WO2025181764A1

WO2025181764A1 - System for determining energy installation configurations for properties on a site

Info

Publication number: WO2025181764A1
Application number: PCT/IB2025/052214
Authority: WO
Inventors: Michael Evans
Original assignee: Octopus Energy Group Ltd
Current assignee: Octopus Energy Group Ltd
Priority date: 2024-03-01
Filing date: 2025-02-28
Publication date: 2025-09-04
Anticipated expiration: 2026-09-01
Also published as: WO2025181762A1; WO2025181766A1

Abstract

A computer-implemented method is disclosed for determining a configuration for an energy system including electricity generator and consumer assets associated with a plurality of properties on a site, the properties connected to a site electricity network coupled to an electricity distribution network, the method comprising: receiving specification data specifying characteristics of energy assets associated with the site, the energy assets comprising electricity generator assets and electricity consumer assets installed or intended for installation at the properties; and performing an evaluation process to evaluate the specification data. The evaluation process comprises simulating electricity generation and consumption by generator and consumer assets at each property in accordance with the specification data; determining flow data for each property indicating electricity drawn by the property from the site electricity network and/or supplied to the site electricity network by the property based on the simulated generation and consumption; determining, based on the flow data, grid flow data indicating electricity imported from the distribution grid to the site electricity network and/or electricity exported from the site electricity network to the distribution grid; and determining a utilisation metric based on the grid flow data. One or more modifications to the specification data are then determined in dependence on the utilisation metric obtained by the evaluation process; and suggested configuration data for one or more assets is output based on the one or more modifications.

Description

System for determining energy installation configurations for properties on a site

FIELD OF THE INVENTION

The present application relates to methods and systems for evaluating and optimising the configuration of energy installations.

BACKGROUND OF THE INVENTION

Trends towards reducing energy consumption and increasing reliance on renewable energy sources with a view to reaching “net zero” have been gathering pace in recent years. In domestic settings, consumers are encouraged to use private solar installations (typically solar panels on the roof) and/or heat pumps to reduce consumption of electricity and gas or other fuels. Excess energy generated in such systems can in principle be supplied back to the electricity distribution grid (e.g. to be offset against household energy costs) but this can place a burden on distribution grids, which were traditionally designed principally for energy flow to consumer households rather than reverse flow from the households.

Furthermore, these systems can be expensive and disruptive to install and the investment required is often only recouped over a long period of time. Even for new-build properties, these limitations may make installing the necessary solar generation capacity and heat pumps less attractive. Difficulties in determining the best system specifications (e.g. in terms of solar generation capacity, battery and heat pump models etc.) can exacerbate this. Underspecified systems may not fully realize the benefits of local generation, whilst excess generation capacity of over-specified systems may place a burden on the distribution grid.

To meet long term net zero goals, it would be beneficial to increase the reliance on local renewable energy sources, which will require new approaches to engineering and controlling home energy systems.

SUMMARY OF THE INVENTION

Aspects of the invention are set out in the independent claims. Certain preferred features are set out in the dependent claims. In a first aspect of the invention, there is provided a computer-implemented method for analysing the configuration of an energy installation, the energy installation being installed at or intended for installation at a first property connected to an energy distribution grid and comprising an energy generation facility arranged to supply energy to energy consumers at the first property, the method comprising: receiving specification data for the energy installation, the specification data including specification attributes indicating: an energy generation capacity of the energy generation facility; and one or more characteristics of the first property; accessing a database of energy consumption data for a plurality of further properties; selecting peer energy consumption data from the database for one or more properties similar to the first property, wherein similarity is determined based on the one or more characteristics of the first property; simulating operation of the energy installation over a simulation period based on the specified energy generation capacity and the selected peer energy consumption data for the similar properties, the simulation comprising determining energy flow data (or energy import/export data) specifying energy imported from the distribution grid or exported to the distribution grid over the simulation period; processing the energy flow data to determine a compliance indication indicating whether the energy installation complies with a predetermined efficiency criterion; and outputting the compliance indication.

The following features may be applied to any of the aspects of the invention set out herein.

The efficiency criterion (also referred to herein as a utilisation criterion or compliance criterion) preferably relates to energy exchange between the energy installation and the distribution grid. More particularly, the efficiency/utilisation/compliance criterion may indicate grid import and/or export constraints, e.g. limits on energy consumed from and/or returned to the distribution grid by an energy installation (or in later examples by a group of energy installations e.g. on a site). Thus, determining compliance with such a criterion may involve evaluating energy installation specifications against grid import/export constraints. Such constraints are preferably expressed in terms of energy (e.g. electricity) but in some examples may be expressed in cost terms as discussed below. Evaluating compliance with such constraints can allow an energy installation to be assessed in terms of how efficiently local generation capacity is being utilised (vs. utilisation of grid-supplied energy). The configuration of an energy installation can then be modified to achieve compliance, as discussed in more detail below, e.g. to optimise against the efficiency criterion (e.g. finding specifications for components of energy installations that result in the grid import/export constraints or limits being met).

Preferably, the simulating step simulates, at each of a series of simulation intervals, energy generated by the energy generation facility and energy consumed by one or more consumers at the property, wherein the energy consumed by the one or more consumers is estimated based on the selected energy consumption data for similar properties. The energy installation preferably includes an energy storage facility, the simulating step comprising simulating storage of energy to and/or release of energy from the energy storage facility over the simulation period (e.g. simulating charging/discharging a battery). The specification attributes may further indicate a storage capacity of the storage facility. Simulating operation may comprise, at each simulation interval, one or more of: determining electricity generated by a generator in dependence on a generation capacity specified in the specification; determining electricity consumed by an HVAC device in dependence on a HVAC consumption specified in the specification; determining electricity consumed or released by the battery asset in dependence on a battery capacity and a simulated battery control algorithm; determining electricity consumed by one or more further consumers at the property.

Preferably, processing the energy flow data comprises deriving from the energy flow data a compliance metric that is related to an amount of energy supplied to and/or consumed from the distribution grid over the simulation period as determined by the simulation, wherein the compliance indication is determined based on the compliance metric. The compliance metric may be based on a difference between a total amount of the energy supplied to the distribution grid and a total amount of energy consumed from the distribution grid over the simulation period or based on a cost or profit value related to the energy supplied and consumed to/from the grid over the simulation period. The method may comprise comparing the compliance metric to at least one threshold to obtain the compliance indication, optionally comprising determining that the energy installation complies with the efficiency criterion if the compliance metric meets the threshold or falls within a predetermined range. The compliance metric may be a measure of an amount of energy (e.g. net import/export of energy) or a measure of cost associated with the amount of energy (e.g. the net import/export). Thus, in some cases the compliance metric (and/or resulting compliance indication) may be determined also in dependence on energy price information. The compliance indication may be a binary classification (e.g. compliant or non-compliant) or may include additional categories (e.g. under-specified, compliant, over-specified).

Preferably, the method comprises determining that the energy installation complies with the efficiency criterion if the total net energy consumption from the distribution grid or an associated energy cost over the simulation period does not exceed a predetermined first threshold, optionally zero, and preferably if a total net energy supply to the distribution grid or an associated energy cost over the simulation period does not exceed a second threshold. Net energy consumption preferably refers to a difference between energy consumed and energy supplied (with net energy supply being the inverse).

Preferably, the database stores energy consumption traces for the further properties, each trace specifying a time series of energy consumption values indicating energy consumed by consumers at a respective property, the selecting step selecting a plurality of peer energy consumption traces from the database based on the characteristics of the first property. The method may comprise: repeating the simulating step for each selected peer energy consumption trace to generate energy flow data using the respective energy consumption trace and the energy generation capacity of the energy generation facility. Preferably, the method includes deriving an energy flow metric from the energy flow data obtained for each selected peer energy consumption trace, and aggregating over the energy flow metrics to obtain a compliance metric, wherein the compliance indication is determined based on the compliance metric. The energy flow metric may e.g. be a total net energy import/export value or associated cost value as described above for the compliance metric, but for the individual trace, with aggregation producing the overall compliance metric. The aggregation step may perform bootstrap sampling to obtain a mean energy flow metric (which may serve as the compliance metric or from which the compliance metric may be derived).

Preferably, the method comprises generating an estimated energy generation trace based on the specified energy generation capacity, the estimated energy generation trace comprising a time series of energy generation values over the simulation period, wherein the simulating step uses the estimated energy generation trace, optionally wherein the estimated energy generation trace is obtained by scaling a default trace based on the specified energy generation capacity.

The energy installation preferably further includes an HVAC (heating, ventilation and/or air conditioning) device, the specification attributes further specifying an energy consumption of the HVAC device and wherein the simulation further simulates energy consumption by the HVAC device in dependence on the specified energy consumption. The HVAC device may comprise a heat pump. The method may comprise generating an estimated HVAC consumption trace comprising a time series of energy consumption values for the HVAC device over the simulation period based on the specified HVAC device energy consumption, wherein the simulating step simulates energy consumption by the HVAC device based on the estimated HVAC consumption trace, the estimated HVAC consumption trace optionally obtained by scaling a default trace based on the specified HVAC device energy consumption.

The energy (or electricity) generation trace obtained for the generator may be determined (e.g. scaled) in dependence on location and/or weather data. The HVAC device consumption trace may be obtained (e.g. scaled) in dependence on one or more of: location data, weather data, a user behaviour modifier indicative of an expected increase in consumption due to user behaviour, and an efficiency modifier indicative of an expected increase in consumption due to heating and/or insulation inefficiency at the property.

Preferably, selecting the peer energy consumption data comprises selecting energy consumption data for properties having property characteristics matching the specified characteristics of the first property. Matching here may mean that the characteristics are identical or similar, e.g. based on some similarity criterion (e.g. numerically specified characteristics may need to match to within some tolerance e.g. a percentage or absolute difference between the matched characteristics to be considered a match). The characteristics of the property may be indicative of a size and/or type of the property, optionally specifying a floor area and/or a number of rooms.

The energy preferably comprises electrical energy (electricity). The energy generation facility preferably comprises an electricity generator, such as a solar photovoltaic (PV) generator (e.g. comprising one or more solar panels). The energy installation preferably comprises one or more of: a battery for storage of electrical energy, and an electrically powered heat pump.

In the examples throughout this disclosure, specification parameters used for evaluation and/or being optimised that relate to an electricity generation capacity (value) for a solar generator may refer to a capacity (e.g. power/wattage) of the whole generator, of one or more individual solar panels, or per unit area of the solar panel(s). In the latter cases additional parameter(s) may specify number of panels and/or a panel area of the individual panel(s) or a total (combined) panel area of the panels, from which total generation capacity (e.g. as a power/wattage value) can be derived. Panel area preferably refers to the effective area (surface area where PV generation occurs).

In a further aspect, which may be combined with any of the other aspects set out herein, the invention provides a computer-implemented method for analysing the configuration of an energy installation, the energy installation being installed at or intended for installation at a property connected to an electricity distribution grid and comprising an electricity generator, a battery and an electrically powered HVAC (heating, ventilation and/or air conditioning) device, the method comprising: receiving specification data for the energy installation, the specification data including attributes indicating: an electricity generation capacity of the electricity generator; and an electricity consumption of the HVAC device; simulating operation of the energy installation over a simulation period, including: determining electricity generated by the electricity generator based on the specified electricity generation capacity; determining electricity consumed by the HVAC device based on the specified electricity consumption for the HVAC device; determining estimated electricity consumed by one or more other energy consumers at the property; simulating charging and/or discharging of the battery; and computing, based on the determined electricity generated by the electricity generator, the determined electricity consumed by the HVAC device and other energy consumers and based on the simulated charging and discharging of the battery, import/export data specifying electricity imported from and/or exported to the electricity distribution grid; deriving from the import/export data a compliance metric; comparing the compliance metric against at least one threshold to determine compliance of the energy installation with an efficiency criterion; and outputting a compliance indication based on the comparison.

The following optional features may be used with any of the aspects set out herein.

Preferably, the simulating step comprises computing electricity flows between components of the energy installation and/or between the energy installation and the electricity distribution grid for each of a series of simulation intervals. Electricity consumed by one or more other energy consumers at the property may be estimated using peer consumption data as set out above, with respective sets of import/export data determined for each set of peer consumption data which are processed to obtain an aggregated compliance metric.

Simulating charging and/or discharging of the battery may comprise determining a battery control schedule for charging and discharging of the battery in dependence on one or more of: electricity generated by the electricity generator, electricity consumed by the HVAC device, electricity consumed by the one or more other consumers, a charge status of the battery, a capacity of the battery, and energy price information; and determining at a given simulation interval an amount of electricity consumed by charging of the battery or supplied by discharging of the battery in dependence on the battery control schedule.

Determining flow data or import/export data may comprise one or both of: determining a shortfall of electricity required from the distribution grid over an simulation period to supplement local generation by the electricity generator, and determining an excess of electricity generated locally by the electricity generator over the simulation period that is not consumed by consumers at the respective property and is available for supply back to the distribution grid.

The simulating may further comprises simulating operation of the HVAC device, preferably based on one or more of: heating set point data, and weather or temperature forecast data; and/or simulating participation of the energy installation in a flexibility service associated with the distribution grid. Simulation may involve simulating control operations according to an algorithm implemented by a controller associated with the energy installation for controlling the energy installation during operation.

The steps of simulating operation of the energy installation, deriving a compliance metric and determining compliance may be repeated for a plurality of specifications to identify at least one compliant specification that meets the efficiency criterion; the method then outputs specification data based on the at least one compliant specification.

In a further aspect (which may be combined with any of the other aspects set out herein), there is provided a system (and corresponding method) for evaluating compliance of energy installations with an efficiency criterion, the system comprising: a trained machine Learning model (e.g. an artificial neural network (ANN)), trained to generate, based on an input to the machine learning model, an output indicative of compliance of an energy installation with the efficiency criterion, wherein the energy installation includes an electricity generator and an electrically powered HVAC (heating, ventilation and/or air conditioning) device and wherein the efficiency criterion relates to electricity consumed from an electricity distribution grid and/or supplied to the electricity distribution grid by the energy installation; wherein the input to the model comprises a set of specification parameters for the energy installation including parameters indicating: a generation capacity of the electricity generator; and an electricity consumption of the HVAC device; the system further comprising an evaluation module arranged to: receive a set of specification parameter values for an energy installation to be evaluated; input the specification parameter values to the machine learning model; and output a compliance indication based on the output of the machine learning model.

The machine learning model may have been trained based on a plurality of training samples, each training sample associated with an energy installation and comprising the specification parameters for the energy installation and a compliance indicator generated based on simulating operation of the energy installation in accordance with the specification parameters over a simulation period. Asimulation module used fortraining the machine learning model by performing simulation as set out above may be provided.

In any of the above examples, the electricity generator may include one or more solar panels, and the specification parameters specify one or more of: an electricity generation capacity value, a panel size or area of the one or more solar panels, and a panel orientation. This may be used as inputs to the machine learning model and/or simulation. Other inputs may include attributes of a property where the energy installation is installed or is intended to be installed, optionally indicative of a size or type of the property. The machine learning model may be used to evaluate compliance of a plurality of different energy installation specifications (e.g. in a search/optimisation process as described elsewhere); and the method may identify one or more of the energy installation specifications classified as compliant based on the output of the machine learning model; and output specification data based on the one or more identified specifications. In a further aspect (which may be combined with any of the other aspects), the invention also provides a corresponding method for training a machine Learning model for evaluating compliance of an energy installation with an efficiency criterion, the method comprising: obtaining a plurality of training samples, wherein each training sample is representative of an energy installation including an electricity generator and an electrically powered HVAC (heating, ventilation and/or air conditioning) device, the training sample including specification parameters for the energy installation and a compliance value, the specification parameters including parameters indicating: a generation capacity of the electricity generator; and an electricity consumption of the HVAC device; wherein the compliance value for each training sample is generated based on simulating operation of the energy installation in accordance with the specification parameters over a simulation period; training the machine learning model using the plurality of training samples; and storing the trained machine learning model for use in evaluating specifications of energy installations.

In a further aspect (which may be combined with any of the other aspects set out herein), there is provided a computer-implemented method of determining a configuration for an energy installation for use in a property connected to an electricity distribution grid, the energy installation comprising a plurality of components including an electricity generator arranged to supply electricity to the property, a battery for storing electricity generated by the electricity generator, and a heating, ventilation and/or air conditioning (HVAC) device, wherein the method comprises: evaluating a plurality of energy installation specifications for compliance with a compliance criterion, wherein each specification includes specification attributes specifying performance characteristics of components of the energy installation and wherein evaluating a specification comprises: computing a compliance indicator based on the specification attributes, wherein the compliance indicator is dependent on flow of electricity between the energy installation and the distribution grid during operation of the energy installation in accordance with the specification attributes; and determining whetherthe specification is compliant based on the compliance indicator; selecting at least one energy installation specification identified as compliant in the evaluating step; and outputting configuration data for the energy installation based on the selected at least one energy installation specification.

The following optional features may be used with any of the aspects set out.

The specification attributes may include attributes indicative of one or more of: an electricity generation capacity of the generator, a storage capacity of the battery, and an electricity consumption of the HVAC device. Outputting configuration data may comprise outputting one or more of the selected specifications and/or one or more of the specification attributes of the selected specifications to a user. Outputting configuration data may comprise determining one or more constraints on one or more specification attributes indicating values of the attributes required for compliance and outputting the determined constraints. Outputting configuration data may comprise outputting one or more of: a suggested type, model and/or generator capacity of the electricity generator; a suggested number, size and/or area of solar panels for the electricity generator; a suggested consumption rating, type and/or model of HVAC device; a suggested capacity rating, type and/or model of battery. The outputting step may output one or more recommendations for modifying an existing or planned energy installation to achieve compliance with the compliance criterion based on the selected compliant specification(s).

The evaluation may be performed iteratively until a compliant specification is identified, with configuration data based on the identified compliant specification provided as output. The method may comprise: receiving an initial specification for the energy installation; wherein the evaluating step comprises evaluating a plurality of further specifications derived from the initial specification by repeating a process of: modifying at least one specification attribute of the initial specification, and evaluating the modified specification. This may involve evaluating a plurality of specifications with modified values of a selected specification attribute while keeping other specification attribute values fixed and then optionally iterating over the selected specification attribute to evaluate further specifications with modified values of further specification attributes.

Evaluating a specification may comprises simulating operation of the energy installation or using a machine learning model as set out elsewhere herein. In a further aspect of the invention (which may be combined with any of the other aspects set out), there is provided a computer-implemented method for determining a configuration for an energy system including electricity generator and consumer assets associated with a plurality of properties on a site (e.g. an existing or planned housing development), the properties connected to a site electricity network coupled to an electricity distribution network, the method comprising: receiving specification data specifying characteristics of energy assets associated with the site, the energy assets comprising electricity generator assets and electricity consumer assets installed or intended for installation at the properties (e.g. a solar photovoltaic PV generator, battery and HVAC at each property); performing an evaluation process to evaluate the specification data, the evaluation process comprising: simulating electricity generation and consumption by generator and consumer assets at each property in accordance with the specification data; determining flow data for each property indicating electricity drawn by the property from the site electricity network and/or supplied to the site electricity network by the property based on the simulated generation and consumption; determining, based on the flow data, grid flow data indicating electricity imported from the distribution grid to the site electricity network and/or electricity exported from the site electricity network to the distribution grid; and determining a utilisation metric based on the grid flow data; determining one or more modifications to the specification data in dependence on the utilisation metric obtained by the evaluation process; and outputting suggested configuration data for one or more assets based on the one or more modifications.

The following optional features may be used with any of the aspects set out.

The specification data may further specify characteristics of one or more shared energy assets connected to the site electricity network, e.g. comprising one or more of: a communal generator asset configured to provide electricity to multiple properties of the site; a communal battery for storing electricity generated by generator assets at the properties and/or by a communal generator asset; a communal HVAC system powered by electricity drawn from the site electricity network, optionally a district heating system. The evaluation process preferably further includes simulating operation of the shared energy assets to determine further flow data indicating electricity drawn by the shared energy assets from the site electricity network and/or supplied to the site electricity network; wherein the grid flow data is further determined in dependence on the further flow data. Generating one or more modifications may comprise modifying attributes of the specification data specifying: characteristics of one or more of the generator and/or consumer assets associated with properties; and/or characteristics of one or more of the shared energy assets.

The method may comprise evaluating the utilisation metric against a predetermined compliance criterion (with the modifications to the specification data determined in response the utilisation metric not meeting the compliance criterion). The utilisation metric may be compared to a predetermined compliance threshold or range, optionally wherein the utilisation metric is determined to meet the compliance criterion if it meets the threshold or falls within the range. The utilisation metric may be as previously set out except applied collectively to the properties / energy installations on the site, e.g. being based on one or more of: a total electricity export to the distribution network over a simulation period; a total electricity import from the distribution network over the simulation period; a difference between the total electricity export and total electricity import. More specifically, the utilisation metric may comprise a net energy import/export value (or an estimated cost or profit value associated with the total energy export and/or import, which may be computed based on time-varying energy cost information over the simulation period).

The evaluation process may be repeated using the modified specification data. These steps may be repeated until the utilisation metric for the modified specification data meets the compliance criterion (e.g. using an iterative search/optimisation as set out elsewhere) with the process outputting configuration data indicating a proposed configuration for one or more assets based on the final (compliant) modified specification data.

Outputting configuration data may comprise outputting one or more proposed specification attribute values for one or more assets and/or determining and outputting one or more constraints on one or more specification attributes indicating values of the attributes required for compliance and outputting the determined constraints. The output may comprise one or more of: a suggested type, model and/or generator capacity of an electricity generator for a property or of a shared electricity generator; a suggested number and/or area of solar panels for an electricity generator for a property or for a shared electricity generator; a suggested consumption rating, type and/or model of a HVAC device for a property of for a shared HVAC system; a suggested capacity rating, type and/or model of battery for a property or for a shared battery. In a further aspect of the invention (which may be combined with any of the other aspects set out herein), there is provided a method of determining configurations for energy installations, comprising: receiving specification data specifying configurations for a plurality of energy installations, wherein each energy installation is associated with a respective property connected to an electricity distribution grid and comprises energy assets including: a solar photovoltaic (PV) generator arranged to supply electricity to one or more electricity consumers at the respective property, a battery and an electrically powered HVAC (heating, ventilation and/or air conditioning) device, wherein the specification data includes specification attributes for each energy installation indicative of an electricity generation capacity of the solar PV generator, a battery capacity of a battery and an electricity consumption of the HVAC device, and wherein the electricity generation capacities indicated by the specification data for two or more of the energy installations differ due to different arrangements of solar panels used by the respective solar PV generators; performing a search process to identify configurations for assets of the energy installations, the search process comprising iteratively modifying the specification data for the energy installations and evaluating the modified specification data until the specification data meets a compliance criterion, comprising, at each iteration: for each energy installation, simulating, based on the modified specification data, electricity generated by the solar PV generator, electricity consumed by the HVAC device and charging and discharging of the battery over a simulation period; based on the simulation, determining, for each energy installation, grid flow data indicating one or both of: a shortfall of electricity required from the distribution grid over the simulation period to supplement local generation by the respective solar PV generator, and an excess of electricity generated locally by the respective solar PV generator over the simulation period that is not consumed by consumers at the respective property and is available for supply back to the distribution grid; and determining whether the modified specification data meets the compliance criterion in dependence on the grid flow data; wherein the iterative evaluation keeps attributes of the specification data indicating the solar generation capacity of each energy installation fixed while modifying specification attributes indicating performance characteristics of other assets; and outputting configuration information for the energy installations based on the modified specification data identified as meeting the compliance criterion by the iterative evaluation.

The following optional features may be used with any of the aspects set out.

In any of the aspects herein, specification attributes for an energy installation indicative of an electricity generation capacity of a solar PV generator may comprise one or more of: an electricity generation capacity value, a number of solar panels, a type of solar panels, a size or effective generation area of solar panels, and an orientation of the solar panels, wherein two or more of the energy installations differ in at least one of said attributes. The two or more energy installations preferably differ in total solar panel effective area (surface area where PV generation occurs) and/or orientation. Modifying the specification data for the energy installations may comprise modifying one or more of: a battery capacity of a battery of at least one energy installation, and an electricity consumption of an HVAC device of at least one energy installation.

Determining whether the specification data meets the compliance criterion may be based on aggregating the grid flow data to determine aggregate electricity flow between the plurality of energy installations and the distribution grid over the simulation period; determining a utilisation metric based on the aggregated grid flow data (e.g. based on a total net export and/or total net export), and determining compliance based on comparing the utilisation metric to a predetermined threshold or range. The compliance metric may be determined and evaluated as described elsewhere herein, aggregated over the energy installations. The search process preferably identifies changes to specifications of batteries and/or HVAC devices of one or more of the energy installations to compensate for different electricity generation capacities of the respective PV generators while maintaining compliance with the compliance criterion (e.g. grid import/export limits).

In a further aspect (which may be combined with any of the other aspects set out herein), the invention provides a computer-implemented method comprising: evaluating specifications for a plurality of energy installations, each energy installation provided at a respective property connected to an electricity distribution grid and comprising a local electricity generator for providing electricity to electricity consumers at the property and a battery for storing electricity generated by the electricity generator, the evaluating comprising, for each energy installation: receiving a set of specification attributes for the energy installation, the specification attributes indicating performance characteristics of the electricity generator and battery; determining, based on the specification attributes, whether the energy installation complies with a compliance criterion, wherein the compliance criterion is related to electricity exchanged between the energy installation and the distribution grid during operation; selecting, based on the evaluation, a group of two or more energy installations of the plurality of energy installations that were identified as meeting the compliance criterion; and making the selected group of energy installations available as an aggregated flexibility asset to a flexibility service associated with the electricity distribution grid.

Making the selected group of energy installations available as an aggregated flexibility asset may comprise transmitting information about the aggregated asset and/or energy installations to the flexibility service to allow the flexibility service to trigger flexibility service provision by the energy installations. The method may comprise configuring each energy installation of the group to alter electricity exported to the distribution grid or imported from the distributed grid in response to flexibility service events. Flexibility service events may comprise one or more of: a flexibility control message received at a controller from a flexibility control system; and a locally detected trigger, optionally comprising a measured deviation of a grid frequency. The flexibility service may comprise a system for balancing electricity supply and demand across the distribution grid or a segment of the distribution grid, a grid frequency control service or a capacity market. Evaluation of specifications may be performed as set out elsewhere herein (e.g. based on simulation or machine learning classification).

In a further aspect (which may be combined with any of the other aspects set out herein), the invention provides a method for controlling an energy installation at a property, the energy installation comprising a solar photovoltaic (PV) electricity generator for providing electricity to one or more consumers at the property and a battery for storing electricity generated by the electricity generator, the method comprising, for each of a series of control intervals: determining a battery charge level of the battery; obtaining generation forecast data indicating an estimate of electricity generated over the control interval by the electricity generator; obtaining consumption forecast data indicating an estimate of electricity consumed over the control interval by the one or more consumers; computing, based on the battery charge level, the generation forecast data and the consumption forecast data, a battery control schedule indicating times at which to charge and/or discharge the battery; and controlling charging and/or discharging of the battery during the control interval in dependence on the battery control schedule.

The generation forecast data may be determined in dependence on one or more of: specification data of the PV electricity generator, information indicative of a current date or available sunlight hours, and weather data. The specification data may include one or more of: a generation capacity of the generator, a size or effective area of one or more solar panels of the generator, and an orientation of the one or more solar panels. The consumption forecast data may be determined based on one or more of: historical consumption data for the property; consumption data for one or more other properties.

The one or more consumers may comprise a heating, ventilation and/or air conditioning (HVAC) device, e.g. a heat pump, and the consumption forecast data may be obtained based on a known electricity consumption of the HVAC device and/or based on an estimated usage of the HVAC device, the estimated usage optionally determined based on one or more of: historical usage patterns for the property, seasonal usage patterns, and weather data. The method may further comprise determining a control schedule for controlling operation of the HVAC device, for example based on one or more of: user-specified heating set points, interior temperature data of the property, and current or forecast exterior temperature data. The consumption forecast data may further be generated based on the determined HVAC device control schedule. One or both of the battery control schedule and the HVAC control schedule may be generated in dependence on energy cost data.

The battery control schedule and/or the HVAC control schedule may be generated in accordance with a utilisation criterion related to utilisation of energy locally generated by the electricity generator. The utilisation criterion may be based on consumption of electricity from the distribution grid and/or supply of electricity to the distribution grid. Preferably, the utilisation criterion corresponds to the efficiency/utilisation/compliance criterion (and/or associated grid import/export constraints or limits) as discussed in relation to other aspects set out herein. This may involve one or both of: maintaining import of electricity from the grid (e.g. in terms of energy or cost) at or below a first threshold, optionally zero; maintaining export of electricity to the distribution grid (e.g. in terms of energy or cost) at or below a second threshold. The utilisation criterion (in any of the described aspects) may comprise maximizing utilisation of electricity generated locally by the generator and/or minimising import of electricity from the distribution grid.

The invention also provides a system having means, optionally in the form of one or more processors with associated memory, for performing any method as set out herein and a computer readable medium comprising software code adapted, when executed on a data processing device or system, to perform any method as set out herein.

Features of one aspect or example may be applied to other aspects or examples, in any combination. Furthermore, method features may be applied to system or computer program aspects or examples (and vice versa).

BRIEF DESCRIPTION OF THE FIGURES

Certain embodiments of the invention will now be described by way of example only, in relation to the Figures, wherein:

Figure 1 illustrates a system for evaluating and optimizing the configuration of an energy installation;

Figure 2 illustrates a process for evaluating and optimizing the configuration of an energy installation in overview;

Figure 3 illustrates evaluation of a configuration in more detail;

Figure 4 illustrates a variation of the evaluation process;

Figure 5 illustrates use of a machine learning model for evaluation;

Figures 6A-6B illustrate application of the evaluation process to a site with multiple properties;

Figure 7 illustrates a process for optimising configurations of multiple properties;

Figure 8 illustrates a control process; and

Figure 9 illustrates a processing device for implementing described techniques. DETAILED DESCRIPTION

Embodiments of the invention provide a system for evaluating a specification of an energy installation that includes local generation capacity against an efficiency criterion to ensure efficient utilisation of that generation capacity and, if required, generating modifications to the specification where the efficiency criterion is not met.

A system in accordance with certain embodiments is illustrated in Figure 1. The system includes an electricity metering system 110 which obtains meter data from a number of properties / households 112 that are supplied with electricity by an electricity distribution grid 102. The meter data measures electricity consumption by those properties. Electricity consumption may be recorded, for example, as electrical energy values recorded at half- hourly intervals or some other appropriate interval, indicating energy consumed during the preceding interval. The consumption data is stored in a consumption database 114.

The term “property” is used herein to refer to any dwelling or other building or part of a building forming a self-contained unit that receives electricity through a connection to an electricity distribution grid. While the description principally focusses on residential properties (e.g. houses, apartments etc.) the described approaches can be applied to commercial or industrial premises as well.

The consumption data is used by an evaluation and optimisation system 120 in the efficiency evaluation of an energy installation 122 that is to be evaluated. The energy installation 122 may be an existing installation provided in a property 121 or may be a planned installation that is intended for deployment in such a property. The energy installation 122 includes various energy assets, including a solar photovoltaic (PV) electricity generator 128 in the form of a set of solar panels (typically installed on a roof of the building) and a battery 126 for storing electrical energy, along with a heat pump 124 for providing space and water heating. The property typically also includes a number of other electricity consumers 132 (such as washing machines/dryers, dishwashers, lights, televisions and other appliances and devices).

A local controller 130 controls operation of the energy installation, including energy storage in the battery and operation of the heat pump. This includes controlling charging of the battery, typically when local generation by generator 128 exceeds consumption by heat pump 124 and other consumers 132, and discharging the battery when consumption by the heat pump and other consumers exceeds Local generation. Additionally, the controller may use other criteria to control battery charging / discharging. For example, the controller may control the system to draw electricity from grid connection 102 during certain times to charge the battery and/or to supply stored energy back to the grid, e.g. based on energy price and tariff considerations or in support of flexibility services (e.g. to compensate for frequencyfluctuations on the grid by increasingsupply to or consumption from the grid). The controller 130 runs a control algorithm which may use Live consumption and generation data (and possibly tariff data) to make these control decisions substantially in real time.

The configuration of the energy installation 122 is captured in a specification 140 which specifies performance characteristics of the individual assets 124, 126, 128 and 130, such as the generation capacity of generator 128 and the electricity consumption of the heat pump 124.

The evaluation and optimisation system 120 (implemented e.g. as software running on a dedicated server or as a cloud-based service) evaluates the actual or planned configuration of energy installation 122 based on the specification 140 to determine its compliance with predetermined efficiency criteria and/or to optimise the specification to improve compliance with the criteria. As noted above, the energy installation 122 may be representative of an existing installation in a building that is to be evaluated, e.g. with a view to identifying upgrades to the system, or may be representative of a planned installation, for example for a property being built, for which a specification is to be evaluated and possibly modified prior to construction.

In the case of a planned installation, the consumption of other consumers 132 is generally not known. Even for an existing installation, precise meter data may not be available (and would include heat pump consumption). Thus, the evaluation system uses consumption data of one or more other properties from the consumption database 114 to estimate the consumption of the other consumers 132.

An aim of some embodiments is to provide energy installations that are specified and/or controlled in such a way as to reduce reliance on energy import from the grid and maximize use of Local generation. For example, it may be desirable for the Local PV generator to provide all of the energy required for a household, with any excess generation supplied back to the grid. This reduces reliance on central (possibly non-renewable) energy sources and reduces demand on the grid. Achieving this typically requires careful balancing of energy generation and storage capacities against expected demand. For example, the solar generator should have sufficient capacity to meet the total local energy requirements over time and the battery should have sufficient capacity to store excess energy during periods of peak generation and/or low demand so that demand at times of low generation and/or higher demand can be met (e.g. charging during the day to support consumption at nighttime).

On the other hand, over-specifying components can lead to unnecessary cost and inefficient operation. For example, if the PV generator generates more energy than can be used or stored by the battery, then the energy is exported back to the grid, and while some supply to the grid may be desirable (e.g. reducing generation capacity needed elsewhere), excessive export back to the grid may place a strain on the grid infrastructure.

Thus, embodiments seek to provide energy installations that are able to supply most or all local demand from local generation whilst limiting oversupply back to the grid to within a certain range. This goal is embodied in the compliance criteria against which an energy installation is assessed by the evaluation system 120.

In one approach, the evaluation system 120 may be used to assess compliance based on specification 140 and, if needed, make adjustments to the specification (e.g. to increase or reduce generation capacity, increase or reduce battery capacity etc.) This may additionally be combined with real-time control of the energy installation to optimise the use of the available resources using controller 130.

In these examples, the heat pump 124 is considered separately from other electricity consumers 132 at the property for a number of reasons. Firstly, it typically uses significantly more electricity than other consumers and the selection of a particular heat pump model may thus have a significant impact on the overall efficiency of the energy installation. Secondly, the consumption by the heat pump can be controlled effectively by modifying the heating schedule. For example, by appropriately altering heating periods and/or target temperatures, the consumption characteristics can be modified without appreciably impacting on the comfort levels required by the occupants. Thus, while the described techniques focus on the heat pump consumption, in principle other household loads could also be considered e.g. where time shifting or other consumption control is possible (e.g. refrigerators). The evaluation and optimisation system 120 is implemented principally using a simulationbased evaluator. However, this may be supported by a machine learning-based evaluator. These approaches will be described in the following sections.

Simulation-based evaluator

Evaluation is based at least in part on predicted energy export by an energy installation to the distribution grid and/or import from the grid.

The evaluation performed by evaluation and optimisation system 120 is illustrated in overview in Figure 2. The process starts in step 202 with receiving a system specification for the energy installation, specifying certain characteristics of the system such as PV generation capacity, heat pump consumption and attributes of the property where the installation is installed, or is planned to be installed.

In step 204, a simulator is run to simulate energy flows within the energy installation, based on the received system specification. The simulation is run in time steps over a simulation period and the energy import and export of the system to/from the distribution grid over the simulation period is computed. In step 206, the system evaluates the import/export data over the simulation period to determine compliance with one or more efficiency criteria. If the process is used only to evaluate compliance, then in step 208, the process outputs the compliance result (e.g. as an indication that the system is either compliant or not) and the process ends.

When used for specification optimisation the process determines in step 210 whether the system is compliant. If not, then the specification is altered in step 212 (e.g. to increase the number/size of solar panels, change the battery or heat pump model etc.) and the simulation process is repeated in step 204. This process then continues iteratively until compliance is achieved at step 210. In that case, a final specification is then output in step 214. This may, for example, form a recommendation fora homeowner to upgrade the energy system of their existing property or a recommendation for a house builder to adopt the altered specification for a planned property.

Figure 3 illustrates the simulation-based evaluation process in more detail. The main input to the simulation is the specification 140 of the energy installation being evaluated which is also referred to as the target system. This includes a set of property attributes 302 specifying characteristics of the property where the system is installed or is intended to be installed. In an embodiment, the attributes are indicative of the size and type of the property and include a floor area (e.g. square footage) and a number of rooms (or number of bedrooms). Other embodiments could include additional and/or different attributes, such as house age, building materials used in construction of the house, insulation, number of occupants, location etc. The property attributes are used to identify similar properties in terms of consumption as described in more detail below, and thus the attributes chosen are preferably useful for grouping properties by similar consumption patterns.

The specification further includes a PV generation capacity 304 and heat pump electricity consumption 306. These may be specified numerically, e.g. as power or energy values (using suitable units such as watts, joules, kilowatt-hours etc). Alternatively, the system may determine the values from other provided information, e.g. by looking up the kW consumption for a specified model of heat pump.

The system uses the specification data to generate simulator inputs as follows.

Firstly, the system computes an estimated PV generation trace 308 over a simulation period. The simulation period is the time period over which the operation of the target system will be simulated. A “trace” here refers to a time series of values (in this case generator energy output values). In one approach, the generation trace 308 is generated by scaling a default trace based on the generation capacity 304. The default trace may be determined empirically from measured generation data of a population of solar generators, or theoretically e.g. based on known generator performance and typical / average weather patterns in a region or at a representative (average) location. In an embodiment, the generation trace (and similarly the default trace on which it is based) includes a time series of energy values at a half-hour time resolution (one value per half-hour period). The trace may be determined for any required simulation period; in an embodiment, the simulation period is 1 year.

An estimated electricity consumption trace 310 for the heat pump is similarly computed from the heat pump consumption 306 given in the specification. Again, this may be obtained by scaling a default trace based on the consumption value 306 given by the specification. As before, the default trace may be obtained empirically. The heat pump consumption trace is defined over the same simulation period and has the same time resolution as the PV generation trace 308 (e.g. comprising half-hourly energy values).

The PV generation and heat pump consumption traces 308/310 may be further modified to take into account various factors. For example, the PV generation trace 308 may be modified (e.g. by further scaling) based on the location and/or local weather patterns at the location of the target system (e.g. by scaling using a time-series of weather-based attenuation values). The heat pump consumption trace 310 may similarly be modified based on location and/or weather patterns (e.g. reflecting different expectations for heating demand in different regions and/or over time) or for user behaviour. As an example of the latter, since an effect of optimising reliance on local solar generation is to reduce energy import from the grid (typically resulting in reduced energy bills), as a side effect of the present optimisation scheme, occupants may choose to use space heating more frequently and/or set higher target temperatures, resulting in higher heat pump consumption than would otherwise be expected. The system can compensate for this behavioural change by increasing expected consumption (e.g. by a fixed scaling parameter such as a 20% consumption increase, or by a season-dependent scaling). Other compensation factors may be used to modify the heat pump consumption trace, such as to account for properties requiring more energy to heat in practice than theoretically expected (pessimism adjustment) due to e.g. substandard construction or insulation provision or occupant behaviour.

Additionally, the system obtains electricity consumption data 312 for other properties/households relating to energy consumption by those households. The consumption data preferably excludes heat pump consumption and so gives representative data of consumption by other appliances in such systems which can thus be used to provide a model of the expected consumption by other consumers 132 (not including the heat pump) in the target system. One way to achieve this is to select energy consumption data for dual-fuel customers of an energy provider (customers using both electricity and gas), since gas is predominantly used for heating and so such customers are likely not to have heat pumps and their electricity consumption can thus be used to estimate consumption of other consumers 132 in the target system.

Preferably, the consumption data for other properties 312 is again in the form of traces (time series energy consumption data) at the same time resolution as the traces 308/310. If necessary, traces at the required time resolution may be derived from the underlying consumption data. A filter 314 selects from this consumption data a set of consumption traces for properties that are similar to the property 121 of the target system, using the property attributes 302 to determine similarity. To enable this, the consumption traces for other properties 312 are tagged in the database with corresponding property attributes (e.g. floor area/number of rooms) of the properties from which the traces were obtained, which are compared to the property attributes of the specification 140. For example, the system may select a number (e.g. 1000) of properties having matching property attributes, where “matching” property attributes may mean identical attribute values or attribute values within some defined tolerance, or attribute values that are closest to the property attributes for the target system. This ensures that the selected consumption data is broadly representative of expected consumption (except for the heat pump) of the target property. Consumption traces are obtained for the selected properties for a time period matching the simulation period (e.g. consumption data from 1 January to 31 December where the simulation period is a calendar year). These are referred to as peer traces 315.

The simulator runs multiple simulations of the target system using PV generation trace 308, heat pump consumption trace 310 and the peer consumption traces 315. In particular, an individual simulation is run for each peer trace to simulate energy flows in the system over time on the assumption that the consumption of other consumers 132 matches the current peer trace 315 being evaluated.

The simulation takes into account charging and discharging of the battery (by simulating the control decisions used to control charging and discharging). The simulation is run at the time resolution of the traces 308, 310, 315, determining at each time instant the net balance between:

• total energy output, by a) the generator 128 (as determined by the generation trace 308), and b) by battery 126 when the battery is being discharged, and

• total energy consumption, by a) the heat pump 124 as determined by the heat pump consumption trace 310, b) the other consumers 132 in the property as determined by the peer consumption trace 315 currently being evaluated and c) any energy used during charging of the battery

The net energy balance determines the amount of energy imported from the grid (corresponding to a local generation shortfall) or exported to the grid (corresponding to local excess generation) at each simulation time instant. For example, this could be specified as a net import value (e.g. total imported minus total exported over the simulation period, where a positive value indicates a net import i.e. more energy imported than exported and a negative value indicates a net export) or conversely as a net export value (e.g. total exported minus total imported). This information is used to produce an energy import/export trace 318 indicating how much energy is imported/exported from/to the grid at different time instants over the simulation period.

The above process is repeated for each peer trace, producing separate import/export traces for each peer trace evaluated.

The resulting import/export traces are then further processed and aggregated in step 320. This involves computing an import/export metric for each peer trace, for example the total net import/export value over the simulation period, and then averaging the results over the peer traces to produce an average metric e.g. average energy import/export. This output provides a compliance metric 321 for the specification being evaluated. The compliance metric may also be referred to as a utilisation metric as it can be used to indicate how well local generation capacity is being utilised by the energy installation. The compliance metric may be a single value, e.g. electricity import/export balance (or an estimated cost/profit value related to electricity import/export as discussed further below). However, in other examples, the compliance metric could include multiple values (e.g. separate import/export energy or cost values).

The resulting compliance metric is then processed to obtain the compliance evaluation in step 322. This may simply indicate compliance or non-compliance with the evaluation criteria. Alternatively, there may be more than two output classifications, for example:

• Non-compliant (system is underspecified) - this may indicate that there is a positive net import from the grid, i.e. the energy demand is not met by local generation and storage, or alternatively that the net import exceeds some specified threshold. This could be the result of insufficient PV generator capacity or an inappropriate heat pump model being selected.

• Compliant - there is no net import from the grid (or alternatively the net import is below the specified threshold) though there may be export to the grid.

• Non-compliant (system is over-specified) - there is net export to the grid that exceeds some specified threshold, which may suggest that excess generation capacity has been specified for the system which may result in a strain on the distribution network. Any other classification scheme could be used or as a further alternative the evaluation output could be a numerical score on a scale (e.g. ranging from underspecified to over- specified).

In one example implementation, for a total net export E over the simulation period as the compliance metric, averaged over the simulations for the selected peer traces, and an export threshold T, the system is classified as

• non-compliant when E < 0 (corresponding to a net import from the grid i.e. more energy is imported than exported)

• compliant when 0 < E < T

• over-specified when E > T

In some embodiments, the final compliance metric may take into account other factors, for example cost information to allow a final evaluation to be made in the cost domain rather than purely energy domain, as described in more detail below.

Instead of averaging results over all peer traces, some embodiment may employ sampling techniques, for example bootstrap sampling. In that approach, subsets of the peer traces are selected randomly (with replacement) and the mean compliance metric (e.g. mean net import/export) computed for each subset. An overall mean is then computed over the individual subset means. This can provide a better reflection of the underlying distribution and also gives a narrower distribution as it is more robust to outliers.

In the approach described above, the PV generation trace 308 and heat pump consumption trace 310 are obtained by scaling a preconfigured default trace based on the PV generation capacity 304 / heat pump consumption 306. However, either or both traces could be computed using more detailed models.

For example, the PV generation trace 308 could be computed as a function of the total generation capacity, as well as other factors of the solar panel installation, such as the panel type/model, orientation, location, weather patterns etc. Instead of simply specifying total generation capacity, the specification could specify the PV generator in more detail, e.g. by explicitly including attributes such as panel type/model, number of panels, (total) panel size/area and panel orientation (facing direction), and that information could then be used to compute the generation trace. Similarly, for the heat pump consumption trace, in addition to the total consumption other factors could be used, such as model, location and associated weather patterns that influence heating demand etc.

In a further modification, instead of deriving a heat pump consumption trace that is used as a static input to the simulator, a (controlled) heat pump model could be included in the simulation, so that the consumption in each half hour simulation interval is determined by an optimiser scheduling the heat pump (determiningeither a temperature setpoint schedule or even lower level parameters such as flow rate + flow temperature of the water through the system).

The simulator 316 simulates the actions taken by controllable assets of the energy installation 122, as determined by the local controller 130. The local controller runs an optimiser algorithm that controls the assets to achieve particular objectives, for example minimising import of energy from the grid. In one embodiment, the optimiser simulation incorporates just the battery -specifically simulating control decisions determining when to charge/discharge based on the current battery status and charge level, current consumption by the heat pump 124 and other consumers 132 and current generation output by the solar generator 128. Energy price information may also be taken into account. In one particular approach, the simulator runs a linear program to determine, on a daily basis, the best periods of charging/discharging based on battery parameters (power and energy limits and midnight state of charge each day); site generation/consumption and import/export limits; and price forecasts.

The Figure 3 examples assumes a fixed (known) battery capacity (e.g. a default battery model) and the simulator simulates battery charging and discharging on the basis of that capacity (and any other relevant battery parameters). However, in other examples, the battery capacity (and possibly other battery parameters) may be variable and also specified in the specification 140, in which case the simulator simulates the charging/discharging schedule for the battery on the basis of the specified information.

The simulator can be extended to incorporate other controllable assets. For example, as mentioned above, if the optimiser also controls the heat pump e.g. to optimise heating schedules, then this may be incorporated into the simulation.

T1 Furthermore, the operation of more complex market mechanics could be incorporated into the simulator, for example, to simulate active participation in demand/flexibility services such as the ESO Balancing Mechanism. In one example this could include simulating control decisions by the controller to increase supply to or consumption from the distribution grid (e.g. by charging/discharging the battery) in response to a request from a control system (or in response to a locally measured fluctuation in grid frequency) in order to counter grid frequency fluctuations and/or ensure demand and supply on the grid are balanced.

While the property attributes 302 are principally used to select consumption data for similar properties, such attributes could also be used to inform the simulation in other ways. For example, information such as property size, type, building materials and the like could be used to estimate thermal properties/heating efficiency of the property which in turn could be used in deriving the heat pump consumption trace, e.g. to scale the expected heat pump consumption in dependence on the thermal characteristics and heat retention efficiency of the property.

Figure 4 illustrates an extension to the above approach in which cost information is further used in the evaluation. Here, the import/export trace is processed using energy price data 402 to generate an energy cost trace 404. The cost information could be static or could define temporally varying pricing, e.g. in the form of one or more price traces over the simulation period, to account for changing energy prices over the period (though this may be provided at a different time resolution, e.g. daily prices). Furthermore, separate price information may be provided for energy import (cost per unit energy consumed from the grid) and energy export (price at which excess energy is sold back to the grid).

The price information is used at each time instant of the import/export trace to determine a cost value for the energy import/export. A positive value of the cost value could e.g. represent a cost charged to the property occupant for energy consumed from the grid and a negative value could represent an amount paid to the property occupant (or offset against energy charges) for energy supplied back to the grid. From this a total net cost over the simulation period is then computed in step 406. The computed cost may be further refined by applying one or more modifiers 408 (e.g. representing additional cost factors, standing charges etc.) The aggregation operation 410 then average the results over the peer traces evaluated (e.g. using bootstrap sampling as previously described) to produce an expected mean cost or profit. The final cost/profit value is then used as the compliance metric and is evaluated against the compliance criterion (322) (for example being considered compliant if it lies in an acceptable range and non-compliant otherwise).

In a particular variation of this approach, the aim may be to minimise (or eliminate) energy bills for a customer and thus a property for which the projected total energy cost for the simulation period (as before, averaged over the peer traces e.g. using bootstrap sampling) is below some threshold (or is below zero) may be deemed compliant. This can allow the energy supplier to offer a special fixed cost tariff or even zero cost tariff to compliant properties: if the customer’s energy installation meets the compliance criterion, then the customer receives energy at fixed cost or no cost (even during times when net consumption exceeds local generation, on the assumption that net consumption from the grid will be close to zero over the longer term). This may be limited by a fair use policy, e.g. by setting a grid consumption limit above which additional charges are applied. In such approaches any excess generation can be monetized by the supplier allowing the supplier to make some profit whilst providing predictability to the customer.

Instead of using the basic energy cost as the evaluation metric a projected profit value to the energy supplier can be derived from the energy cost 406 using the modifiers 408. For example, in some embodiments, a capture rate is applied to the result, modelling the fact that the raw cost data given by the above process implicitly assumes perfect foresight of prices, consumption and generation, which is in reality an upper bound on the realisable profit. Next, the effects of the fair use policy are applied to simulate the energy supplier recouping revenue for any use above the limit. A further adjustment could be made for expected revenue due to participation in grid services (e.g. flexibility provision). Alternatively, this and other modifiers could be incorporated into the simulation model rather than being applied as static compensators to the simulation output.

Specification optimisation

Referring back to Figure 2, the described evaluation algorithm can be extended to an optimisation algorithm (loop 204-206-210-212) whereby, if the compliance evaluation identifies the target system as non-compliant, the specification is modified and reevaluated.

With reference to Figure 3, in this approach, an optimiser 330 iteratively modifies a particular input to the simulation whilst keeping the other inputs fixed. The property attributes are assumed to be fixed and thus are not modified by the optimisation. Thus, the optimisation Loop may repeat the evaluation for a range of different values of PV generation capacity 304 and/or for a range of different values of the heat pump consumption 306. The values considered may be informed by the value of the compliance metric, e.g. net energy import/export or profit. For example, if the net import from the grid is greater than zero then the local consumption exceeds local generation capacity and so this may be addressed by increasing generation capacity (e.g. specifying additional solar panels) or reducing heat pump consumption (e.g. by selecting a heat pump model with a lower consumption specification). Conversely, if the system is over-specified (so that there is a net export to the grid exceeding some threshold), then the local generation capacity may be reduced or more powerful heat pump may be chosen.

The optimiser algorithm may be based on a bisection algorithm, in which one test attribute is modified iteratively, keeping the other specification attributes fixed until the resulting compliance metric reaches a point close to the inflection point. This process can be repeated for other specification attributes until a suitable specification is identified.

The above approaches to optimisation are based on an exploratory approach, in which specifications are modified and reevaluated iteratively, e.g. on step-by-step, attribute-by attribute basis. However, any suitable approaches to optimisation could be employed. For example, simulated annealing or other approaches based on random / probabilistic sampling of the search space (defined by the set of specification attributes being optimised) could be used.

Once at least one modified specification is found that is classified as compliant this may be output as a recommendation to a user. Alternatively, a range of possible options could be recommended.

In some implementations the system may evaluate a range of possible specifications and may find multiple compliant specifications. The system then outputs all compliant specifications. Alternatively, the system selects one or more specifications from the compliant results. In one approach this could be the specifications with lowest generation capacity, battery capacity and/or heat pump consumption as these may correspond to more cost-effective solutions.

Alternatively, the system may use the compliant results to identify one or more constraints on specific specification attributes (e.g. minimum/maximum values or acceptable ranges). These constraints could then be output to the user (e.g. generation capacity should be at least X or battery capacity should be at least Y).

In other implementations, the evaluation process may stop when a first compliant specification is identified which is then output to the user.

Instead of identifying to the user particular values for specification attributes, the attributes may be mapped to specific asset models or other relevant characteristics of the assets of the energy installation. Specific models may be identified from a list of stored models defined with the relevant specification attributes, allowing a suggested specification identifying particular asset models to be proposed to the user.

More generally, the optimisation results may be used to output any useful form of configuration data for the energy installation to the user. This could, for example, include:

• a suggested type, model and/or generator capacity of the PV generator;

• a suggested number and/or size/area of solar panels for the PV generator;

• a suggested consumption rating, type and/or model of HVAC device;

• a suggested capacity rating, type and/or model of battery.

In some examples above, the battery 126 is considered to be fixed - e.g. a fixed capacity battery model is assumed for the energy installation. However, in alternative implementations, the battery characteristics (e.g. charge capacity) may be specified as part of the specification 140. The battery capacity may then be one of the specification attributes modified by the optimiser in the optimisation loop, e.g. to propose replacing a battery with a higher-capacity model to ensure that more locally generated energy can be stored and ultimately used. In that case, the simulator 316 would also adapt the simulation of the control decisions of the energy installation based on the selected battery capacity.

The optimizer may be used to determine a suitable specification for a planned energy installation (e.g. for a new-build property). Alternatively, the optimizer may be used to propose upgrades to existing installations, e.g. to propose a change of an installed battery or an increase in solar capacity in order to meet the compliance criterion. In one example, a web interface is provided where a user inputs the specification of their system (e.g. property attributes, PV generation capacity, heat pump capacity and/or battery capacity) and the system then outputs an indication whether the system is compliant (and thus the user may access any associated fixed or zero bill tariff) or, if not compliant, suggests one or more changes to make it compliant (e.g. a suggestion to install a larger battery).

In addition to PV generators, batteries and heat pumps, other energy assets may be incorporated into the evaluation/optimization process. For example, an electric vehicle (EV) charger (or the EV itself) may be modelled as an additional asset. While an EV is connected to the energy installation it may be treated as an additional battery with its own charging constraints (e.g. full EV charge must be attained by 7am each morning). Charging/discharging of the EV battery may then be simulated by the simulator within those constraints. However, given the large and unpredictable energy demand of EVs the operator may prefer not to include EV charging in a fixed or zero bill tariff and so the optimizer may in that case aim to optimize the specification of the energy installation without regard to EV charging so that the net grid import is below a threshold (or zero) not including EV consumption, with any EV consumption charged to the consumer as an additional cost beyond the fixed/zero bill tariff.

Instead of (or in addition to) PV generators, other types of local electricity generators could also be used in the described system, such as wind turbines, petrol generators etc.

Furthermore, other types of HVAC (heating, ventilation and/or air conditioning) units may be specified and/or simulated instead of (or in addition to) heat pumps, for example boilers, furnaces, air conditioners and the like. Note the term “HVAC” or “HVAC device” herein refers to any device providing any related function such as heating, cooling, ventilation or air conditioning (rather than necessarily a specific combination of heating, ventilation and air conditioning though combined devices are possible).

Machine-learning based evaluation

Once a sufficient number of specifications (e.g. combinations of property attributes, PV generation capacities, heat pump consumption values, battery capacity values etc.) have been evaluated using the simulator, a machine learning model can be trained, using evaluated specifications as training samples together with the assigned compliance outputs determined using the simulator as ground truth labels. The model may be trained to output the final compliance classification (e.g. compliant/non-compliant) or alternatively an intermediate value (e.g. compliance metric such as net energy import/export or cost/profit). As illustrated in Figure 5, such a prediction model 510, once trained, is then able to generate a predicted compliance metric or label 512 for a particular input specification 500 (comprising in this example a particular combination of property attributes 502, PV generation capacity 504 and heat pump consumption 506) without running the simulator. This can be advantageous since the simulation may be compute-intensive (especially when simulation is performed for large numbers of peer traces and/or for an extended simulation period at high temporal resolution).

The predicted compliance label can be used as an initial indication of compliance (e.g. allowing a user to understand quickly whether their energy installation is compliant). If necessary, the simulator may subsequently be used to perform a full evaluation (e.g. before accepting the customer onto a fixed or zero bill tariff). This could be based on operator request, or automatically based on the model prediction. For example, if the compliance label or metric output by the machine learning model meets a criterion or threshold (or is within a certain range), a full evaluation using the simulator may be triggered. As a concrete example, if the model outputs a predicted energy import/export value or an associated cost or profit value that meets a threshold or is within a range, e.g. being close to a target value or range, the simulator may then be run to obtain a more accurate evaluation for the energy installation.

The trained prediction model can similarly also be used to evaluate alternative specifications to more quickly propose modifications to an energy installation by the optimizer 330 (see Figure 3 / the optimization loop of Figure 2).

As noted above, the training samples consist of the specification attributes 500 together with a relevance compliance label (e.g. under-specified/compliant/over-specified). Instead of directly predicting a compliance label, the output predicted by the model could be the compliance metric such as the total energy import/export from/to the distribution grid (or a related cost/profit value). In that case, once a predicted compliance metric has been obtained as output from the model, that output can then be used to obtain the final compliance indication or classification as described previously.

In an embodiment, the machine learning model is an artificial neural network, trained using backpropagation techniques as known to those skilled in the art. However, other types of machine learning model (such as linear regression, support vector machines, decision trees/random forest models etc. ) could be used. While the machine learning inputs are shown in Figure 5 as comprising the property attributes 502, PV generation capacity 504 and heat pump consumption 506, as for the simulator-based approach, these may be varied to include additional and/or different attributes. For example, instead of explicitly specifying the PV generation capacity (e.g. as an energy/power value) the specification could specify a (total) size or area of the solar panels, a total number of solar panels, a generation capacity per unit area or per panel, an orientation of the panel (e.g. with respect to compass directions) or the like. Similarly, the heat pump consumption could be specified byway of a heat pump model identifier instead of an explicit power consumption value. Other attributes such as the battery capacity and/or model could also be included in the specification as described in relation to the simulationbased evaluator.

While the attributes used in the specification 500 as training/classification attributes will typically correspond to those used by the simulator (since the simulator is used to generate the training samples), in some examples the model could be trained and used with a subset of attributes available to the simulator or using derived attributes. For example, while the simulator could use specific generation capacity values, the model could use a set of value ranges represented by different input labels. As another example, the simulator could use a solar panel size (e.g. effective panel area) and orientation as input specification attributes, whilst the machine learning model could use an estimated generation capacity value derived from that information.

Instead of using the simulator to generate training samples, these could be derived from real-world data sets. For example, where data for a set of properties is available indicating specifications of the energy installation assets and specifying import/export from/to the distribution grid, compliance can be evaluated for each property directly from that data (in the manner described above), and the resulting compliance labels together with the specification data can then be used to provide the training samples for training the model.

Site-wide optimization

The techniques described above for evaluating compliance and/or optimizing specifications for energy installations can be extended to cover multiple such installations across a site. This is illustrated in Figure 6A. Figure 6A shows a site 600, for example a housing estate, including multiple properties 601 , connected to the distribution grid 102 via a common grid connection 602. Each property includes an energy installation with a battery, solar generator, heat pump and local controller as shown for property 121 in Figure 1. However, it is also possible that different properties may include different combinations of energy assets. For example, a particular property may not include a solar generation facility or may lack a heat pump etc.

The site may optionally include one or more communal energy assets, such as a communal generation facility 604 (e.g. solar panels), communal storage 606 (e.g. one or more large- capacity batteries) and/or communal heating system 608 (e.g. a communal ground-source heating setup).

Communal generation and storage facilities 604, 606 may generate and/or store electricity for provision to site 600 in a similar way to those provided in individual properties. The communal heating system 608 (also known as a district heating system) may generate a supply of heated air and/or water centrally for delivery to houses via a network of insulated pipes/ducts. The properties 601 and (where provided) communal energy assets 604-608 are interconnected into a local site network or microgrid 610 for transfer of electricity between the assets and properties and to/from the distribution network 102 via grid connection point 602. A control system 612 may perform control of assets (e.g. communal assets 604-608) e.g. to schedule battery charging/discharging and/or heat generation, and may interact with individual control systems in the properties 601 (or instead a single central control system 612 may directly control assets in the individual properties).

The compliance evaluation and optimization techniques discussed above may be adapted to account for differences in properties on a site and their energy installations. One example is shown in Figure 6B. Here, two separate properties 620, 624 are each provided with solar panels 622, 626 on their roofs. However, due to the panels facing in different directions, the energy generated will differ. Such differences affect both generation over time (e.g. properties may generate maximum solar generation output at different times of day) and overall generation (e.g. a property with south facing panels may receive more light and hence have highertotal output than a property with east- or west-facing panels). Other differences may include different panel sizes/quantities (e.g. larger houses may have more roof space for panels), different models with varying efficiency etc. Similarly, different properties may use different heat pump models with different consumption profiles, or may use batteries with different capacities.

Figure 7 illustrates the application of the evaluator and optimizer to multi-property scenarios. In this case, the evaluator 710 receives specifications 702, 704, 706 for any number of properties that are located on a site. The evaluator in this example is implemented based on simulation, as previously described in relation to Figures 3-4, except the simulation in this case simulates each of the energy installations and their individual assets.

In one approach, each energy installation is simulated individually as previously described, and then the resulting import/export flows (as specified by trace 318 in Figure 3) are combined to identify total flows to/from the grid, e.g. as site-wide import/export trace 712. This allows modelling e.g. of excess generation at one installation (property) being consumed by another installation on the site, thus reducing supply from the grid to the site.

In another approach, the simulation simulates the system of multiple installations as a combined system, determining energy flows within and between installations at each simulation interval to obtain the final aggregate energy flows. This approach may be appropriate where there are also one or more communal assets operating at the site (e.g. 604/606/608). For example, this can allow charge levels of communal storage 606 to be tracked over time, allowing excess generation capacity to be captured and utilized at times of low generation. Where communal asset(s) are included, their specification(s) 708 are provided as additional inputs to the simulator. The simulator simulates operation of those assets as needed, in particular to simulate the charging and discharging of a communal battery system 606 (e.g. simulating operation of the site control system 612 in addition to individual controllers at the properties).

Once site-wide aggregated import/export flows to/from the grid have been determined (e.g. as a site-wide import/export trace 712) this data can be further processed as described earlier (e.g. in the energy domain or energy cost domain) to obtain a compliance metric (714) for the site as a whole. As before, the site as a whole may be classified as compliant if total expected import is below zero (or some other threshold) and/or total export is below some (typically non-zero) threshold. Alternatively, the energy flow data may be further processed to obtain cost or profit information with compliance determined based on comparison to a profit threshold/range. This approach thus allows a planner to evaluate a new planned housing development to determine whether the energy installations planned for individual dwellings (and any communal assets) are adequate to meet the efficiency criterion or whether they are under- or over-specified.

As in the single-dwelling approach, the evaluation process may be applied iteratively using an optimizer 716. The optimizer operates as previously described except that it may vary individual specifications of multiple dwellings and/or the specifications of any community assets until suitable specifications are obtained that meet the compliance criterion. The modified specifications can then be provided to the planner as suggestions for revising the plan. For example, this could be a suggestion to increase battery capacities of individual dwellings or to increase a capacity of a communal battery facility 606.

Using this approach, grid consumption for the whole site may be minimized, reducing the required grid connection capacity at connection point 602, whilst still allowing for differences in the energy installations at different properties, for example to take into account that different available roof areas and orientations lead to differences in PV generation capacity of solar panels for different properties.

In a specific example of this, the optimizer may consider properties of solar electricity generators fixed (as they may be constrained by roof area and orientation of the house/roof). As discussed in relation to Figure 6B, generation capacity of individual solar generators may vary e.g. due to panel orientation, panel size/number etc. The optimizer thus keeps the solar generator specification fixed while varying attributes of the other systems assets (e.g. local heat pumps and batteries of individual properties and/or communal assets of the site) to compensate forthe differences in generation capacity, untila suitable specification meeting the site-wide compliance criterion is identified.

Reduction of the required network connection capacity may be specified as an additional constraint forthe optimizer. In that case, the optimizer would then identify specifications for the individual dwelling assets and/or communal assets so as to reduce the required grid connection capacity (e.g. specified in terms of peak import/export flows). This allows upgrade cost to be reduced. Aggregation of compliant properties does not necessarily require the properties to be located on a single site. More generally, the compliance evaluation can be used to confirm that a group of properties comply with the efficiency criterion (e.g. zero grid import). This consumption predictability in turn allows compliant properties to be bundled into an aggregate asset which can then be enrolled into flexibility service markets - such as (in the UK) the Capacity Market, DNO markets and the Balancing Mechanism. Such aggregation may be site-based as shown in Figure 6A but may also be used for properties that are not necessarily located close to each other in terms of geographical or network topological distance.

In an implementation of this approach, the system is used to evaluate specifications for a (potentially large) number of energy installations of the type described above (e.g. each including a local electricity generator, battery and heat pump), using the described simulator-based and/or machine learning based approaches. The system then selects a group of the evaluated energy installations that were identified as meeting the compliance criterion and the group of energy installations is made available as an aggregated flexibility asset to a flexibility service associated with the electricity distribution grid (e.g. a service operated by the grid operator).

In practice this may involve submitting information about the aggregated asset (e.g. combined capacity for increasing or reducing grid consumption or grid export of electricity) to the flexibility service to allow the flexibility service to trigger flexibility service provision by the energy installations as and when required, e.g. by requesting a decrease or increase in consumption from, or supply to, the grid to meet the objectives of the flexibility service.

Furthermore, the controllers of each of the group of energy installations are configured to control one or more assets in response to flexibility service events (e.g. a flexibility control message received at the controller from a flexibility control system or a locally detected trigger, such as locally measured deviation of a grid frequency). This may involve controlling the assets to alter electricity exported to the distribution grid or imported from the distributed grid, for example by charging or discharging the battery or increasing or reducing consumption by the heat pump (e.g. by deviating from the heating set point schedule).

The flexibility service may, for example, be any system for balancing electricity supply and demand across the distribution grid or a segment of the distribution grid, a grid frequency control service, a capacity market etc. The guarantees provided by the compliance evaluation, supported by the real-time control of assets to ensure operation complies with the import/export limits, allows additional services to be offered to DNOs, and can enable smaller aggregate grid connection capacities to be configured. This can be particularly relevant for export to the grid, where the fact that the energy installations absorb solar generation in batteries, and self-consume or export it outside of solar peak periods, can allow the same export limit to be provided at a smaller grid upgrade cost.

Control process

Figure 8 illustrates a control process that maybe used by controller 130 of a deployed energy installation. In this example, both the battery and heat pump are actively controlled by the controller. However, in other examples, additional or fewer assets may be actively controlled (e.g. just the battery).

The depicted control process is repeated for each control interval. In one example, the control interval is one day. In overview, the control process involves scheduling the operation of the heat pump for the next control interval in step 802, scheduling the operation of the battery for the next control interval in step 804 and then controlling the heat pump and battery using the determined schedules in step 806.

In more detail, a heat pump control schedule is determined in operation 802. This may be based on heating set points 810 configured by a user (specifying desired comfort levels) along with interior temperature data 812 of the property (e.g. measured by a thermostat) and possibly current or forecast exterior temperature data 814. This information is used to determine a control schedule indicating when the heat pump will need to be run to achieve the comfort levels required by the heating set points (which may further take into account thermal properties of the building).

In operation 804, the controller uses a current battery charge level 816 (e.g. the charge level at midnight) together with fixed battery parameters (e.g. total capacity) along with various forecast data 818-822 to determine a control schedule for the battery specifying times at which the battery should be charged or discharged. In an example, the forecast data includes generation forecast 818, consumption forecast 820 and cost data 822. The forecast data may be based on historical data, weather data etc. For example, a generation forecast 818 may be obtained based on known generation capacity and/or other specification data of the solar PV generator (e.g. total panel size/area and/or orientation) together with information on the date/time of year (e.g. indicating available daylight hours) and weather forecast data to predict energy generated over the next control period.

The consumption forecast 820 may include estimated consumption of the heat pump and of other consumers at the property. The heat pump consumption may be determined based on the previously determined heat pump control schedule (e.g. on/off times) together with the known heat pump power consumption. Alternatively (e.g. if the heat pump scheduling 802 is omitted from the control process, for example being under control of a separate control system) the heat pump consumption may be estimated based on the known heat pump consumption and a predicted heating/ hot water demand (e.g. based on weather data and/or average seasonal demand patterns for a cohort of properties, and/or based on historical demand patterns). Consumption of other consumers may be estimated based on past consumption patterns at the property or based on consumption data for other (similar) properties as previously described.

Optionally, energy cost data 822 may additionally be used in the scheduling decision (e.g. to enable charging of the battery at times of cheap electricity).

Operation 804 produces a battery control schedule determining when the battery should be charged and/or discharged over the next control interval.

The battery and heat pump control schedules are determined in accordance with a selected optimization criterion. In a preferred embodiment, this relates to increasing utilization of energy generated by the local solar PV generator and reducing use of electricity from the distribution grid. In a specific example, the control schedules are determined to maintain electricity imported from the grid, or an associated energy cost, at or below a certain threshold (which may be zero). Alternatively or additionally, the scheduler may aim to maintain export of excess local energy generation to the grid below a further threshold. These optimization criteria may correspond to those used by the evaluation/optimization system described above to determine compliance of an energy installation. As noted above, the heat pump scheduler (802) uses heating set points 810 specified by a user, e.g. in the form of a heating schedule specifying periods when heating is required and target temperatures for those periods. In some embodiments the scheduler may modify the heating schedule, e.g. changing heating periods and or target temperatures, to alter consumption of the heat pump so as to achieve compliance with the optimization criterion. For example, the scheduler may reduce target temperatures slightly or shorten heating periods to reduce consumption. This may be done within certain constraints (e.g. limiting deviation from the user-specified target temperatures to within a certain range) to ensure that the user-required comfort levels are still broadly being met.

In step 806, the controller applies the battery schedule and heat pump control schedule during the next control period. Specifically, this involves dispatching control commands to the battery (or a battery controller) and/or the heat pump based on the respective schedules.

At the end of the control interval, the process returns to step 802 to perform scheduling for the next control interval.

Whilst in this example, the control intervals are the same for battery and heat pump scheduling, this need not be the case; for example, the battery schedule could be determined on a daily basis while the heat pump control schedule could be configured hourly (e.g. using actual current temperature data rather than forecast data 814). Generally speaking, shorter control intervals may be used to achieve substantially real-time control at the cost of greater computational load on the controller.

Furthermore, in some examples, only the battery scheduling part may be performed by the controller (with heat pump control performed by a separate system, e.g. an HVAC control system).

The control process is merely given as an example and other control strategies may be implemented. The control process may be implemented as software running on a control computer or using dedicated hardware.

The described control process may be used by the local controllers 130 of energy installations 122 during operation (and/or a site-wide control system in the Figure 6A example). Furthermore, the same control process is also used by the simulator 316 (Figure 3) during simulation-based evaluation and optimization. This enables the simulator to simulate the control decisions that the controller would make fora given energy installation, as represented by a specification 140 being evaluated, allowing the simulator to accurately predict energy flows in such a system.

Note the term “battery” as used herein encompasses an individual battery as well as a battery system consisting of multiple individual battery units. Thus batteries provided at properties or as communal facilities could be single batteries or multi-battery systems.

Processing device

Figure 9 illustrates a processing device 900 suitable for implementing processing elements of the system, such as the evaluation and optimisation system 120 of Figure 1 .

The processing device 900 may be based on conventional workstation or server hardware and as such includes one or more processors 908 together with a main memory 902 (e.g. volatile / random access memory) for storing temporary data and software code being executed.

An input/output subsystem 906 includes one or more I/O interfaces for communicating with external devices and peripherals, such as displays, input devices (e.g. keyboard, mouse), external storage devices and the like. A network interface 910 is provided for communication with external systems via network 120 (encompassing e.g. Local and/or Wide Area Networks, including private networks and/or public networks such as the Internet, cellular telephony networks etc.) For example, the server may communicate with the consumption database 114 and with client devices accessing the evaluation/optimisation functions via the network.

Persistent storage 904 (e.g. in the form of hard disk storage, optical storage and the like) persistently stores software and data for performing various described functions (e.g. forthe evaluation/optimisation processes as described in relation to Figures 2-5 and 7).

The persistent storage further includes a computer operating system and any other software and data needed for operating the processing device. The device may include other conventional hardware components as known to those skilled in the art. The various components are interconnected by one or more data buses 912 (e.g. system/memory bus and one or more I/O buses).

While a specific architecture is shown and described by way of example, any appropriate hardware/software architecture may be employed to implement the processing device.

Furthermore, functional components indicated as separate may be combined and vice versa. The processing functions may be performed by a single device or may be distributed across multiple devices (e.g. in a server cluster).

It will be understood that the present invention has been described above purely by way of example, and modification of detail can be made within the scope of the invention.

Claims

1 . A computer-implemented method for determining a configuration for an energy system including electricity generator and consumer assets associated with a plurality of properties on a site, the properties connected to a site electricity network coupled to an electricity distribution network, the method comprising: receiving specification data specifying characteristics of energy assets associated with the site, the energy assets comprising electricity generator assets and electricity consumer assets installed or intended for installation at the properties; performing an evaluation process to evaluate the specification data, the evaluation process comprising: simulating electricity generation and consumption by generator and consumer assets at each property in accordance with the specification data; determining flow data for each property indicating electricity drawn by the property from the site electricity network and/or supplied to the site electricity network by the property based on the simulated generation and consumption; determining, based on the flow data, grid flow data indicating electricity imported from the distribution grid to the site electricity network and/or electricity exported from the site electricity network to the distribution grid; and determining a utilisation metric based on the grid flow data; determining one or more modifications to the specification data in dependence on the utilisation metric obtained by the evaluation process; and outputting suggested specification data for one or more assets based on the one or more modifications.

2. A method according to claim 1 , wherein the energy assets include, at each property, one or more of: a solar photovoltaic (PV) generator; a battery; and an electrically powered heating, ventilation and/or air conditioning (HVAC) device, optionally a heat pump.

3. A method according to claim 1 or 2, wherein the specification data further specifies characteristics of one or more shared energy assets connected to the site electricity network, optionally wherein the outputting step comprises outputting suggested specification data for one or more of the assets associated with the properties and/or one or more of the shared energy assets based on the one or more modifications.

4. A method according to claim 3, wherein the shared assets comprise one or more of: a communal generator asset configured to provide electricity to multiple properties of the site; a communal battery for storing electricity generated by generator assets at the properties and/or by a communal generator asset; a communal HVAC system powered by electricity drawn from the site electricity network, optionally a district heating system.

5. A method according to claim 3 or 4, wherein the evaluation process further includes simulating operation of the shared energy assets to determine further flow data indicating electricity drawn by the shared energy assets from the site electricity network and/or supplied to the site electricity network; wherein the grid flow data is further determined in dependence on the further flow data.

6. A method according to any of the preceding claims, wherein generating one or more modifications comprises modifying attributes of the specification data specifying: characteristics of one or more of the generator and/or consumer assets associated with properties; and/or characteristics of one or more of the shared energy assets.

7. A method according to any of the preceding claims, comprising evaluating the utilisation metric against a predetermined compliance criterion.

8. A method according to claim 7, comprising determining the modifications to the specification data in response to determining that the utilisation metric does not meet the compliance criterion.

9. A method according to claim 7 or 8, comprising comparing the utilisation metric to a predetermined compliance threshold or range, optionally wherein the utilisation metric is determined to meet the compliance criterion if it meets the threshold or falls within the range.

10. A method according to any of the preceding claims, wherein the utilisation metric is based on one or more of: a total electricity export to the distribution network over a simulation period; a total electricity import from the distribution network over the simulation period; a difference between the total electricity export and total electricity import.

11. A method according to claim 10, wherein the utilisation metric comprises or is based on a net energy import/export value.

12. A method according to claim 10 or 11 , wherein the utilisation metric is based on or comprises an estimated cost or profit value associated with the total energy export and/or import, optionally comprising computing the cost and/or profit value based on time-varying energy cost information over the simulation period.

13. A method according to any of the preceding claims, comprising repeating the evaluation process using the modified specification data.

14. A method according to claim 13, comprising: repeating the steps of modifying the specification data and performing the evaluation process until the utilisation metric for the modified specification data meets the compliance criterion; and outputting specification data indicating a proposed configuration for one or more assets based on the modified specification data.

15. A method according to claim 13 or 14, wherein the evaluation process is performed by an optimizer, wherein the optimizer iteratively modifies the specification data and evaluates the modified specification data based on the utilisation metric.

16. A method according to any of the preceding claims, wherein the specification data includes attributes specifying for each asset performance characteristics of the asset, optionally including one or more of: an electricity generation capacity of a generator asset, optionally of a solar PV generator; a storage capacity of a battery, and an electricity consumption of a consumer asset, optionally of a HVAC device such as a heat pump.

17. A method according to any of the preceding claims, wherein outputting specification data comprises outputting one or more proposed specification attribute values for one or more assets.

18. A method according to any of the preceding claims, wherein outputting specification data comprises determining one or more constraints on one or more specification attributes indicating values of the attributes required for compliance, and outputting the determined constraints.

19. A method according to any of the preceding claims, wherein outputting specification data comprises outputting one or more of: a suggested type, model and/or generator capacity of an electricity generator for a property or of a shared electricity generator; a suggested number and/or area of solar panels for an electricity generator for a property or for a shared electricity generator; a suggested consumption rating, type and/or model of a HVAC device for a property of for a shared HVAC system; a suggested capacity rating, type and/or model of battery for a property or for a shared battery or battery system.

20. A method according to any of the preceding claims, wherein the properties are connected to the distribution grid through a common connection point.

21 . A method according to claim 20, wherein the evaluation process comprises optimizing the specification data with respect to a network connection capacity constraint, optionally to reduce a required network connection capacity associated with the connection to the distribution grid at the common connection point.

22. A method according to any of the preceding claims, wherein the properties comprise houses or other dwellings, optionally wherein the site comprises a planned housing development.

23. A method of determining configurations for energy installations, comprising: receiving specification data specifying configurations for a plurality of energy installations, wherein each energy installation is associated with a respective property connected to an electricity distribution grid and comprises energy assets including: a solar photovoltaic (PV) generator arranged to supply electricity to one or more electricity consumers at the respective property, a battery and an electrically powered HVAC (heating, ventilation and/or air conditioning) device, wherein the specification data includes specification attributes for each energy installation indicative of an electricity generation capacity of the solar PV generator, a battery capacity of a battery and an electricity consumption of the HVAC device, and wherein the electricity generation capacities indicated by the specification data for two or more of the energy installations differ due to different arrangements of solar panels used by the respective solar PV generators; performing a search process to identify configurations for assets of the energy installations, the search process comprising iteratively modifying the specification data for the energy installations and evaluating the modified specification data until the specification data meets a compliance criterion, comprising, at each iteration: for each energy installation, simulating, based on the modified specification data, electricity generated by the solar PV generator, electricity consumed by the HVAC device and charging and discharging of the battery over a simulation period; based on the simulation, determining, for each energy installation, grid flow data indicating one or both of: a shortfall of electricity required from the distribution grid over the simulation period to supplement local generation by the respective solar PV generator, and an excess of electricity generated locally by the respective solar PV generator over the simulation period that is not consumed by consumers at the respective property and is available for supply back to the distribution grid; and determining whether the modified specification data meets the compliance criterion in dependence on the grid flow data; wherein the iterative evaluation keeps attributes of the specification data indicating the solar generation capacity of each energy installation fixed while modifying specification attributes indicating performance characteristics of other assets; and outputting configuration information for the energy installations based on the modified specification data identified as meeting the compliance criterion by the iterative evaluation.

24. A method according to claim 23, wherein specification attributes for each energy installation indicative of an electricity generation capacity of the solar PV generator comprise one or more of: an electricity generation capacity value, a number of solar panels, a type of solar panels, a size or effective generation area of solar panels, and an orientation of the solar panels, wherein two or more of the energy installations differ in at least one of said attributes.

25. A method according to claim 23 or 24, wherein the two or more energy installations differ in total solar panel effective area and/or orientation.

26. A method according to any of claims 23 to 25, wherein modifying the specification data for the energy installations comprises modifying one or more of: a battery capacity of a battery of at least one energy installation, and an electricity consumption of an HVAC device of at least one energy installation.

27. A method according to any of claims 23 to 26, wherein determining whether the specification data meets the compliance criterion comprises: aggregating the grid flow data to determine aggregate electricity flow between the plurality of energy installations and the distribution grid over the simulation period; determining a utilisation metric based on the aggregated grid flow data, and determining compliance based on comparing the utilisation metric to a predetermined threshold or range.

28. A method according to claim 27, wherein the utilisation metric is further determined in dependence on energy price information.

29. A method according to claim 27 or 28, comprising determining a net energy import metric based on the aggregated grid flow data, wherein compliance is determined based on comparing the net energy import metric to a predetermined threshold, optionally zero.

30. A method according to claim 29, wherein the net energy import metric is based on a net amount of electricity imported by the plurality of energy installations from the distribution grid over the simulation period or an associated energy cost value.

31 . A method according to any of claims 23 to 30, comprising determining the specification data as compliant if a difference between a total amount or cost of electricity imported from the distribution grid by the energy installations and a total amount or cost of electricity exported to the distribution grid by the energy installations does not exceed a predetermined threshold, optionally zero.

32. A method according to any of claims 23 to 31 , wherein the search process identifies changes to specifications of batteries and/or HVAC devices of one or more of the energy installations to compensate for different electricity generation capacities of the respective PV generators while maintaining compliance with the compliance criterion.

33. A method according to any of claims 23 to 32, wherein outputting configuration data comprises outputting specification attributes for one or more assets of the energy installations given by the modified specification data identified as meeting the compliance criterion.

34. A method according to any of claims 23 to 33, wherein outputting configuration data comprises outputting for one or more of the energy installations, optionally for each energy installation, one or more of: a suggested consumption rating, type and/or model of HVAC device; a suggested capacity rating, type and/or model of battery.

35. A method according to any of claims 23 to 34, wherein the simulating step comprises simulating operation of a battery control algorithm for charging and/or discharging the battery.

36. A method according to any of claims 23 to 35, wherein the simulating step comprises simulating operation of an HVAC device control algorithm for controlling operation of the HVAC device to determine electricity consumption of the HVAC device.

37. A method according to any of claims 23 to 36, wherein the (or each) HVAC device comprises a heat pump.

38. A method according to any of claims 23 to 37, wherein the plurality of energy installations are associated with properties on a site.

39. A method according to claim 38, wherein the properties are connected to the distribution grid through a common connection point, wherein the search process optionally comprises optimizing the specification data with respect to a network connection capacity constraint, optionally to reduce a required network connection capacity associated with the connection to the distribution grid at the common connection point.

40. A system comprising means, optionally in the form of one or more processors with associated memory, for performing a method as set out in any of the preceding claims.

41 . A computer readable medium comprising software code adapted, when executed on at least one data processing device, to perform a method as set out in any of claims 1 to 39.