WO2020014006A1

WO2020014006A1 - Dynamic photovoltaic curtailment for electrical power distribution

Info

Publication number: WO2020014006A1
Application number: PCT/US2019/039709
Authority: WO
Inventors: Ulrich Münz; Xiaofan Wu; Ti-Chiun Chang; Jiaxing PI
Original assignee: Siemens Aktiengesellschaft; Siemens Corporation
Priority date: 2018-07-13
Filing date: 2019-06-28
Publication date: 2020-01-16

Abstract

Dynamic photovoltaic curtailment (48) uses one or more forecasts (40) of photovoltaic power generation with or without a forecast (42) of load and flexibility (44). The forecast power, load, and/or flexibility are used in a solution (46) for power levels from one or more photovoltaic sources (11) meeting constraints of the distribution system (10). The curtailment (48) is adjusted to the power levels from the solution (46), allowing for more optimum levels of power to be in-fed to the distribution system (10) from the photovoltaic sources (11) at any given time.

Description

DYNAMIC PHOTOVOLTAIC CURTAILMENT FOR ELECTRICAL POWER

DISTRIBUTION

RELATED APPLICATIONS

[0001] The present patent document claims the benefit of the filing date under 35 U.S.C. §119(e) of Provisional U.S. Patent Application Serial No. 62/697,759, filed July 13, 2018, which is hereby incorporated by reference.

BACKGROUND

[0002] The present embodiments relate to curtailment of photovoltaic power in-fed to a power distribution system. The increase of renewable power generation by photovoltaics poses challenges to distribution system operators. Renewables are volatile, changing from full or higher power to zero or lesser power within a few minutes (e.g. if a cloud covers a photovoltaic system). Distribution system operators account for this volatility as uncertainty. The larger this uncertainty, the higher the risk of sever distribution system problems due to over-voltage and/or over-loading of transformers and power lines.

[0003] In order to avoid black outs, over-voltage, or over-loading, distribution system operators curtail the power from the photovoltaic systems (i.e. limit the maximal infeed power from solar sources). The curtailment limiting the power accepted from the photovoltaics is sufficient to reduce risk to a low level despite the uncertainties. This curtailment is static (i.e., does not change over time) to avoid jeopardizing the power distribution system. As a result, the power curtailment from photovoltaics is much higher than actually needed.

SUMMARY

[0004] By way of introduction, the preferred embodiments described below include methods, systems, instructions, and computer readable media for dynamic photovoltaic curtailment. In further embodiments, the dynamic curtailment uses one or more forecasts of photovoltaic power generation with or without a forecast of load and flexibility. The forecast power, load, and/or flexibility are used in a solution for power levels from one or more photovoltaic sources meeting constraints of the distribution system. The curtailment is adjusted to the power levels from the solution, allowing for more optimum levels of power to be in-fed to the distribution system from the photovoltaic sources at any given time.

[0005] In a first aspect, a system is provided for dynamic photovoltaic curtailment. An electric power distribution system connects with a

photovoltaic generator at a photovoltaic infeed. A photovoltaic inverter at the photovoltaic infeed has a programmable curtailment of power from the photovoltaic power generator to the electric power distribution system. A controller is configured to dynamically control the programmable curtailment by the photovoltaic inverter. The dynamic control varies a maximum level of the power from the photovoltaic power generator provided to the electric power distribution system over time.

[0006] In one embodiment, the photovoltaic infeed is a bus. The electrical power distribution system includes a transformer and meter connected with the photovoltaic infeed, a substation, and a power wire or wires connected from the photovoltaic infeed to the substation. If a further embodiment, the controller is at the substation. The controller is configured to dynamically control the programmable curtailment by the photovoltaic inverter and other curtailments by other photovoltaic inverters for other photovoltaic power generators connected with the electrical power distribution system.

[0007] In another approach, the controller is configured to dynamically control the programmable curtailment based on limitations of the electric power distribution system and based on a photovoltaic forecast of future power from the photovoltaic power generator. The controller may be configured to dynamically control the programmable curtailment based on the photovoltaic forecast output, which is estimated, using a machine-learned estimator, from sky images acquired from a camera. The controller may be configured to dynamically control the programmable curtailment based on the photovoltaic forecast and a load forecast of future load in the electric power distribution system. The controller may be configured to dynamically control the programmable curtailment based on the photovoltaic forecast and an amount of flexibility from a flexibility market for the electric power distribution system.

[0008] In other embodiments, the controller is configured to dynamically control the programmable curtailment by solution of a chance-constrained optimal power flow to maximize the maximum level of the power. The chance-constrained optimal power flow is constrained by limits of the electric power distribution system and using uncertainty in the power from the photovoltaic power generator. The uncertainty accounts for possible variation from the forecast. For example, the uncertainty includes a probability density function for the power from the photovoltaic power generator. The chance- constrained optimal power flow may be solved in a stochastic optimization or deterministically. In one embodiment, the chance-constrained optimal power flow uses uncertainty in a load of the electrical power distribution system. The limits include voltage and power transfer limits, and the solution provides a reactive power setpoint. The controller is configured to control a reactive power from the photovoltaic power generator based on the reactive power setpoint.

[0009] Any update period or frequency may be used. For example, the controller is configured to update the maximum level at a frequency of one hour or less, such as every 15 or 30 minutes.

[0010] In a second aspect, a method is provided for dynamic photovoltaic curtailment in a power distribution system. A photovoltaic power to be output by a photovoltaic connected with the power distribution system is forecast. A power limit from the photovoltaic is solved for as a constrained optimization based on the forecast photovoltaic power. The power from the photovoltaic to the power distribution system is limited to at or below the power limit.

[0011] In one embodiment, the solution solves a chance-constrained optimal power flow function maximizing the power limit. The solution may be solved as a deterministic function with stochastic uncertainties modeled as deterministic parameters with upper and lower bounds.

[0012] In another embodiment, the solution is based on the forecast photovoltaic power and a forecast of load of the power distribution system and is with constraint based on reactive power, power transfer, phase, and voltage limits of the power distribution system.

[0013] In a third aspect, a method is provided for dynamic photovoltaic curtailment in a power distribution system. Amounts of power to be provided from different solar electricity sources to the power distribution system in a time window of one hour or less are predicted. Limits for the different solar electricity sources are optimized based on the predicated amounts. The optimization is constrained based on operating parameters of the power distribution system. The levels of power from the solar electricity sources are controlled based on the limits during the time window.

[0014] In one embodiment, the limits are optimized as power limits for the time window with a chance-constrained optimal power flow function where the operating parameters constraining the optimization include power transfer, voltage and phase of the power distribution system. In another embodiment, a chance-constrained optimal power flow function is optimized based on the predicated amounts and predicted load for the power distribution in the time window, where the chance-constrained optimal power flow function includes probability distribution functions for the predicated amounts and the predicted load.

[0015] The present invention is defined by the following claims, and nothing in this section should be taken as a limitation on those claims.

Features of one type of claim (e.g., method or system) may be features of another type of claim. Further aspects and advantages of the invention are discussed below in conjunction with the preferred embodiments and may be later claimed independently or in combination.

BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The components and the figures are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the invention. Moreover, in the figures, like reference numerals designate corresponding parts throughout the different views.

[0017] Figure 1 is a block diagram of one embodiment of a system for dynamic photovoltaic curtailment; [0018] Figure 2 illustrates optimization for dynamic curtailment, according to one embodiment;

[0019] Figure 3 shows an example probability distribution function and corresponding curtailment used in a chance-constrained solution; and

[0020] Figure 4 is a flow chart diagram of one embodiment of a method for dynamic photovoltaic curtailment in a power distribution system.

DETAILED DESCRIPTION OF THE DRAWINGS AND PRESENTLY PREFERRED EMBODIMENTS

[0021] Dynamic photovoltaic curtailment is provided for a distribution system with renewable integration. The maximal admissible photovoltaic generation infeed is dynamically (e.g., during operation or varying over time) determined such that power system limits are not violated. The photovoltaic generation infeed is curtailed dynamically based on uncertain photovoltaic generation forecasts and/or uncertain load forecasts.

Flexibility markets consideration may be included in the dynamic

determination. The photovoltaic generation infeed may be maximized while also providing reliable and secure power supply and power system operation. With accurate forecasts and their probability distributions, the maximal admissible PV infeed is provided such that the power system is not at risk.

[0022] Various factors may be included in the optimal power flow formulation. The photovoltaic generation forecast, load forecast, and/or flexibility from the flexibility market may be included. The optimal power flow formulation is constrained by power system constraints. The formulation of the optimal power flow or chance-constrained optimal power flow includes the power system constraints.

[0023] Figure 1 shows one embodiment of a system for dynamic photovoltaic curtailment. The power curtailment for power infeed from photovoltaics is dynamically controlled. For dynamic control, the curtailment changes over time and/or is set based on measurements or feedback during operation or curtailment in a previous period. [0024] The system implements the method of Figure 4, but other methods may be implemented. The system controls the power from one or more photovoltaic generators 11 in a dynamic fashion. The dynamic fashion avoids exceeding constraints of the distribution system 10 with maximizing the amount of power provided at a given time or period as compared to using a static constraint on power infeed.

[0025] The system includes a distribution system 10, camera 16, one or more photovoltaic generators 11 , one or more meters 12, and one or more inverters 13. Additional, different, or fewer components may be provided. For example, any number of photovoltaic generators 11 , meters 12, and/or inverters 13 are provided. Multiple distribution systems 10 operating independently or connected together to share power may be provided. One or more transmission lines for powering the substation 17 may be provided.

[0026] The distribution system 10 includes transformers 15 and power lines or wires 19 connecting power transmission lines through the substation 17 to loads. The distribution system 10 distributes electrical power to customers, such as residential or commercial buildings. Figure 1 shows a distribution system 10 according to one embodiment where the components for photovoltaic infeed control are represented. The distribution system 10 may include other parts.

[0027] In one embodiment, the distribution system 10 has a substation 17 connected by power lines or power wires 19 to transformers 15. The transformers 15 connect directly or through other power wires 19 to buses 14. A controller 18, using a memory 9, controls operation of the distribution system 10. Additional, different, or fewer components may be provided. For example, the inverters 13, camera 16, and/or meters 12 may be part of the distribution system 10. As another example, the controller 18 and/or memory 9 may be separate from the distribution system 10.

[0028] The distribution system 10 connects with the photovoltaic infeed to receive power from the photovoltaic generators 11. The photovoltaic infeed may be the bus 14 or other connection between the photovoltaic generator 11 and the distribution system 10. The bus 14 receives power from the photovoltaic generator 11 through the inverter 13, which is part of the residential or commercial photovoltaic generator 11 , part of the distribution system 10, and/or separate. While the transformer 15, bus 14, inverter 13, meter 12, and photovoltaic generator 11 are shown grouped together, the components may be spaced apart by intervening wires, may be directly connected together, and/or may be in a same housing, panel, or circuit board.

[0029] In the example of Figure 1 , three photovoltaic generators 11 and corresponding components for connecting to and supplying power to the distribution system 10 are provided. Only one, two, or four or more

photovoltaic generators 11 and corresponding components may be used.

[0030] The photovoltaic generators 11 are solar plants or one or more photovoltaic cells or panels. For example, one or more of the photovoltaic generators 11 are formed as arrays of solar panels. Different photovoltaic generators 11 may have different numbers and/or sizes of solar panels. The solar panels may be for supplying power to a regional (e.g., county, country, or state) power distribution system or may be for supplying power to a distribution system of a facility, complex, or building. The solar panels may be for supplying residential or commercial buildings with power, with overflow power supplied to the distribution system 10.

[0031] The photovoltaic generators 11 connect to the distribution system 10 for supplying power to the distribution system 10. Extra power or overflow power is provided. Alternatively or additionally, the photovoltaic generators 11 connect to the distribution system 10 as a power source for the distribution system 10. The photovoltaic generators 11 connect through transformers 15 and/or direct to alternating current converters (e.g., inverters 13) as the photovoltaic infeed to the power distribution network or system 10.

[0032] The photovoltaic inverter 13 connects at the photovoltaic infeed.

The photovoltaic inverter 13 is a circuit converting the direct current output of the photovoltaic generator 11 to alternating current of the distribution system 10. The photovoltaic inverter 13 includes voltage and/or current limiters to limit the power provided from the photovoltaic generator 11 to the distribution system 10, such as at the bus 14. The curtailment of power from the photovoltaic power generator 11 to the electric power distribution system 10 is programmable. The circuit for limiting is programmable. A processor or other controller sets the limit or setpoint for the level of power allowed by the inverter 13 at the photovoltaic infeed to the distribution system 10.

[0033] The meter 12 is an e-meter or other meter for measuring voltage, current, power, reactive power, active power, inductance, and/or other characteristic of the power supplied to the distribution system 10 by the inverter 13 and/or power in the distribution system 10. The meter 12 is a circuit with a processor and transmitter in one embodiment. The processor controls the circuit and transmissions on the transmitter. The measured data from the meter 12 is output to the controller 18 via the power lines 19, separate wired communication, wireless communication, and/or over a computer network.

[0034] The camera 16 is a charge coupled diode (CCD) camera. Other optical cameras may be used. One camera is provided, but an array of cameras capturing the same or different parts of the sky may be used. The camera 16 has any field of view. In one embodiment, a fish-eye lens is provided to extend the field of view close to the horizon in a 360-degree arc around the horizon.

[0035] The camera 16 is positioned at the distribution system 10, such as at the substation 17, along the power wires 19, and/or by one of the

photovoltaic generators 11. For example, the camera 16 is positioned on a tower with a center of the field of view directed straight upward (i.e. , normal to the earth or parallel to gravity) to capture the sky over one or more of the photovoltaic generators 11. Other locations or positions may be used. The camera 16 is positioned to capture images of at least part of the sky. Ground- based objects may be physically or electronically masked from captured images. Non-cloud objects in the sky (e.g., birds or planes) may be removed or reduced by filtering or segmentation.

[0036] As an alternative or in addition to the camera 16, an irradiation sensor, such as a pyranometer or photovoltaic cell, may be used. The irradiation sensor measures a current irradiation from the sun.

[0037] The controller 18 is a computer, workstation, server, industrial process controller, processor of a panel, application specific integrated circuit, field programmable gate array, or other processor or network of processing hardware, such as a hybrid power optimizer (HPO). The controller 18 is a single device or multiple devices operating in serial, parallel, or separately.

The controller 18 may be a main processor of a computer, such as a laptop, server, or desktop computer, or may be a processor for handling some tasks in a larger system, such as a processor of a panel or workstation for power control of the distribution system 10. The controller 18 is configured by instructions, firmware, design, hardware, and/or software to perform the acts discussed herein.

[0038] The controller 18 is shown as part of the substation or a controller remote from but for controlling the power distribution system 10. In other embodiments, the controller 18 is a controller for the photovoltaic generator 11 , such as a controller in a panel with the inverter 13, meter 12, and/or bus 14. The controller 18 is for control of various components or just the inverter 13 or just the components for infeed from the photovoltaic generator 11. In other embodiments, the controller 18 is for control of multiple distributions systems and/or for multiple sub-stations 17.

[0039] The curtailment of power from the photovoltaic generator 11 is controlled. The same or different limit is provided to two or more inverters 13. The controller 18 is configured to set the curtailment of power from the photovoltaic generator 11 by the inverter 13. The controller 18 directly controls the inverter 13 or indirectly controls by communicating a set point to a controller of the inverter 13.

[0040] The curtailment of power implemented by the controller 18 is discussed below for infeed from one photovoltaic generator 11. The same process is implemented for the other photovoltaic generators 11 connected to the distribution system 10. One optimization to solve for the power level may be used to find the same or different power levels for different ones of the photovoltaic generators 11.

[0041] The controller 18 is configured to update or set the maximum level at a frequency of one hour or less. The setting of the curtailment is dynamic, changing over time and/or being based on measures during operation. Any frequency of gathering data, solving for the power level limit, and/or changing the set point for curtailment may be used. In one embodiment, the controller 18 repeats the solving for the power limit and setting the power limit every 15 minutes. More (e.g., every 5 minutes) or less frequent (e.g., 30 minutes or 2 hours) may be used. The frequency may depend on forecasting accuracy, the quality of communication, and/ the ability to provide new maximal infeed values to the photovoltaic generator systems (i.e., the inverters 13). In other embodiments, the update occurs upon one or more triggers. For example, forecasting of solar power and/or load happens at one frequency. When the forecast or forecasts change by a threshold amount from the values of the previous update, then a new update to the set points is triggered. Otherwise, the set point remains the same.

[0042] The set point of the power limit for curtailment is a value for a given future time range, such as 15 minutes. The future time is from a current time or a time beginning in the future (e.g., in one minute from communication of the set point). Any period may be used, such as the same period for updating the curtailment setpoint.

[0043] The controller 18 is configured to dynamically control the

programmable curtailment by the photovoltaic inverter 13. One or more inverters 13 for one or more of the photovoltaic power generators 11 are controlled to infeed power to the distribution system 10 at or below the one or more setpoints (i.e., power level limits for the curtailments). The controller 18 communicates the photovoltaic power limits, Ppvi max, to the inverters 13 and/or photovoltaic power generators 11. Reactive power setpoints, Qi, may also be communicated. The photovoltaic generators 11 are not allowed to feed in more than the calculated limits.

[0044] Since the control is dynamic, the maximum level of the power (i.e., photovoltaic power limit) varies over time. During each period, the limit is the same. For each period, the limit is the same or different than used in the preceding period and/or subsequent period. The power limit is solved for each period, so may be different over time or for different periods. The data used to solve for the power limit is collected, at least in part, from the preceding period. The control is dynamic by changing over time and/or using data recently acquired during operation. The algorithm runs repeatedly every 15 minutes or other frequency to dynamically curtail the photovoltaic infeed based on various inputs, such as the current photovoltaic forecast and the current load forecast.

[0045] The maximum level of the power is based on one or more variables. Example variables include a forecast of the power from the photovoltaic generator 11 , a forecast of a load on the power distribution system 10, a power level available from the flexibility market, and one or more constraints of the power distribution system 10. Other variables or combinations of variables may be used, such as one or more constraints in combination with the photovoltaic power prediction and/or the load prediction. The power from the flexibility market and/or forecast of the load may not be used in some embodiments.

[0046] The maximum level of the power is based on a photovoltaic forecast. The power to be output by the photovoltaic generator 11 in the next period is predicted. The power may be predicted as a median, average, sum, and/or other characteristic of power over the period. For example, the photovoltaic infeed power, Ppvi, for the next 15 minutes (e.g., next period) is forecast.

[0047] The forecast is based on a weather model. For example, the forecast is based on weather predictions from a weather service. The predictions are for the level of solar irradiation. For example, cloud coverage level may be used to indicate the power to be output in the period. In one embodiment, a machine-learned model, such as a machine-learned neural network, outputs the forecast of the power. The machine-learned model or estimator is trained to estimate power based on input information. The input information may be one or more images of the sky from the camera 16 and/or weather predictions. Short-term photovoltaic power output is forecast or estimated based on the images from the camera (i.e., sky imager) and one or more neural networks.

[0048] The maximum level of the power may be based on a load forecast. The load burden in the electric power distribution system 10 for the next period is predicted. For example, the uncertain load, Pu, in the distribution system 10 for the next 15 minutes is predicted. [0049] The load is predicted from the current load. In other

embodiments, the load is predicted based on past power usage. For example, power measurements by the meters 12 and/or other sources for the same period by time (e.g., hour, day, week, month, and/or year) are used to predict the power for this period. In one embodiment, both long and short-term temporal patterns in load are used to predict the load for the next period. The current load may be used in the prediction.

[0050] The maximum level of the power may be based on an amount of flexibility from a flexibility market for the electric power distribution system 10. The operator of the distribution system 10 may have data indicating an amount of power reduction available based on electric vehicle, battery chargers, and/or other sources for immediate reduction in the load.

[0051] The variables, such as the forecast power and load, are used with the constraints to solve for the power level limit. Any solution may be used, such as a numerical optimization. Since the load and/or photovoltaic powers are predictions, the uncertainty may be used in the solution. For example, the forecast load has the uncertainty reflected with a probability density function, PDFPU, showing distribution of possible loads. The load probability density function PDFPU may be included in the optimization problem as a chance constraint. As another example, the forecast photovoltaic infeed (power) has the uncertainty reflected with a probability density function, PDFPVI, showing distribution of possible power. The power probability density function PDFPVI may be included in the optimization problem as a chance constraint. The chance constraint handles a distribution of probability for different levels (e.g., different powers or loads).

[0052] This chance-constrained optimal power flow problem (CC-OPF) minimizes photovoltaic curtailment dynamically by maximizing the admissible photovoltaic infeed power Ppvi,max under uncertain photovoltaic power and load forecasts. The flexibilities, Pn, in the distribution system 10 may also be included in the CC-OPF.

[0053] Figure 2 shows an example. The controller 18 receives the sky image and meter data from the distribution system 10. The photovoltaic power 20 is forecast, the load 22 is forecast, and the amount of flexibility 24 is determined. This power information is used by the controller 18 in the CC-OPF to generate the power limit and the reactive power setpoints. Other optimal power flow problems without the chance constraint may be used, such as using single values for the forecast power and/or load.

[0054] Any of various constraints may be used in the optimization. For example, the voltage and power transfer limits ½ min, V i_:max , Pij,min, Pij,max Of the distribution system 10 are accounted for as constraints. The power flow may be used as the constraint. Various variables may establish the power flow in the distribution system 10, such as the power limit of power from the photovoltaic generator 11 , the load forecast, the available amount of power flexibility, the reactive power, voltage magnitude, phase, and power transfer from bus 14 to bus 14. Additional, different, or fewer variables may be used for the power flow. For stability, the power flow is to equal 0 (e.g., power added equals the power removed while not exceeding the power transmission capabilities of the lines 19 and/or other components of the distribution system 10) as a constraint on the

optimization. The optimization solves the chance-constrained optimal power flow to maximize the maximum level of the power while constrained by limits of the electric power distribution system 10. The uncertainty in the power from the photovoltaic power generator and/or load is used in the solution.

[0055] The chance-constrained optimal power flow (CC-OPF) problem minimizes the photovoltaic curtailment (i.e., maximizes the admissible photovoltaic infeed power Ppvi.max) as well as calculates reactive power setpoints Qi under the given photovoltaic uncertainties while respecting the power system limits.

[0056] The CC-OPF is solved as a stochastic optimization. The CC-OPF uses uncertainty in a load of the electrical power distribution system 10 and/or power from the photovoltaic generator 11. In one embodiment, the limits include voltage and power transfer limits, and the solution provides a reactive power setpoint. The controller 18 is configured to control a reactive power from the photovoltaic power generator 11 based on the reactive power setpoint.

[0057] In one embodiment, the CC-OPF is formulated as:

where P_PV, is the uncertain photovoltaic infeed forecasted for the next 15 minutes with probability density function (PDF) PDF_Ppv. from the

photovoltaic forecasting, P_{PVi ma}x is the optimization variable for maximal photovoltaic infeed which curtails P_PV, such that Pp_Vi is the curtailed (but still uncertain) photovoltaic infeed with PDF PDF_p^, Pn is the available flexibility of the flexibility market, W_PV, and w are the weights of Ppvi.m_ax and Pn given by the flexibility market, Pu is the load forecast for the next 15 minutes with PDF PDF_Pw Q_/, V,- , q,- are the reactive power, voltage magnitude and phase in the distribution system, Pi_j is the power transfer from bus / to busy, Vi_,min, V_imax, P _min, Pg_max are voltage and power line 19/transformer 15 limits, and PF is a function describing the power flow equations. Indices / refer to the corresponding values at bus /, allowing for the solution of the curtailment for one or more photovoltaic generators 11 (i.e., ΐ=l or more). The sum of the powers from the photovoltaic

generators 11 and flexibility is maximized while the power flow is constrained to be 0.

[0058] Other functions or solutions may be provided, such as not including one or more variables listed for the power flow or not including an amount of power from the flexibility market in the maximization.

[0059] The weights are set empirically. The reactive power, voltage magnitude and phase may be measured by one or more meters 12. The PDFs for photovoltaic power and load forecasting are based on past powers given similar cloud coverage, weather conditions, and/or past usage. [0060] Figure 3 shows one example PDF 30 for the photovoltaic power. The optimization of the power limit sets the limit 34, resulting in the distribution of possible powers being limited, as reflected by curve 32. The optimization accounts for the PDF and the alteration of the PDF based on the limit.

[0061] Rather than using a stochastic optimization, the CC-OPF is solved deterministically. For example, the function given above for CC- OPF may be formulated in a deterministic way by modelling the stochastic uncertainties PPVI and Pu as deterministic parameters with upper and lower bound Ppvi ,min , PpVi.max, Pu ,min , Pu.rnax from the respective PDF. The resulting CC-OPF is illustrated as:

where Ppvi.min is provided by the photovoltaic forecast, Pu ,min , Pu.max 3Gb provided by the load forecast, and Ppvi max is the decision variable to curtail photovoltaic infeed. The resulting optimization problem is a min-max problem. Any optimization for a min-max problem may be used.

[0062] Referring to Figure 1 , the memory 9 is a random-access memory, system memory, cache memory, hard drive, optical media, magnetic media, flash drive, buffer, database, combinations thereof, or other now known or later developed memory device for storing data. The memory 9 is part of the controller 18, part of a computer associated with the controller 18, part of a database, part of another system, or a standalone device.

[0063] The memory 9 stores the images, irradiation measures, weather data, meter readings, distribution system limits, forecasts, PDFs, solution or function, and/or solved for power level limits and/or other set points. The memory 9 may store other information. The controller 18 may use the memory 9 to temporarily store information during performance of the method of Figure 4.

[0064] The memory 32 or other memory is alternatively or additionally a non-transitory computer readable storage medium storing data representing instructions executable by the programmed controller 18. The instructions for implementing the processes, methods and/or techniques discussed herein are provided on non-transitory computer-readable storage media or memories, such as a cache, buffer, RAM, removable media, hard drive, or other computer readable storage media. Non-transitory computer readable storage media include various types of volatile and nonvolatile storage media. The functions, acts or tasks illustrated in the figures or described herein are executed in response to one or more sets of instructions stored in or on computer readable storage media. The functions, acts or tasks are

independent of the particular type of instructions set, storage media, processor or processing strategy and may be performed by software, hardware, integrated circuits, firmware, micro code and the like, operating alone, or in combination. Likewise, processing strategies may include multiprocessing, multitasking, parallel processing, and the like.

[0065] In one embodiment, the instructions are stored on a removable media device for reading by local or remote systems. In other embodiments, the instructions are stored in a remote location for transfer through a computer network or over telephone lines. In yet other embodiments, the instructions are stored within a given computer, CPU, GPU, or system.

[0066] Figure 4 shows one embodiment of a method for dynamic photovoltaic curtailment in a power distribution system. Periodically and/or based on a trigger using data obtained during operation, one or more controllers solve for one or more curtailment levels to be applied to one or more solar electric sources for infeeds to the power distribution system. The solution has one or more of various attributes: the solution may use a solar power forecast, a load forecast, and/or a power flexibility to minimize curtailment from the solar electric sources; constraints, such as zero power flow and/or limits of the power distribution system, may be enforced in the solution; and/or probability information may be used in the solution, such as with a chance-constrained optimal power flow function.

[0067] The method is implemented by one or more controllers. The controller is for the distribution system, such as being in a panel at a substation. Alternatively, the controller is for the solar electric source.

Meters, access to a server or feed from a computer network, a memory, and/or other sources of data may be used for power prediction (e.g., load and/or solar source), determining an amount of flexibility, and/or setting constraint values. The system of Figure 1 may be used.

[0068] The acts are performed in the order shown (i.e. , top to bottom or numerical order) or other orders. For example, acts 40, 42, and 44 may be performed in any order or simultaneously. Additional, different, or fewer acts may be provided. For example, act 42 and/or act 44 are not provided where the solution uses the solar source power prediction and constraints without load prediction and/or flexibility.

[0069] In act 40, the controller predicts amounts of power to be provided from different solar electricity sources to the power distribution system. The prediction is performed for each of the solar electricity sources connected to the power distribution system but may be for only a sub-set or only one source. The photovoltaic power to be output (in-fed) by a photovoltaic (solar electricity source) is predicted.

[0070] The prediction is for a future time. The time may be for a window, such as a range of one hour or less. The range or future time is for the immediate future, such as in or starting in less than one, one, five, ten, or fifteen minutes. The amount of power as an average, peak, median, total (e.g., integration or summation), or other amount is predicted for the time window.

[0071] In one embodiment, a machine-learned estimator or model predicts the power based on input of one or more sky images and/or weather data. The prediction may include a probability distribution function for the power to be supplied based on past sampling. [0072] In act 42, the controller predicts the load on the distribution system. The load may include load local to the solar electricity source and/or other loads (e.g., remote from the solar electricity sources).

[0073] The load is predicted for the same time or time window. The load is predicted from current load and/or from historical usage for the same time of day, week, month, or year. Average usage is used as the prediction. The prediction may include a distribution or probability distribution function of the load based on past loading.

[0074] In act 44, the controller determines an amount of flexibility from the flexibility market. The distribution system operator may provide an amount of load that may be removed (i.e. , a gain in power) for the period.

[0075] The solution may account for this flexibility. The amount of power from the flexibility market may be part of the optimization (e.g., maximization or minimization) with the power from the solar sources. A weighting may be used to adjust the level of usage of the flexibility in the optimization.

[0076] In act 46, the controller, server, workstation, computer, and/or panel processor optimizes limits for the different solar electricity sources. The optimization is based on one or both of the predicated amounts (i.e., predicted powers from solar sources and/or predicted load). The optimization is constrained based on operating parameters of the power distribution system and a power flow substantially equal to zero. Substantially is used to account for tolerance in the distribution system and/or the distribution systems ability to handle over or under voltage. The limits are power limits for infeed from the solar sources. The limits are applicable for the time window.

[0077] The optimizing solves for the limits. An optimal power flow function is used. The function includes the constraint or constraints with minimization or maximization of curtailment or power limit for the infeed from the solar sources. The function uses the predicted power with or without predicated load and/or flexibility.

[0078] In one embodiment, a chance-constrained optimal power flow function is used. The operating parameters constraining the optimization include power transfer, voltage and phase of the power distribution system. The predicted infeed power and/or predicted load are represented probabilistically, such as with a probability density function. Other

probabilistic representations may be used.

[0079] The function is solved by optimization for the time window or time. The function includes values for one or more solar sources, such as solving for limits for multiple solar sources as part of the same optimization and function. A stochastic optimization or deterministic optimization accounting for the probabilities in the predictions is performed.

[0080] For a given solar source, the controller solves for a power limit from the photovoltaic as a constrained optimization based on the forecast photovoltaic power. Using a chance-constrained optimal power flow function maximizing the power limit, a deterministic function with stochastic

uncertainties modeled as deterministic parameters with upper and lower bounds or a stochastic function may be used.

[0081] In act 48, the controller controls levels of power from the solar electricity sources based on the limits. The limits are communicated to inverters, the solar sources, or other limiters to limit the infeed powers from the solar sources. The controller controls through communication of set points to other devices or by configuring the limiters directly.

[0082] The control is performed during the time window. The limits are set for the time window and/or until a trigger causes new set points for the power limits from the solar sources to be determined. At each infeed, the power from the photovoltaic to the power distribution system is limited to at or below the power limit.

[0083] By setting the power limits from photovoltaics dynamically, the amount of power input from the photovoltaics may be maximized while avoiding exceeding limits of power transfer by the distribution system. The power distribution system accepts power from solar sources with infeed limits that vary over time and/or based on current or expected operation. Rather than statically limiting, the dynamic limiting may better use available power.

[0084] While the invention has been described above by reference to various embodiments, it should be understood that many changes and modifications can be made without departing from the scope of the invention.

It is therefore intended that the foregoing detailed description be regarded as illustrative rather than limiting, and that it be understood that it is the following claims, including all equivalents, that are intended to define the spirit and scope of this invention.

Claims

I (WE) CLAIM:

1. A system for dynamic photovoltaic curtailment, the system comprising: an electric power distribution system (10) connected with a photovoltaic generator (11 ) at a photovoltaic infeed;

a photovoltaic inverter (13) at the photovoltaic infeed, the photovoltaic inverter (13) having a programmable curtailment of power from the

photovoltaic power generator (11 ) to the electric power distribution system (10); and

a controller (18) configured to dynamically control the programmable curtailment by the photovoltaic inverter (13), the dynamic control varying a maximum level of the power from the photovoltaic power generator (11 ) provided to the electric power distribution system (10) over time.

2. The system of claim 1 wherein the photovoltaic infeed comprises a bus (14), and wherein the electrical power distribution system (10) comprises a transformer (15) and meter (12) connected with the photovoltaic infeed, a substation (17), and a power wire or wires connected from the photovoltaic infeed to the substation (17).

3. The system of claim 2 wherein the controller (18) is at the substation (17), the controller (18) configured to dynamically control the programmable curtailment by the photovoltaic inverter (13) and other curtailments by other photovoltaic inverter (13)s for other photovoltaic power generator (11 )s connected with the electrical power distribution system (10).

4. The system of claim 1 wherein the controller (18) is configured to dynamically control the programmable curtailment based on limitations of the electric power distribution system (10) and based on a photovoltaic forecast of future power from the photovoltaic power generator (11 ).

5. The system of claim 4 wherein the controller (18) is configured to dynamically control the programmable curtailment based on the photovoltaic forecast output by a machine-learned estimator from input to the machine- learned estimator of a camera (16) image of the sky.

6. The system of claim 4 wherein the controller (18) is configured to dynamically control the programmable curtailment based on the photovoltaic forecast and a load forecast of future load in the electric power distribution system (10).

7. The system of claim 4 wherein the controller (18) is configured to dynamically control the programmable curtailment based on the photovoltaic forecast and an amount of flexibility from a flexibility market for the electric power distribution system (10).

8. The system of claim 1 wherein the controller (18) is configured to dynamically control the programmable curtailment by solution of a chance- constrained optimal power flow to maximize the maximum level of the power, the chance-constrained optimal power flow constrained by limits of the electric power distribution system (10) and using uncertainty in the power from the photovoltaic power generator (11 ).

9. The system of claim 8 wherein the uncertainty includes a probability density function for the power from the photovoltaic power generator (11 ).

10. The system of claim 8 wherein the chance-constrained optimal power flow is solved in a stochastic optimization.

11. The system of claim 8 wherein the chance-constrained optimal power flow is solved deterministically.

12. The system of claim 8 wherein the chance-constrained optimal power flow uses uncertainty in a load of the electrical power distribution system (10) and wherein the limits include voltage and power transfer limits and the solution provides a reactive power setpoint, the controller (18) configured to control a reactive power from the photovoltaic power generator (11 ) based on the reactive power setpoint.

13. The system of claim 1 wherein the controller (18) is configured to update the maximum level at a frequency of one hour or less.

14. A method for dynamic photovoltaic curtailment in a power distribution system (10), the method comprising:

forecasting (40) a photovoltaic power to be output by a photovoltaic (11 ) connected with the power distribution system (10);

solving (46) for a power limit from the photovoltaic (11 ) as a

constrained optimization based on the forecast photovoltaic power; and

limiting (48) the power from the photovoltaic (11 ) to the power distribution system (10), the power limited to at or below the power limit.

15. The method of claim 14 wherein solving (46) comprises solving (46) a chance-constrained optimal power flow function maximizing the power limit.

16. The method of claim 15 wherein solving (46) comprises solving (46) as a deterministic function with stochastic uncertainties modeled as deterministic parameters with upper and lower bounds.

17. The method of claim 14 wherein solving (46) comprises solving (46) based on the forecast photovoltaic power and a forecast of load of the power distribution system (10) and solving (46) with constraint based on reactive power, power transfer, phase, and voltage limits of the power distribution system (10).

18. A method for dynamic photovoltaic curtailment in a power distribution system (10), the method comprising:

predicting (40) amounts of power to be provided from different solar electricity sources to the power distribution system (10) in a time window of one hour or less;

optimizing (46) limits for the different solar electricity sources based on the predicated amounts, the optimizing constrained based on operating parameters of the power distribution system (10); and controlling (48) levels of power from the solar electricity sources based on the limits during the time window.

19. The method of claim 18 wherein optimizing (46) comprises optimizing (46) the limits as power limits for the time window with a chance-constrained optimal power flow function where the operating parameters constraining the optimization include power transfer, voltage and phase of the power distribution system (10).

20. The method of claim 18 wherein optimizing (46) comprises optimizing (46) a chance-constrained optimal power flow function based on the predicated amounts and predicted load for the power distribution in the time window, where the chance-constrained optimal power flow function includes probability distribution functions for the predicated amounts and the predicted load.