The kinetic Sunyaev Zel’dovich (kSZ) effect, cosmic microwave background (CMB) anisotropies induced by scattering from free electrons in bulk motion, is a primary target of future CMB experiments. In addition to shedding light on the distribution of baryons and the details of the epoch of reionization, measurements of the kSZ effect have the potential to address fundamental questions about the structure and evolution of our Universe on the largest scales and at the earliest times. This potential is unlocked by combining measurements of small-scale CMB anisotropies with large-scale structure surveys, a technique known as kSZ tomography. Previous work established a quadratic estimator for the remote dipole field, the CMB dipole observed at different locations in the Universe, given a CMB map, and a redshift-binned map of large-scale structure. This previous work did not include gravitational lensing, redshift space distortions, or nonlinear evolution of structure. In this paper, we investigate how well the remote dipole field can be reconstructed in the presence of such effects by using mock data from a suite of simulations of gigaparsec-sized regions of the Universe. To properly model both large and small scales, we develop a novel box-in-box simulation pipeline, where small-scale information is obtained from L-PICOLA N-body simulations, and large-scale information obtained by evolving fields using linear theory and adding the resulting corrections to the N-body particle data. This pipeline allows us to create properly correlated maps of the primary CMB including lensing as well as the kSZ effect and density maps on the past light cone of an observer. Analyzing an ensemble of mocks, we find that the dipole field can be reconstructed with high fidelity over a range of angular scales and redshift bins. However, we present evidence for a bias due to the nonlinear evolution of structure. We also analyze correlations with the primary CMB, investigating the ability of kSZ tomography to reconstruct the intrinsic CMB dipole. Our results constitute a proof-of-principle that kSZ tomography is a promising technique for future data sets.