Nonorthogonal multiple access (NOMA) superposes several user signals for simultaneous transmission, considerably ameliorating spectral efficiency. To exploit efficiently radio frequency energy sources surrounding NOMA transmitters for improved energy efficiency, NOMA networks further employ energy harvesting, which practically possesses nonlinear characteristic. In addition, multiple antennas can be deployed at users to transfer-and-harvest energy efficiently and to transmit/receive information reliably with antenna array processing. As such, this paper evaluates numerically the throughput and the outage probability of NOMA networks accounting for practical factors such as nonlinear energy harvesting, all multi-antenna users, and imperfect channel information. Illustrative results expose that nonlinear energy harvesting and channel information imperfection considerably degrade system performance, which is governed flexibly by multiple parameters. Notably, NOMA networks can prevent a total outage and achieve the best performance with a proper parameter selection. Moreover, their performance is considerably ameliorated with accreting the number of antennas.