The Renewables: Part 1 – Wind Farms

The renewables are here!  Whereas, power plants using renewable
energy sources were not too long ago considered exotic, today they are
the new face of energy — the wind mill replacing the smokestack as the
symbol of electric power generation.  Spurred by governmental
incentives, renewable energy sources are rapidly changing the nature and
composition of power systems.  They are still a fraction of the
overall energy portfolio, but the renewables’ level of penetration of
energy markets is growing.  In most US RTOs and power pools, the
queue for interconnection projects is dominated by renewables, primarily
wind farms.

Our
particular interest in renewables is their unique technical
interconnection and operations impacts on existing power systems.
Also, we are concerned about the long-term impact from the
reliability planning
perspective of integrating the system with high penetration
levels of renewables.  Whereas renewables in practical systems were
primarily hydro-based and geothermal, the new renewables are making
their own statements.  These include wind farms (both onshore and
offshore), solar and
biomass.

For this first part, we will focus on wind farms.

What is unique about wind farms?

The output from a wind farm depends primarily on the
prevailing wind speed. The analogy to a coal power plant, for example,
would be that the coal supply would vary on a minute-to-minute basis.
This characteristic leads to certain features of wind farms that
challenge the typical network integration methodologies.

  • Control function needs to take into account the
    wind speed and direction and the dynamics of how wind is converted into
    electricity.  For network analysis methods such as power flow and
    stability simulations, there are simplifying assumptions that are
    applied.  But even then,
    specialized models
    are required.  Existing simulation software
    may not have the structure and solution methodology to address wind
    farms adequately.

  • Output Unpredictability: The dependable or
    available capacity of wind farms has a broader probability distribution
    than traditional types of power plants.  However, large-scale wind
    farms are easier to predict, using global behavior, than small-scale
    farms.

  • Variability: Time-varying wind speeds presents a
    challenge to planners and operators alike.  In one extreme, the
    operator can simply accept what power is available and balance the
    energy portfolio with other types of resources.  On the other hand,
    there may be a need for firmer supply.

  • Voltage Control.  Wind generation generally
    has limited reactive power control, and marginal voltage ride-through.
    One can liken wind farms to the end of a whip.  When a transient
    disturbance rolls into the power system, the wind farms see the worst of
    it.  Voltage control is thus provided by additional equipment such
    as capacitors and SVCs or via a strong grid interconnection.
    Offshore wind farms have the special characteristics of voltage control
    on a radial feed, where the feeder tends to be submarine with high
    charging capacitance.  This issue has become significant that low
    voltage ride-through (LVRT) capability is proposed by FERC as a
    requirement for future incoming new wind farms.

System Impact

The first aspect of integrating wind farms into existing networks is
to assess the system impact and to ensure that reliability criteria
continue to be met.  Among the technical aspects to consider are:

  • In the steady-state, the wind farm is tested for any impacts on
    power flow and voltage during normal operations and under credible
    contingencies.  This analysis may include any impacts on
    thermally or voltage limited interfaces.  Mitigation of impacts
    may include transmission reinforcement, capacitor addition, remedial
    action schemes and other transmission planning solutions.
  • In the short circuit, the wind farm should demonstrate no impact
    on fault levels.  the typical impact is on circuit breakers
    that exceed rating.  Though nominally, the incremental fault
    current is small, in certain areas, this may be sufficient to result
    in breaker overduty.
  • In
    stability analysis, the wind farm should not introduce
    instability
    to the existing grid.  Further, the wind farm,
    in most regions, is required to remain online for faults that
    do not directly isolate the farm.  The typical wind generator
    is light enough not to impact most stability conditions.  But
    the latter condition, requiring LVRT, is more challenging.
  • If mitigation includes switched capacitors, the size should not
    exceed allowable delta-V at the load and transmission buses.
  • The operations of the wind turbines produce both voltage flicker
    and harmonic distortion.  These can be measured and compared
    against accepted standards such as IEC-41000 or IEEE 519.
  • The wind turbines also introduce excitation currents that may
    excite resonant frequencies in steam generators.  This may
    be accentuated if there is another source of non-fundamental frequency
    close by, such as a DC controller, SVC or series compensation.
    Torsional analysis may be needed to demonstrate potential
    impacts from torsional stresses.

Modeling Issues

Wind
farms pose a significant modeling challenge to power system
analysts.  They do not behave like normal large-scale power
plants, in many aspects.  Even simulation software are affected.
Often, in conducting detailed assessments, the analyst has to check
whether any seeming impacts are due to the predicted physical
response or to modeling error.  For power flow analysis, the issue
may be indicated by a non-converged solution, in which physical reality might
be that voltage is too low at the wind generators’ terminals (or there
is not enough reactive power) that the solution algorithm fails to
converge, or computational reality may be that voltage
sensitivity to the simulated reactive draw from the wind generators is
causing solution divergence.  For stability simulations, an
analyst may see a wind generator tripping off due to low voltage
or frequency excursion (the rotor angle swing is not a stability
indicator for an asynchronous generator), or the dynamic model may be
simulating a false voltage or frequency dip.   The
system may be stable when the simulation says it is not.  These are
just some of the issues.

Power Flow Modeling

For transmission studies, a detailed representation of the wind farm
is not required.  Typically, the farm is modeled in hybrid fashion,
representing the main collector substations, step-up transformation and
the transmission lines that interconnect these, but aggregating the
individual wind turbines into single composite machines at collector
buses.  For power flow-based analysis, each composite wind turbine
at a collector bus is represented by the combined real power generation.
Reactive power is modeled depending on the control strategy.
Often, if capacitors are included in the design, these are netted with
the composite reactive power of the wind turbine themselves.  In
certain situations, it may be sufficient to model the composite wind
generator as a load with negative power.  (This assumption impacts
the way the wind farm would be modeled for dynamic simulation.)

For certain line out contingencies, the voltage at the collector
buses may drop significantly below 0.9 p.u.  Power flow solution
algorithms may diverge in these cases.  Among techniques to confirm
that the issue is a physical rather than a numerical issue are:

  • Adding a fictitious synchronous condenser
  • Changing solution method – a full Newton solution that
    recalculates the Jacobian every iteration would work best in this
    case
  • Changing the solution acceleration factor (in PSLF, the reactive
    solution factor may also be changed)
  • Changing location of the swing bus
Modeling for Dynamic Simulation

An important aspect of WTGs is that these are asynchronous machines.
Conventional aspects of generator performance related to internal angle,
excitation voltage, and synchronism are thus not applicable to WTGs.  Hence,
rotor swing curves and damping simulations, used to determine stability,
cannot be monitored from the WTG.  However, the WTGs’ impact on
stability will show in the swing curves and damping of conventional
synchronous machines – a form of stability by induction.  (This
begs the question — what is system stability when a system
becomes predominantly asynchronous?  There’s a good topic for a
future Techblog!)

Typical dynamic simulation modeling for wind turbine generators (WTG)
include:

  • generator model – typically an induction machine, asynchronous
  • excitation model – representing the method for field control, in
    those designs that have this feature
  • pitch control model – for newer WTGs, this control provides for
    adjustment to high wind speeds
  • wind turbine model – to capture the conversion of wind energy to
    electric energy
  • generator protection – represents the relay package for
    under/over frequency and under/over voltage protection, among others
  • low voltage ride through (LVRT) capability – represents
    the added capability to ride through low voltage conditions normally
    seen during fault conditions on the network
  • wind gust model – representing response to sudden changes in
    wind

There is a wide variety of designs, features and manufacturers, and
the process of selecting appropriate models for simulation can be
challenging.  Some of the features may be combined into one or more
models.  Earlier versions may contain bugs that are addressed in
later versions. Furthermore, there are a variety of model developers
that supply closed form models (see “Open Source or Proprietary Data:
the Model Dilemma
,” from the November 2005 TechBlog).


The
safest bet is to contact the manufacturer for the latest model for
simulation.  Otherwise, you can refer to Pterra’s summary of
wind
farm models by clicking on the pdf link on the right side of
this paragraph.  Also, utilities have started to qualify models for
acceptance in system simulations.  See for instance
reference [1]
.  A comparison of simulated response from
different software packages is given in reference [2].

Several industry groups are busy defining and developing
models that would consistently represent wind turbine generators (WTG), including the WECC WTG and AWEA among others.  One of these would eventually
become a standard.  At present, each manufacturer of wind turbines
supports development for their individual equipment on various software
packages.


Not
a regular induction machine.
  Although WTGs are primarily induction generators, using a generic
induction generator simulation model leads to pessimistic results.
The field and turbine controls significantly alter the dynamic
performance of the wind turbine, and have to be taken into account.
At right is a comparison between a WTG model (red) and a conventional induction machine
(green).  The conventional
machine has identical parameters to the WTG, but without
the field and turbine controls.  The figure
shows the response to a fault at the machine terminals.
[3]  The power swing is more severe and
the voltage recovery is poorer in the plain induction generator
model.

Four Basic Types

Four basic types of WTG are:

  • Fixed-speed” induction generator (FSIG) – Simplest
    design.  Noise level and blade deflection can be
    significant in high wind. Mainly used for smaller turbines (< 3 MW).
    Examples include NEG-Micon,
    Bonus CombiStall (used in King Mountain, Texas), Mitsubishi.
  • Wound rotor induction generator with controlled rotor resistance
    or “variable” slip systems; examples include Vestas V47 and V80
    with OptiSlip.
  • Doubly-fed
    variable speed induction generator (DFIG) –
    Voltage regulation
    similar to that of a synchronous machine but with faster response.
    AC excitation for the generator is supplied through an ac-dc-ac
    converter.  Converter synthesizes an internal voltage behind a
    transformer reactance.  The WTG’s rotor and stator windings are
    primary and secondary windings of the transformer.   Examples include GE 1.5 or
    3.6 MW, Vestas V90, Gamesa.
  • Full conversion variable speed induction generator (VSIG);
    Low noise and reduced blade deflection in high wind.  Examples
    include Bonus CombiPitch, Enercon, GE 2.x.

There are also differences in the method of reactive control, average
wind speed, offshore on onshore, etc. which have some impact on modeling
requirements.  As the industry tries to make sense of the
technologies for wind power generation, new technologies are introduced.

Planning Issues

As we look farther into the future, it is necessary to consider the
possibility of even higher penetrations of wind farms.  How
would power systems handle the special operating nature of wind farms?
Also, there is the possibility that wind turbines’
unique operating characteristics might influence reliability criteria,
especially with respect to acceptable voltage and
stability performance.  The LVRT issue is an example of
performance criteria being introduced as regulation.  The
next issue might well be stability, and the manner in which we simulate
and determine it given a large number of asynchronous generation.

Furthermore,
although wind farms may be treated as peak load capacity for resource
assessment, they are more likely to be near rated capacity during
offpeak
.  This may result in a different method for treating
dispatch
for planning studies.

We need to do a little more planning if we are to make  some
headway with these issues.  As we have often said in
these tech blogs, we cannot ignore planning.
Without planning, we are subject to surprises.  The end
is not simply to develop a plan that may eventually not work out right,
but to develop an understanding of the issues through the application of
the planning process.  “A plan is nothing, but planning is
everything
!”

For questions, comments and further discussion, contact us at
mailto:info@pterra.us

References


  1. Wind Modelling Update
    , ESB National Grid, May 2006.

  2. Sensitivity Analysis on Low-Voltage Ride-Through Requirements
    ,
    ABB, 2004.  A memo comparing simulation results using PSSE and
    PSLF.
  3. PSLF Users’ Manual, March 2003.
  4. Piwko, R.; Miller, N.; Sanchez-Gasca, J.; Xiaoming Yuan,
    Renchang Dai; Lyons, J., “Integrating Large Wind
    Farms into Weak Power Grids with Long Transmission Lines,”

    Transmission and Distribution Conference and Exhibition: Asia and
    Pacific, 2005 IEEE/PES

    15-18 Aug. 2005