these simulation numerique et modelisation de l'ecoulement autour ...
these simulation numerique et modelisation de l'ecoulement autour ... these simulation numerique et modelisation de l'ecoulement autour ...
in the x i direction. Furthermore, both the solid part (S s ) and the aperture (S h ) of the multiperforated plate contribute to the global flux over the x−z planes represented in Fig. 4. The expressions of the different contributions are summarized in Table 1. In the configuration considered, the normal to the homogeneous boundary is along the y-direction: ⃗n = −⃗e y or n 2 = −1 for the injection wall and ⃗n = ⃗e y or n 2 = 1 for the suction wall. S s and S h stand for the solid wall and aperture surfaces respectively (see Fig. 4). solid wall hole viscous non-viscous viscous non-viscous ∫ ∫ ∫ ∫ ρV S s τ 22 n 2 dxdz S s (−P) n 2 dxdz S h τ 22 n 2 dxdz S h (−P − ρV 2 ) n 2 dxdz ∫ ∫ ρV ti ∫S s τ i2 n 2 dxdz 0 S h τ i2 n 2 dxdz S h (−ρV V ti ) n 2 dxdz Table 1. Contributions to the momentum fluxes over a x − z plane just above (injection side n 2 = −1) or just below (suction side n 2 = +1) the perforated plate. V ti is U or W (i = 1 or 3). Small-scale computations allow to assess the different terms of the momentum fluxes on the suction and injection wall planes. Integrations over the solid wall (S s ) and the hole surface (S h ) are performed and the results are reported in Tables 2 and 3 for the streamwise and vertical momentum respectively. Viscous fluxes have not been reported in Table 3 (vertical momentum), as they are negligible compared to inviscid contributions. Region total plate hole solid wall Expression ∫ ∫ ∫ ∫ S W (−ρUV + τ 12 )n 2 dxdz S h −ρUV n 2 dxdz S h τ 12 n 2 dxdz S s τ 12 n 2 dxdz Injection 8.0 × 10 −1 111.5 −0.1 −11.4 Suction −3.11 × 10 −1 90.4 −0.1 9.7 Table 2. Wall fluxes for the streamwise momentum: First column: expression and values of the total flux (in ρ j V j 2 d 2 ) on both sides of the plate (total surface S W ). Columns 2–4: relative contributions (in %) of the terms involved in the wall fluxes. Streamwise momentum ρU: the non-viscous streamwise momentum flux is the main term for both the suction and the injection sides of the perforated plate. The viscous term over the hole surface is very small. The wall friction over the solid wall is approximately 10 times smaller than the non-viscous aperture term for the operating point considered. This means that one can only focus on the inviscid part of the flux when developing a (rough) model for effusion. In other words, assuming that the turbulent transfers scale as the wall friction, turbulence is not a first-order issue when dealing with discrete effusion, which is of course significantly different from the classical case of an attached boundary layer over a solid plate. Vertical momentum ρV : the flux of normal momentum involves a pressure term that is clearly dominant. The velocity term in the hole is small compared to the pressure term. The 11 of 26
Region total plate hole solid wall Expression ∫ ∫ ∫ S W (−P − ρV 2 + τ 22 )n 2 dxdz S h −(P + ρV 2 )n 2 dxdz S s −P n 2 dxdz Injection 3.64 × 10 3 4 96 Suction −3.69 × 10 3 4 96 Table 3. Wall fluxes for the vertical momentum: First column: expression and values of the total flux (in ρ j V j 2 d 2 ) on both sides of the plate (total surface S W ). Columns 2–5: relative contributions (in %) of the terms involved in the wall fluxes. repartition between hole and solid surface fluxes is completely related to the porosity of the plate σ = 0.04: pressure is almost constant over the whole wall. Note that owing to the symmetry of the problem, the spanwise momentum flux should be zero. In the computation, it is not perfectly so but for injection the spanwise momentum flux is approximately 50 times smaller than the streamwise momentum flux. This ratio is even smaller for the suction side. An appropriate model has thus to reproduce the two main effects of the flow around a perforated plate: the non-viscous streamwise momentum flux due to injection and the nonviscous vertical momentum flux that can be reduced to a pressure effect. All the other terms are negligible, at least in a first modeling effort. As it is usual in wall bounded flows, we will consider that the outer pressure is a good measure of the pressure in the vicinity of the wall. This is verified in the small-scale LES results. Thus, the pressure term can be easily related to the pressure values obtained in the calculation. Hence, the main modeling effort consists in obtaining a good estimation of the non-viscous streamwise momentum flux in the hole. We are going to present the models in terms of equivalent boundary condition, answering the following question: what is the equivalent injection/suction over the whole plate surface that better represents the real injection/suction of fluid through cooling holes. Before constructing homogeneous models over the whole surface, an intermediate model that takes two different constant velocity values over the hole zone and the solid wall zone is built. The wall-normal vertical velocity to impose is directly related to the mass flow rate. The ratio between the vertical velocity and the streamwise velocity over the hole surface depends on the side of the plate: indeed, the velocity at the hole outlet is strongly orientated by the hole angle; on the contrary, at the hole entrance, the streamwise velocity highly depends on the crossflow velocity. Let U c2 be the average streamwise velocity on the suction channel. We assume the streamwise The intermediate model we propose reads: Injection side: V = V inj jet = q S h ρ inj over S h and V = 0 over S s , (1) 12 of 26
- Page 97 and 98: Chapitre 5 Simulations numériques
- Page 99 and 100: 2 S. Mendez and F. Nicoud COMBUSTIO
- Page 101 and 102: 4 S. Mendez and F. Nicoud (c) The i
- Page 103 and 104: 6 S. Mendez and F. Nicoud CALCULATI
- Page 105 and 106: 8 S. Mendez and F. Nicoud Name Numb
- Page 107 and 108: 10 S. Mendez and F. Nicoud x/d z/d
- Page 109 and 110: 12 S. Mendez and F. Nicoud main, it
- Page 111 and 112: 14 S. Mendez and F. Nicoud 1.0 a 1.
- Page 113 and 114: 16 S. Mendez and F. Nicoud ROWS 1 3
- Page 115 and 116: 18 S. Mendez and F. Nicoud Figure 1
- Page 117 and 118: 20 S. Mendez and F. Nicoud 30 ◦ 3
- Page 119 and 120: 22 S. Mendez and F. Nicoud & Mahesh
- Page 121 and 122: 24 S. Mendez and F. Nicoud Figure 1
- Page 123 and 124: 26 S. Mendez and F. Nicoud Figure 1
- Page 125 and 126: 28 S. Mendez and F. Nicoud film, th
- Page 127 and 128: 30 S. Mendez and F. Nicoud Region t
- Page 129 and 130: 32 S. Mendez and F. Nicoud Expressi
- Page 131 and 132: 34 S. Mendez and F. Nicoud geometri
- Page 133 and 134: 36 S. Mendez and F. Nicoud (f) Two
- Page 135 and 136: 38 S. Mendez and F. Nicoud Mendez,
- Page 137 and 138: Chapitre 6 Un modèle adiabatique p
- Page 139 and 140: ρ Mass density, kg/m 3 P T V i τ
- Page 141 and 142: cence of the jets. This is particul
- Page 143 and 144: Simulations. The small-scale LES re
- Page 145 and 146: chosen to compare with numerical re
- Page 147: pressure drop of 41 Pa is effective
- Page 151 and 152: This equality is almost verified at
- Page 153 and 154: V U σ V inj jet cotan(α′ ) V U
- Page 155 and 156: PERIODIC Z INFLOW 1 WALL 24 OUTFLOW
- Page 157 and 158: part of the plate. Downstream of th
- Page 159 and 160: streamwise momentum flux. As a cons
- Page 161 and 162: 8 Fric, T. and Roshko, A., “Vorti
- Page 163 and 164: Discussion sur la modélisation de
- Page 165 and 166: 20 15 10 y/d 5 0 -5 -10 0.0 0.2 0.4
- Page 167 and 168: Chapitre 7 Simulations numériques
- Page 169 and 170: Primary flow (burnt gases) Secondar
- Page 171 and 172: Figure 7: Nusselt number over the s
- Page 173 and 174: the perforated plate. A Large-Eddy
- Page 175 and 176: Les tableaux de flux permettent de
- Page 177 and 178: Conclusion générale Dans cette é
- Page 179 and 180: Conclusion générale En ce qui con
- Page 181 and 182: Bibliographie AKSELVOLL, K. & MOIN,
- Page 183 and 184: BIBLIOGRAPHIE CORON, X. 2001 Étude
- Page 185 and 186: BIBLIOGRAPHIE HALE, C. A., PLESNIAK
- Page 187 and 188: BIBLIOGRAPHIE LIEUWEN, T. & YANG, V
- Page 189 and 190: BIBLIOGRAPHIE MOST, A. & BRUEL, P.
- Page 191 and 192: BIBLIOGRAPHIE SCHILDKNECHT, M., MIL
- Page 193: Annexes
- Page 196 and 197: ARTIFICIAL VISCOSITY MODELS A.2.1 T
Region total plate hole solid wall<br />
Expression<br />
∫ ∫<br />
∫<br />
S W<br />
(−P − ρV 2 + τ 22 )n 2 dxdz<br />
S h<br />
−(P + ρV 2 )n 2 dxdz<br />
S s<br />
−P n 2 dxdz<br />
Injection 3.64 × 10 3 4 96<br />
Suction −3.69 × 10 3 4 96<br />
Table 3. Wall fluxes for the vertical momentum: First column: expression and values of the<br />
total flux (in ρ j V j 2 d 2 ) on both si<strong>de</strong>s of the plate (total surface S W ). Columns 2–5: relative<br />
contributions (in %) of the terms involved in the wall fluxes.<br />
repartition b<strong>et</strong>ween hole and solid surface fluxes is compl<strong>et</strong>ely related to the porosity of the<br />
plate σ = 0.04: pressure is almost constant over the whole wall.<br />
Note that owing to the symm<strong>et</strong>ry of the problem, the spanwise momentum flux should<br />
be zero. In the computation, it is not perfectly so but for injection the spanwise momentum<br />
flux is approximately 50 times smaller than the streamwise momentum flux. This ratio is<br />
even smaller for the suction si<strong>de</strong>.<br />
An appropriate mo<strong>de</strong>l has thus to reproduce the two main effects of the flow around a<br />
perforated plate: the non-viscous streamwise momentum flux due to injection and the nonviscous<br />
vertical momentum flux that can be reduced to a pressure effect. All the other terms<br />
are negligible, at least in a first mo<strong>de</strong>ling effort. As it is usual in wall boun<strong>de</strong>d flows, we will<br />
consi<strong>de</strong>r that the outer pressure is a good measure of the pressure in the vicinity of the wall.<br />
This is verified in the small-scale LES results. Thus, the pressure term can be easily related<br />
to the pressure values obtained in the calculation. Hence, the main mo<strong>de</strong>ling effort consists<br />
in obtaining a good estimation of the non-viscous streamwise momentum flux in the hole.<br />
We are going to present the mo<strong>de</strong>ls in terms of equivalent boundary condition, answering<br />
the following question: what is the equivalent injection/suction over the whole plate surface<br />
that b<strong>et</strong>ter represents the real injection/suction of fluid through cooling holes.<br />
Before constructing homogeneous mo<strong>de</strong>ls over the whole surface, an intermediate mo<strong>de</strong>l<br />
that takes two different constant velocity values over the hole zone and the solid wall zone is<br />
built. The wall-normal vertical velocity to impose is directly related to the mass flow rate.<br />
The ratio b<strong>et</strong>ween the vertical velocity and the streamwise velocity over the hole surface<br />
<strong>de</strong>pends on the si<strong>de</strong> of the plate: in<strong>de</strong>ed, the velocity at the hole outl<strong>et</strong> is strongly orientated<br />
by the hole angle; on the contrary, at the hole entrance, the streamwise velocity highly<br />
<strong>de</strong>pends on the crossflow velocity. L<strong>et</strong> U c2 be the average streamwise velocity on the suction<br />
channel. We assume the streamwise The intermediate mo<strong>de</strong>l we propose reads:<br />
Injection si<strong>de</strong>:<br />
V = V inj<br />
j<strong>et</strong> =<br />
q<br />
S h ρ inj<br />
over S h and V = 0 over S s , (1)<br />
12 of 26