WO2014135219A1

WO2014135219A1 - System and method for distributing electrical power

Info

Publication number: WO2014135219A1
Application number: PCT/EP2013/054767
Authority: WO
Inventors: Maja ETINSKI; Anett Schuelke
Original assignee: NEC Europe Ltd
Current assignee: NEC Europe Ltd
Priority date: 2013-03-08
Filing date: 2013-03-08
Publication date: 2014-09-12
Anticipated expiration: 2015-09-08
Also published as: EP2965273A1; JP6162263B2; JP2016515373A

Abstract

A system for distributing electrical power of a local power generation facility (1), preferably RES (Renewable Energy Sources) facility, among a group of competing consumers (3), comprises a first interface for receiving power availability information from said local power generation facility, a second interface for enabling consumers (3) to submit jobs to be executed, wherein each job relates to an appliance having a particular power profile, and wherein each job is assigned a deadline until which the job must finish, a queue (4) for buffering submitted jobs that still wait to be executed, and a scheduler that is configured to perform online scheduling of queued jobs by matching the respective appliance power profiles and job deadlines with power availability according to predefined rules, wherein a differentiation among the jobs of different consumers (3) is introduced depending on the consumers' (3) energy usage patterns.

Description

SYSTEM AND METHOD FOR DISTRIBUTING

ELECTRICAL POWER

The present invention relates to systems and methods for distributing electrical power of a local power generation facility, preferably RES (Renewable Energy Sources) facility, among a group of competing consumers.

Power systems have been undergoing substantial changes for more than a decade resulting in a wide use of concepts such as distributed generation (DG) and distributed energy resources (DER). In particular, distributed renewable generation (DRG) has become ubiquitous due to a demand for higher renewable penetration motivated by the limited amount of fossil fuels and a need to reduce green house gas emission. Governments used various incentives, such as feed-in tariffs guarantee return on investment through fixed rates per kWh, to encourage local renewable generation and increase of renewable energy sources (RES).

However, there are multiple technical, commercial and regulatory challenges to increased penetration of DG in general. DRG arises additional concerns due to the intermittent nature and the unpredictability of availability of RES such as solar and wind whose power output varies significantly depending on weather conditions. Various load balancing schemes based on time shifting of controllable loads as well as energy storage systems have been considered to circumvent these issues related to the integration of RES.

Though feed-in tariffs have been very successful measure to promote use of solar energy, recently the incentives changed starting favoring on-site consumption of locally generated energy over the feed-in into the power grid. Since the renewable energy is going to be consumed locally, there is a need for new load balancing schemes that address issues of local sharing of an intermittent resource avoiding the costly solution of comprising of energy storage means. For this, the consumption rules for sharing the renewable resources are open. While on the one hand, a high usage of the generation source is desired, the fluctuation in supply and the variations in demand are not matching. It is therefore an object of the present invention to improve and further develop a system and a method for distributing electrical power of the initially described types in such a way that renewable energy is shared in a fair manner while still achieving the main goal of load balancing, ensuring an efficient use of renewables.

In accordance with the invention, the aforementioned object is accomplished by a system comprising the features of claim 1. According to this claim, such a system comprises a first interface for receiving power availability information from said local power generation facility, a second interface for enabling consumers to submit jobs to be executed, wherein each job relates to an appliance having a particular power profile, and wherein each job is assigned a deadline until which the job must finish, a queue for buffering submitted jobs that still wait to be executed, and a scheduler that is configured to perform online scheduling of queued jobs by matching the respective appliance power profiles and job deadlines with power availability according to predefined rules, wherein a differentiation among the jobs of different consumers is introduced depending on the consumers' energy usage patterns.

Furthermore, the aforementioned object is accomplished by a method for distributing electrical power of a local power generation facility comprising the features of claim 7. According to this claim, such a method comprises

retrieving power availability information from said local power generation facility,

collecting from consumers jobs to be executed, wherein each job relates to an appliance having a particular power profile, and wherein each job is assigned a deadline until which the job must finish,

buffering collected jobs that still wait to be executed in a queue, and by a scheduler, performing online scheduling of queued jobs by matching the respective appliance power profiles and job deadlines with power availability according to predefined rules, wherein a differentiation among the jobs of different consumers is introduced depending on the consumers' energy usage patterns. According to the invention it has been recognized that load balancing, ensuring an efficient use of renewable energy, can be combined with a fair sharing of renewable energy by introducing a differentiation among different consumers depending on the consumers' energy usage patterns. Previously, load balancing has been explored, normally proposing an optimization based scheduling of controllable loads. This approach comprises of load and generation followed by an optimization under given constraints trying to maximize the use of renewables by time-shifting of loads. While this approach can be efficient for larger-scale systems, it is not suitable for the integration of local renewable energy production, e.g. at the level of a residential building, with highly fluctuating supply environment. Not only that accurate fine-grain generation forecast would be needed but also load would have to be known in advance for an optimization over a finite horizon. While it is not difficult to accurately predict load averaged over a larger community such as it is done in grid-level balancing, load prediction at residential building level would be much more challenging.

Embodiments of the invention are related to a load balancing mechanism for high RES utilization, which applies fairness-driven on-line scheduling of stochastic discrete controllable residential loads respecting the variation in local generation against the aggregated usage pattern of the consumers. In particular, embodiments of the invention are targeted to consumer groups with a dedicated grid scope: e.g. residential/ commercial buildings with involvement of the consumers in the production level (solar panels on roof, with local system connection), commercial business parks with wind park supply involving tenant participation models, community grids with consumer participation models. Competition among the users is solved through a cooperative online scheduling approach among all active users against variable supply profile. A benefit of the present invention is the increased local RES efficiency and the increased renewable penetration for the given system, while enabling a high load serving factor.

According to a preferred embodiment an appliance power profile is given as an ordered array of subtasks specified with their duration and power consumption. Subtasks can be selected with a higher degree of flexibility than entire jobs, thereby improving load balancing, which results in a more efficient use of locally generated energy. Furthermore, it is noted that the online scheduling approach in accordance with embodiments of the present invention considers the current generation and assumes that there will not be sudden abrupt drop over the course of the execution of the currently active subtasks. Even if there is a substantial decrease in generation, the negative impact of misprediction is not severe if subtasks are given with sufficiently fine granularity.

According to an embodiment the scheduler may be invoked upon termination of the execution of a job or a subtask of a job, or upon an increase in energy generated by said local power generation facility. In both cases, there are new available resources that might be sufficient to start jobs or subtasks of jobs waiting in the queue.

According to a preferred embodiment the scheduling of jobs waiting in the queue to be executed may be realized by the scheduler assigning each job a priority. The scheduler may then sort the jobs within said queue by their priority in decreasing order, and it may iterate through the queue in decreasing order for identifying those jobs for which the power consumption of the next subtask to be started is not higher than the available power.

Generally, pure load balancing may be realized by means of a scheduling function that calculates the priority for each job by relating the sum of subtask durations of that job that have not been executed to the time left until the deadline of that job.

On the other hand, the differentiation among consumers may be realized by a prioritization function that assigns queued jobs a priority depending on said consumers' past energy usage patterns. According to a preferred embodiment the prioritization function may perform a priority calculation by taking into consideration said consumers' energy consumption accumulated over a predefined past time period and/or consumers' job deadline flexibility within said predefined past time period. Alternatively or additionally to the above analysis of historic consumer behavior, it is also possible to consider consumers' current energy usage patterns. Different priorities result in different job waiting times, wherein the scheduler may implement a fairness concept through variability of wait times applying higher wait times (resulting from lower priorities) for bigger consumers of renewable energy.

In case a subtask of a job gets activated it may be provided that the available power for other jobs in the queue is decreased by the power consumption of the activated subtask. Further, the job may be removed from the wait queue since subtasks of a single job can not be executed in parallel. Still further, the start time of a job that guarantees that the deadline of that job will not be violated is updated as soon as a subtask of that job gets activated.

Depending on the size of the local power generation facility, it may be provided that some of non-controllable loads are served from the local generation by setting the deadlines so that there is no slack time. Such integration of non-controllable loads under sufficient renewable generation would allow, e.g., for the strongly demanded concept of zero energy buildings.

In order to ensure seamless operation it may be provided that energy from the grid is allocated to an active job in case there is not sufficient local power generation and a deadline specified for the job must be satisfied.

Embodiments of the present invention can be implemented in different scenarios. For instance, tenants of a residential building can invest into a rooftop-mounted photovoltaic (PV) system whose generation is used locally to serve heavy loads that are controllable such as washing machines, tumble dryers and dish washers. Other application of RES sharing in accordance with embodiments of the present invention might include group of commercial park tenants serviced by a park of wind power plants or charging of electrical vehicles served by local wind generation that can also provide energy over night. As will be appreciated by those skilled in the art, further application scenarios can benefit from the advantages of the present invention such that applications are not limited to the examples given above. There are several ways how to design and further develop the teaching of the present invention in an advantageous way. To this end it is to be referred to the patent claims subordinate to patent claims 1 and 7 on the one hand and to the following explanation of preferred embodiments of the invention by way of example, illustrated by the figure on the other hand. In connection with the explanation of the preferred embodiments of the invention by the aid of the figure, generally preferred embodiments and further developments of the teaching will be explained. In the drawing

Fig. 1 is a schematic view illustrating an application scenario of the present invention in connection with a residential PV system,

Fig. 2 is a flow diagram illustrating the basic concept of scheduling performed for the purpose of load balancing in a power sharing system according to an embodiment of the present invention,

Fig. 3 is a flow diagram illustrating various aspects of prioritized scheduling performed in a power sharing system according to an embodiment of the present invention,

Fig. 4 is a diagram illustrating simulation results for power generation and baseline load without applied load scheduling,

Fig. 5 is a diagram illustrating simulation results for power generation and baseline load with applied load balancing,

Fig. 6 is a diagram illustrating simulation results for power generation and baseline load with applied load balancing together with fairness-driven scheduling,

Fig. 7 is a diagram illustrating simulation results of local versus grid energy consumption, and Fig. 8 is a diagram illustrating average job waiting times obtained by simulation.

Fig. 1 schematically illustrates a preferred application scenario of embodiments of the present invention. The illustrated application scenario relates to a local power generation facility 1 realized as a rooftop-mounted PV system 2 comprising solar panels at the level of residential building. It is assumed that a total of six tenants of the residential building - user1 -user6 - investigated/participate in the system and are therefore permitted to submit power loads to the PV system 2 as competing consumers 3. Hereinafter, consumers 3 will be considered on a per-flat basis, which means that the term consumer can either refer to a single person (i.e. an individual like userl and user6), two persons (e.g. a couple like user2 and user5), or more than two persons (e.g. a family like user3 and user4).

In the following description scheduling terminology is adopted from computing systems. A power load that corresponds to a single use of an appliance will be referred to as a job. When submitting a job, the consumer 3 specifies an appliance and a hard deadline until which the job must finish. Furthermore, an active appliance will be referred to as a job execution. Since each appliance has a different power profile, the implemented power-sharing system assigns power data to the job at the job submission time when the appliance and its program are selected. It is assumed that job power signature is given as an ordered array of subtasks specified with their duration and power consumption. Each of the subtasks corresponds to different phase of the appliance use such as water heating or centrifugation and each subtask is considered to be a non-preemptive (i.e. not interrupted) unit of the job execution. Nevertheless, pauses between two subtasks are allowed.

Each job also contains information on the consumer 3 to which it belongs so that the scheduler can take into the account consumer-specific data when making scheduling decisions. For instance, as explained in detail below, according to an embodiment of the invention renewable energy consumption per consumer 3 over a specific time period, e.g. the current or a past accounting period, influences the job execution order. Moreover, it is possible to prioritize consumers 3 based on their shares in the installed PV system 2 or based on their deadline flexibility shown in past.

Fig. 2 is a flow diagram illustrating the basic concept of scheduling performed for the purpose of load balancing in a power sharing system according to an embodiment of the present invention. Upon a job arrival 201 , which may be submitted to the system by a consumer via a dedicated interface, the system, at 202, sets the deadline the user had specified for that job and calculates a guaranteed start time that ensures that the deadline will not be violated. Next, shown at 203, the system evaluates whether there is enough locally generated renewable energy to start the first subtask of the job. In case there is not enough locally generated renewable energy, shown at 204, the job is added to a wait queue 4. Otherwise, shown at 205, the first subtask of the job is started and the deadlines and the starting times that guarantee observance of the deadlines are adjusted accordingly.

In case of a subtask termination or upon an increase in power generation, as shown at 206, a scheduler is invoked for selection of new subtasks to be executed, as will be explained in more detail in connection with Fig. 3.

Fig. 3 is a flow diagram illustrating various aspects of prioritized scheduling performed in a power sharing system according to an embodiment of the present invention. More specifically, Fig. 3 illustrates an exemplary scheduling policy deployed for the process of the selection of subtasks of jobs to be run next once there is an increase in available resources that might be sufficient to start jobs waiting in the queue 4.

In the illustrated embodiment, when the scheduler is invoked, it uses the algorithm given below to select which jobs will be activated. When selecting a job from the wait queue 4, only its first subtask that has not been executed so far is considered for activation: Input: jobs from wait queue WQ, power generation, power consumption, historic per-consumer data on renewable energy usage

begin

ComputeJobPriorities(\NQ)

SortWQByPriorities(\NQ)

ComputeAvailablePower(generaWon,consump.\on)

for all job J from sorted WQ do

if PowerRequested{J)≤ AvailablePower then

StartNextSubtask(J)

ReduceAvailablePower(PowerRequested(J))

Re mo ve Job From WQ{J )

UpdateDeadlineGuaranteeStartTime(J)

end if

end for

end

The embodiment of the scheduling scheme illustrated in Fig. 3 gets as inputs current local power generation and consumption of active appliances, 301 , queued jobs, 302, and per-consumer data on previous renewable energy usage, 303. Then, as shown at 304, the jobs are assigned a priority and are sorted by their priority in decreasing order. The process of priority calculation will be described in detail below.

As shown at 305, the scheduler iterates through the queue trying to start each job what is possible if the power consumption of the next subtask of the job to be started is not higher than the available power, which is assumed to be difference between the current generation and consumption. If a job gets activated, first the available power for other jobs in the queue is decreased by the power consumption of the subtask that just started. Next, the job is removed from the wait queue since subtasks of a single job can not be executed in parallel. Finally, the job start time that guarantees that deadline will not be violated is updated, as shown at 306. The scheduler keeps track of these times so that all deadlines are respected. If a job does not get enough resources before to finish before this time, it will be activated at this moment even if that would mean that energy will be consumed from the grid.

Considering in a first step only load balancing issues, the following scheduling function for calculation of job priorities, which only depends on the job deadlines, may be employed:

P deadline (J) = ExecutionRemaining Time(J)/DeadlineRemaining Time(J) ( 1 ) where ExecutionRemainingTime(J) is the sum of subtask durations of the job J that have not been executed and DeadlineRemainingTime(J) is the time left until the deadline of the job J. Generally, consumers should be encouraged to give flexible deadlines in order to allow for higher load balancing capability.

The scheduling scheme described so far does not differentiate between different consumers. Besides appliance power profiles, only job deadlines affect the scheduling order. In order to achieve fairness, a differentiation among the jobs of different consumers is introduced in accordance with the present invention depending on the consumers' energy usage patterns. Specifically, scheduling could be influenced by per consumer renewable consumption over an accounting period, previous or a current one, and/or other consumer specific parameters such as user deadline flexibility shown in past.

For simplicity, in connection with the illustrated embodiment it is assumed that all participating consumers have equal shares in the respective local energy production system. Further, in the illustrated embodiment user priority is calculated depending on the previous consumption of renewable energy over the current accounting period. The accounting period can be set to one year, month or a week. After each accounting period the renewable energy counters are set to zero for the beginning of the next account period. A fairness-driven scheduling scheme may compute the job priorities as follows: Pri^'ority(J) = P deadline (J) * Pfair (J) (2) where P deadline (J) is the priority function used in the simple balancing scheme described above (see eq. 1 ) and Pfair (J) is defined as:

Pfair (J) = 1/[1 + UserRenewableConsumption(J)] (3) where UserRenewableConsumption(J) represents the renewable energy consumption of the consumer that submitted the job J. In this way, the consumers that consumed more energy over the current accounting period get lower priorities. Thus, fairness as seen in connection with embodiments of the present invention has an important role in user comfort. Consumers that request a reasonable share of the locally generated energy should not be penalized by consumers with higher loads. This is reflected through higher wait times for users submitting more load. Furthermore, in this way users are motivated to cooperate through loose deadlines since they know that fairness is considered in scheduling.

Figs. 4-7 are related to simulation results of a scheduling scheme in accordance with an embodiment of the present invention for a setup of 5 consumers with different load demands in load tasks, volume and time. The example assumes a business model, where the produced solar energy (e.g. joint solar panel on a residential building with N parties) is equally distributed between all users as much as possible. Besides the absolute sharing rule, the aim of load scheduling in a typical power grid scenario for micro-grids will be to avoid a loss of load. Within a grid connected scenario, the reduction of non-RES energy intake as well as the spillage of locally produced RES energy into the grid is the main scope.

The simulation setup is as follows:

- Number of users: 5

- Number of tasks per job: 1

- Job duration: between 5 and 15 min

- Job power consumption: between 100 and 600 W Average slack time (difference between the time until the deadline and the job duration): 100 min

Simulated time: 10 hours

Fig. 4 illustrates the solar power generation (solid line) in comparison to the baseline load (dashed line) without applying any scheduling mechanism, i.e. without introducing any load shifts. As can be easily seen, the efficiency of solar energy utilization is rather poor, which is expressed by the strong deviations between the two curves. For instance, in the early hours (in particular between 0 and 150 min) the solar power generation strongly stays behind the actual load. To meet this overshooting demand, energy has to be retrieved from the grid.

Figs. 5 and 6 illustrate the solar power generation (solid line) in comparison to the load (dashed line) when applying load balancing with (Fig. 6) or without (Fig. 5) fairness-driven component as described above. In both cases, compared with Fig. 4 both curves show a much better fitting, which indicates a significantly improved efficiency of local energy utilization. As can be obtained from Fig. 7, in both cases the solar/grid energy ratios are very similar, both in total as well as per user.

When applying the fairness-driven scheduling policy, the users' different usage patterns that can be obtained from Fig. 7, automatically reflect in different waiting time of scheduled loads of the different users. As can be seen in Fig. 8, high load demanders (e.g. user 1 ) are penalized by higher flexibility on their waiting times against low demanders, while still the requested energy is surely delivered as needed.

Many modifications and other embodiments of the invention set forth herein will come to mind the one skilled in the art to which the invention pertains having the benefit of the teachings presented in the foregoing description and the associated drawings. Therefore, it is to be understood that the invention is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended claims. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.

Claims

C l a i m s

1. System for distributing electrical power of a local power generation facility (1 ), preferably RES (Renewable Energy Sources) facility, among a group of competing consumers (3), comprising:

a first interface for receiving power availability information from said local power generation facility,

a second interface for enabling consumers (3) to submit jobs to be executed, wherein each job relates to an appliance having a particular power profile, and wherein each job is assigned a deadline until which the job must finish, a queue (4) for buffering submitted jobs that still wait to be executed, and a scheduler that is configured to perform online scheduling of queued jobs by matching the respective appliance power profiles and job deadlines with power availability according to predefined rules, wherein a differentiation among the jobs of different consumers (3) is introduced depending on the consumers' (3) energy usage patterns.

2. System according to claim 1 , wherein an appliance power profile is given as an ordered array of subtasks specified with their duration and power consumption.

3. System according to claim 1 or 2, wherein said scheduler is invoked upon termination of the execution of a job or a subtask of a job, or upon an increase in energy generated by said local power generation facility.

4. System according to any of claims 1 to 3, wherein said differentiation is realized by a prioritization function that is configured to assign queued jobs a priority.

5. System according to claim 4, wherein said prioritization function is configured to analyze said consumers' (3) past energy usage patterns by considering said consumers' (3) energy consumption accumulated over a predefined past time period and/or consumers' (3) job deadline flexibility within said predefined past time period.

6. System according to any of claims 1 to 5, wherein said local power generation facility is a photovoltaic system (2) or a wind power plant, preferably associated with a group of consumers (3) in a residential building or a group of commercial park tenants.

7. Method for distributing electrical power of a local power generation facility, preferably RES (Renewable Energy Sources) facility, among a group of competing consumers (3), in particular for execution by a system according to any of claims 1 to 6, comprising:

collecting from consumers (3) jobs to be executed, wherein each job relates to an appliance having a particular power profile, and wherein each job is assigned a deadline until which the job must finish,

buffering collected jobs that still wait to be executed in a queue (4), and by a scheduler, performing online scheduling of queued jobs by matching the respective appliance power profiles and job deadlines with power availability according to predefined rules, wherein a differentiation among the jobs of different consumers (3) is introduced depending on the consumers' (3) energy usage patterns.

8. Method according to claim 7, wherein an appliance power profile is given as an ordered array of subtasks specified with their duration and power consumption.

9. Method according to claim 7 or 8, wherein said scheduler is invoked upon termination of the execution of a job or a subtask of a job, or upon an increase in energy generated by said local power generation facility.

10. Method according to any of claims 7 to 9, wherein said scheduler assigns each job a priority.

1 1. Method according to claim 10, wherein said scheduler sorts the jobs within said queue (4) by their priority in decreasing order.

12. Method according to claim 10 or 1 1 , wherein said scheduler iterates through said queue (4) in decreasing order for identifying those jobs for which the power consumption of the next subtask to be started is not higher than the available power.

13. Method according to any of claims 10 to 12, wherein load balancing is realized by means of a scheduling function that calculates the priority for each job by relating the sum of subtask durations of that job that have not been executed to the time left until the deadline of that job.

14. Method according to any of claims 7 to 13, wherein said differentiation is realized by a prioritization function that assigns queued jobs a priority depending on said consumers' (3) past energy usage patterns

15. Method according to claim 14, wherein said prioritization function performs a priority calculation by taking into consideration said consumers' (3) energy consumption accumulated over a predefined past time period and/or consumers' (3) job deadline flexibility within said predefined past time period.

16. Method according to any of claims 7 to 15, wherein, in case a subtask of a job gets activated, the available power for other jobs in said queue (4) is decreased by the power consumption of said activated subtask.

17. Method according to any of claims 7 to 16, wherein a job is removed from said queue (4) as soon as a subtask of that job gets activated.

18. Method according to any of claims 7 to 17, wherein the start time of a job that guarantees that the deadline of that job will not be violated is updated as soon as a subtask of that job gets activated.

19. Method according to any of claims 7 to 18, wherein jobs with non- controllable loads are served from said local power generation by setting the job deadlines so that there is no slack time.

20. Method according to any of claims 7 to 19, wherein energy from the grid is allocated to an active job in case there is not sufficient local power generation and a deadline specified for the job must be satisfied.