Patent Grant
10439401
Electric utilities need reliable sources of energy. Because generation must match electrical load, which varies throughout the day, they want generation resources that can be turned on or off, or up or down anytime, like a water faucet. They call such sources of energy dispatchable.
They consider a hydro-electric plant to be dispatchable because the river is always flowing. They consider a natural gas ‘peaker’ plant to be dispatchable because it can be fired up at any time.
The conventional view of photovoltaic (PV) solar arrays is that they are not dispatchable, because the sun is not always shining on them. Solar is an intermittent energy source, which is not what utility operators want when they need more energy immediately.
The benefits of utilizing solar power include lower fuel costs, lower carbon emissions, and other benefits associated with more distributed, less centralized sources of generation. While many utilities acknowledge these benefits, the inventors have recognized that the intermittent quality of solar generation limits their ability to maximize these benefits.
The inventors have noted that utilities often proceed from a sense of obligation to use every Watt of solar output that is available, not wanting to leave any solar energy unused. The inventors have recognized that this approach tends to exacerbate the very characteristic of solar that utilities find most troubling—its volatility. The inventors have recognized that by carefully discarding some of the solar energy that is available, the solar energy that remains can much more reliably be used—and cost-effectively provide greater overall benefit to the power system.
Moreover, dispatchability, as a characteristic of energy resources, is not binary. All resources vary in how reliably they are available, or can be controlled—and even those traditionally considered to be dispatchable are not absolutely so. All resources have limits, and equipment can break down at any time. Resources like hydro and natural gas are considered dispatchable—not because they are without limit or can never break down—but because, over time, their limitations have become understood, accepted, and managed.
In view of the foregoing, the inventors have conceived and reduced to practice a software and hardware facility for understanding, accepting, and managing the limitations of a PV solar array (“the facility”). The facility enables solar to be considered a significantly more dispatchable source of energy by utilities than it currently is, making solar energy workably dispatchable.
By realistically determining the available capacity of solar output, the facility makes that capacity more useful and reliable to the electrical grid and its operators. The more the natural limitations of solar output are understood, accepted, and managed, the more confidently the portion of solar output that can be utilized will be utilized.
In some embodiments, the facility uses “sentinel” PV panels placed amid or adjacent to the bulk of the solar array as a basis for assessing the maximum available output of the entire array from moment to moment.
In some embodiments, the facility calculates a usable range of solar output that is below the maximum available output.
In some embodiments, in factoring into these calculations the interruption of sunlight falling on the panels of an array by clouds or other intervening objects (“solar blockage”), the facility does not attempt to predict whether or not blockage will occur. Rather, it calculates a probability of blockage based on recent and current solar data.
In some embodiments, the facility employs user-settable parameters informed by such factors as local climatological and seasonal patterns, the relative efficiency of local PV equipment, and the desire to decrease the volatility of solar energy output.
In some embodiments, the facility enables a PV solar array to be paired with any electrical grid control scheme that issues commands for greater or lesser active power to regulate other system parameters.
Such grid control schemes typically include:
First, Automatic Generation Control (AGC), a system used to control frequency within a Grid Balancing Authority. Fluctuations in AC frequency on the grid can indicate whether present power supplies are meeting present power demands. An AGC system typically responds to such fluctuations by sending commands to dispatchable sources of energy to increase or decrease their real energy output.
Second, any other source of a Frequency Response signal that responds to fluctuations in grid frequency by sending commands to dispatchable sources of energy to increase or decrease their real energy output.
Third, a control scheme (such as Volt/Watt) that responds to fluctuations in grid voltage by sending commands to dispatchable sources of energy to increase or decrease their real energy output.
Fourth, a control scheme (such as Power Smoothing) that responds to fluctuations in system load or generation by sending commands to dispatchable sources of energy to increase or decrease their real energy output.
Fifth and finally, in some embodiments, the facility determines locally how to vary the output of the PV array within the output range determined by the facility, such as by observing the frequency of the grid at the array and applying a “droop curve”-a function relating frequency to power.
The facility makes it practical for a PV solar array to be among the sources of energy called upon by any of the above grid control schemes.
By operating in some or all of the ways discussed above, the facility transforms solar energy from a source of volatility into a solution to volatility—whether that volatility is in frequency, voltage, or real power—and whether that volatility comes from some other source, or from the PV array itself.
The inventors believe that in many situations this facility is a more cost effective way to manage grid volatility than many of the resources used today to do so, such as traditional ancillary services. Although this facility may result in a reduction of the overall energy utilized from an individual solar array, it can result in overall economic benefit to the grid by displacing a more expensive means of managing grid volatility.
The small “sentinel” PV array 110 sends its energy to a solar inverter 115, whose output is not controlled by any grid control scheme or algorithm. The larger PV array 160 sends its power to a separate inverter 170, whose output is controlled by the responsive solar algorithm 135 that is at the heart of the facility.
The responsive solar algorithm 135 is a software program run on a computer 130. The algorithm gets information from the uncontrolled solar inverter 115, and sends commands to the controlled solar inverter 170. The algorithm can also send information to and receive commands from an external computer 140 running an electrical grid control scheme 145 such as AGC or any of several other grid control schemes that need to control real power from the PV array to condition other grid characteristics, such as voltage or frequency.
The external electrical grid control scheme 145, in turn, can get information from the electrical grid 150 about the grid characteristics that it is assigned to condition.
In some embodiments, the electrical grid control scheme 145 gets all the information it needs to issue commands to the responsive solar algorithm 135 by reading changes in the electrical grid 150.
In some embodiments, as a more specific basis for the electrical grid control scheme 145 generating its commands, 145 receives indications of the present usable range of output of the PV array directly from the responsive solar algorithm 135. In this embodiment, the responsive solar algorithm 135 generates information and data for the electrical grid control scheme 145 from information received from solar inverter 115 and from other sources.
In some embodiments, there are user-definable settings and other inputs such as weather forecasts, load volatility or other relevant data 120, which affect the amount of PV output that is allowed, and in some embodiments those settings are calculated from weather forecasts and other inputs that will increase or decrease buffer size and sensitivity based on those forecasts.
The flow of solar energy in
The flow of information in
The flow of commands, feedback loop, and resulting operation of the larger PV array in
To maximize the efficient use of solar power, the responsive solar algorithm 340 first answers the question: “Moment to moment, how much solar power is the large controlled array 320 capable of providing?” In order to answer this question, the algorithm gets a report 335 of the actual output of the smaller array from the uncontrolled solar inverter 330, uses that information to calculate a Maximum Available Solar Output, and then uses that value to send appropriate commands 345 to intelligently control the output of the larger array's solar inverter 350.
The Maximum Available Solar Output is a dynamic benchmark of the energy that the larger solar array 320 could provide at every moment. Using that benchmark, the facility then defines a subset of the Maximum Available Solar Output that is most reliable to use.
The amount of solar power that the larger PV array 320 is capable of providing at a given moment—its Maximum Available Solar Output—is given by:
For example, at a particular moment, the smaller “sentinel” array is providing 200 kilowatts of energy. The rated output capacity of the larger array is 1200 kilowatts. The rated output capacity of the smaller array is 300 kilowatts. Thus:
The facility therefore determines the Maximum Available Solar Output of the larger PV array at this moment to be 800 kilowatts.
The smaller “sentinel” array is needed to determine the Maximum Available Solar Output of the larger array because the larger array's inverter will be busy attenuating its power output per commands from the facility. Inverters, as presently built, are not able to determine the maximum potential of its solar input at the same time it is attenuating that solar for output.
A first problem is described in callout 510: The upper range of the Maximum Available Solar Output is vulnerable to sudden, unpredictable solar blockage. As illustrated on the chart, the line of solar output can “cut down” and reduce the Maximum Available Solar Output at any time.
This is the reason electric utilities do not consider solar energy to be dispatchable. As utilities seek to increase solar's contribution to the power grid, the common assumption that they must at all times use a solar array's entire Maximum Available Solar Output brings them face-to-face with this problem.
A second problem is described in callout 520: It is typically difficult for solar inverters to operate within the lower range of the Maximum Available Solar Output because of the way they work. Some inverters can become instable (and liable to shut-down) when asked to attenuate large amounts of solar energy at low power output levels relative to their rated power capacity.
The facility addresses the first problem 510 by calculating an Upper Buffer to put distance between the volatility of solar blockage and a Usable Range of solar output. Defining this Upper Buffer between the Maximum Available Solar Output and the Usable Range reduces the likelihood that the next solar blockage will infringe upon that Usable Range.
No buffer can reduce that likelihood to zero. If a thick solitary cloud comes “out of nowhere” and blocks the PV array, nothing can stop the Maximum Available Solar Output from falling to zero. However, in calculating an Upper Buffer, the facility greatly increases the likelihood that the solar energy it defines as being within the Usable Range will be available.
The facility addresses the second problem 520 by calculating a Lower Buffer that ensures that the Lower Limit of Usable Range stays above the point of potential inverter instability.
By addressing these two problems, the facility enables electric utilities to define a range of available solar output within which they can control the solar array output with a reasonable degree of reliability. This enables the solar array to deliver grid-supportive functions that require such dispatchable flexibility. Thus, the facility enables new benefits from solar that are available even when it is not necessarily delivering the maximum amount of solar energy.
As the sun climbs the sky, the Maximum Available Solar Output 605 gradually increases, subject to varying degrees of solar blockage (the dips in the line).
By calculating the size of an Upper Buffer 610, the facility defines the Upper Limit 615 of a Usable Range 620 of solar output.
By calculating the size of a Lower Buffer 640, the facility defines the Lower Limit 635 of that Usable Range 620.
Once Usable Range has been defined, in some embodiments the facility determines a Midpoint 625, which is useful for grid control schemes like AGC that request a delta of real power output from a baseline.
In short, the facility calculates the amount of less reliable capacity to cut off each end (610 and 640) of the Maximum Available Solar Output, leaving the remainder—the Usable Range 620—to be called upon with a degree of confidence much greater than is conventionally granted to solar energy sources.
In calculating the size of the Upper Buffer, in some embodiments the facility uses five tools: A Shorter Moving Average, a Longer Moving Average, a Moving Average Delta, a Minimum Size, and a Sensitivity Ratio.
Here, a Minimum Size of 100 kilowatts is specified for the Upper Buffer 920, and the Upper Buffer is pinned directly to the Maximum Available Solar Output 910.
Here, a Minimum Size of 100 kilowatts is specified for the Upper Buffer 970, and the Upper Buffer is pinned to the Shorter Moving Average 960 (i.e. 100 kilowatts less than the Shorter Moving Average).
As shown, in some embodiments, the facility pins the Upper Buffer to various values, including the Maximum Available Solar Output (
Pinning the Upper Buffer to one of the Moving Averages increases the risk of interruptions from solar blockage, but yields a more stable Usable Range (and Usable Range Midpoint). The facility's preferred approach is to pin the Upper Buffer to the Shorter Moving Average (
The next tool used by the facility in some embodiments to calculate the size of the Upper Buffer is the Moving Average Delta.
While these figures and text use the words “cloudy” or “cloudier” to represent solar blockage, many things can prevent sunlight from reaching a solar array—clouds, smoke, fog, haze, birds, airplanes, helicopters, hot air balloons, hang gliders, leaves, dust and other debris. Sources of blockage are typically difficult to predict with the degree of certainty required by grid control schemes.
Similarly, the words “sunny” and “sunnier” are used to represent moments of greater delivery of the sun's energy to the solar panel; these moments may or may not be moments when the sun appears to be higher, or brighter, or hotter.
The purpose of the Moving Average Delta is to make the size of the Upper Buffer more responsive to the volatility of solar blockage. To determine how likely it is that the sun will be blocked in the next moment, the facility calculates:
Longer Moving Average−Shorter Moving Average
In example 1010: The Maximum Available Solar Output is trending gently downward. That is, it is getting a bit cloudier. The resultant Moving Average Delta is a small positive number:
500 kilowatts−475 kilowatts=25 kilowatts
In example 1020: The Maximum Available Solar Output is trending severely downward. That is, it is getting a lot cloudier. The resultant Moving Average Delta is a larger positive number:
500 kilowatts−400 kilowatts=100 kilowatts
In example 1030: The Maximum Available Solar Output is trending gently upward. That is, it is getting a bit sunnier. The resultant Moving Average Delta is a small negative number:
500 kilowatts−525 kilowatts=−25 kilowatts
In example 1040: The Maximum Available Solar Output is trending severely upward. That is, it is getting a lot sunnier. The resultant Moving Average Delta is a larger negative number:
500 kilowatts−600 kilowatts=−100 kilowatts
In applying Moving Average Delta to the size of the Upper Buffer, in some embodiments the facility ignores negative values. That is, if it is trending sunnier (Moving Average Delta is negative), the facility determines the size of the Upper Buffer based upon Minimum Size alone. Only if it is trending cloudier (Moving Average Delta is positive) does the facility increase the size of the Upper Buffer beyond the Minimum Size.
In some embodiments, the facility uses a Sensitivity Ratio to control the degree of effect that Moving Average Delta has on the size of the Upper Buffer.
On all six charts shown in
On the three left-most charts, a simple, linear Moving Average Delta and the resultant effect of Sensitivity Ratio illustrate the relationship between the two values.
On the three right-most charts, Moving Average Delta over a portion of a sample solar day is used to illustrate the effect of Sensitivity Ratio in real time.
On the top row of charts, the Sensitivity Ratio is 0.50. On the middle row of charts, the Sensitivity Ratio is 1.00. On the bottom row of charts, the Sensitivity Ratio is 1.50.
In various embodiments, these parameters (Minimum Size and Sensitivity Ratio) are settable by the user of the facility. For example, in some embodiments the facility adjusts these settings daily. In some embodiments, the facility adjusts these settings based on current conditions on both the supply side (how much volatility in solar output is anticipated) and the demand side (how much volatility in circuit load is anticipated).
On the supply side, in various embodiments the facility adjusts these settings in response to the climatological and seasonal weather patterns at the location of the PV array; based on weather forecasts and personal observations of weather patterns, such as “What type of cloud day is this?”
For example, if local weather patterns are indicating either a full-sun day or a low-blanket-cloud-cover day, the volatility on the supply side is likely to be low, so in some embodiments the facility sets the parameters Minimum Size and Sensitivity Ratio to lower values. If local weather patterns are indicating a come-and-go-cloud day, in some embodiments the facility sets the parameters Minimum Size and Sensitivity Ratio to higher, more conservative values.
On the demand side, in various embodiments the facility adjusts these parameters higher or lower to response to the “neediness” of the electrical grid on a given day, and/or the relative “stiffness” or resilience of the local grid in dealing with the volatility of customer load.
The Maximum Available Solar Output 1310 is charted to the right-side axis. The associated Moving Average Delta 1320 is charted to the left-size axis. Given a Minimum Size of 150 kilowatts and a Sensitivity Ratio of 1.20, the resultant size of the Upper Buffer is also charted to the left-size axis.
Note that, in this example, the Upper Buffer is not perfect in preventing solar blockage from cutting into the Usable Range, but that it is substantially successful in doing so.
The facility calculates the Lower Buffer as a percentage of the rated capacity of the PV solar array. Here, the rated capacity of the controlled solar array is 1000 kilowatts, and the Lower Buffer is set at 5% of that capacity, thus 50 kilowatts. In various embodiments, the facility sets this Lower Buffer at various percentages of capacity, based on the relative efficiency of the inverter(s) being used, such as 5% of rated capacity.
In some embodiments, the facility includes use of the Usable Range Midpoint.
For example, a typical AGC command is “give me 20 more kilowatts of real power.” The AGC program is not asking for a specific quantity of total solar output, it is saying “whatever you were giving me a moment ago, now give me 20 kilowatts more than that.” By pinning its response to AGC commands to the Midpoint of the Usable Range, this facility both (a) delivers to the electric grid the most solar output it can reliably deliver, and (b) responds to the specific requests of the AGC program to help condition frequency on the electrical grid.
The Usable Range Midpoint on which the facility bases its responses is itself a continuously moving value. Depending on solar conditions and parameter settings, a grid control scheme (like AGC) can receive more or less energy than it requested. This is an essential trade-off necessary to utilize the best of solar. Solar will not always be available, and the amount of solar that is available will inevitably incorporate a degree of up-and-down variability. The facility quantifies, manages and minimizes this “sometime” quality of solar, but it does not eliminate it.
Typically, grid control schemes like AGC calculate the needs of the system, and then distribute the resultant commands (requests for more or less real power) among multiple energy resources. Typically, these commands are divided in fixed proportions among the available resources. The facility enables grid control schemes to divide their commands in dynamic proportions, responding to solar availability as the facility defines it throughout the day.
In some embodiments, the facility is incorporated into the grid as a “black box” that responds to requests for power as described above. The facility can also effectively engage power control schemes like AGC in a two-way dialogue. In basing its responses to AGC commands on the Midpoint of the Usable Range, the facility accomplishes three things: (a) delivering the most solar output it can reliably deliver, (b) responding typically to the specific requests of the AGC program, and importantly (c) giving the AGC program dynamic feedback about the latest availability of solar energy, which the AGC program can then incorporate into formulating its next request.
It is a distinct advantage of the facility that, in some embodiments, it can respond to AGC requests in a nuanced way, delivering specific deltas of real power (e.g., 10 kilowatts more, 25 kilowatts less) in real time. Often, AGC requests are sent to resources (like energy storage systems) that are programmed to respond only in an all-or-nothing fashion, turning on when system frequency is in need of conditioning, and then turning off when system frequency has returned to an acceptable range.
It is a distinct advantage of the facility that, in some embodiments, it can respond to any electrical grid control scheme (AGC being only one of them) that issues commands for greater or lesser real energy to regulate other system parameters.
It is a distinct advantage of the facility that, in some embodiments, in responding to electrical grid control schemes like Frequency Response and Volt/Watt, it enables the PV array itself to respond dynamically to the frequency and voltage volatility that it itself is potentially introducing into the grid.
Those skilled in the art will recognize that other electrical grid control schemes may request real power in other ways. The facility makes available to any such control scheme a Usable Range of real power available, and a Midpoint of Usable Range as a baseline.
Key elements for all four of these charts are identified on
The top-most line 1605 is the Maximum Available Solar Output. The Upper Buffer 1610 is defined to reduce the likelihood that solar blockage will affect the Usable Range. The Usable Range 1620 identifies the most reliable portion of solar energy available. The Midpoint of the Usable Range 1630 is useful to grid control schemes such as AGC as a baseline for requesting greater or lesser real power output from the PV arrays through the facility. The Lower Buffer 1640 distances the Usable Range from solar inverter instability.
For each sample day, the figures in the following table indicate how reliably available solar energy is within the Usable Range as defined by the facility:
“Overall availability of energy within the Usable Range:” Looking at the entire day, what percentage of the Usable Range as defined by the facility (the medium grey area on the charts) was actually available for dispatch (stayed within Maximum Available Solar Output)? That is, how effective is the facility in filtering out solar intermittency?
“Percentage of time the full Usable Range was available:” If, every five seconds throughout the day, a grid control scheme requested the full Usable Range as defined by the facility, what percentage of those requests would be successfully fulfilled?
For an ideal solar day (like Sample Solar Day 1), is it easy to achieve 100% availability in the two categories described above. For more complex solar days (like Sample Solar Days 2, 3, and 4), achieving high availability is much more challenging.
In comparison, the figures in the following table indicate how reliably available solar energy would have been without using the facility, and thus without defining a Usable Range.
Having no facility for doing otherwise, the utility can only observe how actual solar availability falls far short of an ideal solar curve (as represented by Sample Solar Day 1):
“Overall availability of solar energy:” Looking at the entire day, what percentage of an ideal solar curve (equal to Sample Solar Day 1) was available for dispatch (stayed within Maximum Available Solar Output)?
“Percentage of time solar energy was fully available:” If, every five seconds throughout the day, a grid control scheme requested all of an ideal solar curve (equal to Sample Solar Day 1), what percentage of those requests would be successfully fulfilled?
Key elements for all four of these charts are identified on
For each sample day, the figures in the following table indicate how reliably these energy requests were honored:
“Overall availability of requested energy:” Looking at the entire day, what percentage of the requested energy (the dark grey area on the charts) was actually available for dispatch (stayed within Maximum Available Solar Output)? That is, how effective is the facility in reliably delivering requested energy to the grid?
“Percentage of time the requests were fully honored:” If, every five seconds throughout the day, a grid control scheme made typical energy requests (the dark grey area on the charts), what percentage of those requests would be successfully fulfilled?
Even when the availability of solar energy is low, the performance of the responsive solar algorithm in responding to energy requests can be high.
By using the facility, the applied performance of solar can be workably dispatchable.
In various embodiments, the facility performs in similar ways to manage the output of energy generation resources of a variety of other types, including renewable energy generation resources such as wind, geothermal, hydro, wave, and tidal energy generation resources.
It will be appreciated by those skilled in the art that the above-described facility may be straightforwardly adapted or extended in various ways. While the foregoing description makes reference to particular embodiments, the scope of the invention is defined solely by the claims that follow and the elements recited explicitly therein.
| Number | Name | Date | Kind |
|---|---|---|---|
| 20090295162 | Oohara | Dec 2009 | A1 |
| 20110019447 | Olsen | Jan 2011 | A1 |
| 20110227417 | Walker | Sep 2011 | A1 |
| 20130018516 | Chee | Jan 2013 | A1 |
| 20130054044 | Shaffer | Feb 2013 | A1 |
| 20130073099 | Bronicki | Mar 2013 | A1 |
| 20130131884 | Jain | May 2013 | A1 |
| 20130231797 | Li | Sep 2013 | A1 |
| 20130305528 | Anderson | Nov 2013 | A1 |
| 20140074311 | Kearns | Mar 2014 | A1 |
| 20140095076 | Marwah | Apr 2014 | A1 |
| 20150234943 | Brier | Aug 2015 | A1 |
| 20160118803 | Takahashi | Apr 2016 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20170149247 A1 | May 2017 | US |
