Experimental and Numerical Study of Swirling ... - Solid Mechanics
Experimental and Numerical Study of Swirling ... - Solid Mechanics Experimental and Numerical Study of Swirling ... - Solid Mechanics
Experi imental and Numerical N Stud dy of Swirling g Flow in Scaveenging Processs for 2-Stroke Marin ne Diesel Engin nes Figure 2.14: Transit tion of helical vortices s (a) L-Transition n (b) R-T Transition (Okulo ov et al., 2002). 2 In real confined swir rling flows, thhere can existt many exampples where in a single vortex v chambe er a helical swwirling flow exists where tthe vortex corre filamen nt not only precess aroundd the vortex cchamber axis but also has a periodic c behavior an nd undergoes vortex breakddown at somee cross-sectionnal position n. The presen nce of wall will make thhe overall beehavior of thhis confine ed swirling flo ow even more complex as tthe ‘mirror voortex’ effect wiill make th he flow behave as if two pparallel helicaal vortex systeems are existeed inside the t vortex cha amber. In casse of highly tturbulent swirrling flows, thhe empiric cal models dev veloped (see AAlekseenko et al., 2007) are compared witth the ave eraged values s obtained fr from the expperimental reesults. For thhe comput tational mode els, the swirlinng flows are also one of tthe challenginng areas. 2.4 (a) CFD Mo odeling CChallenge es There is a large amount of scientific liiterature avaailable on thhe Compu utational Fluid d Dynamics (CCFD) modelinng of swirling flow. Howeveer, the pur rpose of this se ection is to discuss some off the aspects of swirling flowws that ma ake it very di ifficult or in oother words, require special treatment to model them t using standard turbullence models i.e. Two-equation turbulence models and Reynolds s Stress Modells. Swirl flows has some challenging ffeatures that aare absent in siimple flows e. .g. streamline curvature which is viewwed as an extraa strain rate (reelative to simpple shear) if the velocities s would be exppressed in Carrtesian rather tthan cylindriccal coordin nates (Moene e A.F., 2003) ), strong departure fromm local energgy equilibr rium and effec cts of turbulennce anisotropyy (Jakirlic´ et aal., 2002). The sta andard two eq quation modeels such as k , based oon Boussinesqq’s eddy vi iscosity hypoth hesis have faiiled to reprodduce importannt swirling floow features s like velocity component deecay, jet spreaading or diffussion rate, degree of entrainment, characteristics off recirculationn zones and Reynolds streess (b) Chapter 2 28 Swirling Flows
Experimental and Numerical Study of Swirling Flow in Scavenging Process for 2-Stroke Marine Diesel Engines Chapter 2 levels etc. (Sloan et al., 1986). The adoption of a relation that local stresses being directly proportional to local strain rates rules out any special sensitivity to the streamline curvature (Craft et al., 2008). The assumption of turbulent/ eddy viscosity as isotropic cannot be valid in flows influenced by body forces acting in a preferred direction, such as buoyancy, rotation and streamline curvature (Sloan et al., 1986). Algifri et al. (1988) proposed, using experimental data on confined swirling flow, that the eddy viscosity is function of Reynolds and swirl numbers. Further, the k model is a simplistic function of a single time scale (proportional to a turbulence energy turnover time k ) which implies that the spectral energy transfer rates for production of turbulent kinetic energy by mean shear, the intermediate transferal regime and the viscous dissipation regime are equal without any characteristic lag (Sloan et al., 1986). The second order closure schemes having transport equations for individual components of Reynolds Stress tensor account for the curvature of streamlines however the effect is at least an order of magnitude large than expected (Moene A.F., 2003) i.e. a 1% curvature strain produces for a boundary layer a 10% or greater effect on the turbulent stresses (Bradshaw, 1973). The source terms in dissipation equation, to a significant degree control the magnitude of turbulent kinetic energy, mean velocity decay and the spreading rates (Sloan et al., 1986). However, the transport equations for k and have no specific terms (source) to account for sensitiveness to rotation and swirl (Jakirlic´ et al., 2002). Since both of these transport equations are common in two-equation and second order closure schemes, the performance of second order closure schemes are also affected to some degree for modeling the swirling flows. Numerical issues also have significant impact on predicting the behavior of swirling flows (Craft et al., 2008). Different modeling approaches so far have been adopted to improve the performance of the aforementioned turbulence models and some of them are discussed here in brief. In order to include a rapid response to streamline curvature, the coefficient C in the turbulent viscosity equation (2.12) has been altered to a functional form (see Leshcziner et al., 1981). Sloan et al., (1986) reports that more complex versions of C have been developed that include non-equilibrium turbulence and wall damping. 2 k t C. . (2.12) Where t is turbulent viscosity, C is a constant and has empirically determined value of 0.09 and is the fluid density. Yakhot et al., (1992) developed the RNG k model which, unlike the standard k model, is derived using a statistical technique called 29 Swirling Flows
- Page 1: Experimental and Numerical Study of
- Page 5 and 6: Experimental and Numerical Study of
- Page 7 and 8: Experimental and Numerical Study of
- Page 9 and 10: Experimental and Numerical Study of
- Page 11 and 12: Experimental and Numerical Study of
- Page 13 and 14: Experimental and Numerical Study of
- Page 15 and 16: Experimental and Numerical Study of
- Page 17 and 18: Experimental and Numerical Study of
- Page 19 and 20: Experimental and Numerical Study of
- Page 21 and 22: Experimental and Numerical Study of
- Page 23 and 24: Experi imental and Numerical N Stud
- Page 25 and 26: Experi imental and Numerical N Stud
- Page 27 and 28: Experimental and Numerical Study of
- Page 29 and 30: Experimental and Numerical Study of
- Page 31 and 32: Experimental and Numerical Study of
- Page 33 and 34: Experimental and Numerical Study of
- Page 35 and 36: Experi imental and Numerical N Stud
- Page 37 and 38: Experi imental and Numerical N Stud
- Page 39 and 40: Experi imental and Numerical N Stud
- Page 41 and 42: Experi imental and Numerical N Stud
- Page 43 and 44: Experimental and Numerical Study of
- Page 45 and 46: Experi imental and Numerical N Stud
- Page 47: Experi imental and Numerical N Stud
- Page 51 and 52: Experi imental and Numerical N Stud
- Page 53 and 54: Experi imental and Numerical N Stud
- Page 55 and 56: Experimental and Numerical Study of
- Page 57 and 58: Experimental and Numerical Study of
- Page 59 and 60: Experi imental and Numerical N Stud
- Page 61 and 62: Experi imental and Numerical N Stud
- Page 63 and 64: Experi imental and Numerical N Stud
- Page 65 and 66: Experi imental and Numerical N Stud
- Page 67 and 68: Experi imental and Numerical N Stud
- Page 69 and 70: Experimental and Numerical Study of
- Page 71 and 72: Experimental and Numerical Study of
- Page 73 and 74: Experi imental and Numerical N Stud
- Page 75 and 76: Experi imental and Numerical N Stud
- Page 77 and 78: Experi imental and Numerical N Stud
- Page 79 and 80: Experi imental and Numerical N Stud
- Page 81 and 82: Experi imental and Numerical N Stud
- Page 83 and 84: Experimental and Numerical Study of
- Page 85 and 86: Experimental and Numerical Study of
- Page 87 and 88: Experi imental and Numerical N Stud
- Page 89 and 90: Experi imental and Numerical N Stud
- Page 91 and 92: Experi imental and Numerical N Stud
- Page 93 and 94: Experi imental and Numerical N Stud
- Page 95 and 96: Experi imental and Numerical N Stud
- Page 97 and 98: Experi imental and Numerical N Stud
Experi imental <strong>and</strong> <strong>Numerical</strong> N Stud dy <strong>of</strong> <strong>Swirling</strong> g Flow in Scaveenging<br />
Processs<br />
for 2-Stroke<br />
Marin ne Diesel Engin nes<br />
Figure<br />
2.14:<br />
Transit tion <strong>of</strong> helical<br />
vortices s (a) L-Transition n<br />
(b) R-T Transition (Okulo ov<br />
et al., 2002). 2<br />
In real confined swir rling flows, thhere<br />
can existt<br />
many exampples<br />
where in a<br />
single vortex v chambe er a helical swwirling<br />
flow exists where tthe<br />
vortex corre<br />
filamen nt not only precess<br />
aroundd<br />
the vortex cchamber<br />
axis but also has a<br />
periodic c behavior an nd undergoes vortex breakddown<br />
at somee<br />
cross-sectionnal<br />
position n. The presen nce <strong>of</strong> wall will make thhe<br />
overall beehavior<br />
<strong>of</strong> thhis<br />
confine ed swirling flo ow even more complex as tthe<br />
‘mirror voortex’<br />
effect wiill<br />
make th he flow behave<br />
as if two pparallel<br />
helicaal<br />
vortex systeems<br />
are existeed<br />
inside the t vortex cha amber. In casse<br />
<strong>of</strong> highly tturbulent<br />
swirrling<br />
flows, thhe<br />
empiric cal models dev veloped (see AAlekseenko<br />
et al., 2007) are compared witth<br />
the ave eraged values s obtained fr from the expperimental<br />
reesults.<br />
For thhe<br />
comput tational mode els, the swirlinng<br />
flows are also one <strong>of</strong> tthe<br />
challenginng<br />
areas.<br />
2.4<br />
(a)<br />
CFD Mo odeling CChallenge<br />
es<br />
There is a large amount <strong>of</strong> scientific liiterature<br />
avaailable<br />
on thhe<br />
Compu utational Fluid d Dynamics (CCFD)<br />
modelinng<br />
<strong>of</strong> swirling flow. Howeveer,<br />
the pur rpose <strong>of</strong> this se ection is to discuss<br />
some <strong>of</strong>f<br />
the aspects <strong>of</strong><br />
swirling flowws<br />
that ma ake it very di ifficult or in oother<br />
words, require special<br />
treatment to<br />
model them t using st<strong>and</strong>ard<br />
turbullence<br />
models i.e. Two-equation<br />
turbulence<br />
models <strong>and</strong> Reynolds s Stress Modells.<br />
Swirl flows<br />
has some challenging ffeatures<br />
that aare<br />
absent in siimple<br />
flows e. .g.<br />
streamline<br />
curvature which is viewwed<br />
as an extraa<br />
strain rate (reelative<br />
to simpple<br />
shear) if<br />
the velocities s would be exppressed<br />
in Carrtesian<br />
rather tthan<br />
cylindriccal<br />
coordin nates (Moene e A.F., 2003) ), strong departure<br />
fromm<br />
local energgy<br />
equilibr rium <strong>and</strong> effec cts <strong>of</strong> turbulennce<br />
anisotropyy<br />
(Jakirlic´ et aal.,<br />
2002).<br />
The sta <strong>and</strong>ard two eq quation modeels<br />
such as k , based oon<br />
Boussinesqq’s<br />
eddy vi iscosity hypoth hesis have faiiled<br />
to reprodduce<br />
importannt<br />
swirling floow<br />
features s like velocity component deecay,<br />
jet spreaading<br />
or diffussion<br />
rate, degree<br />
<strong>of</strong> entrainment,<br />
characteristics<br />
<strong>of</strong>f<br />
recirculationn<br />
zones <strong>and</strong> Reynolds streess<br />
(b)<br />
Chapter 2<br />
28<br />
<strong>Swirling</strong> Flows