This document presents a new approach for arc fault analysis in DC distribution systems using wavelet transform-based spectral energy calibration. It involves modeling DC arcs using the Cassie arc model and analyzing the arc voltage signatures. The arc voltage signals are decomposed using wavelet multiresolution analysis to extract detail and approximation signals. Spectral energy values are then calculated from these signals for arc fault classification and identification of the fault location. MATLAB simulations are performed to validate the proposed methodology. The results show increases in the arc voltage magnitude and higher frequency components in the wavelet decomposed signals when arcs occur, allowing identification of normal and fault conditions.