuals for mass, momentum and energy,
various other quantities such as the thrust and mass flow rate at the exit plane are monitored for

12
each iteration and the residuals are shown in Figure 3.6. Twenty Intel Xeon (2.6 GHz) processor
cores are used for each simulation which took about 9-15 hours to complete depending on the
mesh size which was approximately 4 million cells for the coarse grid level.

Figure 3.6: Residuals from CFD computation.

3.1.3 Experimental validation

Experimental data for the baseline geometry is taken from Baier et. al [14]. Figure 3.7 shows
the comparison of velocity magnitudes along the jet centerline from nozzle exit to 20 De down-
stream for the experimental and numerical results. Note that the experimental measurements
start from x/De ≈ 0.82 and not at the nozzle exit i.e. x/De = 0. The error in the experimental
measurements is ±1%. Figure 3.7 shows that the error between experimental data and simulated
data is within 5% until 10 De , which is calculated using the following equation:

Vexp − Vsim
e = ( )x100 (3.1)
Vexp

where, e is the percent error, Vexp is the experimentally calculated velocity in m/s and Vsim
is the CFD calculated velocity in m/s.

13
 Figure 3.7: Experimental validation.

JPL circular conical nozzle [33] is also validated numerically which serves as a second
witness to the CFD methodology and boundary conditions. Figure 3.8 shows the validation for
the conical nozzle, where, the ratio of static to total pressure (Ps /Pt ) along the nozzle walls is
compared with the experimental data. It is in a good agreement with the experimental data and
is sufficient for the validation purpose of this research.

14
 Figure 3.8: Conical nozzle validation with JPL experiment [33].

3.1.4 Grid resolution study

Grid resolution study is conducted by keeping the total thickness of prism layers and base size
the same and by varying the refinement cell size for each grid level.

15
 (a) Coarse – 4 million.

(b) Medium – 9 million.

(c) Fine – 12 million.

Figure 3.9: 3D perspective of grid shown on XY and XZ symmetry plane (quarter view).

16
 The details of the grids and mass averaged quantities at the throat and exit are presented in
Table 3.1. Figure 3.10 shows the comparison of velocity magnitudes along the jet centerline
for coarse, medium and fine grid levels with the experimental data from Baier et. al [14]. The
velocity magnitudes and mass averaged quantities do not change significantly with each grid
refinement. Hence, coarse grid level is used for all simulations to save computational time.

Units Coarse Medium Fine

Cell count million 4.56 9.56 12
Base size mm 50 50 50
Refinement size mm 0.5 0.35 0.3
Throat Pt Pa 370884 370887 370805
Mach 0.8834 0.8832 0.8859
ṁ kg/s 0.2398 0.2397 0.2394
Exit Pt Pa 363594 363707 363437
Mach 1.4857 1.4865 1.4894
ṁ kg/s 0.241 0.241 0.2404

Table 3.1: Grid details.

Figure 3.10: Grid resolution study.

17
3.1.5 Turbulence model study

Three turbulence models namely Spallart-Allmaras, k-epsilon and Menter’s k-omega SST are
used [34]. Figure 3.11 shows the comparison of three turbulence models with the experimental
data. k-omega SST predicts the velocity magnitudes more accurately where as other turbulence
models show large deviation in velocity magnitudes from 10 De to 20 De . k-omega SST is
known for its better performance under adverse pressure gradients in boundary layers [34],
hence it is used for all simulations.
The literature by Georgiadis et al. [35], Truemner et al. [36] highlights that capturing initial
jet growth region is still a difficulty for RANS because the computed rates of jet mixing are
generally lower than that of the experimental data, turbulence intensities are underestimated,
and potential core lengths are overestimated by RANS for free jets. The turbulence model study
(Figure 3.11) shows that k-epsilon and Spallart-Allmaras (SA) predict the shock cell amplitude
better than k-omega SST, until a few diameters downstream the nozzle exit. However, they
do not capture the velocity decay accurately further downstream i.e. 10 De onward. Hence,
k-omega SST is chosen for this study as it gives a good approximation of the overall trend in
velocity decay. Previous literature on turbulence models and corrections [37] talks about the
effects of compressibility on the dissipation rate of turbulent kinetic energy. Turbulence model
and compressibility correction play a key role in capturing the shock cell amplitude. The error
in validation is likely due to the underestimated dissipation rate of turbulent kinetic energy in
the current RANS model.

18
 Figure 3.11: Turbulence model study.

3.2 Multiphysics modeling

3.2.1 Governing equations

3D steady RANS equations [38] are solved for compressible flow of air modeled as an ideal gas.
k-omega SST turbulence model is used for CFD simulations. Fourier’s steady heat conduction
equations [39] are solved for the nozzle. Since this work is a preliminary investigation of FTSI
effects on the structural integrity and boundary layer flow, all simulations represent steady
state operation. Twenty Intel Xeon (2.6 GHz) processor cores are used for each simulation
which took about 15-20 hours to complete depending on the CFD mesh size of approximately
1 million elements.
Equations 3.2-3.6 [38] represent the governing equations for conservation of mass, momen-
tum and energy, where e is the internal energy per unit mass, P is the pressure, ρ is the density
and u, v, w are the x, y, z velocity component vectors. The viscous stress tensors, σij and τij ,
are defined in Equation 3.7–3.9. A density-based solver is used, and a second order upwind

19
scheme is used for spatial discretization. Roe-FDS is used for convective fluxes. Fourier’s
steady heat conduction Equation 3.10 is solved to account for the heat transfer.
Equations 3.11-3.13 represent the equation of motion, strain-displacement relationship and
constitutive equation (Hooke’s law) [39], where xi , i = 1, 2, 3 are the components of the position
vector x, t is the time, σijs is the Cauchy stress tensor, f i are the body forces, ρs is the mass
density, ui is the displacement vector, eij is the strain, eth is the thermal part of the strain tensor
and Cijkl is the elasticity tensor.

δρ δ δ δ
+ (ρu) + (ρv) + (ρw) = 0 (3.2)
δt δx δy δz

δ δ δ δ δ δ δ
(ρu) + (ρu2 + P) + (ρuv) + (ρuw) = (σxx ) + (τxy ) + (τxz ) (3.3)
δt δx δy δz δx δy δz

δ δ δ δ δ δ δ
(ρv) + (ρuv) + (ρv2 + P) + (ρvw) = (τxy ) + (τyy ) + (τyz ) (3.4)
δt δx δy δz δx δy δz

δ δ δ δ δ δ δ
(ρw) + (ρuw) + (ρvw) + (ρw2 + P) = (τxz ) + (τyz ) + (σzz ) (3.5)
δt δx δy δz δx δy δz

δ δ δ δ δ δ δ
(ρe) + (ρue + ρu) + (ρve + ρv) + (ρwe + ρw) = (θ x ) + (θy ) + (θz )
δt δx δy δz δx δy δz
(3.6)

δu δv δw δu δu δv
σxx = λ( + + ) + 2µ & τxy = µ( + ) (3.7)
δx δy δz δx δy δx

δu δv δw δv δw δu
σyy = λ( + + ) + 2µ & τxz = µ( + ) (3.8)
δx δy δz δy δx δz

δu δv δw δw δw δv
σzz = λ( + + ) + 2µ & τyz = µ( + ) (3.9)
δx δy δz δz δy δz

δT
qi = − k (3.10)
δxi

20
 δσijs D 2 ui
+ ρs f i = ρs (3.11)
δx j Dt2

1 δui δu j
∈ij = ( + ) (3.12)
2 δx j δxi

σijs = Cijkl ∈kl + ∈ijth (3.13)

3.2.2 Challenges in FTSI

Data exchange and data mapping are important during FTSI simulations. Appropriate surface
selection while importing the loads is equally important because it serves as boundary condi-
tions to the thermal and structural simulations. Mesh sizing and quality is crucial while map-
ping CFD calculated loads. Mechanical and thermal loads affect the nozzle during supersonic
operation. There are two ways to account for this – 1) Fluid-Structure Interaction (FSI) where
the temperature and pressure on the nozzle walls are imported directly in the structural solver
and then thermal strains and equivalent stresses are calculated. 2) Fluid-Thermal-Structural In-
teraction (FTSI) where the temperature and pressure on the nozzle walls is imported in thermal
solver to solve Fourier’s equations for steady state heat conduction. Temperatures calculated
by thermal solver are then imported to the structural solver. Accurate prediction of heat transfer
plays a key role as the static temperature gradient (Tin – Tex ) in fluid flow through the existing
nozzle design is approximately 92 K. Heat transfer due to conduction is pronounced and not
as trivial. Therefore, FTSI is the right physics since it accounts for the heat transfer through
conduction. Convective heat transfer is not considered in the thermal analysis as conduction
is the main contributor of thermal loads. Moreover, imported pressure loads represent the net
force due to shear and normal pressure [30] hence using appropriate y+ is essential.

21
3.2.3 FTSI workflow

Figure 3.12: One-way coupled FTSI flowchart.

Figure 3.12 shows one-way FTSI methodology. This is a 1D to 3D approach for rectangular
nozzle design and the philosophy is also applicable to any other product design/development.
Gencurve (refer Chapter 4) is used to add smooth curves to the nozzle walls which are then
lofted/extruded in Solidworks to obtain a 3D CAD model. It is simplified by removing bolt
holes, drafts, fillets and other components to avoid unrealistic values of stresses and strains.
Figure 3.13 illustrates the FTSI methodology with appropriate data exchange for multiphysics
modeling. The fluid domain extends 50 De downstream the nozzle exit. 15 prism layers with
tetrahedral unstructured mesh are used to ensure y+ < 5 on the nozzle walls. Mass flow rate at
inlet and exit plane is monitored to ensure convergence. CFD calculated thermal loads along the
nozzle walls are imported and used as boundary conditions to the steady-state thermal solver
in ANSYS Workbench. Total heat flux and temperature on the nozzle are evaluated. Results
from CFD and thermal simulations are then imported to the static structural module of ANSYS
Workbench. Tetrahedral mesh is used for structural analysis. Along with the imported pressure
and temperature loads, gravity and a fixed support at the nozzle inlet is considered to simulate

22
the experimental set up for cold jet i.e. TR of 1. Structural deformation, equivalent stress, strain
and thermal strain are evaluated at design conditions i.e. NPR 3.67 and off-design conditions
i.e. NPR 4.5, the highest NPR in the experimental facility for this design. Structural steel is
used for all simulations.

23
Figure 3.13: FTSI methodology with the sequence of steps numbered as shown in the image.

24
Chapter 4

Gencurve – Python based smooth curve

generator

4.1 Wall curve

Wall curvature is crucial while modeling internal and external flows as it sets up the pressure
gradient which affects the boundary layer [3]. Previous literature talks about the effect of cur-
vature on turbulent boundary layer [40], wall bounded turbulent flows [41]. Having a smooth
curve is important from aerodynamics perspective. There are many ways to model the wall
curves - parametric B-splines [42], super-ellipse method [16], polynomials [26], as shown pre-
viously. Curves can also be modeled directly in a CAD package using splines but it becomes
iterative and time-consuming process especially when the curves are not predefined and the user
has to model them manually. The aim of this work is to have a predefined curve. Python based
general-purpose curve generator - Gencurve is developed which is a polynomial based smooth
curve generator for nozzles, inlets, diffusers, automobile exhaust ducts, arteries, etc. The goal
is to develop a general purpose design tool which could not only model smooth nozzle wall
curves but also other devices with smooth curves.
The wall curve for the improved design (Figure 4.1 shown in green) is generated using this
tool. The tool reads an input text file containing the inlet, throat and exit (X,Y) coordinates and
the angles. It gives a text file as its output with two curves for converging and diverging sections

25
by matching their slope at the throat and by solving a cubic polynomial equation. These curves
share a common tangent at the throat and have G1 continuity. The curves are then imported to
a CAD package. Note that the throat angle is zero in Figure 4.1 for both curves to match the
slope. Gencurve demonstrates the slope matching technique for generating curves as opposed
to the super-ellipse method used by Dippold III [16]. Figure 4.1 shows wall coordinates of the
existing and improved design. Slope matching reduces throat shocks (refer Chapter 6).

Figure 4.1: Improved nozzle design with contoured walls shown in green.

(a) Existing nozzle geometry. (b) Improved nozzle geometry.

Figure 4.2: Nozzle geometry created using cubic polynomial.

26
 Although a cubic polynomial is used for this curve, lower/higher order polynomials (straight
line, second, third, fourth and fifth order) can also be used. Figure 4.3 shows one such exam-
ple which is created by solving a fifth order polynomial and then lofted in CAD by having
appropriate cross-sections in streamwise direction. The user can choose the order of polyno-
mial depending on the number of known parameters. This makes the tool more flexible. The
objective is to make this tool more general and add more features.

Figure 4.3: Fifth order polynomial wall curve.

The input parameters to Gencurve are listed in Table 4.1. The subscripts in , th , ex represent
inlet, throat and exit respectively. Note that the angle is measured with respect to the horizontal.

X co-ordinate Y co-ordinate angle

xin yin θin
xth yth θth
xex yex θex

Table 4.1: Input parameters to Gencurve – cubic polynomial.

27
4.2 Parametric geometry with domain

Another application of this tool is to circumvent or minimize the interaction with a CAD pack-
age while modeling axisymmetric/2D geometries with fluid domain for CFD simulations. This
is shown in Figure 4.4. The goal here is to import the geometry directly for meshing and
circumvent the CAD modeling. This tool has parametric geometry and domain modeling capa-
bility. It can be used to directly obtain a fluid volume by performing various operations such as
rotate, revolve, translate, loft, etc in CAD. The domain size is a function of equivalent diameter
which makes the tool fully parametric and gives quicker and efficient geometry modeling fea-
ture. It gives multiple output files at the same time and the output file formats are compatible
with Solidworks and ANSYS Workbench.

28
 (a) 5x8 domain.

(b) 10x15 domain.

Figure 4.4: Parametric domain as a function of equivalent diameter.

29
Chapter 5

Isentropic nozzle flow and oblique shock

calculator

A Quasi 1D tool is developed using Python to calculate the isentropic flow properties at nozzle
exit and the oblique shock relations. There is no heat addition or work done by the nozzle on
the fluid. Hence, total enthalpy of the flow remains constant and isentropic expansion of the
flow occurs [43]. Also, the area ratio, pressure ratio and temperature ratio become a function
of Mach number [44].

Figure 5.1: Area ratio vs Mach [1].

This is given by the following equations.

30
 Pe γ − 1 2 γ−−γ1
= (1 + M ) (5.1)
Pt 2

Te γ − 1 2 −1
= (1 + M ) (5.2)
Tt 2

ρe γ − 1 2 γ−−11
= (1 + M ) (5.3)
ρt 2
γ +1
γ −1 2 2( γ −1)
Ae γ + 1 2−(γγ−+11) (1 + 2 M )
= ( ) (5.4)
A∗ 2 M
r
APt γ γ − 1 2 2−(γγ−+11)
ṁ = √ M (1 + M ) (5.5)
Tt R 2

p
Ve = Me γRTe (5.6)

F = ṁVe + ( Pe − P0 ) Ae (5.7)
Oblique shock relations [45] are given by the following equations –

Figure 5.2: Oblique shock wave [45].

( γ + 1) M 2
cot a = tan s[ − 1] (5.8)
2( M2 sin2 s − 1)

31
 T1 [2γM2 sin2 −(γ − 1)][(γ − 1) M2 sin2 s + 2]
= (5.9)
T0 (γ + 1)2 M2 sin2 s

p1 2γM2 sin2 s − (γ − 1)
= (5.10)
p0 ( γ + 1)

pt1 (γ + 1) M2 sin2 s γ−γ 1 ( γ + 1) 1

= [ ] [ ] γ −1 (5.11)
pt0 (γ − 1) M2 sin2 s + 2 2γM2 sin2 s − (γ − 1)

ρ1 (γ + 1) M2 sin2 s
= (5.12)
ρ0 (γ − 1) M2 sin2 s + 2

32
Chapter 6

Results

6.1 3D steady RANS CFD

6.1.1 Performance at various operating conditions – sharp vs smooth noz-

zles

Nozzle aspect ratios 2 and 3 with sharp and smooth throats are compared in this section. All
configurations have same design nozzle pressure ratio for better reference of comparison which
yields same area ratio and Mach number from the isentropic relations. It is important to evalu-
ate the performance map of these nozzles at design and off-design (under and over expanded)
conditions. Figures 6.1–6.3 show the mass averaged quantities at exit planes for the four nozzle
configurations at various operating conditions. Smoothly contouring the nozzle walls reduces
the blockage to the flow because of the streamline curvature [3]. Hence, smooth nozzle per-
forms better than an equivalent sharp nozzle. All nozzle configurations show higher thrust
coefficients at design conditions. Although sharp AR 3 nozzle performs better than sharp AR 2
nozzle in terms of thrust and discharge coefficient, it has lesser total pressure recovery because
there are more number of shock reflections inside the nozzle due to the reduced shock cell size
as shown in Figure 6.8. This is because there is a loss of total pressure associated with each
shock wave.

33
Figure 6.1: NPR vs discharge coefficient – sharp vs smooth.

Figure 6.2: NPR vs pressure recovery – sharp vs smooth.

34
 Figure 6.3: NPR vs thrust coefficient – sharp vs smooth.

6.1.2 Wall curve and shock boundary layer interactions

Figures 6.4a and 6.4b show Mach contours on major and minor axis symmetry planes for sharp
and smooth AR 2 nozzle configurations respectively. A very weak shock at the throat is present
in smooth nozzle because the pressure gradient normal n to the wall depends upon the radius of
curvature R, and the streamwise velocity u shown in streamline curvature equation below [3].

∂p
= ρu2 /R
∂n
This affects the boundary layer profile which can be seen in Figures 6.6a and 6.6b which
represent the velocity vectors along with total pressure contours on minor axis symmetry planes
for sharp and smooth AR 2 nozzles. Boundary layer thickness for sharp AR 2 nozzle is less
near the throat and faces adverse pressure gradient (Figure 6.6a) imposed by an oblique shock
wave and becomes thicker. This creates more blockage to the flow which affects the mass flow
rate and the discharge coefficient of the nozzle as shown in Figure 6.1. Apart from boundary
layer thickening, a shock wave is a major source of drag in nozzles because there is always a

35
total pressure loss associated with shock waves [4]. This is quantified by total pressure recovery
coefficient and is shown in Figure 6.2. Figure 6.5 shows the wall shear stress magnitudes on
sharp vs smooth nozzles. It has a higher magnitude at the throat due to the presence of shock
wave, however, it decreases when the walls are contoured.

(a) Sharp AR 2.

(b) Smooth AR 2.

Figure 6.4: Mach contours on XY and XZ symmetry planes.

36
Figure 6.5: Wall shear stress – sharp vs smooth.

37
 (a) Boundary layer – sharp AR 2.

(b) Boundary layer – smooth AR 2.

Figure 6.6: Velocity vectors with total pressure contours – sharp vs smooth.

6.1.3 Symmetry boundary conditions and asymmetric flow

Typically, symmetric geometries (major and minor axis symmetry) are modeled as half or quar-
ter sections by imposing symmetry boundary conditions. This section shows the difference in
results of modeling quarter domain and full domain of the nozzle geometry. Figure 6.7 shows

38
the stream wise vorticity contours on exit plane for smooth AR 2 nozzle - quarter domain and
full domain respectively. The section highlighted with mesh is modeled as a quarter of the
full flow domain and the remaining three sections are obtained by applying symmetry trans-
forms. The figure does not show the asymmetry in streamwise vorticity which is seen in the
full flow domain. This is because the solution is mirrored about the symmetry planes and does
not account for asymmetry in the flow.

39
 (a) Quarter section of the domain.

(b) Full flow domain.

Figure 6.7: Contours of streamwise vorticity on exit plane.

6.1.4 Aspect ratio effects

This section is divided into three categories - emphasizing specifically on low and high aspect
ratio nozzles and their wake profiles to understand the mixing downstream the nozzle exit.

40
6.1.4.1 Low aspect ratio nozzles

The focus of this subsection is to investigate the low aspect ratio nozzle configurations i.e. AR
1, 2 and 3. Figure 6.8 shows the shadowgraph on XY symmetry plane for the three nozzle
configurations. It can be seen that the shock cell size reduces (shown in red dotted line) with
increase in aspect ratio. This causes more shock reflections inside the nozzle and less total
pressure recovery. When the shock travels from the wall to the centerline, the boundary layer
becomes thicker and when it interacts with the shock wave from the opposite wall and travels
back to the wall, the thickness of boundary layer reduces. This repeats as the shock waves
travel from the nozzle throat to the exit and causes blockage to the flow. Table 6.1 shows the
CFD calculated values of boundary layer thicknesses on exit planes for the three ARs along
minor axis. AR 1 nozzle has the highest value of boundary layer thickness on the exit plane
along the minor axis. The boundary layer thickness reduces as the AR increases. Hence, AR 1
nozzle has less discharge coefficient and thrust recovery coefficient than AR 3 nozzle shown in
Figure 6.11.

Figure 6.8: Shadowgraph – XY symmetry plane.

41
 Figure 6.9: Contours of wall shear stress.

AR Boundary layer thickness (mm)

Minor axis Major axis
1 2.74 0.32
2 0.485 0.647
3 0.475 0.554

Table 6.1: CFD calculated boundary layer thickness on exit plane.

42
 Figure 6.10: AR vs boundary layer profiles on exit plane along minor axis.

Figure 6.11: AR vs performance coefficients.

Figure 6.12 shows the velocity decay from nozzle exit to 20 De downstream along the jet

43
centerline. AR 1 nozzle has the smallest potential core and decays faster. This could be related
to the thicker boundary layer. Fontaine et al. [11] states that a thick boundary layer on the
nozzle exit removes the initial region of shear layer growth which reduces the total acoustic
power and the potential core closes earlier. Hence, the potential core of AR 1 nozzle is shorter.

Figure 6.12: AR vs PC.

Figure 6.13 shows the iso-surfaces of Mach 0.3 from x/De = 3 to x/De = 20 with cross-flow
planes showing the contours of Mach number. It is evident from Figure 6.15 that all three jets
spread more in minor axis and eventually become circular a few diameters downstream.

44
 (a) AR 1.

(b) AR 2.

(c) AR 3.

Figure 6.13: Iso-surfaces of Mach 0.3 shown in green with Mach contours from x/De = 3 to
x/De = 20.

45
 (a) All AR minor axis jet spread.

(b) All AR major axis jet spread.

Figure 6.14: Jet spread.

Figure 6.15 shows the jet spread in minor axis vs major axis for the three aspect ratios.
AR 2 shows axis switching at 7 De . AR 1 and 3 show faster jet spread in minor axis than in
major axis. This is due to the difference in boundary layer thicknesses in minor and major axis
as listed in Table 6.1. All jets have a greater spread in minor axis because the nozzles have
converging-diverging walls in the minor axis plane and parallel walls in the major axis plane.
This affects the jet growth downstream the nozzle exit.

46
 Figure 6.15: Comparison of jet spread – AR 1, 2, 3..

6.1.4.2 High aspect ratio nozzles

This subsection presents a comparison of low (1,2,3) vs high (8,12) aspect ratio nozzles with
sharp throats.
Figures 6.16-6.20 illustrate the throat shock formation on XY and XZ symmetry planes and
the propagated shock train. The shock strength and the shock cell size reduce with increasing
aspect ratio.

47
Figure 6.16: Mach contours on XY and XZ symmetry plane (quarter view) – AR 1.

Figure 6.17: Mach contours on XY and XZ symmetry plane (quarter view) – AR 2.

48
Figure 6.18: Mach contours on XY and XZ symmetry plane (quarter view) – AR 3.

Figure 6.19: Mach contours on XY and XZ symmetry plane (quarter view) – AR 8.

49
 Figure 6.20: Mach contours on XY and XZ symmetry plane (quarter view) – AR 12.

The wall shear stress on the nozzle walls decreases with increasing aspect ratio due the
reduced shock strength as shown in Figure 6.21.

Figure 6.21: Contours of wall shear stress.

The discharge coefficient increases due to the reduced thickness of the boundary layer with
increasing the aspect ratio.

50
 Figure 6.22: AR vs discharge coefficient.

6.1.4.3 Wake profiles

A comparison of the velocity decay and jet mixing with ambient air downstream the nozzle exit
are presented here. Figure 6.23 shows the velocity decay and the potential core length of each
AR. The velocity decays faster in AR 12 nozzle and has a smallest length of potential core.
Since the nozzle height decreases as the aspect ratio increases, the potential core length of AR
8 and AR 12 is smaller than the low AR nozzles.

51
 Figure 6.23: AR vs PC.

Velocity profiles downstream the nozzle exit at various x locations are plotted to understand
the mixing of the nozzle configurations.
x/De = 5

Figure 6.24: Velocity profile at x/De = 5 – minor axis symmetry plane.

52
 Figure 6.25: Velocity profile at x/De = 5 – major axis symmetry plane.

x/De = 10

Figure 6.26: Velocity profile at x/De = 10 – minor axis symmetry plane.

Figure 6.27: Velocity profile at x/De = 10 – major axis symmetry plane.

x/De = 20

Figure 6.28: Velocity profile at x/De = 20 – minor axis symmetry plane.

53
 Figure 6.29: Velocity profile at x/De = 20 – major axis symmetry plane.

6.2 Multiphysics modeling

In addition to CFD, it is equally important to explore the impact of pressure and temperature
loads imposed by the in-nozzle shocks on the structural integrity of the nozzle. Multiphysics
modeling is very crucial while exploring this because the real-world operation of a nozzle
is truly ’Multiphysics’ and becomes necessary to replicate this phenomenon when perform-
ing simulations as shown by the author [46]. Although investigating the noise characteristics
is important, the scope of this work is limited to fluid-thermal-structural interaction. Hence,
aeroacoustics will not be discussed here considering the fact that much work has been done in
that area & the timeline of this research.
A unidirectional steady state 3D Fluid-Thermal-Structural Interaction (FTSI) is chosen to
begin with the preliminary analysis. The influence of FTSI as a diagnosis of existing rectangu-
lar nozzle geometry, and for the preliminary investigation of the improved design is evaluated.
The structural deformation and boundary layer thickness are computed for the existing design
to study the effect of internal shock formation on boundary layer flow. Improved design is
further evaluated at off-design conditions.

6.2.1 Structural deformation and boundary layer thickness

Figure 6.30 shows Mach contours on major and minor axis symmetry planes in the exist-
ing/baseline nozzle. Oblique shock formation at the throat can be seen, which induces pressure
loads on the nozzle due to shock reflection from the wall. The boundary layer becomes thicker
due to the adverse pressure gradient imposed by an oblique shock wave. However, the bound-

54
ary layer does not separate from the nozzle wall, hence the shocks are mild and the effect of
SBLI is not as severe. Therefore, it does not cause large deformations in the nozzle. This
is quantified by comparing the boundary layer thickness δ and maximum deformation ∆. δ
calculated from CFD is 1.23 mm at the nozzle exit and maximum ∆ is 0.005 mm. Hence,

δ  ∆ (6.1)
However, this becomes crucial when stronger shocks are present and shows the importance
of FTSI when structural deformations are in the order of boundary layer thickness, i.e.,

δ ≈ ∆ (6.2)

Figure 6.30: Mach contours on symmetry planes in existing nozzle shown on a quarter section.

6.2.2 Existing (sharp AR 2) and improved (smooth AR 2) design

Very weak shocks are observed in the improved design due to the smooth nozzle wall curve
which makes the CFD-computed loads lesser in this case. In contrast, these loads are greater in
the existing design due to the internal shock formation, as shown in Figure 6.30. Figure 6.31
compares the wall shear stress on the existing and improved design. It is higher on the existing
nozzle due to shock formation. Figure 6.32 shows total pressure contours on the existing and
improved designs. Figure 6.34 shows Von-Mises stress at NPR 4.5 for the improved design.
It is maximum at corners since there are no fillets or drafts. It can be seen from Figure 6.34

55
that the structural integrity of this design remains intact at off-design conditions. Although
maximum stress is induced on sharp corners at the exit, the design does not fail as the stress is
less than the material strength. This makes the design safe at NPRs < 4.5. Total deformation
and thermal strain are illustrated in Figure 6.33. Note that the deformations are exaggerated
by 3x for better visualization. FTSI modeling (Figure 6.35) indicates that the pressure and
temperature loads are compressive in nature. This is because the simulations represent nozzle
operation at atmospheric pressure and temperature. They are greater than the static pressure
and temperature of the fluid at the nozzle exit. Hence, the loads are directed away from the
nozzle walls towards the fluid, which reduces the cross-sectional area at the nozzle exit.

(a) Wall shear stress – sharp AR 2

(b) Wall shear stress – smooth AR 2

Figure 6.31: Contours of wall shear stress – sharp vs smooth AR 2.

56
 (a) Sharp AR 2.

(b) Smooth AR 2.

Figure 6.32: Contours of Pt .

57
Figure 6.33: Total deformation (exaggerated 3x for better visualization) and thermal strain in
improved design.

58
Figure 6.34: Off-design analysis.

59
 Figure 6.35: Imported pressure loads on nozzle walls.

6.2.3 Multiphysics modeling as a prelude to prototype

The existing design is improved using 1D to 3D design analysis, where FTSI modeling gives an
insight into the structural integrity at off-design conditions. Although a supersonic rectangular
nozzle is investigated, this approach can be applied in general to demonstrate the benefits of
FTSI in the initial stages of product development cycle, as shown in Figure 6.36.

60
Figure 6.36: Product development cycle.

61
Chapter 7

Conclusions

The aim of this work is threefold - 1. To develop and implement a general-purpose smooth
curve generator to eliminate the sharp throats in the existing nozzle geometries, 2. To address
the shock boundary layer interactions in a series of low to high aspect ratio nozzles 3. To
evaluate the influence of fluid-thermal-structural interaction on the boundary layer flow.
These are further discussed as below -

7.1 Gencurve - Python based curve generator

Geometry modeling using CAD becomes an iterative process which makes it time-consuming.
A Python based design tool is developed and implemented to model the wall curves which are
imported directly to a CAD package. This makes the process quicker and efficient by minimiz-
ing the user interference with CAD. Slope-matching technique is implemented to remove the
sharp throats which gives a smooth and predefined wall curve with G1 continuity. This tool also
generates 2D and axisymmetric fluid domains that can be imported directly to a CAD package
for further geometry modifications or for meshing. Low to high order (upto 5) polynomials
are solved using this tool. The tool is flexible in a way that a user can choose the order of the
polynomial depending on the number of known parameters.

62
7.2 Aspect ratio and wall curvature effects

Increasing the aspect ratio decreases the shock strength and the shock cell size, hence discharge
coefficient improves due to the less blockage offered by the boundary layer. This increases the
thrust coefficient. Although there are improvements in the thrust and discharge coefficients, the
total pressure recovery coefficient decreases as the aspect ratio increases. This is because there
are more shock reflections inside the nozzle due to the reduced shock cell size.
All jets spread more in the minor axis than in the major axis and eventually become circular
a few diameters downstream the nozzle exit. Sharp AR 1 nozzle has a greater boundary layer
thickness while sharp AR 3 nozzle has a smaller value of boundary layer thickness on exit plane
along minor axis.
The throat shock formation is mitigated by smoothly contouring the nozzle walls using
Gencurve. This gives improved discharge coefficients up to 0.99 due to the reduced blockage
offered by the boundary layer. The thrust and total pressure recovery of the smooth nozzle also
increases.

7.3 Fluid-thermal-structural interaction simulations

The novelty of this work lies in the fluid-thermal-structural interaction simulations, which are,
for the first time, performed on rectangular supersonic nozzles. This work also connects the
structural deformation and boundary layer flow which is a novel approach in supersonic flows.
An existing design is improved using an in-house Python based design tool - Gencurve, which
demonstrates the benefits of the slope-matching technique for generating nozzle wall curves.
Aerodynamic performance improvements in terms of discharge coefficient and total pressure
recovery coefficient are achieved due to the reduced in-nozzle shocks. Although the current
RANS modeling (k-omega SST) increases the amplitude of the first shock cell, the amplitude
is still in the acceptable range, although not tight enough. It gives a better approximation of
the overall velocity decay than other turbulence models. Multi-physics modeling approach is
used to evaluate the feasibility of the existing and improved design. This work shows that

63
FTSI becomes essential during the initial design phase of supersonic rectangular converging-
diverging nozzles. Observed mild throat shocks do not affect the boundary layer flow in the
existing design. Hence, the impact of SBLI on the existing nozzle is not as severe because
the structural deformation is far less than the boundary layer thickness. The improved design
does not fail at highest NPR, making it safer at low NPRs. A general 1D to 3D design analysis
approach is demonstrated through the rectangular nozzle design case, which highlights the
significance of FTSI and shows that it can also be used in any other product development cycle.

7.4 Symmetry boundary conditions

Streamwise vorticity is induced due to the corners of the rectangular cross-section. High and
low regions of streamwise vorticity can be captured accurately by modeling the full domain
which has a greater impact on predicting the asymmetry in the flow accurately and it is the
right physics.

64
Chapter 8

Discussion & future work

Considering the time, cost and energy associated with experimental testing, it is crucial to
manufacture a design which can withstand pressure and temperature loads at various operating
conditions in rectangular supersonic nozzles. This work highlights that the accuracy of every
step matters since the structural deformation is affected by the shock loads which are affected
by the boundary layer which makes the choice of y+ essential. The choice of turbulence model
plays a key role in capturing the shock cell amplitude, as can be seen from the current validation.
A detailed scrutiny with compressibility correction will reduce the error margin, however, for
the objective of showing FTSI, the current validation is sufficient to move ahead. An effort
is made to connect the flow physics with the structural integrity, which serves as a prelude
to building prototypes. Exploiting the advancements in simulation technology is essential to
showing the proposed design is feasible for the real-world operation. The results at nozzle
pressure ratios 3.67 and temperature ratio 1 are presented. This means that the pressure loads
have a greater impact on the nozzle geometry than the temperature loads. Performing the
simulations at higher temperature ratios (> 1) would truly address the effects of temperature
loads. Current diagnosis demonstrates the analysis for the improved design as a precursor
to prototyping. Hence, one-way steady-state FTSI is performed, which serves as a preliminary
investigation. However, evaluating the impact of structural deformation on boundary layer flow
during unsteady operation is crucial for SBLI with tight coupling, and will be addressed in the
next phase of this research. In addition to this, fatigue induced due to the number of operating

65
cycles will also be part of future work.
Improvements in the nozzle performance using Gencurve are presented in this work. How-
ever, the wall curve needs to be optimized further to have an optimum contour by implementing
a higher order curvature continuity. Hence, adding more capabilities to the existing tool and
making it more general will be part of the future work. Furthermore, side plate thickness and
nozzle wall thickness could also be optimized to reduce the overall weight and material costs.
Along with rectangular cross-section, other cross-sections need to be explored. Effect of asym-
metric flow and secondary flows from corners will be explored in future work. Along with
aerodynamics, acoustics and modal analysis will be part of future work.

66
Chapter 9

Contributions

Journal article

1. Kalyani Bhide, Kiran Siddappaji, and Shaaban Abdallah. Influence of fluid–thermal–

structural interaction on boundary layer flow in rectangular supersonic nozzles. Aerospace,
5(2):33, 2018.

Conference paper

1. Kalyani Bhide, Kiran Siddappaji, and Shaaban Abdallah. A combined effect of wall curva-
ture and aspect ratio on overall performance of rectangular supersonic nozzles. In proceedings
of the 6th International Conference on Jets, Wakes and Separated Flows, October 2017.

Presentations

1. Kalyani Bhide, Kiran Siddappaji, and Shaaban Abdallah. “On shock boundary layer in-
teractions and multiphysics modeling of rectangular supersonic nozzles”, AIAA 43rd Dayton-
Cincinnati Aerospace Sciences Symposium (DCASS), February 2018.
2. Kalyani Bhide, Kiran Siddappaji, and Shaaban Abdallah. “A combined effect of wall cur-
vature and aspect ratio on overall performance of rectangular supersonic nozzles”, In 6th
International Conference on Jets, Wakes and Separated Flows, October 2017.

67
Software disclosure

1. Kalyani Bhide, Kiran Siddappaji, and Shaaban Abdallah. “Gencurve - Python based smooth
curve generator”, University of Cincinnati, September 2017, UC 118-021.

68
 Bibliography

[1] John David Anderson. Modern compressible flow: with historical perspective, volume 12.
McGraw-Hill New York, 1990.

[2] Saeed Farokhi. Aircraft propulsion. John Wiley & Sons, 2014.

[3] Edward M Greitzer, Choon Sooi Tan, and Martin B Graf. Internal flow: concepts and
applications, volume 3. Cambridge University Press, 2007.

[4] Holger Babinsky and John K Harvey. Shock wave-boundary-layer interactions, vol-
ume 32. Cambridge University Press, 2011.

[5] Hermann Schlichting, Klaus Gersten, Egon Krause, and Herbert Oertel. Boundary-layer
theory, volume 7. Springer, 1955.

[6] http://shodhganga.inflibnet.ac.in/bitstream/10603/6989/8/08c hapter

[7] Pablo A Mora Sánchez. Investigation of the Noise Radiation from Heated Supersonic Jets. PhD
thesis, University of Cincinnati, 2016.

[8] KBMQ Zaman. Axis switching and spreading of an asymmetric jet: the role of coherent
structure dynamics. Journal of Fluid Mechanics, 316:1–27, 1996.

[9] KBMQ Zaman. Spreading characteristics and thrust of jets from asymmetric nozzles. In 34th
Aerospace Sciences Meeting and Exhibit, page 200, 1995.

[10] E Gutmark, KC Schadow, and CJ Bicker. Near acoustic field and shock structure of rectangular
supersonic jets. AIAA journal, 28(7):1163–1170, 1990.

69
[11] Ryan A Fontaine, Gregory S Elliott, Joanna M Austin, and Jonathan B Freund. Very near-
nozzle shear-layer turbulence and jet noise. Journal of Fluid Mechanics, 770:27–51, 2015.

[12] https://aviation.stackexchange.com/questions/21081/advantages-of-square-over-circular-
engine-nozzle.

[13] Kamal Viswanath, Ryan F Johnson, Andrew T Corrigan, Kailas Kailasanath, Pablo A Mora,
Florian Baier, and Ephraim J Gutmark. Noise characteristics of a rectangular vs circular nozzle
for ideally expanded jet flow. In 54th AIAA Aerospace Sciences Meeting, page 1638, 2016.

[14] Florian Baier, Pablo Mora, Ephraim J Gutmark, and Kailas Kailasanath. Flow measurements
from a supersonic rectangular nozzle exhausting over a flat surface. In 55th AIAA Aerospace
Sciences Meeting, page 0932, 2017.

[15] A Krothapalli, D Baganoff, and K Karamcheti. On the mixing of a rectangular jet. Journal of
Fluid Mechanics, 107:201–220, 1981.

[16] Vance F Dippold III. Design and analyses of high aspect ratio nozzles for distributed propulsion
acoustic measurements. In 34th AIAA Applied Aerodynamics Conference, page 3876, 2016.

[17] Tanmay J Tipnis. Effects of upstream nozzle geometry on rectangular free jets. 2010.

[18] Junhui Liu, Kailas Kailasanath, Ravi Ramamurti, David Munday, Ephraim Gutmark, and
Rainald Lohner. Large-eddy simulations of a supersonic jet and its near-field acoustic proper-
ties. AIAA journal, 47(8):1849–1865, 2009.

[19] Nicholas S Heeb, Pablo Mora Sanchez, Ephraim J Gutmark, and Kailas Kailasanath. Inves-
tigation of the noise from a rectangular supersonic jet. In 19th AIAA/CEAS Aeroacoustics
Conference, page 2239, 2013.

[20] David Munday, Ephraim Gutmark, Junhui Liu, and K Kailasanath. Flow structure of supersonic
jets from conical cd nozzles. In 39th AIAA Fluid Dynamics Conference, page 4005, 2009.

[21] David Munday. Flow and Acoustics of Jets from Practical Nozzles for High-Performance Mil-
itary Aircraft. PhD thesis, University of Cincinnati, 2010.

70
[22] Franco C Frate and James E Bridges. Extensible rectangular nozzle model system. AIAA paper,
975:2011, 2011.

[23] Zachary Riley, Jack McNamara, and Heath Johnson. Hypersonic boundary layer stability in
the presence of thermo-mechanical surface compliance. In 53rd AIAA/ASME/ASCE/AHS/ASC
Structures, Structural Dynamics and Materials Conference 20th AIAA/ASME/AHS Adaptive
Structures Conference 14th AIAA, page 1549, 2012.

[24] Kalyani Bhide, Kiran Siddappaji, and Shaaban Abdallah. A combined effect of wall curvature
and aspect ratio on overall performance of rectangular supersonic nozzles. In 6th International
Conference on Jets, Wakes and Separated Flows, October 2017.

[25] John W Slater. Supin: A computational tool for supersonic inlet design. In 54th AIAA
Aerospace Sciences Meeting, page 0530, 2016.

[26] Karla K Quintao. Design optimization of nozzle shapes for maximum uniformity of exit flow.
2012.

[27] Solidworks.

[28] Python 3.5.

[29] STAR-CCM+: User guide.

[30] ANSYS Workbench User’s guide.

[31] Tecplot, Inc., Bellevue, WA. Tecplot 360 User’s manual, 2013.

[32] Milovan Peric and Stephen Ferguson. The advantage of polyhedral meshes. Dynamics, 24:45,
2005.

[33] LH Back, HL Gier, and PF Massier. Comparison of measured and predicted flows through
conical supersonic nozzles, with emphasis on the transonic region. AIAA Journal, 3(9):1606–
1614, 1965.

[34] David C Wilcox et al. Turbulence modeling for CFD, volume 2. DCW industries La Canada,
CA, 1998.

71
[35] Nicholas J Georgiadis and James R DeBonis. Navier–stokes analysis methods for turbulent
jet flows with application to aircraft exhaust nozzles. Progress in Aerospace Sciences, 42(5-
6):377–418, 2006.

[36] Jens Truemner and Christian Mundt. Total temperature based correction of the turbulence pro-
duction in hot jets. In ASME Turbo Expo 2017: Turbomachinery Technical Conference and Ex-
position, pages V02BT41A003–V02BT41A003. American Society of Mechanical Engineers,
2017.

[37] Nico Gross, Gregory Blaisdell, and Anastasios Lyrintzis. Evaluation of turbulence model cor-
rections for supersonic jets using the overflow code. In 40th Fluid Dynamics Conference and
Exhibit, page 4604, 2010.

[38] Jonathan Simon Geerts. Shock Train/Boundary-Layer Interaction in Rectangular Scramjet

Isolators. PhD thesis, University of Maryland, College Park, 2016.

[39] Plamen Pironkov. Numerical simulation of thermal fluid-structure interaction. PhD thesis,
Technische Universität, 2010.

[40] Virendrakumar Chaturbhai Patel. The effects of curvature on the turbulent boundary layer.
TIL, 1968.

[41] Robert D Moser and Parviz Moin. The effects of curvature in wall-bounded turbulent flows.
Journal of Fluid Mechanics, 175:479–510, 1987.

[42] Kiran Siddappaji, Mark G Turner, and Ali Merchant. General capability of parametric 3d blade
design tool for turbomachinery. In ASME Turbo Expo 2012: Turbine Technical Conference and
Exposition, pages 2331–2344. American Society of Mechanical Engineers, 2012.

[43] Ascher H Shapiro. The Dynamics and Thermodynamics of Compressible Fluid Flow: In Two
Volumes. Wiley, 1953.

[44] https://www.grc.nasa.gov/www/k-12/airplane/isentrop.html.

[45] https://www.grc.nasa.gov/www/k-12/airplane/oblique.html.

72
[46] Kalyani Bhide, Kiran Siddappaji, and Shaaban Abdallah. Influence of fluid–thermal–structural
interaction on boundary layer flow in rectangular supersonic nozzles. Aerospace, 5(2):33, 2018.

73
Appendix A

Jet spread

Figures A.1–A.5 show the jet spread. The spread in minor axis is more than the spread in
the major axis. This is because the nozzle walls are converging-diverging in the minor axis
and parallel in major axis. The sharp AR 2 nozzle shows axis switching at approximately 7
De downstream the nozzle exit.

Figure A.1: Jet spread – AR 1.

74
Figure A.2: Jet spread – AR 2.

Figure A.3: Jet spread – AR 3.

Figure A.4: Jet spread – AR 8.

75
Figure A.5: Jet spread – AR 12.

76