Académique Documents
Professionnel Documents
Culture Documents
BC Hydro has identified Joint Var Control (JVC) as the chepaest means to extend
transmission capability limits to meet load forecasts. JVC controls generators such that a
common point voltage is maintained while coordinating machine var output in a multi-
machine plant. It does not involve installation of additional var capacity.
In this article, some of BC Hydro's experiences in designing the JVC scheme for its Mica
generation station are presented.
Another thing to consider when setting the LDC is to allow the generating units to be as
responsive to system conditions as possible. This lets each unit to contribute as much VAR as
possible to the system in the case of an outage; hence the higher the LDC setting, the more
output VAR becomes available to aid the system when it is most needed.
A compromise between these conflicting objectives leads to a solution that ranges between
the terminal and the high side of the unit transformer. BC Hydro has determined that an LDC
setting of 60 per cent of the step-up transformer impedance allows all machines to control
their own output voltages.
The setting is also dependent on how accurately the voltage regulator is able to control
voltage. A voltage regulator that drifts or oscillates could create large VAR changes even at
relatively low LDC settings. The 60 per cent trade-off in setting allows the local operating
constraint to be satisfied but limits generator participation in system-wide events such as
voltage instability.
Low side JVC adjusts the voltage setpoint of all units and as such does not require any
additional telemetered inputs beyond what is already provided for in each unit's voltage
regulator. No actual regulation to a common point is provided. By simulating a common
voltage point using LDC for the lead unit's regulator, coordinated plant control beyond the
limitation of single unit LDC is theoretically possible. This form of low side JVC is not used
by B.C. Hydro. High side JVC requires a voltage signal from the point where coordinated
plant voltage control is needed and hence the common point voltage can be physically
controlled, providing tighter voltage regulation than can be achieved with low side JVC.
Assuming all the combined machine and transformer impedance of unit i at a n-unit plant is
denoted by Xti , then the equivalent impedance seen by a machine looking out from it's
internal voltage is given by:
where Xeqi is the common system impedance beyond the transformer high side to the point
where JVC intends on controlling the voltage .
The resulting feedback voltage at VB is then the contribution from both the plant generating
units and the infinite voltage which can be derived from the high side voltage:
The approximate total VAR demand from all plant units is just the sum of all the droop
approximations or
Comparing the output VAR demand with the schedule needed to maintain the common point
voltage gives a VAR feedback error. This error is then added to the VAR schedule and
distributed amongst all units on joint var control.
The units may not be identical in design; it is entirely possible for one generator to reach its
VAR limits before the others exhaust their reserves. There are several ways in which the
reactive power can be dispatched to each of the plant units: apportioning vas such that all
units reach their limits simultaneously dividing the total VAR schedule equally amongst all
units scheduling according to predetermined rules.
In the case of the Mica plant, all units are of the same design and rating and thus the second
allocation rule the simplest was used. Varying allocations may be necessary for derated units
or to prevent some units from operating too close to their thermal limits. If one or more units
reach their limits before the others, the design will automatically reapportion the remaining
VAR to the units with spare capacity. Figure 2 shows a 4 machine high side JVC
configuration with transformers having equal impedance which allocates VARs equally
amongst the units (ie n=4 and Xti =Xt).
The first concern was that if the JVC response was very rapid, it would have the tendency to
cancel out PSS action. Time simulations and eigenanalysis later proved that this interaction
did not manifest itself at the proposed JVC response time.
The second concern was related to the power setpoint ramp ups which result in temporary
VAR disturbances as a result of PSS action. These VAR disturbances, it was feared, would
cause undesirable JVC action. Field tests later proved that judicious choice of JVC gain and
deadband eliminated this problem.
This only causes problems if the generator's operating point is in region A as shown in figure
4.
Initially it was feared that contingencies resulting in operation beyond the OEL would cause
the AGC to ramp the generator power setpoint down significantly or even to zero.
Simulations of this event showed that AGC action causing real power output reductions
would also lower the reactive output power by a similar proportion. Thus the trajectory of the
operating point following a contingency (e.g. line outage) resulting from AGC power
reduction would angle downwards from point 2 to point 3 of figure 4.
It was also determined, that there was only one single 500kV line outage which would cause a
plant's output to rise above the rotor limit for the heavy winter base case. This single
exception was already being handled by an operating order which would trip one unit off-line
effectively rectifying the situation. The rarity of this situation and the effect of AGC on unit
reactive output effectively allayed all concerns that plant operation under these exceptional
conditions on JVC would behave rationally.
Ideally no VAR output from all plants would give the maximum protection from JVC under
voltage collapse scenarios.
It was also apparent that plants which are in or adjacent to the main load center do not require
JVC for voltage collapse prevention. Under collapse conditions these machines will output
the maximum amount of reactive power almost regardless of the voltage setpoints of the
units. Moreover, the time simulations demonstrated that this reactive support is highly
responsive to system conditions and is quickly extracted because of the voltage droop
experienced by each of the plant's units.
The time frame of a voltage collapse scenario establishes the primary design criteria that the
JVC must meet. The final design target established from simulations for the BC Hydro JVC
was that the voltage setpoint should be re-established within 15 to 25 seconds for a 3%
change in system voltage. The main constraint to this requirement is the mechanical
potentiometer based voltage setters used by the generator regulators. This mechanical setpoint
control is common to the regulator of most B.C. Hydro machines and it's ramp rate varies
between plants but is generally capable of rates of 30 per cent per minute.
Studies demonstrated the equivalent amount of VAR support that JVC provided over the base
case. Figure 5 shows the interpolation of a series of time simulations giving the equivalent
amount of Mvar capability that JVC represents for two severe contingencies.
According to the figure JVC provides about an equivalent of 75 Mvars and 50 Mvars of
voltage support to the system at the load center (Ingledow Substation) for a 5L42 and 5L82
outage respectively. This is the difference in the amount of extra shunt capacitor support
needed between similar cases with and without JVC to halt voltage collapse. The power flow
base case used for these simulations did not have minimal VAR output from all plants - when
most plant output was reduced as much as possible, the JVC was equivalent to about 180
Mvars at the load center.
Controlling voltage beyond the plant's substation somewhere into the power system (over-
reaching control) makes the system even more responsive to voltage disturbances. If the over-
reaching control is set to a point beyond a network branch forking into two or more separable
and distinct electrical paths then no distinct physical control point will be recognizable;
consequently this type of control is a coordinated form of LDC making a plant look like a
single unified VAR source; unless a remote telemetered voltage from a load center is used as
the JVC controlling bus voltage input.
Over-reaching control of this type is antithethical to the design of the power system. Voltage
profiles for a given base voltage level are meant to be flat or have a minimum amount of
voltage droop. It is the change in voltage angles that make for real power transfers across the
system. Transporting VAR across the system works on the basis of a discernible transmission
voltage gradient: the higher the gradient the higher the plant VAR demand. The over-riding
physical constraint on generator support is of course the amount of VAR that a plant can
supply.
Transmission equipment BIL limits will generally not be a concern during heavy loading
conditions that characterize voltage collapse. Another constraint on over-reaching voltage
control is for the case of two plants within short electrical distance of each other; the B.C.
Hydro GMS and Peace Canyon plants qualify in this regard. Two such plants both equipped
with JVC can miscoordinate VAR allocation in much the same way as two units under LDC
control with high compensation settings This case can be dealt with by placing both plants on
the same JVC system and this would require telemetry between both plants.
Results of a study attempting to quantify the increase in VAR support using over-reaching
control is shown in table 1. Each of the plants were independently placed on over-reaching
control. There are two lines in the table for each plant one each for two stressed system
planning base cases. The amount of impedance that is being over-reached is given as a
percentage of the equivalent total line length. The amount of increase in the system MW
transmission capability for over-reaching control is compared to 100 per cent LDC (or high
side JVC). There is a significant increase in the system's power transfer capability under this
type of control from 144 to 254 percent of what high side JVC can provide. Note that the
plants are generally constrained by the terminal voltage limit of 1.1 per unit.
Over-reaching control if implemented may mean the difference between system voltage
stability and its collapse; if not it may provide a short time period for the EMS or system
operators to assess the situation and act accordingly.
Conclusions
The support studies for preceding the Mica JVC design and implementation demonstrated the
feasibility of using high side JVC without any adverse generator controller or AGC
interactions. A measure of the amount of equivalent VAR support that JVC provides has also
been calculated; moreover, time simulations have also demonstrated that JVC can delay
voltage collapse or even prevent it. Operating experience to date on the Mica JVC system has
shown it to be reliable without creating any adverse affects on installed equipment. ET
C. Fuchs, D.Apps, and N. Chopra are with B.C.Hydro. W. Xu is with the University of
Edmonton. This paper was originally presented at CEA's Electricity'97 Conference held in
Vancouver last April.