We present a detailed model for the ionized absorbing gas evident in the 900 ksec Chandra HETGS spectrum of NGC 3783 .
The analysis was carried out with PHASE , a new tool designed to model X-ray and UV absorption features in ionized plasmas .
The 0.5-10 keV intrinsic continuum of the source is well represented by a single power law ( \Gamma = 1.53 ) and a soft black-body component ( kT \sim 0.1 keV ) .
The spectrum contains over 100 features , which are well fit by PHASE with just six free parameters .
The model consists of a simple two phase absorber with difference of \approx 35 in the ionization parameter and difference of \approx 4 in the column density of the phases .
The two absorption components turned out to be in pressure equilibrium , and are consistent with a single outflow ( \approx 750 km s ^ { -1 } ) and a single turbulent velocity ( 300 km s ^ { -1 } ) , and with solar elemental abundances .
The main features of the low ionization phase are an Fe M-shell unresolved transition array ( UTA ) and the O vii lines .
The O vii features , usually identified with the O viii and a warm absorber , are instead produced in a cooler medium also producing O vi lines .
The UTA sets tight constraints on the ionization degree of the absorbers , making the model more reliable .
The high ionization phase is required by the O viii and the Fe L-shell lines , and there is evidence for an even more ionized component in the spectrum .
A continuous range of ionization parameters is disfavored by the fits , particularly to the UTA .
Our model indicates a severe blending of the absorption and emission lines , as well as strong saturation of the most intense O absorption lines .
This is in agreement with the O vii ( \tau _ { \lambda } = 0.33 ) and O viii ( \tau _ { \lambda } = 0.13 ) absorption edges required to fit the spectrum .
The low ionization phase can be decomposed into three subcomponents , based on the outflow velocity , FWHM , and H column densities found for three out of the four UV absorbers detected in NGC 3783 .
However , the ionization parameters are systematically smaller in our model than derived from UV data , indicating a lower degree of ionization .
Finally , our model predicts a Ca xvi 
line for the feature observed at around 21.6 Å ( a feature formerly identified as O vii ) , constraining the contribution from a zero redshift absorber .