test3d26, dielectric, cylinder

Test3d26.dat, 26th 'test' file for CPO3DS

Field inside a dielectric cylinder

Care must be taken in setting up dielectric simulations, so it is very important to read the guidance.

A uniform field of 1V/mm is created by applying a transverse linear field to a cylinder. (Technical note: the option at the bottom of the page

/databuilder/segments/advanced options must be used.)

A cylinder radius 0.05mm is put at the axis. It has a dielectric constant K = 2.

The inaccuracy used for evaluation of the fields at the dielectric interfaces is 1.E-7 (but see below).

As the text-books say, the field inside the cavity should be uniform, of strength

E_cavity = E_ext*2/(1 + K)

In the present simulation, with the present number of segments in the outer sphere, E_ext is 0.998456, slightly different from the 1V/mm. Therefore we should have E_cavity = (2/3)*0.998456 = 0.665637. The program gives 0.665871, which is in error by 0.00023, which is 0.04%.

Here are some results for other values of K and other values of N, the number of subdivisions (compared with the number in the present simulation).

K	N/Npresent	program	correct	%rel error
2	0.5	0.662945	0.662574	0.06
2	1	0.665871	0.665637	0.04
2	4	0.666400	0.666369	0.005
10	1	0.181917	0.181537	0.21
1E6	1	0.000713	0.000002	-
0.5	1	1.33145	1.33128	0.01
0.2	1	1.66577	1.664093	0.10
0.001	1	1.99822	1.99492	0.17

The entries above that have K < 1 effectively simulate a cavity in a dielectric of constant 1/K.

The errors do not change monotonically with K or N because an empirical correction is used in the program to try to minimise the dependence on these parameters.

Here are some results for different values of the user-selected inaccuracy for the dielectric field evaluations. This inaccuracy is used for the evaluation of the fields at the dielectric interfaces can be selected by the user (at the bottom of the 'tracing accuracy' sheet). The default value is 0.001 (which can be changed, see the bottom of the 'tracing accuracy' sheet). Note that this inaccuracy is not the same as the inaccuracy used in calculating the surface charges of the conducting electrodes, which is always 1.E-7 and which cannot usually be changed by the user.

In all cases here K = 2, N = Npresent.

inacc	program	correct	%rel error	time, sec
1.E-7	0.665871	0.665637	0.04	13.5
1.E-5	0.665445	0.665637	-0.03	9.7
1.E-3	0.665938	0.665637	0.05	2.9
1.E-1	0.666006	0.665637	0.06	2.2

The default value therefore seems to be adequate for most purposes. The actual inaccuracy is only loosely related to the obtained inaccuracy.

Note that dielectric calculations with CPO3D are always much slower than the analogous non-dielectric calculations.

Note that, apart from water, most materials have K < 2.