Although this might indicate a bad solution, it usually is an artifact of the nature of the contour plot which is exagerated by the plot being too coarse with respect to the geometric detail.
To illustrate the nature of contour plots and artifacts such as this, consider the electrode (light yellow area bounded by black lines) below. To generate a voltage contour plot over an area surrounding the electrode, a grid is set up and the voltage is calculated at the cross-points of the grid. The points closest to the electrode are indicated in red as V1-V9:
In total the sampled points will produce voltage in the range Vmin to Vmax. Now, we request some number of contours, and hence assign contours to the levels
Vmin + n*(Vmax – Vmin)/number
Suppose that process results in the contour which will be colored blue being assigned to a voltage V’ which is close to the voltage of the electrode shown. What we do to plot the blue contour is to interpolate between neighboring sampled points and put a blue dot wherever a value of V’ is achieved.
Next, we connect-the-dots with blue lines.
You can see that this process will produce contour plots, but that they will intrinsically cross the electrode boundaries in a case like that shown above no matter how accurately the values are obtained at V1..V9. However, by making a finer density plot, there will be more detail around the electrode and the contours will wrap better around them. In IES software you can make finer density plots by either using the Coarse/Medium… switch, or by making a more local plot on any given setting.
In fact, our Coarse/Medium/Fine/Very Fine options are a bit more intelligent than this explanation. You will note that they take longer to set up the points than the User Grid option, because they do attempt to locate the points intelligently with respect to surfaces (i.e. it is taking extra time to adjust positions in a non-uniform mesh). This is one aspect of the software that tends to improve from version to version. However, no matter how sophisticated this routine becomes, if there are not sufficient sampled points then one thing or another will look odd in a contour plot. Also note that, you cannot tell by looking at a contour plot where the sampled points are, only where the vertices obtained by interpolation are.