This data base stems from a more recent and broader multi-contingency study screening 26 different disturbances for 5000 operating states of the Brittany system [2]. As above, the studied disturbance corresponds to the loss of 600MW of local generation. Its corresponds to 4041 relevant states, i.e. states among the 5000 where there is a least one such generator in operation. Two sets of pre-disturbance candidate attributes are used : (i) 136 basic ones, comprizing EHV voltages, active and reactive generations, reactive reserves, power flows, and various topological indicators; (ii) 138 attributes including in addition to the 136 previous ones, the pre-disturbance active and reactive load power margins.
In addition to a larger data base size, a major difference with the previous study concerns the level of modelling detail which is much higher here. The electrical static and dynamic models, the voltage security criterion and the load power margin computation procedure used are discussed in detail in [28]. We merely outline their salient characteristics to indicate their highly realistic character.
A 1250-bus modelling including subtransmission levels and distribution feeders is used while secondary voltage control is modelled both in the static and dynamic computations. As concerns the latter, a mid-term voltage stability simulation is carried out, as above, to track the on-load-tap-changers, automatic shunt compensation, machine overexcitation limiters and coordinated secondary voltage control behavior, subsequently to a disturbance.
Pre- and post-disturbance load power margins are computed by
simulating the system response if submitted to a steady load increase
until a voltage instability is reached. The latter is identified
according to the change in sign of the sensitivity of total reactive
power generation to individual HV reactive demands
. With respect to the previous
study, the numerical load power margin discretization error has been
significancly reduced (from 150MW to about 15MW).
Another feature of the data base is the very high diversity of operating conditions which it contains. To fix ideas, let us merely quote the independent variables used during the random sampling of the pre-disturbance states : topology (single or double line (400kV or 225kV) or transformer outages); load level (including randomization of individual HV load distribution, power factors, compensation and voltage sensitivities); regional unit commitment (with variable active power generation levels in some plants); reactive support (synchronous condenser; EHV and HV shunt compensation; gas turbines) and secondary voltage control set-points.