With full-bridge, primary voltage is across 2 transistors at any time. There will be two on-resistances to primary current, and twice the heat loss than if one were used. This heat loss is lower for mosfets than for bipolar junction transistors (BJTs).
In the push-pull circuit, with a split primary, there can be a voltage across the whole primary equal to twice the primary applied volts. Each transistor has to handle twice the primary volts when switched off (voltage is induced in that winding which is switched off. This voltage adds to the primary volts). There will be one on-resistance to primary current at any time, and heat loss for one only.
Thanks for explanation,so what the factors which limits the maximum power can be obtained from a Full Bridge converter (i believe the maximum limit is about 3KVA) ?
This is a very old problem, which used to occur in RF valve circuits. Mosfets have a a very wide frequency response. The components in the circuit can form a number of resonant circuits, all different spurious frequencies (otherwise known as parasitics) which can be shock excited by the large circulating currents and amplified by the mosfets (during the rise-time and fall-time, when the mosfet can act as an amplifier). Also the ferrite cores, because of their size and the currents flowing, can have variable non-linear magnetic fields within the material. The current waveform can be distorted and generate harmonics, again creating parasitics. Amplified parasitics can generate high voltages and large local power.
I suspect there has been little research to find ways to clean up these parasitics. With RF valve circuits, the valves would continue to work, and you could find the source of the parasitics, and suppress them. Mosfets die extremely quickly, as the current is concentrated and sudden surges of heat cannot escape quickly, and they cannot withstand sudden large voltage increases. With valves you could have a momentary arc-over, and the valve would still continue to work.
If you want to build a LARGE power SMPS, operate at a low frequency, high voltage input with low current, and put in suppressors to limit the frequency response, with mosfets or BJTs having low frequency response. Unfortunately the efficiency will not be good, so you will have to dissipate a lot of heat in the transistors. I don't like your chances, but because I say this doesn't mean you shouldn't try it.
The ripple current on the main buss will be 1/2 and at twice the frequency so you save size and area on filter capacitor sizes (1/4 the size). The total area of transistor die will be the same for the same power whether you operate half bridge or full bridge. Half bridge runs at twice the current to that of a full bridge.
I ran 3kW per section at 130kHz using MOSFETs about 15 years ago. Using current technology 6kW per section is now possible at 200kHz or higher.
I used a 3 phase poly-phase power supply design for 18kW.