As you probably realize, if you were making a cylinder or a cone, you would (with many materials anyway) not need the press. This is because those surfaces,unlike a sphere, have no "intrinsic curvature" so you can bend the pieces to shape without deforming them. For a sphere this is impossible. As a result, if you are pressing your piece into a spherical mold, the material will be deformed.

If the segments were being bent into place without any intrinsic deformation (you would not get a perfect sphere of course) then the total height of the segment would be (pi/2)r [you can use 1.5708 r, good to a millimeter] while (with 18 segments) the width of each would be 2r x sin(10 degrees) or about 0.3473 r at the base. Call that width W.

Now, as you go up in 2 degree increments (that is, increments of height/45, or 0.0349 r) the width at the nth site will be cos(2n degrees) W. The first few of these values will be very close to 1:

n=0: width = W (= .3473 r)
n=1: width = 0.9994 W (= .3471 r)
n=2: width = 0.9976 W (= .3465 r)
...
n=43 width = 0.0698 W (= .0242 r)
n=44 width = 0.0349 W (= .0121 r)
n=45 width = 0 ("north pole")

If you are using a spherical mold and pressing causes your elements to be the wrong shape, you will probably have to resort to ad hoc methods. Try first adjusting the height till it presses to the right final height. Divide the height into 45ths and press another piece; then measure the widths. At each site, compute a new width using the formula

new width = (computed width)^2 / (pressed width of test piece)

and it ought to be pretty close.

Good Hunting!
-RD