Heat Transfer Coefficients — External Natural Convection
Vertical Wall
Figure 4-30: Schematic representation of geometry and parameters for the heat transfer coefficient correlation applied to natural convection on a vertical wall.
The following correlations correspond to equations 9.26 and 9.27 in Ref. 21:
(4-182)
where the height of the wall, L, is a correlation input.
The Rayleigh number is defined as:
(4-183)
if the density depends on temperature only, or as:
(4-184)
if the density depends on other parameters.
All material properties are evaluated at (T + Text) ⁄2, except ρs, which is evaluated at the wall temperature, T, and g is the acceleration of gravity equal to 9.81 m/s2. This correlation is valid for 104RaL 1013. The laminar-turbulent transition at RaL = 109 is handled by the use of a smoothed Heaviside function with a transition of size 108, which corresponds to 10% of the threshold value.
Inclined Wall
Figure 4-31: Schematic representation of geometry and parameters for the heat transfer coefficient correlation applied to natural convection on an inclined wall.
The following correlations correspond to equations 9.26 and 9.27 in Ref. 21 (the same as for a vertical wall):
(4-185)
where the length of the wall, L, is a correlation input and φ is the tilt angle (the angle between the wall and the vertical direction; φ = 0 for vertical walls). These correlations are valid for 60° < φ < 60° and 104RaL 1013.
The definition of the Raleigh number, RaL, is analogous to the one for vertical walls and is given by the following:
(4-186)
if the density depends on temperature only, or as
(4-187)
if the density depends on other parameters.
g denotes the gravitational acceleration, equal to 9.81 m/s2.
For turbulent flow, 1 is used instead of cos φ in the expression for h, because this gives better accuracy (see Ref. 41).
According to Ref. 21, correlations for inclined walls are only satisfactory for the top side of a cold plate or the down face of a hot plate. Hence, these correlations are not recommended for the bottom side of a cold face and for the top side of a hot plate.
The laminar-turbulent transition depends on φ (see Ref. 41). Unfortunately, little data is available about transition. There is some data available in Ref. 41 but this data is only approximate, according to the authors. In addition, data is only provided for water (Pr around 6). For this reason, the flow is defined as turbulent, independently of the φ value, when
All material properties are evaluated at (T + Text) ⁄ 2, except ρs, which is evaluated at the wall temperature, T. The laminar-turbulent transition at RaL = 109 is handled by the use of a smoothed Heaviside function with a transition of size 108, which corresponds to 10% of the threshold value.
Horizontal Plate, Upside
Figure 4-32: Schematic representation of geometry and parameters for the heat transfer coefficient correlation applied to natural convection on the top surface of an horizontal plate.
The following correlations correspond to equations 9.30–9.32 in Ref. 21 but can also be found as equations 7.77 and 7.78 in Ref. 41.
If ρ < ρext, then
(4-188)
while if ρ  ρext, then
(4-189)
RaL is given by Equation 4-183 or Equation 4-184, and L, the characteristic length (defined as area/perimeter, see Ref. 41) is a correlation input. The material data are evaluated at (T + Text) ⁄2, except ρswhich is evaluated at the wall temperature, T.
When the density depends only on temperature, the conditions ρ < ρext and ρ ≥ ρext can be replaced by T > Text and T ≤ Text respectively.
The laminar-turbulent transition at RaL = 107 is handled by the use of a smoothed Heaviside function with a transition of size 106, which corresponds to 10% of the threshold value.
Horizontal Plate, Downside
Figure 4-33: Schematic representation of geometry and parameters for the heat transfer coefficient correlation applied to natural convection on the bottom surface of an horizontal plate.
Equation 4-188 is used when ρ ≥ ρext (or  Text) and Equation 4-189 is used when ρ < ρext (or T > Text). Otherwise it is the same implementation as for Horizontal Plate, Upside.
Long Horizontal Cylinder
Figure 4-34: Schematic representation of geometry and parameters for the heat transfer coefficient correlation applied to natural convection on a long horizontal cylinder.
The following correlations correspond to equations 9.34 in Ref. 21. It is validated for RaD ≤ 1012.
(4-190)
Here D is the cylinder diameter and RaD is given by
if the density depends on temperature only, or as:
if the density depends on other parameters.
The material data are evaluated at (T + Text) ⁄2, except ρswhich is evaluated at the wall temperature, T.
Sphere
Figure 4-35: Schematic representation of geometry and parameters for the heat transfer coefficient correlation applied to natural convection on a sphere.
The following correlations correspond to equation 9.35 in Ref. 21. It is validated for RaD ≤ 1011 and Pr ≥ 0.7.
(4-191)
Here D is the cylinder diameter and RaD is given by
if the density depends on temperature only, or as:
if the density depends on other parameters.
The material data are evaluated at (T + Text) ⁄2, except ρswhich is evaluated at the wall temperature, T.
Vertical Thin Cylinder
Figure 4-36: Schematic representation of geometry and parameters for the heat transfer coefficient correlation applied to natural convection on a vertical thin cylinder.
The following correlation corresponds to equation 7.83 in Ref. 41. It evaluates the heat transfer coefficient on the lateral surface of the thin cylinder. The horizontal disks (top and bottom) should be treated as horizontal plates.
For the lateral surface, the heat transfer coefficient reads:
where D is the cylinder diameter, H is the cylinder height, and RaH is given by
if the density depends on temperature only, or as:
if the density depends on other parameters.
The material data are evaluated at (T + Text) ⁄2, except ρswhich is evaluated at the wall temperature, T.
If the thermal boundary layer thickness δT is much smaller than the cylinder diameter D, the curvature does not play any significant role, and vertical wall correlations should be used. In practice, the (δT < D) criterion requires