Radiation View Factors
Radiation View Factors
Strip to strip......................................................................................................................................... 18
Tilted strip configurations ....................................................................................................................... 18
Equal adjacent strips ........................................................................................................................... 18
Triangular prism .................................................................................................................................. 18
Numerical computation ............................................................................................................................... 19
References ................................................................................................................................................... 20
L d d cos ( 1 ) cos ( 1 )
cos ( 1 ) d2 cos ( 2 ) cos ( 1 ) cos ( 2 )
d 212
d=
12
d2
= 1 12 1
=
=
M 1d1
M 1d1
r122
r122
(1)
F12 =
1
A1
cos cos
1
2
A
d
dA1
2
r122
A1 A2
(2)
Recall that the emitting surface (exiting, in general) must be isothermal, opaque, and Lambertian (a
perfect diffuser for emission and reflection), and, to apply view-factor algebra, all surfaces must be
isothermal, opaque, and Lambertian. Finally notice that F12 is proportional to A2 but not to A1.
One should stress the importance of properly identifying the surfaces at work; e.g. the area of a square
plate of 1 m in side may be 1 m2 or 2 m2, depending on our considering one face or the two faces. Notice
that the view factor from a plate 1 to a plate 2 is the same if we are considering only the frontal face of 2
or its two faces, but the view factor from a plate 1 to a plate 2 halves if we are considering the two faces
of 1, relative to only taking its frontal face.
For an enclosure formed by N surfaces, there are N2 view factors (each surface with all the others and
itself). But only N(N1)/2 of them are independent, since another N(N1)/2 can be deduced from
reciprocity relations, and N more by closeness relations. For instance, for a 3-surface enclosure, we can
define 9 possible view factors, 3 of which must be found independently, another 3 can be obtained from
Ai Fij = Aj Fji , and the remaining 3 by Fij = 1 .
j
and the 1 and 2 angles when z=0, 10 and 20), and the depth z of the dA2 location. The relationship are:
r12= x 2 + y 2 + z 2 = a 2 + z 2 , cos1=cos10cos, with cos1=y/r12=(y/a)(a/r12), cos10=y/a, cos=a/r12,
and cos2=cos20cos, therefore, between the two patches:
cos ( 1 ) cos ( 2 )
a 2 cos ( 10 ) cos ( 20 )
a 2 cos ( 10 ) cos ( 20 )
dF12 =
d2
d2
d2
=
=
2
r122
r124
a2 + z 2
(3)
Fig. 2. Geometry for view-factor between two patches in parallel strips: a 3D sketch, b) profile view.
Expression (3) can be reformulated in many different ways; e.g. by setting d2A2=dwdz, where the d2
notation is used to match differential orders and dw is the width of the strip, and using the relation
ad10=cos20dw. However, what we want is to compute the view factor from the patch dA1 to the whole
strip from z= to z=+, what is achieved by integration of (3) in z:
d 2 F=
12
a 2 cos ( 10 ) cos ( 20 )
( a2 + z 2 )
dwdz dF=
12
d F
2
12
d=
z
cos ( 10 ) cos ( 20 )
2a
d=
w
cos ( 10 )
2
d10
(4)
For instance, approximating differentials by small finite quantities, the fraction of radiation exiting a
patch of A1=1 cm2, that impinges on a parallel and frontal strip (10=20=0) of width w=1 cm separated a
distance a=1 m apart is F12=s/(2a)=0.01/(21)=0.005, i.e. a 0.5 %. It is stressed again that the exponent in
the differential operator d is used for consistency in infinitesimal order.
Now we want to know the view factor dF12 from an infinite strip dA1 (of area per unit length dw1) to an
infinite strip dA2 (of area per unit length dw2), with the geometry presented in Fig. 2. It is clear from the
infinity-extent of strip dA2 that any patch d2A1=dw1dz1 has the same view factor to the strip dA2, so that
the average coincides with this constant value and, consequently, the view factor between the two strips is
precisely given by (4); i.e. following the example presented above, the fraction of radiation exiting a long
strip of w1=1 cm width, that impinges on a parallel and frontal strip (10=20=0) of width w2=1 cm
separated a distance a=1 m apart is F12=w2/(2a)=0.01/(21)=0.005, i.e. a 0.5 %.
Notice the difference in view factors between the two strips and the two patches in the same position as in
Fig. 2b: using dA1 and dA2 in both cases, the latter is given by the general expression (1), which takes the
form dF12=cos10cos20dA2/(a2), whereas in the two-strip case it is dF12=cos10cos20dA2/(2a).
multiply by their respective areas (in our case L1, L2, L3, by unit depth length); next, we apply some
reciprocity relations, and finally perform de combination of equations as stated:
0 + F12 + F13 = 1 L1 F12 + L1 F13 = L1
F21 + 0 + F23 =
1 L2 F21 + L2 F23 =
L2
(5)
L1 F12 + L2 F23 =
L2
(6)
(7)
(8)
We see how easy it is now to recover the result for perpendicular bands of width L1 and L2, F12=
L1 + L2 L12 + L22 ( 2 L1 ) ; e.g. the view factor between equal perpendicular bands is F12= 2 2 2
=0.293, i.e. 29 % of the energy diffusively outgoing a long strip will directly reach an equal strip
perpendicular and hinged to the former, with the remaining 71 % being directed to the other side 3 (lost
towards the environment if L3 is just an opening).
Even though we have implicitly assumed straight-line cross-sections (Fig. 3), the result (8) applies to
convex triangles too (we only required Fii=0), using the real curvilinear lengths instead of the straight
distances. As for concave bands, the best is to apply (8) to the imaginary straight-line triangle, and
afterwards solve for the trivial enclosure of the real concave shape and its corresponding virtual straightline. For instance, if in our previous example of two equal perpendicular straight strips (F12= 2 2 2 ),
we substitute these planar strips for equal concave semi-cylinders with the same end points, the new view
Radiative view factors
5
factor between semi-cylinders is F12= 2 2 4 =0.146, i.e. now only 15 % of the radiations diffusively
emanating from concave semi-cylinder 1 arrives directly to concave semi-cylinder 2, another 15 %
impinges on its own surface (F11), and the remaining 70 % impinging on the third side, as before.
Now we generalise this algebraic method of computing view factors in two-dimensional geometries to
non-contact surfaces.
=
F12
L4 + L5 L3 L6
=
2 L1
(9)
Fig. 4. Sketch used to deduce F12 in the general case of two infinitely long bands.
The result (9) is deduced by applying the triangular relation (8) to triangle 134 (shadowed in Fig. 4)
and triangle 156, plus the closure relation to the quadrilateral 1326 (F13+F12+F16=1), namely:
L1 + L3 L4
2 L1
L + L5 L3 L6
1 F12 F12 = 4
F12 =
L + L6 L5
2 L1
F16 = 1
2 L1
F13 =
(10)
This procedure to compute view factors in two-dimensional configurations is known as the crossed-string,
first developed by H.C. Hottel in the 1950s. The extension to non-planar surfaces 1 and 2 is as already
presented for triangular enclosures. A further extension is possible to cases where there are obstacles
(two-dimensional, of course) partially protruding into sides 3 and/or 6 in the quadrilateral 1326 (Fig.
4); it suffices to account for the real curvilinear length of each string when stretched over the obstacles.
Example 1. Find the view factor between two long parallel cylinders of equal radii R, separated a distance
2 2 R between centres, using the crossed-string method.
Sol.: With this clever separation, angle in Fig. E1 happens to be =/4 (45), making calculations
simpler. We get F12 from (10) by substituting L1=2R (the source cylinder), L4 and L5 (the
crossing strings) each by the length abcde, and L3 and L6 (the non-crossing strings) each by 2 2 R
between. The length abcde is composed of arc ab, segment bc, and so on, which in our special
Radiative view factors
6
case
is
ab=R=(/4)R
bc=R,
and
abcde=2(ab+cd)=(/2)R+2R,
and
finally
F12=(L4+L5L3L6)/(2L1)= (2abcde 4 2 R)/(4R)=(R+4R 4 2 R)/(4R) =1/4+(1 2 )/=0.12,
as can be checked with the general expression for cylinders in the compilation following.
Fig. E1. Sketch used to deduce F12 between two infinitely long parallel cylinders.
WITH SPHERES
Patch to a sphere
Frontal
Case
From a small planar plate
facing a sphere of radius
R, at a distance H from
centres, with hH/R.
View factor
F12 =
Plot
1
h2
Level
Case
View factor
From a small planar plate
1
1 x
F12
level to a sphere of radius =
arctan 2
x h
R, at a distance H from
2
with x h 1
centres, with hH/R.
1 2 2
( F12 h1
h 1 )
2
Plot
Tilted
Case
View factor
Plot
+ arctan
with x h 2 1,
y x cot ( )
View factor
Plot
F12 = sin 2
View factor
1
=
F12 2r22 1
1
1+ 2
Plot
1
1
=
F12
1
2
1
1+ 2
View factor
Coaxial (=0):
arcsin ( s )
1
s
F12 =
2 (1 + h )
Plot
2h + h 2
From a small cylinder
with s
1+ h
(external lateral area
Perpendicular (=/2):
only), at an altitude
1
H=hR and tilted an angle
1+ h
xE ( x ) dx
4
, to a large sphere of
F12 = 2
0 1 x2
radius R, is between
the cylinder axis and the with elliptic integrals E(x).
line of centres).
Tilted cylinder:
F12 =
1
arcsin
1+ h 2
sin ( ) 1 z 2 d d
0=
0
=
with
z cos ( ) cos ( ) +
+ sin ( ) sin ( ) cos ( )
View factor
F11 = 1
From a finite cylinder
(surface 1) of radius R
and height H, to its
hemispherical closing
cap (surface 2), with
r=R/H. Let surface 3 be
the base, and surface 4
the virtual base of the
hemisphere.
F21 =
F31 =
4r
2r
, F=
F=
F=
12
13
14
, F22 =
1
1
,
, F23=
2 4r
2
, F32 = 1
with =
Plot
2r
, F34 = 1
2r
4r 2 + 1 1
r
Sphere to sphere
Small to very large
Case
From a small sphere of
radius R1 to a much
larger sphere of radius R2
at a distance H between
centres (it must be H>R2,
but does not depend on
R1), with hH/R2.
View factor
F12 =
Plot
1
1
1 1 2
2
h
Equal spheres
Case
From a sphere of radius
R to an equal sphere at a
distance H between
centres (it must be
H>2R), with hH/R.
View factor
F12 =
Plot
1
1
1 1 2
2
h
Concentric spheres
Case
Between concentric
spheres of radii R1 and
R2>R1, with rR1/R2<1.
View factor
Plot
F12=1
F21=r2
F22=1r2
(e.g. for r=1/2, F12=1, F21=1/4, F22=3/4)
Hemispheres
Case
From a hemisphere of
radius R (surface 1) to its
base circle (surface 2).
View factor
F21=1
F12=A2F21/A1=1/2
F11=1F12=1/2
Plot
10
From a hemisphere of
radius R1 to a larger
concentric hemisphere of
radius R2>R1, with
RR2/R1>1. Let the
closing planar annulus be
surface 3.
, F13 =
,=
F21
F31 =
2R
, F32 = 1
2 ( R 2 1)
with
1
1 ,
4
4
4
R2
1 1
=
F
22
1 2 ,
2
R
1
1
,
F23 =1 2 1
2 R 2 R2 1
F12 = 1
1 1 2
1
R 1 ( R 2 2 ) arcsin
2
R
WITH CYLINDERS
Cylinder to large sphere
See results under Cases with spheres.
View factor
Plot
F12=1
F21=r
F22=1r
11
View factor
Plot
View factor
F12 =
arcsin
Plot
1
h
View factor
Plot
h 2 4 h + 2 arcsin
F12 =
2
h
View factor
F12 =
F21 =
2r
, F13 = 1
, F22 = 1
with =
2r
Plot
,
, F23 =
2
4
4r 2 + 1 1
r
12
View factor
f
f
1
F12 =
1 arccos 2 4 , F13 = 1 F12 ,
f1 2h
F22 =1
Plot
1
2
2 R 2 1 hf 7
+
arctan
,
R R
h
2 R
F23 =
1 F21 F22
with f1 = h 2 + R 2 1 , f 2 = h 2 R 2 + 1 ,
f3 =
f 4 = f 3 arccos
=
f5
( A + 2)
4R2 ,
f2
1 f
+ f 2 arcsin 1 ,
Rf1
R
2
4R2
2h 2
+
1
,
,
=
f
1
6
h2
R 2 ( h2 + 4R 2 4 )
=
f 7 f5 arcsin f 6 arcsin 1 2 + ( f 5 1)
R 2
View factor
1
x4
Between two identical
=
+ 4 wy
F12
ln
2
2
parallel square plates of
w 1 + 2w
Plot
View factor
Plot
13
=
F12
1 p
ln + s t , with
2
w1 q
p ( w2 + w2 + 2 ) 2
1
2
q ( x 2 + 2 )( y 2 + 2 )
x w2 w1 , y w2 + w1
s u x arctan x y arctan y
u
u
t v x arctan x y arctan y
v
v
u x 2 + 4, v y 2 + 4
View factor
From an external-box face:
=
F12 x=
, F13 y=
, F14 x,
F11 0,=
=
, F16 x=
, F17 za 2=
, F18 r ,
F15 x=
=
F1,10 r=
, F1,11 r=
, F1,12 r
F19 0,=
Plot
1 a) p
(
ln + s + t
z= F71=
2
4 a q
2
3 2a + 3a 2
p 2
2
(1 a )
2
q 2 18 + 12a + 18a
2
(1 a )
2
w
s u 2 arctan w arctan
#1, and its corresponding
u
u
2
w
8 (1 + a 2 )
1+ a
,w2
u 8, v
1 a
1 a
and:
14
r a 2 (1 z ) 4
y 0.2 (1 a )
2
x (1 y za 4r ) 4
(e.g. for a=0.5, F11=0, F12=0.16, F13=0.10,
F14=0.16, F15=0.16, F16=0.16, F17=0.20,
F18=0.01, F19=0, F1,10=0.01, F1,11=0.01,
F1,12=0.01), and (F71=0.79, F72=0.05, F73=0,
F74=0.05, F75=0.05, F76=0.05, F77=0, F78=0,
F79=0, F7,10=0, F7,11=0, F7,12=0).
Notice that a simple interpolation is proposed
for yF13 because no analytical solution has
been found.
View factor
F12 =
Plot
x y
1
ln 2
xy x1 + y 1
2 2
1 1
2
1
x
+ 2 x y1 arctan arctan x
y1
y
+ 2 y x1 arctan arctan y
x1
with x1 1 + x 2 and y1 1 + y 2
(e.g. for x=y=1, F12=0.1998)
Equal discs
Case
Between two identical
coaxial discs of radius R
and separation H, with
r=R/H.
View factor
F12 = 1 +
Plot
1 4r 2 + 1
2r 2
Unequal discs
Case
From a disc of radius R1
to a coaxial parallel disc
of radius R2 at separation
H, with r1=R1/H and
r2=R2/H.
View factor
Plot
x y
2
with x =
1 + 1 r12 + r22 r12 and
F12 =
=
y
x 2 4r22 r12
15
Strip to strip
Note. See the crossing-string method, above, for these and other geometries.
Case
View factor
Between two identical
parallel strips of width W
and separation H, with
h=H/W.
Plot
F12 = 1 + h 2 h
( w1 + w2 )
F12 =
+4
2 w1
( w2 w1 )
+4
2 w1
View factor
Plot
1 + cos
2
1 cos
Back side: F12 =
2
Patch to disc
Case
From a patch to a parallel
and concentric disc of
radius R at distance H,
with h=H/R.
View factor
F12 =
Plot
1
1 + h2
16
Perpendicular configurations
Square plate to rectangular plate
Case
From a square plate of
with W to an adjacent
rectangles at 90, of
height H, with h=H/W.
View factor
Plot
1 1
1
1 h
F12 =
+ h arctan h1 arctan ln h2
4
h
h1 4
4
h
with h=
1 + h 2 and h2 = 2 1 2
1
h 2+h
2
rectangles at 90,
of
h
2h
height H and width L,
1 h h
with h=H/L.
+ ln 1 2
4h 4
with =
h1 2 (1 + h 2 )
1
and h=
1
2
h1
Plot
2 h 2 1
View factor
=
F12
From a horizontal
rectangle of WL to
adjacent vertical
rectangle of HL, with
h=H/L and w=W/L.
Plot
1
1
1
h arctan + w arctan
w
h
w
1
h 2 + w2 arctan
2
2
h +w
2
2
1
+ ln ab w c h
4
(1 + h )(1 + w ) ,
with a =
2
1 + h 2 + w2
w2 (1 + h 2 + w2 )
h 2 (1 + h 2 + w2 )
, c=
b=
(1 + h2 )( h2 + w2 )
(1 + w2 )( h2 + w2 )
(e.g. for h=w=1, F12=0.20004)
Radiative view factors
17
From non-adjacent
rectangles, the solution
can be found with viewA
A
F12 = F12+ 2' F12' = 2+ 2' F2+ 2'1 2' F2'1 =
factor algebra as shown
A1
A1
here
A
A
= 2+ 2' ( F2+ 2'1+1' F2+ 2'1' ) 2' ( F2'1+1' F2'1' )
A1
A1
Strip to strip
Note. See the crossing-string method, above, for these and other geometries.
Case
View factor
F12 =
(e.g. F12
Plot
1 + h 1 + h2
2
H =W
=
1
2
=
0.293 )
2
View factor
F12 = 1 sin
(e.g. F12
1
=
Plot
2
0.293 )
=
2
Triangular prism
Case
View factor
Between two sides, 1 and
2, of an infinite long
triangular prism of sides
L1 + L2 L3
F12 =
L1, L2 and L3 , with =
2 L1
h=L2/L1 and being the
1 + h 1 + h 2 2h cos
angle between sides 1
=
and 2.
2
Plot
18
NUMERICAL COMPUTATION
Several numerical methods may be applied to compute view factors, i.e. to perform the integration
implied in (2) from the general expression (1). Perhaps the simpler to program is the random estimation
(Monte Carlo method), where the integrand in (2) is evaluated at N random quadruples, (ci1, ci2, ci3, ci4)
for i=1..N, where a coordinates pair (e.g. ci1, ci2) refer to a point in one of the surfaces, and the other pair
(ci3, ci4) to a point in the other surface. The view factor F12 from surface A1 to surface A2 is approximated
by:
F12 =
A2
N
i =1
cos 1 cos 2
r122
(11)
i
where the argument in the sum is evaluated at each ray i of coordinates (ci1, ci2, ci3, ci4).
Example 2. Compute the view factor from vertical rectangle of height H=0.1 m and depth L=0.8 m,
towards an adjacent horizontal rectangle of W=0.4 m width and the same depth. Use the Monte
Carlo method, and compare with the analytical result.
Sol.:
The analytical result is obtained from the compilation above for the case of With plates and
discs / Perpendicular configurations / Rectangular plate to unequal rectangular plate, obtaining,
for h=H/L=0.1/0.8=0.125 and w=W/L=0.4/0.8=0.5 the analytical value F12=0.4014 (mind that
we want the view factor from the vertical to the horizontal plate, and what is compiled is the
opposite, so that a reciprocity relation is to be applied).
For the numerical computation, we start by setting the argument of the sum in (11) explicitly in
terms of the coordinates (ci1, ci2, ci3, ci4) to be used; in our case, Cartesian coordinates (xi, yi, zi,
yi) such that (xi, yi) define a point in surface 1, and (zi, yi) a point in surface 2. With that choice,
2
cos1=z/r12, cos2=x/r12, and r12 =
x 2 + z 2 + ( y2 y1 ) , so that:
A2 N cos 1 cos 2
WL N zx
WL N
zx
WL N
=
=
=
fi
r 2=
4
2
2
2
N i1
N
r
N
N
i
i
i
=
=
=
1
1
1
+
+
x
z
y
y
'
12
12
(
)
i
i
i
where fi is the value of the function at a random quadruple (xi, yi, zi, yi). A Matlab coding may be:
W=0.4; L=0.8; H=0.1; N=1024;
%Data, and number of rays to be used
f= @(z,y1,x,y2) (1/pi)*x.*z./(x.^2+z.^2+(y2-y1).^2).^2; %Defines the function
for i=1:N fi(i)=f(rand*H, rand*L, rand*W, rand*L);end; %Computes its values
F12=(W*L/N)*sum(fi)
%View factor estimation
=
F12
Running this code three times (it takes about 0.01 s in a PC, for N=1024), one may obtain for F12 the three
values 0.36, 0.42, and 0.70, but increasing N increases accuracy, as shown in Fig. E2.
19
Fig. E2. Geometry for this example (with notation used), and results of the F12-computation with a
number N=2in of random quadruplets (e.g. N=210=1024 for in=10); three runs are plotted, with the
mean in black.
REFERENCES
Howell, J.R., A catalog of radiation configuration factors, McGraw-Hill, 1982. (web.)
Siegel, R., Howell, J.R., Thermal Radiation Heat Transfer, Taylor & Francis, 2002.
Back to Spacecraft Thermal Control
20