Figure 1

Wind farms are unique to power systems in that the construction and development time is much shorter than that of transmission lines and other bulk system facilities.  Wind farms can be placed into service well ahead of any planned upgrades, or even proposed non-wind power plants.  In these situations, the wind farms may be allowed to interconnect on a conditional basis or an energy basis; i.e., if congestion is present, they may be first to lose transmission access or have to share the available capacity with other generators, including other wind farms.  Hence, it is important to be able to estimate potential curtailment subject to transmission congestion.  In a previous article, we introduced the raw elements of the methodology for estimating curtailment of wind farms due to transmission congestion.  (See A Methodology for Estimating Potential Curtailment of Wind Farms, Pterra Tech Blog, September 2010).   We now look at the overall methodology applied for the purposes of making annual or seasonal projections of curtailment. 

Chronological Sequences

Loads and dispatch patterns vary on daily and seasonal cycle.   The transmission capacity available to a wind farm is a function of the same cycle.  Hence, when loads are high, transmission capacity is lower.  The transmission capacity is of course determined by security constraints, and typically the same flowgates (combinations of contingency and monitored transmission element) appear as limiting conditions for a specific wind farm.  In general, the transmission capacity available to a wind farm is constrained by a few flowgates; and at any one time, only one flowgate is the most limiting.  Based on this observation, if we had a projection of any combination of chronological system demand, energy interchange and import/export use of the transmission system, we can superimpose on this chronological sequence the available transmission capacity determined by a set of flowgates.

The sequences can be applied deterministically or stochastically.  In the deterministic application, the chonological projections of transmission use and wind availability are considered to be definitive resulting in a direct determination of curtailment for each sequential time period, say, an hour.  This results in projections of total and periodic wind energy curtailed, number of curtailments and magnitude of curtailments.  It further identifies periods, such as specific months or seasons of the year, when curtailments are high relative to other periods.

However, the deterministic approach does not capture highly uncertain events such as unplanned, prolonged outages of transmission lines, transformers and conditioning equipment, and generator outages.  Outage frequency for such events may be in the range from twice a year to once in a hundred years, far exceeding the capture rate of the typical chronological period of one year.  By modeling such low frequency events with a probability distribution, including failure bunching events such as storms and heat waves, we are able to include their effects in a Monte carlo simulation.  By replicating several years of operation taking into account random, uncertain events, we area able to determine probabilitic indices of curtailment such as expected wind energy curtailed, mean and standard deviation of MW curtailed and duration of curtailments.  The stochastic method improves upon the accuracy of the deterministic method.

Case Study Continued

In the previous Blog (see Reference 1) we presented the case study of a 350 MW wind farm.  Additional information on this wind farm: It has an annual capacity factor of 41% with maximum energy available during the months of January, March, April and November.  The available wind energy on an hourly basis projected over a one year period is shown in Figure 1, which also includes the matching chronological system load variation.  The load peaks during the summer months, and this coincides with the period when available transmission capacity is lowest.

Four cases are examined:

  • Case 1: Only the study wind farm is online.
  • Case 2: A second wind farm with a capacity of 160 MW is online and curtailments are shared between the two wind farms.
  • Case 3: Same as case 2 but with a gas turbine rated 80 MW online.  The GT has reserved transmission for its power and any curtailments are applied primarily to the two wind farms.
  • Case 4: Same as Case 1 but a transmission upgrade is delayed to the middle of the year.

Figure 2

The results of simulation are summarized in Figure 2.

  •  In case 1, the expected wind energy curtailed is 8.3% of the available wind energy with a curtailment probability of 24%.   In case 2, with a competing wind farm that has very similar wind supply characteristics, expected wind energy curtailed for the study wind farm increases to 24.2% with the average MW curtailed increasing from 49 to 75.4 MW.
  • In case 3, the gas turbine adds to the curtailment of the wind farm increasing the expected wind energy curtailed for the study wind farm to 26.2% annually.  The impact on curtailment of the wind farm is relatively small since the total wind curtailment during the summer months when the GT is expected to be dispatched is small compared to curtailment during fall and winter periods.
  • In case 4, the delayed transmission upgrade significantly impacts available transmission capacity for the wind farm.  For the year, the expected wind energy curtailed increases from 8% with the upgrade in place to 32% with the upgrade delayed, with average MW curtailed increasing from 49 to 95 MW.

The resulting indices for each of the four cases are shown in the following table below.

  •  The available wind energy peaks in the months of January, April, May and November and is lowest during the summer months of July and August.  In case 1, the expected curtailed wind energy is highest in November and January, and lowest in July and August.  This indicates that curtailments are less likely in the summer months and more likely during the fall and winter months.  In case 2, with a competing wind farm there is an increase in energy curtailed during the months of January, March, April and November.
  • In case 3, the gas turbine adds to the curtailment of the wind farms during the summer months when the GT is expected to be dispatched during peak hours.  In case 4, during the months that the upgrade is delayed, the average monthly wind energy curtailed is 50% of the study wind farm’s capacity.
Index Case 1 Case 2 Case 3 Case 4
Expected Wind Farm Energy Curtailed in GWh 103.8 302.5 327.2 401.3
Expected Wind Farm Energy Curtailed in Percent of Available Wind Energy 8.3 24.2 26.2 32.1
Probability of Wind Farm Curtailments in hrs/yr 2120 4009 4380 4222
Probability of Wind Farm Curtailments in Percent of Total Available Hours 24 46 50 48
Average MW Curtailed 49.0 75.4 74.7 95.0

The data setup and simulation time for the sample system are significantly less than those for a comparable production costing simulation.  Less data is required and computational demand is much less. 

A more rigorous description of the methodology is provided in a forthcoming technical paper.  Please subscribe to this Blog for the latest information.

References

  1. “A Methodology for Estimating Potential Curtailment of Wind Farms,” Pterra Tech Blog, September 2010
  2. “Wind Farm Integration: On the Use of Agreggate Models“, By J. Chen, M. Gutierrez, R. Austria, Pterra Tech Blog, Oct 14, 2009.
  3. “Wind Farm Integration: Analytical Requirements“, Pterra Tech Blog, Oct 26, 2009.
  4. “High Voltage Concern at Wind Farms?” Pterra Tech Blog, May 18, 2010.
  5. “On the Use of Aggregate Models of a Wind Farm,” J. Chen, M. Gutierrez, R. Austria, October 2009.
  6. “The Coincidence of Wind,” Pterra Consulting, May 2009.
  7. “The Renewables: Part 1 – Wind Farms,” May 2006.