Radiation hydrodynamics (RHD) simulations are used to study many astrophysical phenomena, however they require the use of simplified radiation transport and thermal prescriptions to reduce computational cost. In this paper we present a systematic study of the importance of microphysical processes in RHD simulations using the example of D-type H ii region expansion. We compare the simplest hydrogen-only models with those that include: ionisation of H, He, C, N, O, S and Ne, different gas metallicity, non-LTE metal line blanketed stellar spectral models of varying metallicity, radiation pressure, dust and treatment of photodissociation regions. Each of these processes are explicitly treated using modern numerical methods rather than parameterisation. In line with expectations, changes due to microphysics in either the effective number of ionising photons or the thermal structure of the gas lead to differences in D-type expansion. In general we find that more realistic calculations lead to the onset of Dtype expansion at smaller radii and a slower subsequent expansion. Simulations of star forming regions using simplified microphysics are therefore likely overestimating the strength of radiative feedback. We find that both variations in gas metallicity and the inclusion of dust can affect the ionisation front evolution at the 10-20 per cent level over 500 kyr, which could substantially modify the results of simplified 3D models including feedback. Stellar metallicity, radiation pressure and the inclusion of photodissociation regions are all less significant effects at the 1 per cent level or less, rendering them of minor importance in the modelling the dynamical evolution of H ii regions.