Mass and energy balance errors noted in a number of IBM-executed problems are caused by the lack of precision in computing total mass and energy values for a domain. This problem is evident in domains constructed with highly variable mesh sizes during the early time of simulation. The machine round-off was corrected by double-precisioning certain calculations for mass and energy balance. Small differences that exist between the improved INTERCOMP code operating on an IBM machine and the old version on a CDC machine seem unimportant. The noted differences are greatest at an onset of physical system perturbation. These differences diminish rapidly with each succeeding time step. Comparisons with numerical and analytical solutions appear to prove authenticity of code results. Numerical comparisons with the CCC computer code on the Mobile experiment data demonstrate the advantage of using aquifer influence functions in place of an infinitely large mesh. The one-dimensional heat transfer in the overburden and underburden appears sufficiently accurate to describe aquifer heat losses.

We provide a synthesis of our Thomson-scattering measurements of electron temperature (T/sub e/) and density (n/sub e/) for the Tandem Mirror Experiment (TMX). TMX operated in two modes - high and low T/sub e/. When performing in the high T/sub e/ mode (in general > 100 eV), heating the central-cell ions with neutral beams raised T/sub e/ in the end plug. We achieved a maximum T/sub e/ of 260 eV in the east end plug. Specifically, our experiments demonstrated that in the end plug, the radial T/sub e/ profiles were flat to r = 5 cm; the ratio of potential (phi/sub p/) to T/sub e/ ranged between four and six. In addition, we found that although T/sub e/ in the central cell was generally comparable to that in the plug, it was often not constant along a magnetic field line. Under some conditions a non-Maxwellian electron distribution may have been present.