You are asking if real power can be interconverted to reactive power. Reactive power is power flow through a pure reactance, so your generator must contain either such a reactance of a means of simulating one, and this might mean the generator has to be bidirectional,  so it can absorb power during one quarter cycle then return it in the next quarter cycle, then do the same with reversed voltage on the next half cycle. It makes the design more difficult.

As 'ideal devices' do not exist: the formal answer has to be: NO. (Some 'parasitic' active power loss will always occur.)

"(Power) distribution substation" seems to be an inadequate term. I know that in the past - to compensator for high inductive loads - capacitor 'banks' could be switched parallel to the grid to improve the 'power factor'. Not sure whether this technique is still in use now as 'active PFC' is becoming ubiquitous - greatly reducing the effect of inductive load vs. the grid.

You could as well think of "inductor banks", though I never heard of. Maybe due to 2 reasons:

purely capacitive loads were less frequent   and

inductor banks would be quite large and have significant losses as inductive storage is always associated with high currents - creating high I2R losses. Simply impractical.

'Generators' (as the physical device) will only act temporarily as reactive loads during synchronization (snchronizing to the grid). And - as Tony Maine already stated - purely reactive 'generators' (thinking of inverters with significant short-term power storage capacities) are possible but impractical as well.

The trend is to get renewable energy substations (photovoltaic, wind energy and others having already powerful control electronics) 'matured' to supply some out-of-phase current - effectively combining active power feed with reactive power feed - to improve grid supply quality.

For supplying and absorbing reactive power in power system, devices called inductive reactor and capacitor banks are used, as active power loss is very small could be neglected compare to supplied or absorbed reactive power, yet the active power(W) cannot be zero. Synchronous converter could supply and absorb reactive power from system, without much, active power consumption.....

No not possible for any case some active power will absorb or inject. We can't see the reactive power alone even for the capacitive bank or reactors ......

Thank you very much for all answers.

Your answers are so helpful.

Best regards

Ideally it may be possible but the transformers itself in the substation need active power for winding ohmic losses .further pure inductive loads are not possible completely.what can be done is active power can be reduced to a large extent but it can't be eliminated completely 

Your question also made me wonder why you would need to do this. Some loads, especially large motors, tend to have lagging power factors because of large inductances, so if your system contains many of these then you 'tune out' the inductance with capacitance banks - but at the motor, not elsewhere in the system. It should be made very costly for users to draw power factors much different from unity then the situation won't arise that your generators have to handle such situations!! The higher tariffs would also pay for your special generator!

Dear Dr. Maine,

In fact, I examine a  test distribution system with a high penetration of renewable DGs.

In my tets system, DGs can only inject active power to the distribution system. If generated energy by DGs is enough for supplying system demand along with losses, the active power of distribution system substation will be zero. However, reactive power may not be zero. I implement an optimal power flow and obtain such results. Therefore, since this case is not practical (according to the answers), I will modify my OPF model.

Thank you.

Best regards

yes! in some cases it is possible

Dear Prof. Haghifam,

Thank you very much for the answer.

Your answer is so helpful.

Best regards

Reactive 'power' isn't really power in the sense that it does work. It is the transfer of energy between the network and a load at twice the supply frequency. While a conventional generator can supply power during the quarter cycle of the supply, it has to absorb power during the very next quarter cycle. The only way a real power generator can do that is if it is either truly bidirectional or it has an energy store. Like a reactance or a battery.