Q389 : How to handle cases with very large numbers of conductor segments.
Question How to handle cases with very large numbers of conductor segments.
Answer SES software users frequently ask about the limits imposed on the total number of segments that can be modelled by the MALT, MALZ and HIFREQ modules: why are there limits and what can be done when more segments are required for a model?
When numerical methods, such as the Moment Method, are used to model a physical system such as a grounding system, accuracy initially improves with an increasing degree of segmentation of the model elements (i.e., conductors), whereas speed decreases and memory requirements increase with increasing segmentation. One might therefore expect with faster and faster computer CPUs and greater available memory, that it should be possible to accommodate ever increasing numbers of segments (and conductors), with the only limitations being computer memory and user patience. This is not the case, however: as the number of segments continues to increase, small numerical errors, which are not apparent for small systems, build up and eventually have noticeable effects. As a result, SES is very cautious about increasing the conductor segment limits in its grounding software: careful testing must be performed each time a new increase is contemplated. Providing a version of the software with no limit at all (except for the host computer's available resources) is therefore out of the question!
As a result, the user must work with the available segment limit and make a judicious selection of conductors when the limit is exceeded. This can be done by considering the following:
1. In the MALT module, it is unnecessary to enter short conductors which interconnect other conductors: all conductors belonging to the same grounding system (e.g., Main Ground or Return Ground or Buried Structure) are automatically assumed to be interconnected by MALT. Short conductors will have a negligible effect on the ground resistance of the structure to which they belong (and therefore on the GPR) and usually a minor effect on touch and step voltages. In MALZ, a similar effect can be achieved by energizing disparate portions of the same grid from the same bus (note, however, that segments connected in this way will be held at similar potentials).
2. Typically, if a grounding grid is safe with low conductor density, then it will be even safer (at least within the grid perimeter) with all conductors present. At some point, addition of conductors will yield diminishing returns in terms of touch and step voltage reductions. For this reason, it is wasteful of computer resources, for example, to model all the rebar in a concrete foundation or to model very dense conductor mats.
3. The ground resistance of a grid (and therefore its GPR) is determined predominantly by the area spanned by the grid and to a smaller degree by the density of the grid conductors... up to a point. First, model all conductors defining the perimeter of the grid and any protruding conductors, along with representative conductors within the grid perimeter: i.e., enter conductors such that the entire area occupied by the grounding system is as uniformly represented as possible. This will provide a good approximation of the grounding system resistance and GPR. Next, add to this model details of the system in one or more areas for which touch and step voltages are to be investigated in detail in the first run: add the required potential profiles. Run and analyze your results. Do the same thing successively in the different zones in the grounding system area where touch and step voltages are of interest. In this way, you can break down one run into a series of runs, each of which focuses on a different part of the system.