Ion cyclotron emission (ICE) comprises strongly suprathermal emission, which has spectral peaks at multiple ion cyclotron harmonics. ICE is driven by the magnetoacoustic cyclotron instability, resulting from an inversion in the velocity-space distribution of the minority energetic ions. Recent experimental studies on DIII-D (K E Thome et al 2019 Nucl. Fusion 59 086011) have observed spectral peaks at the deuteron ion cyclotron frequency fci and its harmonics, mainly driven by fast ions from neutral beam injection (NBI). In the present study, ICE from deuterium plasmas in DIII-D heated by deuterium NBI is simulated for the first time. We use the EPOCH particle-in-cell code to solve the self-consistent Maxwell-Lorentz system of equations for fully kinetic electrons and thermal ions, together with the minority energetic NBI ions whose velocity distribution is modelled as a ring-beam. Gyro-orbits are fully resolved, hence all aspects of cyclotron resonance phenomenology are captured from first principles. High performance computations give rise to high-resolution spatiotemporal Fourier transforms of the spectral density of the excited electric and magnetic fields which are driven by collective relaxation of the NBI ions. We obtain spectral peaks at integer harmonics up to 7 f_ci, with the second harmonic peak strongest, consistent with the experimental observations of edge ICE in a DIII-D H-mode plasma. The growth rates of the fields at early times agree quantitatively with the theory of the magnetoacoustic cyclotron instability (MCI) and of linear mode conversion. We have compared these DIII-D ICE results and simulations to previous results for a classic edge ICE case, from JET deuterium-tritium plasma 26148. The DIII-D and JET ICE cases display distinct features, which are captured by our simulations; thus we are exploring a different regime of the MCI in DIII-D. We extend our study to include a realistic distribution function for the energetic ion population. Guided by the distribution function of fast ions computed from TRANSP for these plasmas, we construct an analytical approximation to the initial distribution function and load it into EPOCH. We examine how different distribution functions affect the characteristics of the simulated ICE, creating a pathway to enhanced diagnostic exploitation of ICE in relation to fast ions and their radial transport.