The project picture is the spice schematic.  The spice file is also loaded in the files section along with the required TRIAC model.

Concept:  The circuit switches between the primary (V1, utility) and alternate (V2) power sources.  The load, L+, is powered by the alternate whenever it is energized.  Both AC sources must be tied to a common (earth) return.

Operation:  identical negative capacitive supplies are used for both the primary & alternate control circuits.  The primary's supply consists of D5, C2, D3, C1, and R10.  Half-wave rectification & filtering is done by D3 & C2.  D5 is a zener shunt to regulate the supply voltage.  C1 is the voltage divider and R10 provides current limiting.

Both the primary & alternate TRIACs trigger in quadrants two & three.

Primary source:  when the alternate is off, the primary supply biases Q1 to sink current from the gate of U1 and the load is powered from the primary.

Alternate source:  when power is applied to the alternate, the opto-coupler, U2, is immediately energized and charges the gate of M4 which turns Q1 off.  TRIAC U1 will commutate off at the next current zero-crossing.  The time constant of C16 and R16 insert a turn-on/off delay.  Note:  U2 is required since p+/- & a+/- are operating at different potentials.

Simultaneously, C15 begins charging and Q3 turns on when its base is forward biased.  U4 commutates on.  The time constant of R7/R1/C15 must be sufficiently greater than R16/C16 to prevent both TRIACs from being on at the same time.

When alternate power is lost, R16/C16 determine the delay before the primary source resumes powering the load.

Conclusion:  the initial design is intended to switch loads less than 100W so no heat sinks are required.  Snubbers aren't included in the spice model but will be used in the built prototype although the intent is to switch capacitive (SMPS) or light inductive loads.

In simulation, depending on phase/amplitude/timing of the alternate source, you will see a large current spike when the primary TRIAC conducts in the first half-cycle.  I've played with this quite a bit and tried snubbers, adjusting phase of the two sources, etc., but I believe this is an anomaly.  I'll find out soon enough; prototype is next!