Network integration of wind power - Wind Power Integration: Connection and System Operational Aspects - page 105

In summary, there is a sending-end voltage that rises and falls depending on
the LDC perception of load, and a receiving-end voltage that increases or decreases
depending on load level and on how strongly the wind is blowing. How can the
receiving-end customers be protected from widely fluctuating voltages? In the
absence of IT-based solutions, utilities install lower impedance circuits to reduce
the range of voltage effect. They do not pretend that this is efficient, just expedient
in fulfilling their licence conditions.
Clearly a controller incorporating a load model for the line, and which has
knowledge of the receiving-end voltage and wind farm infeed, can estimate the
sending-end voltage. However, since the load varies widely and randomly, this is
dangerous. To be reliable, the controller should also know the sending-end voltage
or, as a minimum, receive a signal of overall system load as a percentage of full
load. Even the latter is somewhat unsatisfactory, as demand in tourist areas may
peak at lunch-time, counter to the overall system demand pattern.
The objective of the controller would be to keep the voltage at the wind farm
within the statutory limits, by adjusting the sending-end voltage and any reactive
power available at the wind farm. If the wind farm is not at the end of the line the
voltage will fall between the wind farm and more distant customers.
To maximise wind power generation potential, a voltage controller should first
control reactive power, and only when this fails to achieve a satisfactory voltage
level should active power be reduced, with the circuit reverting finally to a single-
ended grid supply as originally designed.
A problem exists when more than one embedded generator is connected to a
single circuit: the controller described above would allocate line capacity to the wind
farm nearest the source. A wide-area controller would then be required, incorporat-
ing a function to optimise physical capacity while respecting commercial rules.
If several circuits at one node have embedded generation then the control may
need to be wider again and take into account the source transformer limits.
The above is technically feasible, but the cost and reliability of communication
and the establishment of agreed standards/protocols should not be underestimated.
Embedding local network models in voltage controllers implies updating those
models as connected load/generation changes. This may be resource intensive work
unless it can be automated. Software and controllers may need to be changed
several times within the life of a wind farm. Constraining wind generation to match
network capacity would need to be a carefully quantified risk if wind farm eco-
nomics are not to be undermined.
Clearly a local control system such as that described above has the potential
to avail of demand-side management for network, as opposed to overall system,
objectives.
In-line voltage regulators can be used. These are controlled, close ratio,
variable tap transformers, which in the experiments are positioned at the electrical
load centre of the circuit. When the wind farm is outputting strongly, the voltage
on the wind farm side is higher than on the grid side. This triggers a change in
the tap ratio of the voltage regulator. Figure 4.8 shows the concept, but note that the
sending-end voltage is here set to 1.0 pu, which may pose a problem for other