Versions 96 beta 5 and later of Cloudy have a state of the art treatment of grain physics (this was included in previous betas of C96 but was not used by default). The implementation was by Peter van Hoof and was done in collaboration with Joe Weingartner, Kevin Volk, and Peter Martin. A description of the new implementation can be obtained in PDF or ps format.
Grains were a major pacesetter for the code when there were first introduced in the late 1980s (see Baldwin et al. 1991; ApJ, 374; 580). That treatment assumed that each grain species (silicate, graphite) had a single mean size and was based on classical heating and cooling. This remained the default grain type through version C96b4, although by that time the old style grain physics added little additional overhead to a calculation.
The current grain physics, described in the article mentioned above, is today's state of the art. It fully resolves the grain size distribution, explicitly allowing the grain temperature, charge, energy exchange with the gas, photoelectric heating, and drift velocity to be determined as a function of grain radius. Additionally quantum (or single photon) heating is treated.
This additional physics comes at a price. By default, when grains are specified, both quantum heating and size resolution are included. This will slow down the calculation considerably (just as the old grains did in the late 1980s). Eventually machines will catch up, but in the mean time it is useful to think about ways to speed up the calculation.
Quantum heating is the slowest part of the new physical treatment. This follows the temperature pulse that occurs after a grain absorbs a photon and is most important for smaller grains. The main effect is to increase the predicted continuum emission in the Wien tail of the grain thermal emission. This emission is only output by the code if "punch continuum" or "punch spectrum" output is requested, or when the "print continuum" command given. If this output is not desired then it may be appropriate to turn off quantum heating. This is done with the "no qheat" option on the grains command. The only significant effect of turning off quantum heating, when these observables are not printed, might be in the effects of grain thermal emission directly pumping lines that lie on the Wien tail of the thermal energy distribution. In most cases this should be small.
A further speedup can be achieved by using single-sized rather than size-resolved grains. This is done with the "single" option on the grains command and will make the grain treatment about as fast (and accurate) as the old style grains. This is not recommended since it does affect the predicted emission line spectrum. Resolving the grain size distribution is necessary to recover the thermal interactions between grains and the surrounding gas as well as the photoelectric heating of the gas by the grains.