We present X-ray spectra spanning 18 years of evolution for SN 1996cr, one of the five nearest SNe detected in the modern era. Chandra HETG exposures in 2000, 2004, and 2009 allow us to resolve spectrally the velocity profiles of Ne, Mg, Si, S, and Fe emission lines and monitor their evolution as tracers of the ejecta-circumstellar medium (CSM) interaction. To explain the diversity of X-ray line profiles, we explore several possible geometrical models. Based on the highest signal-to-noise 2009 epoch, we find that a polar geometry with two distinct opening angle configurations and internal obscuration can successfully reproduce all of the observed line profiles. The best fit model consists of two plasma components: (1) a mildly absorbed (2×10 21 cm −2 ), cooler (≈2 keV) with high Ne, Mg, Si, and S abundances associated with a wide polar interaction region (half-opening angle ≈58 • ); (2) a moderately absorbed (2×10 22 cm −2 ), hotter ( > ∼ 20 keV) plasma with high Fe abundances and strong internal obscuration associated with a narrow polar interaction region (half-opening angle ≈20 • ). We extend this model to seven further epochs with lower signal-to-noise ratio and/or lower spectral-resolution between 2000-2018, yielding several interesting trends in absorption, flux, geometry and expansion velocity. We argue that the hotter and colder components are associated with reflected and forward shocks, respectively, at least at later epochs. We discuss the physical implications of our results and plausible explosion scenarios to understand the X-ray data of SN 1996cr.