Planning and operating an electrical vehicle charging center using a linear model

Abstract

A charging network is provided for charging EVs of an EV fleet. The EVs have respective predicted charging-demand function collectively defining predicted energy and / or power requirements of the fleet. The charging network, which is operative to receive electric power from a plurality of electricity providers each associated with a respective predicted electricity-provider availability and collectively defining electricity- provider constraints, comprises: an EV-charging hub comprising respective producer-facing and consumer-facing interfaces, the producer-facing interfaces arranged to receive electric power from the electricity providers, the consumer-facing interfaces comprising a heterogeneous array of EV charging ports, and an EV charging-network controller configured to performing all of first, second and third charging-network regulation operations for the EV charging-hub so as to maximize a value function constrained by the EV-fleet energy-requirements and by the electrical-producer-constraints.

Description

[0001] PLANNING AND OPERATING AN ELECTRICAL VEHICLE CHARGING CENTER USING A LINEAR MODEL CROSS-REFERENCE TO RELATED APPLICATIONS This invention claims priority from United States Provisional Patent Application No. 63 / 649,480, filed on May 20, 2024, which is incorporated herein by reference for all purposes as if fully set forth herein. FIELD OF THE INVENTION The invention relates to matching power and energy resources with power and energy demand for charging electric vehicles (EVs) of an EV fleet, and in particular to designing and operating charging networks on the basis of a linear model. BACKGROUND As we move toward a greener future, developing infrastructure for electric vehicle (EV) charging becomes increasingly important. Fleet operators and owners are considering the pros and cons of switching their vehicles to electrical. These include operators of last-mile delivery vans, public buses, school buses, garbage collection trucks, transport trucks, rent-a-car operators, and more. Typically, their vehicles return daily to base, i.e., a charging center, weekends possibly excepted, after having completed their route. Until the EV leaves the center, when the daily cycle starts anew, the vehicle is to be parked at its designated parking spot, connected to its designated EVSE (electric vehicle supply equipment) connector, and is to be charged up to its required SOC (state of charge). All this is to be done as much as possible under the control of an automated charging plan, with no human intervention (the EV driver is often off-duty during the vehicle’s parking, and she or her replacement may return only when the vehicle’s is ready to leave parking). Whether planning or day-to-day operating an EV charging center, fleet owners face a number of questions, many of which are unfamiliar to them. Questions include the number of EVSE connectors needed, which is often the maximum number of vehicles that are simultaneously parked at the center at any one time; the number and type (AC vs. DC) of EVSE chargers needed, along with respective sizes and numbers of connectors; transformer size; and sourcing, sizing and daily / seasonal availability of renewable energy, including for mitigating use of non-renewable energy from the grid. Additional owner uncertainty surrounds deployment of a BESS (battery energy storage system), i.e., an on-site large battery, not belonging to an EV. A BESS is known to be useful, for example, for regulating solar output. If EV charging takes place overnight, solar output goes to waste if it cannot be stored. A BESS is also known to be useful in bridging over high-tariff hours, using energy stored in low-tariff hours. If a decision is made to deploy a BESS, which is typically very expensive, only part of its capacity can be used. This is because a 100% duty cycle, using 100% DOD (depth of discharge) brings on faster degradation of the battery, potentially limiting battery life to only a few years. Limiting duty cycles to, say, 60% DOD, means a larger BESS, but it will last much longer. Putting this question into a concrete calculation requires adopting one of several competing battery degradation models promoted in the art and academic literature. EV charging curves show the dependence of a battery’s maximum charging power on its state of charge. It is wrong to assume that the charging curve is linear, i.e., that a fixed current charges the vehicle’s SOC proportionally to the time it is applied. The maximum EV battery current (or, equivalently, power) is typically maximal at low SOC and drops sharply near 70%-80%. Charging a vehicle to 100% might need more time than the vehicle is parked. Furthermore, no generalization safely applies here: Each EV make has its particular charging curve. Ignoring it risks either charging an EV to the wrong SOC, or miscalculating the total power used by the center. These, and other questions, are not the sort of questions that can be handled by rules of thumb, or by a back-of-the-envelope calculation. Fleet owners face a complex decision with many variables needing optimization, and the design’s optimality can be verified by nothing less than a years-forward simulation. SUMMARY OF THE INVENTION According to embodiments of the invention, an EV charging network is adapted for charging EVs of an EV fleet. Each EV has a respective predicted charging-demand function such that the predicted charging-demand functions of the EV fleet collectively define predicted energy and / or power requirements of the fleet. The charging network is operative to receive electric power from a plurality of electricity providers each associated with a respective predicted electricity-provider availability and collectively defining electricity-provider constraints. The EV charging network comprises: (a) an EV- charging hub comprising respective producer-facing and consumer-facing interfaces, the producer-facing interfaces arranged to receive electric power from the electricity providers, the consumer-facing interfaces comprising a heterogeneous array of EV charging ports; and (b) an EV charging-network controller configured to performing all of first, second and third charging-network regulation operations for the EV charging-hub so as to maximize a value function constrained by the EV-fleet energy-requirements and by the electrical-producer-constraints. The first charging-network regulation operation includes instructing, at various times, on pairing and de-pairing individual EVs and individual charging ports of the heterogeneous array of charging ports. The second charging-network regulation operation includes regulating and / or requesting incoming electrical power into the charging hub from each of the electrical providers at absolute and relative magnitudes which fluctuate in time. The third charging-network regulation operation includes regulating respective magnitudes of power delivered via each charging port. In some embodiments, it can be that the EV charging-network controller performs all of the first, second and third charging-network regulation operations in a manner which weighs against each other the respective predicted charging-demand functions for the EVs of the EV fleet, and the respective predicted electricity-provider availability, over time, of the electrical producers of the plurality of electrical-producers. In some such embodiments, it can be that the performing is in a linear model with no binary or integer variables, all of the respective predicted charging-demand functions for the EVs of the EV fleet; and the respective predicted electricity-provider availability, over time, of each electrical producer of the plurality of electrical-producers. In some embodiments, the weighing against each other can be simultaneous and integrative. In some embodiments, it can be that (i) each electricity provider is characterized by a respective renewable content indicator, and / or (ii) maximizing the value function includes prioritizing renewable energy at the expense of non-renewable energy. A method is disclosed, according to embodiments, for operating a fleet of electric vehicles (EVs). The method comprises: at a first time, (a) accessing, for each EV of the fleet, a respective predicted charging-demand function for the operating period, the respective predicted charging-demand functions of the fleet collectively defining predicted energy and / or power requirements of the fleet for the operating period; (b) identifying a set of electricity providers operative to provide electricity in accordance with a set of respective electricity-provider parameters, each set of parameters including at least a respective predicted electricity-provider availability parameter and at least one of a financial parameter and an environmental parameter; and (c) designating an array of charging ports characterized by at least indirect electrical communication with at least one electricity provider of the set of electricity providers, wherein the designating includes designating at least one of available power and time-related availability at each charging port. The method further comprises, at a second time: (d) assigning charging ports to chargers. The method further comprises, at a third time during the operating period: (e) regulating pairing and de-pairing of individual EVs and individual charging ports; and (f) regulating respective magnitudes of electrical power transmitted to individual charging ports. In some embodiments, it can be that the first and second times are before an operating period, and that the third time is during the operating period. In some embodiments, it can me that the first time is before an operating period, and that the second and third times are during the operating period. A method is disclosed, according to embodiments, for operating an EV charging network adapted for charging EVs of an EV fleet. According to the method, the EV charging network comprises (i) an EV-charging hub comprising respective producer- facing and consumer-facing interfaces, the producer-facing interfaces arranged to receive electric power from a plurality of electricity providers, the consumer-facing interfaces comprising a heterogeneous array of EV charging ports, and (ii) an EV charging-network controller configured to regulate operations of the EV charging-hub. Each EV of the EV fleet has a respective predicted charging-demand function such that the predicted charging-demand functions of the EV fleet collectively defines the set of predicted energy and / or power requirements of the fleet. Respective predicted electricity-provider availability of the plurality of electricity providers collectively defines the set of electricity-provider constraints. The method comprises: (a) instructing, at various times, on pairing and de-pairing individual EVs and individual charging ports of the heterogeneous array of charging ports; (b) regulating and / or requesting incoming electrical power into the charging hub from each of the electrical providers at absolute and relative magnitudes which fluctuate in time; and (c) regulating respective magnitudes of power delivered via each charging port. Performance of the method is effective to maximize a value function constrained by the set of EV-fleet energy and / or power requirements and by the set of electrical-producer-constraints. In some embodiments, it can be that each electricity provider is characterized by a respective renewable content indicator, and / or that maximizing the value function includes prioritizing renewable energy at the expense of non-renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS The invention will now be described further, by way of example, with reference to the accompanying drawings, in which the dimensions of components and features shown in the figures are chosen for convenience and clarity of presentation and not necessarily to scale. In the drawings: Fig.1 is a schematic block diagram of an EV charging network and environs, according to embodiments of the present invention. Fig.2 shows a flowchart of a method for operating a fleet of electric vehicles, according to embodiments of the present invention. Fig.3 shows a flowchart of a method for operating an EV charging network adapted for charging EVs of an EV fleet, according to embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS The invention is herein described, by way of example only, with reference to the accompanying drawings. With specific reference now to the drawings in detail, it is stressed that the particulars shown are by way of example and for purposes of illustrative discussion of the preferred embodiments of the present invention only, and are presented in the cause of providing what is believed to be the most useful and readily understood description of the principles and conceptual aspects of the invention. In embodiments, an EV charging network for charging an EV fleet 50, for example the EV charging network 100 shown schematically in Fig. 1, comprises a plurality of EVs 55. Each EV 55 has a respective individual EV-specific predicted- charging demand-function. The predicted-charging demand-functions collectively define EV-fleet power requirements. The EV charging network includes and / or is operative with a plurality of electricity providers 35 (e.g., producers) which collectively define electricity provider constraints such that each electricity provider 35 is associated with a respective individual provider-specific predicted availability. Each electricity provider 35 is characterized by an individual provider-specific renewable / non-renewal status. In embodiments, the EV charging network 100 comprises an EV charging-hub 40 having producer-facing and consumer-facing interfaces. The producer-facing interface receives electrical power into the EV charging hub from each electricity provider 35 such that over time the relative magnitudes of electric power inflow provider change relative to each other. The consumer-facing interface comprises a heterogeneous array of EV charge ports 45 configured for EV charging. In embodiments, the EV charging network 100 also includes an EV-charging network controller 150 configured to perform first, second and third charging network regulation operations for the EV charging hub 40 in order to maximize a value function. An example of a value function is one that maximizes renewal usage at the expense of non-renewal usage for incoming electricity. Other examples of value functions are those which include a beneficial combination of environmental and financial parameters. The maximizing of the value function is constrained by the EV fleet power requirements and by the electricity-provider constraints. As shown in Fig. 1, in some embodiments, the EV-charging network 100 can include dispatchable storage devices such as battery energy storage system (BESS) 105. In some embodiments, the EV-charging network 100 can include dedicated and / or co- owned renewable energy sources such as the photovoltaic (PV) system 109 of Fig. 1. Fig. 1 shows the use of numerous protocols and standards related to electrical transmission and / or data transmission. The protocols and standards are non-limiting examples shown for enablement and in some implementations other protocols or standards can be used. These include: LeMS 130 - Local energy management system, for example, a device that can be programmed with different rules on how to route energy. OCPI - Open Charge Point Interface, a communication standard for 3rd party and roaming platforms to communicate. In exemplary applications, OCPI is used to communicate to the CMS and transmit OCPP compliant smart charging profiles to the charger. API - Application Programming Interface, in the case of fleet vehicles that have telemetry devices reporting data to a back office. CMS - Content management system for the EVSE (the electric vehicle charger). In exemplary implementations, a CMS is employed to communicate and control the charge point. Modbus - is a de facto industry-standard communication protocol that has been used in the electrical industry for over 45 years OCPP - Open Charge Point Protocol, is a standard commonly used by charger manufacturers and CMS providers to connect to and control the chargers. Modbus TCP is used to communicate with and control the BESS 105 and an inverter (not shown) associated with the PV system 109. DNP3 and IEC61850 are less communication protocols sometimes used for BESS systems. IEC61850 is based on Modbus TCP SunSpec. The first charging-network regulation operation includes instructing individual EV-charge port pairing and de-pairing at various times between individual EVs and EV charging ports. The second charging-network regulation operation includes regulating and / or requesting incoming electric power into the EV charging hub from each electricity providers at absolute and relative magnitudes which fluctuate in time. The third charging- network regulation operation includes regulating respective magnitudes of power delivered via each EV charge-port of the heterogeneous array of EV-charge-ports. In some embodiments, the EV charging-network controller performs all of the first, second and third charging-network regulation operations in a manner which simultaneously and integratively weighs against each other, in a linear model with no binary or integer variables, all of (i) the individual and respective individual predicted charging demand functions for each of the EVs; and (ii) the respective predicted availabilities, over time, for each of the electricity providers. A method is disclosed herein to optimize the operation of an EV charging center in two stages: (i) A scheduling stage assigns EV’s to time-slots and charging connectors, producing a load (in kW) to be served in each time-slot. (ii) A dispatch stage optimizes the way this required load is satisfied, determining energy flows to and from all available energy resources (grid, main battery, renewable sources). Optimizing in stages is inherently sub-optimal, as it necessarily holds many variables constant while it optimizes others, but an integrated optimization of all stages seems difficult, and has not been described in the art. Moreover, the number of variables to be optimized together is large, and is probably impractical for known optimization methods. An exception is linear programming (LP), which is scalable to millions of variables, and especially so in pure linear programming, without binary or integer variables. Furthermore LPs are solvable absolutely and precisely by many off-the-shelf tools. In embodiments, an optimized plan for the charging center starts with formulating a detailed model for it. This model will be simulated years ahead during planning, or optimized a day or two ahead, with real-time changes, to optimize the day-to-day operation of the center. The optimized solution of the model must determine two main aspects of the center’s operation: 1. Scheduling - A charging plan, detailed by vehicle and time-slot, specifying the energy provided to each electrical vehicle, and summing to the energy needed to bring it to the target SOC, so that it is able to carry out its route on leaving parking. 2. Dispatch - the energy provision plan, detailing how the energy needed by the EV’s is provided, and how it is divided among all the energy assets at the center. These plans are detailed down to time-slots, typically a quarter of an hour, and spanning the lifetime of the center, possibly lasting decades. Furthermore, the coordinated plan for scheduling and dispatch must be optimal, in the sense defined by the fleet owner. The modeling advisor is faced with simultaneous optimization of many decision variables, i.e., variables whose value is determined by optimization. Even for a single day, this is a large optimization problem. Its size is dominated by the size of the charging plan for the vehicles, whose number is marked by m. This should be multiplied by the number of time slots, mark as n, to m · n decision variables. Using a time slot resolution of 15 minutes, n = 96 in a day, and if, say, there are m = 50 vehicles, m· n = 4800. Consideration of EV charging curves may require SOC values to be determined per time slot, doubling the number of decision variables to 2mn. Other decision variables are energy inflows or outflows to each energy asset: The grid, the BESS, solar, wind energy assets, and a sundry other variables may inflate this number by an additional 20% − 30%, reaching > 104decision variables. If optimization is performed for a timespan of months, or for a full year, one easily surpasses a million (106) decision variables, with a million constraints. Obstacles and Solutions The Non-Linearity of Energy Dispatch For millions of decision variables, and even for much less, non-linear optimization methods do not scale well. The alternative, linear optimization, better known as linear programming (LP), requires a linear model. Energy dispatch, sometimes called Economic Dispatch, is a well-studied problem, and is often modeled as a non-linear problem. However, non-linear terms turn out not to be relevant to our problem, at least in part because renewable energy sources are modeled as fully owned by the center’s owner, so that their operational costs, including any non- linear elements, are irrelevant to their owner, and because the only energy source that is not owned, the grid (owned by the utility), has costs that are phrased as formal tariffs. These tariffs may have several components, each, in general specified as a time series (the per kWh cost as a function of time of use). These are not inherently non-linear. While utility tariffs have great variety, and sometimes great complexity, this complexity is unrelated to the economic dispatch problem. The Discrete Nature of Charger Assignment A different obstacle is the apparent discreteness of the question of chargers: EVs are modeled as connected to one of a charger’s connectors throughout its parking period, and to receive charging current according to an optimized charging plan. In embodiments, it can be desirable to ignore the discreteness of chargers in the optimization, treating them as a continuous resource. The cost attributed to chargers is modeled as proportional to their maximum power output. Under this assumption, it does not matter which chargers are employed in the center, and how many of each are employed. Market realities, and buyers power, dictate that charger cost will be broadly proportional to their main benefit, which is their maximum power. (An exception is the distinction between AC and DC outputs. AC- output chargers carry much lower costs than DC-output ones, but fit only AC-input EV’s. The solution, to separate between EV’s according to their AC / DC input type, is not detailed here, for brevity.) In some embodiments, assigning actual chargers is left to a post-optimization procedure or a second operation. Too Many Decision Variables or Constraints – Even with a pure linear model, the model that is appropriate for the intended center may be too large to be solved over the entire span of time (years, possibly decades) that is to be simulated during planning. To handle this possible complexity overflow, a flexible model is presented herein that may be segmented into any number of daily cycles, starting at 1, to be optimized at a time. Our model is “almost-disjoint” among different daily cycles, in the sense that the optimization of a daily cycle does not depend on the optimization of another daily cycle, whether prior or subsequent, except for one attribute: The battery (BESS) state. The battery capacity must be continuous for a feasible solution, so that the beginning battery capacity of one daily cycle equals the ending battery capacity of the previous daily cycle. Any long-term optimization may then be segmented into “batches” of days, or periods, to limit the number of decision variables & constraints into a manageable one. The implementor should insure battery continuity across these periods. This sacrifices some accuracy of the overall optimization, but, if the batches are long (the implementor should experiment with the batch size to discover what long means for the specific context), the impact should be minor to negligible. (The battery capacity in February should not matter much in June, for example). The Linear Model The following paragraphs describe here a linear model of an EV charging center, which may be solved (i.e., optimized under constraints) by an off-the-shelf LP solver such as for example the COIN-OR Foundation’s open source and free-license solvers, IBM’s Cplex, and Gurobi’s professional solvers. The model described does not need any binary or integer decision variables. The model, whether run for the entire expected lifetime of the center, or whether separated into “batch” periods individually optimized, produces a complete charging plan for all EV’s of the center’s fleet, together with a power dispatch plan, detailing the flow of energy to and from each of the energy assets used in the center. This plan is produced for each daily cycle in the center’s expected lifetime. Additionally, the model may be set up for sizing, i.e., determining the optimal size of each or all of the center’s energy assets. Chargers are modeled as available in continuous maximum power output, costing proportionally according to their maximum power. A post-optimization procedure, not described in this paper, pairing EV’s with actual chargers available on the market, is necessary for an implementation with physical chargers. This step is needed only for planning runs, as in operation runs, it is known which chargers the center uses. The model may be run, using real-time data, whether reported or measured by sensory equipment, for the next day, or next several days, to generate an operation plan of the center for the upcoming day or days. This plan may be manually executed, or fed to a computer system automatically controlling the EVSE and energy assets of the center. Parameters The following is a list of parameters of the model, meaning values that are constants of the model, and are not decision variables. The charging center works in daily cycles, a 24-hour period during which all vehicles “return to base”, i.e., to the center, each at its own fixed daily time, and departs some fixed hours later. The arrival and departure times are specified by fi, tifor each vehicle i. The daily cycle is, by default, midnight-to-midnight, but, if any vehicle is parked across midnight, the starting hour may be moved to a different hour of charging inactivity, say, 09:00 to the following 09:00. The daily cycle occurs within a weekly cycle comprised of activity days, and, possibly, “off” days, typically the weekend. The center has no charging activity during “off” days, but other energy components may continue to be active. For example, the solar panels may be generating energy, the battery may be charging, and energy may be sold to the grid. The largest arrays in the model are matrices of size (m + 1) ×n. m is the number of EV’s, indexed by i, i = 1, . . . , m. To this is added the battery (BESS), in its role as a consumer with i = 0, for a total of m + 1 energy consumers. n is the number of time slots in the optimization period. The optimization period ranges from a single daily cycle, g = 1, to an entire year g = 365 or g = 366 for a leap year. ∆t is the length of the time slot in hours. It is typically 1 / 4 of an hour (15 minutes), since most utilities measure consumption in 15-minute resolution. Alternatively, it may be 1 hour (60 minutes), since most utility tariffs vary at an hourly rate, at most. At 15-minute resolution, there are h = 96 time slots in a daily cycle, and gh = 96g = n time slots in our optimization period. This is the second dimension in our (m + 1) × narrays. ^ ≃ 35000 for an entire year.The presence matrix {ai,j, i = 0, . . . , m, j = 0, . . . , n − 1} plays an important part in our model: It specifies, during our g number of daily cycles optimized, the presence of each of the m + 1 energy consumers. i = 0 is for the battery (BESS), which is always present (a0,j= 1 for all j). For EVi, the i’th row of the a matrix specifies the exact parking times of the vehicle, in each working day, and marks time slots of “off” days with ai,j= 0. Table 1. General Parameters Notation meaning

[0002] ^^^ consumer (0 = battery, other = vehicle) ^ ^ ti Table 2. Energy A ssets Parameters Notation meaning The set of energy providers parameters (Table 2) includes cj, our utility’s tariff for a kWh at time slot j. If our utility is willing to purchase excess electricity from us, its offered price is given by dj, for every time slot j. Otherwise, dj is set to 0. Here it can be assumes that tariffs are specified as a time series for an entire year. Tariff time series are usually specified in 1-hour resolution for an entire year. Here one may oversample this time series to match our resolution ∆t. Rjis another time series, detailing the energy output, for a year, of any renewable resources available, such as wind turbines (WT). Some preprocessing may have preceded the creation of this time series, using geographical data, summation for different resources, and oversampling to match our resolution ∆t. Table 3. Energy Assets Cost and Maintenance Parameters Notation meaning The parameters νmin, νmaxare battery percentages set to limit the battery cycles DOD. For example, νmin= 20, νmax= 80 limit DOD to 80% − 20% = 60%. All energy assets carry expenses. The expenses are divided into one-time expenses, which, while being made in advance, are attributable to the entire project period of T years. In addition, there are maintenance expenses, which are given in terms of annual costs, and should so be weighted. Table 4. Vehicle Parameters Notation meaning The EV parameters include, among others, parameters that enable handling the vehicles’ charging curves. For every vehicle, a convex hull of the charging curve is described by a list of line equations of the form y = ux + v, where x is the state of charge of the vehicle, in percentages of its maximal battery capacity, and y is the maximal charging power for SOC x. If Si,jis the EV’s SOC at time slot j, then the charging power is bounded by uSi,j+ v, for every segment of the convex hull. (Note that the charging curve is entirely below its convex hull, and the two may coincide, but never intersect). If a charging curve is not available, power is still limited by the vehicle battery’s maximum rated power Vi, independently of SOC. Decision Variables The decision variables (Table 5) are variables whose values are set by the solver to values that simultaneously produce an optimum, while fulfilling all the constraints. The model assumes that the charging center has a battery (i.e., a BESS). Its maximal capacity, in kWh, is given by B. At optimum, the value of B will be the optimal size of the battery. Note that it might result in B = 0, which means that, optimally, the center should not have a BESS. Such a model optimization can be designated as a “sizing run”, as it determines the optimal size of one or more system components. It is possible to override this sizing, by defining the model with a given battery size, turning B into a preset parameter, and taking it off the decision variable list. Similar treatment may be given to W , the maximal power of the grid connection (transformer), H the maximal power of the battery, and Z the total power of all chargers. They are “sized” in the current model, but may, alternatively, be turned into preset parameters. The rest of the decision variables can be essential for the model, and determine the charging plan for the EV’s, as well as the energy flow to and from all energy assets. These include Bj, the capacity of the battery at time slot j. B0 is typically set to bstart, a parameter (Table 2), equaling the last battery capacity from a previous optimization run, to insure battery continuity. Table 5. Decision Variables Notation meaning 2^amount of energy (kWh) discharged from the battery at time j Xjand Yjare the resulting energy charged or discharged, respectively, by the battery at time slot j. Only one of every such pair can be positive. Pi,j and Qi,j determine the amount of energy flow, in time slot j, to each of the consumers, including the battery, i = 0, or the EV’s, i = 1, . . . , m. Pi,j is the part of this energy drawn from the grid, while Qi,jis the part drawn not from the grid, i.e., from the battery, or from renewable energy sources. Si,j is a matrix of each EV’s SOC at every time slot. It is needed as the argument to the charging curve set of line equations (see Table 4). Note that for a vehicle i for which a charging curve is not specified, the variables Si,jare not needed. Objective Function The objective function describes that total cost attributable to the period which is to be optimized. “Cost” should be understood in a broad sense, as including money- equivalent expenses and income, as well as carbon emission and similar measures. They may also include penalties and incentives to guide the optimization result to have some desired property. A negative value signifies a profit. The solver therefore seeks to minimize the objective function. The exemplary objective function used in the model is: Minimize TC = GE - GP + Cex + Oex, where GE is the expense of buying from the grid, GP is profit from selling to the grid, Cexrepresents one-time expenses of assets, and Oexrepresents maintenance expenses of assets. The skilled artisan will understand that this is merely an illustrative example, and other value functions can be used. The terms are calculated as follows: ^ ^−1 5& = ∑∑ ^^^^,^⋅

[0003] Constraints The minimization of the objective function TC above is subject to linear constraints, that insure that the solution is feasible, i.e., fulfills the requirements for the operation of the charging center. All constraints must be simultaneously fulfilled. Their specification follows: Non-negativity All decision variables in Table 5 are greater than or equal to 0. Energy Provided to Vehicles These constraints verify that every vehicle receives exactly its energy requirements during each daily cycle. Note that the parameter Ei,k is set to 0 on “off”-days. :C+1<⋅ℎ−1 ∑^D^^,^ + 4^,^E = &^,C ∀^ ∈ G^H, C ∈ I0,… , ^ − 1K matches the degraded charged energy. ^+4 10,^ 0,^−;2^ = 0 ∀^ ∈ G0, ^ − 1H Xj≤ H · ∆t∀j∈[0, n − 1] 1] Battery Balance Across Periods This constraint sets the final and initial battery capacities to be equal. ^−1 ^−1 ^= ^ ^ − 1] Alternatively, the to preset parameters. This may be used to insure battery continuity. ^−1 ^−1 ∑2^ +^$ ^^ = ∑ 3^ +^#^^^=0 ^=0∀^ ∈ [0, ^ − 1] Battery Time-Dependent Capacity 10 = ^$ ^^Bj = Bj−1 + Xj − Yj ∀j ∈ [n − 1] Battery Capacity Limits This enforces DOD limits adopted. νmin· B ≤ Bj≤ νmax· B ∀j ∈ [0, n − 1] Sale to Grid Excess This constraint ensures that electricity sold to the grid is above consumer needs at each time slot. See the equation for GP in Section 2.3. ^ ∑≤ ^^ +;3^ ∀j ∈ [0, n − 1] provided to the battery or EV in a time slot does not exceed its power limit. Pij + Qij ≤ Vi · ∆t ∀i ∈ [0, m], ∀j ∈ [0, n − 1] Grid Power Limit This constraint ensures that total grid power provided to consumers does not exceed the grid connection limit in any time slot. ^ ≤= ⋅M ∀ ∈ [0, n − 1] Chargers does not exceed the total power of chargers at any “open hours” time slot. ^ ⋅^ Vehicles State of Charge Initial SOC is set to “arrival” SOC siat start of each day. Otherwise SOC is accumulated according to energy provided..^,^ = $^ ∀(j mod h) = 0.^^,+4^,^ = .^,^−1 + ^ ^,^0^⁄ 100otherwise Limits The charging power is below the vehicle’s convex hull of its charging curve. ^^,^ +4^,^ ≤ (*^,ℓ.^,^ +^^,ℓ) ⋅ M ∀ℓ ∈ *',ℓA a fleet 50 of electric vehicles 55. As shown in the flowchart of Fig. 2, the method includes at least the six steps S01, S02, S03, S04, S05 and S06. Steps S01, S02 and S03 are carried out at a first time, Step S04 is carried out at a second time, and Steps S05 and S06 are carried out at a third time. Step S01 includes accessing, for each EV 55 of the fleet 50, a respective predicted charging-demand function for an operating period, i.e., a future operating period. The respective predicted charging-demand functions of the fleet collectively defines predicted energy and / or power requirements of the fleet for the operating period. Step S02 includes identifying a set of electricity providers 35 operative to provide electricity in accordance with a set of respective electricity-provider parameters. Each set of parameters includes at least a respective predicted electricity-provider availability parameter and at least one of a financial parameter and an environmental parameter. Step S03 includes designating an array of charging ports 45 characterized by at least indirect electrical communication with at least one electricity provider of the set of electricity providers. The designating includes designating at least one of available power and time-related availability at each charging port 45. Step S04 includes assigning charging ports 45 to chargers. In some embodiments, Step S04 is optional. Step S05 includes regulating the pairing and de-pairing of individual EVs and individual charging ports 45. Step S06 includes regulating respective magnitudes of electrical power transmitted to individual charging ports 45. In some embodiments, the first and second times are before an operating period, and the third time is during the operating period. In some embodiments, the first time is before an operating period, and the second and third times are during the operating period. A method is disclosed, according to embodiments, for operating an EV charging network 100 adapted for charging EVs 55 of an EV fleet 50. The EV charging network 100 comprises an EV-charging hub 40 comprising respective producer-facing and consumer-facing interfaces. The producer-facing interfaces are arranged to receive electric power from a plurality of electricity providers 35, while the consumer-facing interfaces comprise a heterogeneous array of EV charging ports 45. The EV charging network 100 also comprises an EV charging-network controller 150 configured to regulate operations of the EV charging-hub 40. According to the method, each EV 55 of the EV fleet 50 has a respective predicted charging-demand function such that the predicted charging-demand functions of the EV fleet 50 collectively defines q set of predicted energy and / or power requirements of the fleet 50. Respective predicted electricity-provider availability of the plurality of electricity providers 35 collectively defines a set of electricity-provider constraints. As shown in the flowchart of Fig. 3, the method includes at least the three steps S11, S12 and S13. Step S11 includes instructing, at various times, on pairing and de-pairing individual EVs 55 and individual charging ports 45 of the heterogeneous array of charging ports 45. Step S12 includes regulating and / or requesting incoming electrical power into the charging hub 40 from each of the electrical providers 35 at absolute and relative magnitudes which fluctuate in time. Step S13 includes regulating respective magnitudes of power delivered via each charging port 45. Performance of the method is effective to maximize a value function constrained by the set of EV-fleet energy and / or power requirements and by the set of electrical- producer-constraints. In some embodiments, each electricity provider is characterized by a respective renewable content indicator, and maximizing the value function includes prioritizing renewable energy at the expense of non-renewable energy. Any steps of any of the methods disclosed herein can be carried out in any combination and in any order while remaining within the scope of the disclosed embodiments. The present invention has been described using detailed descriptions of embodiments thereof that are provided by way of example and are not intended to limit the scope of the invention. The described embodiments comprise different features, not all of which are required in all embodiments of the invention. Some embodiments of the present invention utilize only some of the features or possible combinations of the features. Variations of embodiments of the present invention that are described and embodiments of the present invention comprising different combinations of features noted in the described embodiments will occur to persons skilled in the art to which the invention pertains.

Claims

CLAIMS 1. A charging network adapted for charging EVs of an EV fleet, each EV having a respective predicted charging-demand function, the predicted charging-demand functions of the EV fleet collectively defining predicted energy and / or power requirements of the fleet, the charging network operative to receive electric power from a plurality of electricity providers each associated with a respective predicted electricity-provider availability and collectively defining electricity- provider constraints, the EV charging network comprising: a. an EV-charging hub comprising respective producer-facing and consumer- facing interfaces, the producer-facing interfaces arranged to receive electric power from the electricity providers, the consumer-facing interfaces comprising a heterogeneous array of EV charging ports; and b. an EV charging-network controller configured to performing all of first, second and third charging-network regulation operations for the EV charging-hub so as to maximize a value function constrained by the EV- fleet energy-requirements and by the electrical-producer-constraints, wherein i. the first charging-network regulation operation includes instructing, at various times, on pairing and de-pairing individual EVs and individual charging ports of the heterogeneous array of charging ports, ii. the second charging-network regulation operation includes regulating and / or requesting incoming electrical power into the charging hub from each of the electrical providers at absolute and relative magnitudes which fluctuate in time, and iii. the third charging-network regulation operation includes regulating respective magnitudes of power delivered via each charging port.

2. The charging network of claim 1, wherein the EV charging-network controller performs all of the first, second and third charging-network regulation operationsin a manner which weighs against each other the respective predicted charging- demand functions for the EVs of the EV fleet, and the respective predicted electricity-provider availability, over time, of the electrical producers of the plurality of electrical-producers.

3. The charging network of claim 2, wherein the performing is in a linear model with no binary or integer variables, all of the respective predicted charging-demand functions for the EVs of the EV fleet; and the respective predicted electricity- provider availability, over time, of each electrical producer of the plurality of electrical-producers.

4. The charging network of either one of claims 2 or 3, wherein the weighing against each other is simultaneous and integrative.

5. The charging network of any one of claims 1 to 4, wherein (i) each electricity provider is characterized by a respective renewable content indicator, and (ii) maximizing the value function includes prioritizing renewable energy at the expense of non-renewable energy.

6. A method of operating a fleet of electric vehicles (EVs), the method comprising, at a first time: a. accessing, for each EV of the fleet, a respective predicted charging- demand function for the operating period, the respective predicted charging-demand functions of the fleet collectively defining predicted energy and / or power requirements of the fleet for the operating period; b. identifying a set of electricity providers operative to provide electricity in accordance with a set of respective electricity-provider parameters, each set of parameters including at least a respective predicted electricity- provider availability parameter and at least one of a financial parameter and an environmental parameter; and c. designating an array of charging ports characterized by at least indirect electrical communication with at least one electricity provider of the set ofelectricity providers, wherein the designating includes designating at least one of available power and time-related availability at each charging port, the method further comprising, at a second time: d. assigning charging ports to chargers, and the method further comprising, at a third time during the operating period: e. regulating pairing and de-pairing of individual EVs and individual charging ports; and f. regulating respective magnitudes of electrical power transmitted to individual charging ports.

7. The method of claim 6, wherein the first and second times are before an operating period, and the third time is during the operating period.

8. The method of claim 6, wherein the first time is before an operating period, and the second and third times are during the operating period.

9. A method of operating an EV charging network adapted for charging EVs of an EV fleet, the EV charging network comprising: i. an EV-charging hub comprising respective producer-facing and consumer-facing interfaces, the producer-facing interfaces arranged to receive electric power from a plurality of electricity providers, the consumer-facing interfaces comprising a heterogeneous array of EV charging ports, and ii. an EV charging-network controller configured to regulate operations of the EV charging-hub, wherein: i. each EV of the EV fleet has a respective predicted charging-demand function such that the predicted charging-demand functions of theEV fleet collectively defines a set of predicted energy and / or power requirements of the fleet, and ii. respective predicted electricity-provider availability of the plurality of electricity providers collectively defines a set of electricity- provider constraints, the method comprising: a. instructing, at various times, on pairing and de-pairing individual EVs and individual charging ports of the heterogeneous array of charging ports; b. regulating and / or requesting incoming electrical power into the charging hub from each of the electrical providers at absolute and relative magnitudes which fluctuate in time; and c. regulating respective magnitudes of power delivered via each charging port, wherein performance of the method is effective to maximize a value function constrained by the set of EV-fleet energy and / or power requirements and by the set of electrical-producer-constraints.

10. The method of claim 9, wherein each electricity provider is characterized by a respective renewable content indicator, and maximizing the value function includes prioritizing renewable energy at the expense of non-renewable energy.

Images