Like a lot of the circuits posted to the internet, it has a serious design flaw. How do I know? I recently made the same mistake as discussed on
this long running thread on another forum.
The circuit is just a shunt regulator, which prevents overcharging the battery after the battery voltage rises to ~14.2V. The LEDs are purely incidental to the operation of the charge regulator. It would clamp the voltage across the battery to the voltage determined by the R1/R2 voltage divider regardless of what is connected between the collector of T1 and ground. You could connect a power resistor, lamp, or nothing at all...
The glaring error that the author made is that even when the TL431 cathode pin switches to the high state (when the voltage at the Ref pin < 2.495V), the current into the cathode pin is still about 400uA, which is more than enough to partially turn on T1. The solution is to add a 1K resistor from the base of T1 to its emitter (Panel +).