anl_ini_1.dat (download) is the input.dat file adapted for trans/ini mode.
the initial core configuration is specfied by means of zgri= .., although, in trans/ini mode, core configuratin at sec= 0 only is used.
Best choice of beg requires some caution: for dsec= 1, a too low value such as .001 which was acceptable for the uniform core case results here in an unstable transient and negative flux.
A lower bound for beg of about .05 leads (anl_ini_ch1 ) to rog= -0.1016 reached after 5 s, after a violent undershoot of f2c at first iteration (time-step) to level .2. This may be explained by the flux profile at sec=1 is still of bad quality as shown on graph (anl_ini_ch2) of the f2ck(1:3) /f2c profile at successive iterations.
Remember that at sec= 0 (ini, core) assumes flat power distribution.
For lower values of beg, the first iteration f2c becomes negative but rOg still converges towards the correct eigen reactivity (homogeneous system)! For still lower beg, the transient may "capture" the first harmonic mode.
We observed a similar behaviour when applying Wielant acceleration in classical diffusion program, in case the guess of Wieland upper bound amec is taken lower than fondamental eigenvalue.

The reference r0g is -.109 and was calculated with a fine-mesh two-group diffusion program;
Repeating the exercice with 90 meshes provies r0g= -0.10596 closer to the ref
Still refining the meshing would however not help reaching the theoretical value.
This can be explained by comparing (anl_ini_ch3) fci with the reference for 90 meshes. It is observed that, whilst the overall agreement is quite good, ther are still some deviations at configuration interface due to the thermal goup corrections leakage effects.

anl_trans_1.dat (download) is the transient data file derived from anl_ini_1.dat and making use of the anl_f2c1.dat (download) initial f2ci distribution saved at last step.

anl_trans_1_ch1 compares the trans/ini initialized safpwr results with the reference. The residual deviation from the refenrence, and from the 2-group fitted results can be explained by the residual thermal group effect.
Of course, this effect is particularly sensible here because of the huge inter configuration reactiviy jump which is never encountered in practice.
This is highighted by plotting the refenrence sn0ciwhich show how sn0ci corrections must be introduced in the 1-goup model to account for the thermal group effect.