Matching Utility Loads with Solar and Wind Power in North Carolina
Open Access Document
Executive Summary
Matching Utility Loads with Solar and Wind Power in North Carolina
Dealing with Intermittent Electricity Sources
by John Blackburn, Ph.D.
March 2010
Those reluctant to endorse a widespread conversion to renewable energy sources in the U.S. frequently argue that the undeniably intermittent nature of solar and wind power make it difficult, if not impossible, to provide reliable power to meet variations in demand without substantial backup generation. Several studies, concentrating on areas with ample sources of both wind and solar power have suggested that a combination of the two, when spread over a sufficiently wide geographic area, could be used to overcome the inherent intermittency of each separately, reducing the need for backup generation. Moreover, since the backup power is required at more or less randomly distributed times, the availability of baseload power, so strongly entrenched in utility circles, becomes more or less irrelevant.
This study examines these ideas with data gathered in the state of North Carolina. Contrary to the idea that such an arrangement will be subject to heavy backup requirements from conventional sources, the clear conclusion of the study is that backup generation requirements are modest and not even necessarily in the form of baseload generation.
In North Carolina the two largest potential renewable electricity sources are solar and wind generation. The former is the case almost everywhere in the U. S., the latter is also the case in North Carolina, given wind resources in the mountains, along the coast, and offshore, both in the Sounds and in the ocean. Hydroelectricity (now 2,000 megawatts (MW) and potentially 2,500 MW) and biomass combustion represent the other renewable sources available in the State. Solar and wind generation have some obvious complementarities. Wind speeds are usually higher at night than in the daytime, and are higher in winter than in summer. Solar generation, on the other hand, takes place only in the daytime and is only half as strong in winter as in summertime.
The study described here used hourly North Carolina wind and solar data for the 123 days of the sample seasonal months of January, July, October, and April. This entailed making 2,952 observations at each of three wind sites and three solar sites or 17,712 entries in all. In the absence of actual kilowatt-hour output data for long periods from functioning installations in widely separated locations, wind speed and solar irradiation were taken at the three sites each and converted to presumed wind and solar power outputs. Wind data was converted using the specifications of the wind turbines chosen for the study, shown below. Actual power readings for shorter periods from solar installations at two sites (from readings made in different years), were used to calibrate the presumed solar output at the chosen sites.
The generation patterns given by these sites were, for this initial exploration, taken to be representative of all of the sites in North Carolina. Solar and wind power generation constructed as outlined above were then scaled up to represent 80% (40% each) of average utility loads for the four sample months, with the remainder coming from the hydroelectric system (8%) and assumed biomass cogeneration (12%). The annual utility load was taken to be 90 billion kWh, a somewhat more energy-efficient version of the present 125 billion kWh load. Average hourly loads in each of the four seasons were taken from Duke Energy’s 2006 load profile. These were modified to show some reduction in summer and winter peaks as structures become more energy-efficient and enjoy disproportionate reductions in heating and especially cooling energy demands. The reductions were based on the author’s data set of measured energy use in more than one hundred North Carolina homes.
Wind generation was calculated from wind speeds using the cut-in, cut-out speeds and power curve for the General Electric 1.5 MW turbine (model 1.5xle). Solar generation was taken to be proportional to solar radiation at a ground level flat surface. Not surprisingly, wind generation from the three wind sites combined showed less variability than at each site separately. Solar generation did likewise, but with less variation to begin with. The literature suggests that day-by-day and hour-by hour wind variation would be further reduced by adding many more sites far enough apart to have somewhat different hourly wind regimes.
North Carolina has several means of evening out differences between variable generation and load from hour to hour within days, but very limited ability to carry stored energy forward from day to day. The hydroelectric system is already used as a means to meet peak demands with a generation system heavily oriented toward baseload generation. In addition, there is pumped storage capacity in the Duke Energy system amounting to 2,100 MW, of which 1,360 MW has up to 24 hours of storage. In the summer, hourly storage is supplemented by the capacity of some large commercial customers to make ice in off-peak times and then run air conditioning systems without running the chillers at peak times in the afternoon and evening. In addition, the two largest utilities now have some 2,000 MW of load control arrangements.
As smart grids are developed, some customers will be able to respond to real-time pricing, offering still more opportunities to shift loads during the day. Still other storage opportunities may arise when plug-in hybrid vehicles are in use and have two-way communications with grid operators. With these possibilities in view, days and hours were examined in the data set in order to determine how many days and hours would need auxiliary generation, either by purchase from other systems or by (probably gas turbine) back-up generation within the system.
As the day totals in each of the four sample months were examined, it was apparent that the sum of solar and wind generation, day by day in each month were approximately normally distributed, with standard deviations running about one-fourth of the mean. In January, for example, mean daily generation for the month was about equal to the 80% specified above. Daily total power generation for the sample month of January as well as the hourly power generation for a sample day in January are shown here. Larger versions of these charts as well as charts for other sample months and another sample day are shown in the main text. Day totals varied, with about half the days showing above average generation and half below average. Within the below-average days, two-thirds were below average by a quarter or less of the mean. Only very rarely was the shortfall more than half the mean. Some days with above-average wind and solar generation still had hours when supplemental generation would be needed, but not often.
When all the days and all of the hours were considered, it appeared that auxiliary generation amounting to 6% of the annual total generation would be sufficient to fill in nearly all of the gaps between hourly renewable generation and hourly utility loads. The backup generation amounted to purchases from other systems up to 5% of hourly loads, and 2,700 MW of gas-fired capacity. There were 17 hours in the four months considered when still more backup power would be needed, or a loss-of-load probability of .0058
The out-of-system purchases or back-up generation in the system dropped the wind-solar contribution to 78% of the load. These results are shown in Table 1 of the main text (online at www.ieer.org/reports/NC-Wind-Solar.pdf. )
The important conclusion is that intermittent solar and wind energy, especially when generated at dispersed sites and coupled with storage and demand-shifting capacities of a system like North Carolina’s, can generate very large portions of total electricity output with rather minimal auxiliary backup.
Printer-friendly format
Email this thread to a friend
Bookmark this thread
(61 posts)
Send PM |
Profile |
Ignore
