A taylor series expansion of a function V(x,y,z) for a 'step' of size a in x gives
V(a,0,0) = V(0,0,0) + a ¶ V/¶ x|o + 1/2 a2 ¶2V/¶ x2|o+ terms in a3 and higher .
In slightly different notation we could write
V(a,0,0) = Vo + aVx + 1/2 a2 Vxx + ...
V(-a,0,0) = Vo - aVx + 1/2 a2 Vxx + ... .
Then we could solve for Vxx and find
Vxx = ( V(a,0,0)+V(-a,0,0) - 2Vo )/a2 .
Likewise
V(0,b,0) = Vo + bVy + 1/2 b2 Vyy
+ ... and
Vyy = ( V(0,b,0)+V(0,-b,0) - 2Vo )/b2 .
Laplace's equation is Vxx + Vyy+ +Vzz = 0. For a step size of ± a in the x-direction, a step of ± b in the y-direction and ± c in the z-direction we may write down laplace's equation and solve it for Vo:
2
Vo (1/a2 + 1/b2 + 1/c2) = ( V(a,0,0)+V(-a,0,0)
)/a2 + ( V(0,b,0)+V-,-b,0) )/b2 + ( V(0,0,c)+V(0,0,-c)
)/c2 ,
or
Vo
= Cx (V(a,0,0)+V(-a,0,0)) + Cy (V(0,b,0)+V-,-b,0))
+ Cz (V(0,0,c)+V(0,0,-c)) ,
where Cx = 1/(2a2)/( 1/a2 + 1/b2 + 1/c2) = 1/2 b2 c2 /(a2b2+ b2c2+ c2a2 )etc.
If a = b = c (same step sizes in x, y, and z) we find Vo is the average of all the cells around it:
Vo = 1/6 ( V(a,0,0) + V(-a,0,0)) + V(0,b,0) + V(0,-b,0) + V(0,0,c) + V(0,0,-c) )
(In any case, Vo is the average of all the cells around it, but it's a weighted average over the different step sizes if a¹ b, or b¹ c, or a¹ c)
Where there is no z-dependence, V(0,0,c) = V(0,0-c) = Vo. Then if a = b,
Vo = 1/4 ( V(a,0,0) + V(-a,0,0)) + V(0,b,0) + V(0,-b,0)) .
The plan is to repetitively solve for Vo in each cell by using the iterate feature in Excel under Tools/Options/Calculation. By setting 1000 iterations, you get that number each time you press F9, in manual calculation mode.