Uniform planar antenna (UPA) structures are expected to become popular for massive multiple-input multiple-output (MIMO) systems because they enable deployment of a large number of antennas in limited space. In frequency division duplexing (FDD) systems, quantized channel state information (CSI) should be fed back from users to base stations to make adaptive transmission possible. However, it is difficult to accurately quantize massive MIMO channels due to their large dimensions. In this paper, we propose practical CSI quantizers for full-dimension (FD) MIMO systems employing UPAs. We focus on quantizing and combining dominant beams in channels by keeping realistic channel properties and antenna structures into account. To scan a channel space, we also develop a multi-round beam search approach. Numerical results verify that the proposed quantizers show better quantization performance than the previous channel quantization techniques.