Page 2 of 4
These studies also capture the fuel cost impacts related to Ramp Cycling, within the accuracy of the current generation of production cost models. While we believe these studies are not yet able to capture some of the generation issues we identify later in this paper, we consider several of these studies to be valuable, reasonably accurate and representative of actual operating production cost impacts.
A portion of wind integration costs can be approximated using actual historical wind data to forecast hourly wind production.
The modeling begins with the assumption that we have “perfect knowledge” of the Balancing Area variability, levelized for each hour. Then a standard production cost model is used to evaluate the State Cycling attributable to the higher level of uncertainty when forecast wind is added. The cost difference between this case and a production cost simulation without any variable resources is used to estimate the best-case, lowest potential production costs.
Actual costs are higher than the approximated case described above for
Xcel Energy is focusing on cutting edge wind output forecasting improvements in an effort to hold Balancing Area integration costs as close as possible to the “perfect knowledge” impact level. The company has initiated a project with the National Center for Atmospheric Research (NCAR) in Boulder, Colo., in an attempt to have the most accurate wind output forecasts possible. The project involves supplying available generation resources and full meteorological data to NCAR, which processes the information in an advanced computing system. The company expects the use of this advanced software will help keep State Cycling costs to the lowest achievable levels.
Work on the NCAR project is underway with completion scheduled in 2010. The output from the project will be ‘tuned, site-specific’ forecasts with intra-hour updates. This high level of forecast resolution will bring a new level of detail available for Xcel Energy to manage wind integration impacts for its operating companies.
By the year 2020, with even higher levels of wind penetration, the magnitude of the hourly variability will increase. Table 3 shows our planning forecast for the PSCo system.
Figure 1 shows a typical day on the PSCo system. The top line is the load, the bottom line is the wind output and the center line shows the net of the wind output and load. From this illustration one can see the increased variability presented to the Balancing Area. The net line (native load minus wind) has a steeper slope than the load line; this means the Balancing Area must supply higher levels of ramping flexibility in order to maintain balance during the swings in output and incurs increased Ramp Cycling. The net line also shows greater separation between maximum and net minimum; this means the Balancing Area might have to increase State Cycling of dispatchable units in order to meet operating reliability criteria.
There is an aspect of the system issues that is not immediately apparent from Graph 1 above. The net variability can impact the Balancing Area much more significantly during the off-peak hours. In the days before wind generation, the off-peak hours were generally the tranquil hours for most Balancing Areas.