Checking for Equilibrium in Each Island

An a.c. load flow is performed for each island to check the equilibrium. There are three possible outcomes of each of power flow computation.

• The power flow converges and the resulting state of the system does not exhibit any violation of normal operating limits (i.e., thermal limit violations, or voltage limit violations). This contingency state therefore does not require any corrective action. However, if there are some loads that are not served due to outages of busses or tripping of line causing the de-energisation of busses and hence the disconnection of loads, then there will be a cost of security. Otherwise the cost of security is zero. • The power flow converges but there are some violations of normal operating

constraints. Corrective actions must be taken to bring the system back within acceptable limits. For example, generation may be re-dispatched, gas turbines may be started, or, as a last resort, loads may be shed to correct line overloads.

• The power flow diverges. This is taken to indicate that the occurrence of this contingency state would result in severe problems. In such situations, as an emergency measure, load shedding is carried out to prevent a complete voltage collapse in the system. Based on conversations with experienced operators, a heuristic technique has been developed to determine the ways and percentage of load that is typically dropped to restore the network equilibrium.

According to the heuristic approach, the load shedding is carried out in 5% block of the load in the zone where the bus with the largest mismatch is located. Load shedding is carried out until convergence is achieved. The situation is also deemed as an emergency. If the load shedding is carried out 20 times the zone is blacked out. Then the load disconnections extend to the neighbouring zones where they proceed again in blocks of 5% of the area load. If the convergence has not been achieved after 100 load-shedding steps, the system is deemed to have collapsed. 3.3.7 Operator Action

When the system has reached an equilibrium point (i.e., when convergence of the load flow has been achieved), the resulting state of the system may exhibit violations of normal operating limits. Corrective actions must be taken to bring the system back within acceptable limits. Corrective actions include re-dispatching generation, changing voltage set points, and tap ratios. Load is not shed to remove violations of operating limits if these violations can be corrected using normal controls, i.e., active and reactive despatch. A fuzzy expert system with embedded power flow and linear sensitivity analysis [6] determines the extent and location of these corrective actions.

There are three types of controls available in eliminating violations with the fuzzy expert system in VaSA. They are:

• Active dispatch

 Dispatches settings of active power generation, shedding of loads (active and reactive components in proportion) and changes to phase shifter settings in order to relieve overloads of transmission lines

• Reactive dispatch

 Dispatches settings of reactive control devices in order to correct violations of voltage limits

• Dispatch of active controls for correction of voltage problems

 This is activated to change the active generation and, if necessary, shed load in order to remove any outstanding violations of voltage limits. However, this type of control measure is very costly and is the last option among control actions.

In setting the dispatch in each island, the operator’s judgement is simulated by balancing the criteria of control effectiveness, control margin, cost, simplicity and the possibility of unwanted secondary effects.

At first the expert system checks the sensitivity of the control measure in eliminating or reducing the violations. If it is successful in eliminating or reducing the violations then the corresponding control measure is applied. Effectiveness of the control actions are always checked by applying a control action and then performing an a.c. load flow. 3.3.8 Modelling Time Dependent Phenomena

Cascading tripping of lines, sympathetic tripping and transient instability of generators are due to the results of time dependent phenomena. VaSA is capable of modelling these types of outages and the corresponding modelling strategies are described in the following sections.

Cascading tripping of lines is modelled in VaSA assuming that the overloaded lines are tripped in %r of the situations encountered [7] where the r is the probability that the operator is unable to eliminate the overloads before the protection operates [2].

Sympathetic tripping is modelled in VaSA using the probability of malfunction of the element’s protection system when a failure occurs in its vulnerability region. The vulnerability region defines as the region of the system where a fault may provoke the

tripping of the element. A Monte Carlo trial indicates whether a line is tripped by sympathy using the probability of sympathetic tripping.

Random sampling using the set of probabilities (i.e., probability of stability due to a fault on line k ) determines whether a fault provokes instability in the system. If a stability problem is simulated, it is necessary to determine the affected generators. Offline stability studies are used to determine the vulnerability region associated with the stability of each generator. Thus, if a fault on line k provokes instability and this line is in the vulnerability region of a generator, then this generator is disconnected. 3.3.9 Checking for New Equilibrium

On the basis of models described above, these additional time-dependent outages can also be simulated in a probabilistic manner. Then an ac load flow is performed and, if required, further loads are disconnected to achieve the new equilibrium (i.e., convergence of the power flow)