EP4728607A1 - Faults in electricity supply networks - Google Patents

Abstract

Faults in Electricity Supply Networks A method of determining faults on an electricity supply network is disclosed which includes determining which consumers are supplied by a particular substation. A consumer is selected and, if their substation data is incomplete, a set of supply quality information is obtained in respect of their premises. The process is repeated for a number of other consumers and their respective data is compared with that of the first consumer. Once the sub-network is correctly characterized, further supply quality information is analysed to provide warnings of network problems, to provide short- and medium-term fixes for such problems and also provide long-term solutions involving re-design(s) of the network.

Description

Faults in Electricity Supply Networks

Electricity distribution networks distribute energy from electricity producers such as power stations and wind farms to industrial and domestic users. In order to distribute electricity efficiently, it is transmitted at high voltages, for example at 132kV or higher which reduces ohmic losses in the conductors. The electricity is usually distributed as alternating current (AC) which permits efficient transformation of the voltage. The voltage is transformed or stepped-down, usually in a number of discrete steps, before being provided to consumers at 415V, 230V, 110V or thereabouts. The higher voltage portions of the network are usually called the transmission network and the lower voltage portions (that supply consumers) are usually referred to as the distribution network.

During the past half century, the demands placed upon the distribution network have changed beyond recognition. Some changes have been beneficial, such as changes in working patterns tending to spread demand throughout the day but most changes have been deleterious. Among these are the proliferation of solar panels (or other microgeneration) installed at domestic premises and the increasing desire to charge electric vehicles. When domestic solar panels are producing more electricity than required at their own premises, they may be arranged to output electricity to the distribution network. This is done by increasing the voltage provided at the output of the inverter that converts the direct current (DC) generated by the solar panels into AC. This results in local fluctuations in the voltage on the network. For example, in the UK the nominal supply voltage is 240V but local variations as low as 235V and as high as 248V have been observed. Fluctuations in frequency also occur - the frequency reduces when the grid is heavily loaded and increases when the grid is lightly loaded. Proposals for discharge from electric vehicles will also increase the local voltage.

Such fluctuations in voltage and frequency can be exacerbated by islanding. This is where a region of the grid may need to disconnect due to issues, resulting a smaller grid or island. The smaller the island, the greater the volatility of voltage and frequency. Islanding can enhance the stability of the grid as can the use of interconnects between countries.

Electricity suppliers are obliged to keep the voltage and frequency of supply within a given specification and are often subject to sanctions (e.g. fines) when they fail to do this. The variations in voltage and/or frequency can be enough to upset the operation of appliances, shorten their lives and/or to trip protection devices which is likely to be inconvenient or even dangerous for the consumer. Network operators may need to form islands in the network or use interconnectors to maintain the electricity supply within specification.

Distribution network operators (DNO) also sometimes refuse permission for a consumer to output electricity from solar panels (or other electricity micro-generation sources) to the distribution network because the DNO deems that there is a likelihood that the local network may be destabilised. However, such refusals (via a G99 form in the UK) may often be unnecessary- the DNO simply doesn't have accurate and recent data regarding the network.

Fast charging of electric vehicles also places high demands on the distribution network because each vehicle can present a load of 7kW or more and this exceeds typical domestic demand by a factor of two or three times. As a consequence, network operators have also been forced to refuse permission to consumers to install fast-charging points for vehicles on the grounds that they are likely to destabilise the local grid. This is a particular problem as electric cars proliferate - in the UK, for example, there are expected to be 15 million such vehicles by 2030. Again, these refusals may be unnecessary.

Air conditioning places a substantial load on electricity supply. Another growing issue is the use of heat pumps, such as air-source heat pumps, as these typically consume two to three kilowatts continuously.

In the USA, load zones and hubs are used to set the wholesale price of electricity locally but this is usually done on a fairly crude basis.

It is an object of the present invention to ameliorate these problems.

According to a first aspect of the present invention, there is provided a method of predicting a fault in an electricity supply network, the supply network comprising a distribution network and a plurality of sub-networks each comprising at least one step-down transformer having a primary winding connected to the distribution network and a secondary winding connected via a plurality of conductors to a plurality of consumer premises, the method comprising: receiving, from a plurality of detectors located at respective consumer premises within the supply network, a plurality of samples of at least one electricity supply parameter selected from voltagefrequency, power factor and power consumption the value of the at least one parameter being sampled at a plurality of instances over a period of time, correlating the samples of the at least one supply parameter received from different consumer premises to determine which premises are supplied by each transformer, and for at least one premises determined to be supplied by a particular step-down transformer, comparing at least one value of at least one electricity supply parameter selected from voltage, frequency, power factor and sine wave quality, with a predetermined threshold or thresholds to generate at least one of: an alarm, an instruction to equipment connected to the sub-network to alter its behaviour, a projected life of the sub-network, and a modified design of the sub-network.

In preferred embodiments of the invention, one or more of these four outcomes may be combined in any combination. The electricity supply parameter used for the correlation step is preferably voltage. The plurality of samples preferably comprise the 10-second data available from smart meters. The 10- second data is a data value provided by the smart meter every 10 seconds but the precise timing may vary between regions and jurisdictions. Here, 10-second data means a frequently-supplied data that is distinct from the billing data transmitted every 24 hours or so by the smart meter. The 10 second data is thus preferably provided at least every 5 minutes, preferably at least every minute and more preferably at least every 30 seconds.

At the root of the problems identified above is a lack of visibility into local behaviour of the electricity distribution networks. There is some monitoring of the distribution network at the intermediate (e.g. llkV) levels but monitoring at the lowest voltage level is either non-existent or very limited. The status of the network also varies significantly with time, being influenced by weather, TV schedules and so on. By utilizing the data received from a network of detectors, a precise view of the distribution network can be obtained. This permits both short term fixes and longer-term planning to be effected. For example, it may indicate to the DNO that the addition of solar feed-in or electric vehicle charging infrastructure will not cause destabilization of the network.

Surprising though it may seem, mature distribution networks have typically grown over many years and it is often the case that the distribution network operator (DNO) does not know exactly which properties are supplied by a particular transformer or substation. By correlating parameter measurements according to embodiments of the invention, an accurate understanding of the network layout can be obtained.

The solution may comprise an alarm warning of a near-term assessment such as "the transformer is close to blowing up" or a longer-term assessment such as "the sub-network will require a transformer upgrade in 3 years". The assessment may also provide alternatives such as "if x amount of battery storage is provided now, a transformer upgrade can be delayed for y years".

Preferably the step of correlating the samples comprises comparing respective values of the at least one electricity supply parameter received from a first consumer premises over the plurality of instances in time with respective values of the supply parameter received from at least one other consumer premises. Preferably data is correlated from thousands or even hundreds of thousands of different premises.

The predetermined threshold may comprise a plurality of samples exceeding a predetermined value during a particular time duration. In other words, a brief excursion beyond limits may not, of itself, indicate a failure of the network. A more sophisticated arrangement could be based on a logical expression such as if parameter X exceeds Xi AND parameter Y falls below Yi OR if parameter Z exceeds ZiTHEN take appropriate action.

Since all networks behave differently, the predetermined threshold or thresholds may preferably be derived from historical data. This could be effected by a machine learning system to account for particular idiosyncrasies of the network being monitored.

When instruction to equipment within the network is generated, this may comprise an instruction to at least one electric vehicle to reduce charging current or an instruction to at least one microgenerator to reduce feed-in to the network. In order to fairly distribute the inconvenience of such measures, the equipment may be selected randomly within the sub-network.

The instruction to equipment preferably comprises an instruction to battery storage. Battery storage is charged when there is excess electrical energy available (detected by the voltage and/or frequency being above the desired levels) and is discharged to support the network when demand is high (detected by the voltage and/or frequency being below the desired levels). Battery storage thus operates to "smooth out" demand. Sometimes the operator of the battery storage is a separate commercial entity from the distribution network operator (DNO). In such cases, the DNO will usually want to minimize the use of the battery storage (to reduce cost). Consequently, the distribution network operator may be prepared to allow the grid parameters to fluctuate more widely (subject to meeting legal obligations) than if they owned the battery storage themselves.

When the outcome of embodiments of the present invention is a projected life, the projected life of the sub-network is derived by extrapolation of the electricity supply parameters using a predetermined algorithm. Such algorithm may be based on previous behaviour of the sub-network modified by population-wide factors such as expected penetration of electric vehicles and proliferation of energy-efficient appliances.

When the outcome of embodiments of the present invention is a modification of the network design, this may comprise an increase in local battery storage, fitting an uprated transformer or both an increase in local battery storage and a later transformer upgrade. It may also comprise a cabling upgrade. Requirement for such an upgrade can be better determined when a plurality of detectors are provided distributed throughout the network. In particular, a comparison of voltage levels at a premises near to the transformer (in terms of cable length) and a premises distant from the transformer may indicate that the transformer capacity is adequate but the cabling is underprovisioned.

The responsibility for quality of supply is at the point of delivery so it is advantageous to make the measurements as close as possible to the domestic termination point. Since the behaviour of the local grid can vary within a single street, it is advantageous to measure the quality of supply at as many properties as possible. However, the cost could become prohibitive unless the detectors are both cheap and easy to install.

The detectors preferably comprise consumer access devices (CADs) as these devices are capable of receiving frequent data from a smart meter and also have internet connectivity to report the data to a remote server. The data from the smart meter preferably comprises 10 second data. Sensing may be less frequent subject to a minimum sensing frequency. In order to reduce demand on communication and storage infrastructure, the data is preferably compressed or summarized prior to transmission to the server, in other words many data points are conveyed in a combined message.

The methods according to the first aspect of the present invention are preferably computer- implemented.

According to a second aspect of the present invention there is provided a method of operating a consumer access device, comprising receiving at least one sampled parameter from a smart meter, determining if the at least one sampled parameter is compliant with a predetermined threshold or thresholds, if the parameter is compliant with the threshold or thresholds, storing the value of the at least one sampled parameter, if the parameter is not compliant with the threshold or thresholds, transmitting a message to a server, and in response to a plurality of stored values of the at least one sampled parameter, transmitting the plurality of stored values to the server.

The CAD in this aspect operates in two modes - firstly it accumulates data from a smart meter and sends the data (which may be a compressed or summarized version) to a server and secondly, if the data is out-of-specification it sends an urgent message to the server.

According to a third aspect of the present invention, there is provided a consumer access device for a smart energy system, the consumer access device comprising: a processor, memory for storing at least one application, first communication means for receiving a plurality of samples representing at least one electricity supply parameter selected from voltage, frequency, power factor and sine wave quality, second communication means for communicating with at least one server to upload data representing quality of electricity supply.

In order to save transmission and storage overhead, the data representing quality of electricity supply preferably comprises a summary of the data.

The processor is preferably programmed to upload data urgently if an electricity supply parameter is not compliant with a predetermined threshold or thresholds. While it is known for devices in the consumer's home to report electricity usage in kWH (usually for billing purposes), this information is insufficient to provide the precise mitigation and planning opportunities afforded by embodiments of the present invention. At best, electricity usage is typically reported to electricity suppliers by smart meters once every 24 hours. The usage data also does not include voltage and frequency of supply information.

According to a fourth aspect of the present invention there is provided a computer implemented method of determining the extent of an electricity distribution network, comprising: receiving a least first, second and third sets of voltage values over time from respective premises, allocating the first set of voltage values to a first electrical supply region, processing the first or the second set of voltage values to permit a comparison, comparing the first set of voltage readings with the second set of voltage readings, in response to a match within a predetermined threshold, allocating the second set of voltage values to the first electrical supply region and in response to a failed match, comparing the second set of voltage values with at least the third set of voltage values.

Aspects and embodiments of the invention refer to a consumer access device or CAD.

The Consumer Access Device referred to herein is distinct from a smart utility meter and an in-home display (IHD). A smart meter is concerned with accurately and securely uploading data collected regarding utility usage within premises. Since it would be beneficial for utility customers to modify this data (in order to defraud the utility company) the smart meter must be a secure device. Typically, various tamper-proof and encryption techniques are deployed in the smart meter to ensure that accurate and unadulterated data are transmitted to the utility supplier. For example, the utility customer is never able to update the data or software resident on the smart meter. While over-the- air (OTA) updates of smart meters may be performed, this is a secure process managed entirely by, or on behalf of, the utility company. However, the smart meter may provide data locally on a read-only basis, for example over ZigBee™.

An in-home display is a device that receives usage data, typically from a smart meter, and displays this to the utility customer. Such devices may have different display modes, such as current usage of electricity, usage so far on a daily, weekly or monthly basis and so on. They may also indicate when a usage threshold has been exceeded. However, they are only capable of receiving data from the smart meter (such as over ZigBee TM) and displaying it to the customer. In contrast to the smart meter, a Consumer Access Device or CAD is capable of modification by the consumer. The CAD does not need to be arranged for secure relaying of utility usage data to a utility company server as this will be performed by the smart meter directly. This means that the CAD may be modified (for example by way of a software download) in response to consumer request.

In contrast to the in-home display, a CAD is a smart device capable of being programmed at the direction of the consumer. In this context, the term "programmed" means the installation of software that was not included in the device when provided to the consumer. This is distinct from "programming" a device by selecting the display of certain information, for example from a menu. This selection or customization is a "dumb" process since the software to provide all such displays is already installed or at least not installed at the request of the consumer. The software for installation is preferably developed by third-party developers unrelated to the utility company and selected by the consumer from a remote source such as an application store. The software is preferably downloaded via the internet (such as via a WiFi router in the premises) and is not provided via the smart meter in anyway. This has the benefit of not employing the secure connection between a smart meter and a utility company for data traffic that does not require such a secure connection.

The CAD is capable of receiving utility consumption information directly from a smart meter, for example over ZigBee™. This distinguishes the CAD from another device, such as a tablet or smartphone, that receives utility consumption information from a utility company server (i.e. information already uploaded by a smart meter and provided by way of an app on the consumer's device). This allows the CAD to operate using near-real-time data which would be impossible using the existing infrastructure since the smart meter usually only uploads data every 24 hours (typically for 48 half-hourly time periods). While it would, in theory, be possible to upload data from the smart meter (e.g. via DCC) more frequently, the load on the communications network would be unsustainable.

The software selected by the consumer preferably utilises utility consumption data provided by a secure portion of the CAD that receives the data directly from a smart meter. This allows third-party software developers to develop applications that are directly relevant to the consumer, for example an improved display of consumption data or an energy-reduction game. By providing a CAD with "secure" and "open" portions, security of utility usage data is preserved while allowing third party developers access to the platform. By empowering the development community to provide apps from consumers, customer engagement is enhanced and this could also provide a revenue stream to the utility provider (if consumers pay for apps).

The present invention will now be described by way of example, with reference to the following figures, in which:

Figure 1 is a simplified diagram of an electricity transmission and distribution network,

Figure 2 is a block diagram of a portion of a local distribution network,

Figure 3 is a block diagram of a consumer access device (CAD),

Figure 4 is a flow chart of the operation of the CAD,

Figure 5 is a block diagram of a network of CADs and a server,

Figure 6 is a flow chart of steps performed by a server to remediate problems detected on the local distribution network,

Figure 7 is flow diagram for execution by a server to determine which consumers are connected to a particular sub-network,

Figure 8 is a block diagram of three local distribution networks connected to the transmission network via respective transformers,

Figure 9 shows graphs of data collection from a single local distribution network and from multiple local distribution networks, and

Figure 10 is a flow chart of steps performed by a server to forecast the life of a local distribution network.

Figure 1 shows a simplified diagram of an electricity transmission and distribution network 100. A power station 102 generates electricity, for example from fossil fuels, nuclear generation or renewable sources such as wind farms. The power station outputs an alternating voltage which is fed to a transformer 104 (or series of transformers) to step the voltage up to a higher level. In the UK this voltage is 400kV or 275kV and this may be achieved in more than one step (i.e. two or more transformers increase the voltage in a number of discrete steps). Because power dissipated (i.e. lost) in the transmission and distribution networks is proportional to the square of the current flowing, it is important to reduce the current and this is done by increasing the voltage. The voltage is later stepped-down to suitable levels for supply to consumers as close as practicable to the consumer's premises.

The high voltage is distributed by a transmission network 106 to a number of substations 108 of which one is shown. The term substation comes from a historical pre-grid era when electricity would be generated at a small generating station and consumed within a limited local area. Such substations now house step-down transformers rather than any generating plant. The substations 108 include transformers that step-down the voltage to a first intermediate level, which is 132kV in the UK.

The 132kV output of each of the substations 108 is supplied to a number of further substations 110 of which only one is shown for clarity. These substations step the voltage down to a second intermediate level (the penultimate level), which in the UK is typically IlkV. Sometimes a further intermediate level of 33kV is also used. The IlkV output of each of the substations 110 is supplied to a number of substations 112 of which only one is shown. This is the final substation in the sequence and, in the UK, has an output of 415V phase-to-phase (three phase) which is around 240V phase-to-neutral. This voltage output is provided to consumer premises such as houses 114, 116, and apartment block 118. The final link is usually provided by buried cables in towns and cities.

While this example has focused on the UK, other networks will operate at different voltages and use different numbers of step-up and step-down transformers. However, the principle is the same, i.e. a high voltage core portion of the network supplies successive step-down transformers which ultimately supply consumers.

At each step the substations become smaller and more numerous which means that there is considerable fan-out of the network. There are around 230,000 of the final step-down substations in the UK (population 67 million) each serving up to around 500 properties. In isolated areas there are even more (around 350,000) small, pole-mounted transformers, that each serve a few, or even only one, properties.

The challenge for operators of the distribution network is to provide reliable electricity supply within defined parameters of voltage and frequency, 24 hours a day, 365 days a year. To do this effectively, the operators have to forecast or estimate information about demand and consumer behaviour. Monitoring is provided on the network at the higher voltage levels (IlkV and above) and large consumers such as factories are supplied at this voltage. These consumers are usually also obliged to provide detailed and frequent information to the operator. However, the network operator is typically blind when it comes to the behaviour of the local 415/240V portion of the network. This means that there might be serious and/or imminent problems that are entirely invisible to the operator. Also, in order for network operators to meet their supply obligations, they may consider battery storage and load-shedding. Both of these options have drawbacks. Battery storage can be connected to the distribution network at each voltage level to smooth out demand on the network. The battery storage is arranged to charge when demand for electricity is low and then discharge to provide electrical power onto the network when demand is high. It is vital, of course, to provide sufficient battery storage capacity to meet the load-smoothing demand. However, the requirement for battery storage at the local 415/240V portion of the network is very hard to determine. Batteries are expensive and so over-provisioning battery storage will increase costs unnecessarily. Battery storage can also be costly in terms of real estate. However, under-provisioning battery storage will not solve the problem of variation in quality of electricity supply.

One solution to excessive demand is load-shedding which can occur when certain high-usage customers such as factories are requested to reduce demand for a period. This is something that electricity suppliers and network operators clearly wish to keep to an absolute minimum.

Figure 2 shows a part 200 of the local portion of the distribution network, i.e. the 415/240V portion in the UK. A sub-station 202 (equivalent to the substation 112 in Figure 1) provides electricity to consumer premises of which seven 210, 212, 214, 216, 218, 220, 222 are shown. As mentioned earlier, such a sub-station may typically serve up to 500 properties. The overall monitoring system may include thousands or even hundreds of thousands of properties. Each property includes a respective smart electricity meter 230, 232, 234, 236, 238, 240, 242, that provides frequent supply information ("10 second data") over a short-range ZigBee™ network. ZigBee™ is a secure wireless communication standard and more details can be found . The information available typically includes voltage, frequency, power consumption and power factor sampled and transmitted every 10 seconds. Other means of measuring and communicating such information may equally be used.

Each property is also provided with a respective Consumer Access Device (CAD) 230, 232, 234, 236, 238, 240 and 242. The CADs are arranged to receive the information from their respective smart meters over the ZigBee™ wireless connection and are also connected to the Internet (not shown), for example via a WiFi™ router located in the property. The CADs are each arranged to be able to send the 10 second data to a remote server (Figure 5) for processing. This data may be sent in real-time or near real-time as will be discussed below. CADs will also be described in more detail with reference to Figure 3. Three of the premises shown in Figure 2 have an additional feature which differentiates them from the basic premises. Property 216 has a solar panel 270 together with the ability to feed excess solar electricity to the distribution network. This is performed by an inverter that converts the DC generated by the solar panel to AC. In order to feed the excess electricity onto the distribution network, the output voltage of the inverter must be raised above the voltage level of the local portion of the network. During times of low demand for electricity, this could cause the voltage to be greater than the maximum permitted level. Other types of microgeneration (such as wind power) also cause the same problem. Properties 218 and 220 each have an electric car 280, 282 which is connected to the distribution network via a fast-charging point. Such fast-charging points typically allow vehicle charging at 7kW or more so that, during charging, the load presented by an electric vehicle is around 2 to 3 times the normal domestic load from a property. Where several electric cars are being fast- charged within a local distribution network, this could cause the voltage and/or frequency of the supply to drop below the minimum permitted level. Other high-consumption devices such as air- conditioning may also cause the same problem.

The operator will not have any information as to when consumers are charging, or intending to charge, their cars, although they do have some limited information (from weather forecasts) as to when there may be significant feed-in activity from solar panels.

Figure 2 also shows battery storage 290 connected to the network at the substation 202. This battery storage has the ability to store excess energy when the local network is lightly loaded (and voltage and frequency tend to rise) and to return electricity to the network when demand is high. This type of local storage is discussed further below.

A further complexity may arise in deregulated electricity markets such as the UK. The electricity supplier for an individual consumer is a distinct entity from the distribution network operator (DNO). In other words, the DNO has no direct relationship with the consumers. Where the consumer has a smart meter (or other monitoring device) in their premises, it is the electricity supplier that is provided with the usage data, not the DNO.

The consequence of this is that the local portion of the distribution network fed by a single sub-station will serve customers from a number of different electricity providers. However, each electricity provider will usually have enough customers within each local portion of the network for the present methods to function. Agreements between providers may also ameliorate this problem. Figure 3 is a block diagram of a Consumer Access Device (CAD) 300.

A processor 302 comprises a System-on-Chip (SoC) including memory and one or more processors as are available from a number of semiconductor manufacturers. For communication, the processor interfaces with a ZigBee™ transceiver 304, a WiFi transceiver 306 and a Bluetooth™ transceiver 308. Although each of these transceivers is shown as having its own antenna for clarity, the transceivers may share an antenna to save cost and space. Certain manufacturers also provide WiFi and Bluetooth™ transceivers on a single chip. To interact with a consumer, the CAD also comprises a display 310 and an activate button 312. In this example the display is a matrix of 12x24 LEDs but other displays will be equally appropriate. The display shows images known as "cards". In some embodiments the display is optional since information can be displayed to a consumer using their user device. In addition, the CAD may be provided with a voice activated interface (not shown). A 5-volt power supply 314 is provided via a USB socket 315 which may be provided with a battery back-up as desired. When the CAD is provided with a powerline transceiver (not shown) the power supply is also capable of sending and receiving data, for example with appliances on the premises, via the electricity cabling in the premises. A secure memory 320 is provided either on the SoC or as a stand-alone entity and an optional alarm 322, such as an audible alarm, is provided for warning a consumer.

The figure also shows, within the dotted lines, a number of applications 318 stored in the memory of the processor 302. These applications are managed and executed by Docker™ 316 which allows different software (arranged in "containers") to be run separately while sharing kernel functions in the processor. A key point to note is that data can be provided to a particular application by placing that data in a particular application's container. Applications may not access the contents of other application's containers, preserving security. Further information may be found at https://www.docker.co /. Docker™ saves computer resources when compared with a virtual machine (VM) architecture but a VM architecture may equally be used.

Each CAD has a MAC address, install code, serial number and security certificate (for ZigBee™ communication) that are all unique to the device and generally stored in the secure memory 320.

The processor 302 assumes at least two roles - firstly a secure role in which the consumer consumption or usage data is received via the ZigBee™ transceiver 304 and stored in the secure memory 320. Secondly, an application execution role in which it executes the applications 318 (in cooperation with Docker™ 316).

The CAD comprises an activate button 312 which is pressed to exit a sleep mode of the CAD. Alternatively, the CAD may operate at all times in which case the activate button may be omitted. The activate button may also be used by the consumer to instruct the processor to toggle to execute a desired application or to instruct pairing to a user device.

Once activated, the processor of the CAD executes the applications according to a policy set by the supplier, although this may be customised by the consumer. In the simplest example, the applications are executed in rotation - each application is given sufficient time to display any output to the consumer before the display goes blank and another application is given access to the processor and display. Alternatively, the applications may be executed in order of priority and/or provided with an override function in which an application having crucial information to display is given exclusive access to the processor. A number of "background" applications may be arranged to execute continuously until overridden by a "foreground" application such as a game. Any policy may be governed to some extent by the arrival of the 10 second data from the smart meter.

The display screens generated by the applications are known as cards. The consumer has control (for example via the user interface on a paired user device) of which cards will be displayed by each application and which data is used to generate that card. For example, a consumer has three applications installed in their CAD:

- 10 second data

- electric car battery status

- weather forecast.

The 10 second data application displays "live" energy consumption and includes a number of different cards which use different graphical displays. The user selects one of these cards and arranges for the 10 second application to be displayed for at least 20 seconds.

The charge/discharge condition of the car battery changes less frequently and so the user arranges the car battery status application to execute every minute and display the battery status for 10 seconds. The weather forecast changes less frequently but requires more attention from the consumer. In this case the consumer arranges for the weather forecast application to be displayed every two minutes for 30 seconds. The exact scheduling may have to vary slightly to accommodate the consumer's requests.

The apps may be installed or removed as desired.

Preferably, the CAD is pre-programmed with at least one application such as a Snake game. The consumer may control the game using controls on the CAD (not shown) such as buttons or a touchscreen. Preferably, however, the user interface is provided using another device such as a smartphone which includes a CAD application and has been paired to the CAD. Further CAD functionality will be described with reference to the following flow diagrams.

While a CAD has been described which has a display 310 and alarm 322, these elements may be omitted.

The CAD may be provided with an interface to allow hardware upgrades such as additional sensors and user interface functions. The skilled reader will be familiar with such arrangements.

The CAD according to embodiments of the present invention is programmed to upload quality of supply information to a server as described with reference to Figures 4 and 5.

Figure 4 shows a flow diagram 400 of the operation of a CAD in accordance with an embodiment of the present invention. The CAD is located within one of the consumer premises of Figure 2. In addition to other functions, the CAD accumulates 10 second data from the smart meter (or other source) and, in a first mode, provides this to a supplier server for subsequent processing which is discussed below. In this example, the CAD is also operable in a second mode to warn the supplier immediately if the voltage or frequency of supply stray outside predetermined limits. These two modes are preferably provided as an application on the CAD as discussed above, particularly a "background" application of which the consumer is largely unaware. Alternatively, the functionality could be provided at a firmware level.

The process starts at step 402 and 10 second data is received from the smart meter at step 404. At step 406, the CAD compares the voltage and frequency values with predetermined limits to monitor the quality of supply. If the values meet the predetermined acceptable limits then processing proceeds to step 408 and the parameters are stored locally. At step 410, the CAD determines if it has accumulated enough data to transmit to the server. If so, the data is transmitted at step 412 and processing reverts to step 404; if not, processing reverts directly to step 404. If the parameters were found to be out-of-specification at step 406, the data is transmitted immediately at step 412. In this case, the message transmitted to the server is preferably identified as urgent in some way.

Sometimes a brief excursion beyond the limits of voltage and frequency can be tolerated, provided that it does not occur for longer than a predetermined time. In another example, therefore, the voltage and the frequency may be compared with thresholds for a period of time before triggering a transmission to the supplier.

In this example, the 10 second data is transmitted every 30 minutes so it will include data from 180 instances in time. In order to reduce the load on the reporting network, the CAD may further be arranged to compress the data or generate a summary for transmission to the server, rather than transmit the raw data.

The CAD thus performs a periodic upload of data and an urgent upload of data when the quality of supply fails to meet a predetermined parameter or parameters. While the CAD described effectively has two modes of operation, it will be appreciated that this need not be the case in practice and the CAD could be arranged to operate continually in either of the two modes or other variants. In addition, the data collected and relayed to the server may differ from just voltage and frequency. For example, the data may also include consumption and power factor and/or be collected and transmitted at different intervals from those described.

Figure 5 shows an arrangement 500 of CADs and a server used for processing the data collected from the CADs. CADs 510, 512, 514, 516, 518, 520 and 522 correspond with CADs located at different premises as in Figure 2. Each CAD is connected to a server 530 via in the internet 532, such as via a WiFi router located at their respective premises. Other transmission technologies may equally be used such as SMS messages. It is conceivable that data is transmitted to the server directly from the smart meter (i.e. no separate device such as a CAD is required). The server may comprise the smart meter infrastructure already in place for billing consumers, programmed to provide the additional functionality described here. The server is also connected via the Internet to control battery storage 540 which corresponds to the battery storage 290 in Figure 2. The battery storage comprises a battery, an AC-to-DC converter (or rectifier) which is used to charge the battery when there is a surplus on the local sub-network and a DC-to-AC converter (or inverter) which returns the stored energy to the local network when there is a shortfall. While stand-alone battery storage is shown as being associated with the substation, the battery storage may be distributed throughout the network, may be located in consumer's premises or may even comprise the battery of an electric vehicle.

Alternatively, the battery storage may be controlled locally in response to locally-sensed behaviour of the network. This will be the case when the battery operator and the network operator are distinct entities. In other words, the battery storage may be operated autonomously.

Over time the server will accumulate data such as shown in Table 1:

And so on. This could, of course, be stored in a database which also includes customer information such as address, credit and banking details etc. Further columns may be provided in the table, such as an indication that an electric vehicle is being charged or that microgenerated electricity is being fed from that premises to the network. Figure 6 shows a flow chart 600 of the operation of the server (530, Figure 5) in remediating problems on the local distribution network. The server will be receiving data from at least one CAD (or equivalent device) located in a premises supplied by the local network. Which premises are supplied by which substations (i.e. are located in a particular local network) has been determined beforehand, for example as described with reference to Figure 7 below.

The process starts at step 602 at which it receives live data from a CAD (or other device) which represents the behaviour of the local distribution network. Whether the data complies with the requirement(s) is determined at step 604 and if there are no problems, processing returns to step 602. If a problem is detected, processing proceeds to step 606 which determines whether the problem is due to too high a load or too low a load.

If the problem is determined to be due to an excessive load (such as if the voltage and/or frequency are too low) then processing proceeds to step 608 where the server determines whether any electric vehicles are being charged in the local distribution network. This may be done by interrogating CADs within the same local network or by referring to the data collected from premises and stored in Table 1. If there are one or more electric vehicles being charged then the server can, at step 610, instruct that the charge rate be reduced (or "throttled") to relieve the load on the local network. The throttling may be to zero (i.e. charging ceases) if variable throttling is not possible or if the network is under severe strain. Where only some electric vehicles are affected, they could be selected on a random basis to share any inconvenience fairly among consumers. Processing returns to step 602 to receive the next data update from the CAD and the performance of the local network will be checked again at step 604.

If no electric vehicles are found to be charging at step 608, processing proceeds to step 612 at which the server instructs local battery storage (Figure 2, 290; Figure 5, 540) to provide some support for the network in terms of voltage. Processing then returns to step 602.

It is worth noting that battery storage may operate autonomously to increase the grid voltage should it fall below a target value but this can be a rather blunt instrument. The battery storage is preferably arranged to increase the voltage of the local distribution network to between predetermined parameters, rather than just increase the existing voltage by a certain amount. If, at step 606, the problem is determined to be insufficient load (such as if the voltage and/or frequency are too high) at step 606 then processing proceeds to step 614. At step 614 the server determines whether any microgeneration is being performed and fed into the local network. It may do this by interrogating CADs located within the local network or by consulting a database as discussed above. If microgeneration feed-in is identified then the server instructs a reduction in the feed in (possibly to zero) at step 616 and processing returns to step 602. If no microgeneration feed-in is identified at step 614 processing proceeds to step 618 at which battery storage is instructed to load the local network (i.e. charge its batteries) in order to bring the network operating parameters back within specification. Control then reverts to step 602 and the process repeats.

Whatever remediation is applied, since the process always reverts to step 604, a feedback loop is provided to ensure that the local distribution network is returned to operation within the specified parameters (and will re-activate any interrupted devices when it is possible to do so).

While electric vehicles have been used as an example of a significant load on the network, the principle applies equally to other high-consumption devices such as air conditioning.

The amelioration steps discussed above are purely exemplary. For example, it may be preferable to utilise battery storage in the first instance. Throttling consumption or feed-in from microgeneration could then be used when battery storage capacity is insufficient or has already been exhausted. It is also be possible to encourage consumption at times of light loading such as via a reduced price tariff and vice versa. For example, if the voltage and/or frequency are too high and this could be remedied by suspending feed-in from micro-generation, the feed-in tariff could be adjusted to be zero. This means that consumers with micro-generation facilities receive no payment to feed electricity onto the grid - this is intended to discourage the feed-in.

The extent to which this process is automated can also be varied. In the simplest scenario, the server could just raise an alarm that the local network was not performing to specification. Remediation actions could then be effected by a human operator.

The server may also receive data from two or more premises served by the same transformer to derive a more accurate understanding of the performance of the local network. Properties further from the substation (in terms of a longer run of cable, rather than strict geography) will usually register a lower voltage than those closer to the substation due to ohmic losses in the conductors. This is particularly evident when the network is heavily loaded. By comparing the voltage reported at a premises near to the substation and at a premises further from the substation, the server can estimate the load on the local network. This may be useful in at least two scenarios. Firstly, the server may not be provided with consumption data from the properties (possibly due to data-protection concerns) and can thus infer the load from the voltage drop. Secondly, even where the electricity supplier does receive consumption data, it may only receive the information from a proportion of the premises supplied by the local network. This second scenario can arise due to low penetration of CADs among consumers or in a deregulated electricity market (where a supplier can only receive information from its own customers).

While Figure 6 shows a server performing remediation on a single distribution network basis, the steps may be performed across multiple distribution networks.

Figure 7 shows a flow chart 700 of another process performed by the server that receives the information from the CADs. This process is performed to determine which consumers are being supplied by particular substations. As noted above, and surprising though it may seem, DNOs often do not know exactly which premises are supplied by a particular substation.

The process starts at step 702 and at step 704 a first consumer is selected from the supplier database. This can be done in any suitable manner although it may be preferred to select the first consumer whose premises are geographically close to a particular substation on the understanding that the particular consumer is likely to be supplied by that substation. At step 706 it is determined whether the relevant substation for the consumer has already been identified. If so, the process returns to step 704 to select another consumer. Provided that the relevant substation has not yet been identified for that consumer, processing proceeds to step 708 at which the performance data received from the consumer's CAD is downloaded. At step 710 another consumer is selected. Any suitable selection criteria could be used but it is preferable to select a consumer within a given geographical distance of the first consumer, i.e. one that could conceivably be supplied by the same substation. At step 712 it is checked whether that consumer's substation has been identified and, if so, processing returns to step 710.

Provided that the another consumer's substation has not yet been identified, processing continues to step 714 at which their performance data is downloaded. At step 716 the two sets of performance data are compared and if there is a match, processing continues to step 720. If there isn't a match, processing returns to step 710.

Unfortunately, it is not guaranteed that the data samples are taken at the same instants in time so some extrapolation and/or interpolation may be performed between samples to permit comparison of data from different premises. A straight-line interpolation may be used or different approaches will be known to the skilled reader. In principle the smart meters could be instructed to take the samples at the same instant (for example, aligned to Network Time Protocol (NTP)) but there would still be processing and communication delays so it is likely that some further processing of the data would still be required.

At step 720, the database is updated to confirm that there is a match between the data, i.e. that both of the selected consumers are supplied by the same substation. At step 722 it is determined whether there is data for more consumers to be compared with that of the first selected consumer. If so, then processing returns to step 710 and if not, processing continues to step 724. At step 724 it is determined whether there are more comparisons to be made or whether all of the necessary consumers have had their substation identified. If more comparisons are required then processing returns to step 704 and if not, processing ends at step 726.

In summary, a first consumer is selected and the characteristics of their supply (e.g. as received from their CAD) are compared with those of other consumers to determine which consumers share a particular sub-station. The 10-second data for the first consumer is compared with 10-second data for a plurality of other consumers. Where there is a high degree of correlation, it is assumed that the currently-considered consumer is supplied by the same substation as the first consumer. This can be further confirmed using location data such as postcode (zip code) data for the consumers.

One possibility is to compare the voltage at each 10 second interval for the various consumers. It is possible, however, that there is a degree of variation in the voltage detected at premises supplied by the same sub-station as discussed above. Properties further from the substation will likely detect a lower voltage than those closer to the substation due to ohmic loss in the distribution network. This is more likely if one or more consumers distant from the substation have high electricity usage at the point in time that the voltage was sampled. To accommodate this variation, the comparison at step 716 may be arranged to compare variations in voltage between the 10 second data from different consumers rather than absolute voltage. Imagine that a consumer starts to charge an electric car during the sampling period. This is likely to cause a reduction in voltage detected at every property supplied by the same sub-station. Where such a reduction is detected at the same time and is of a similar magnitude (or proportion), it can reasonably be assumed that the properties are supplied by the same substation, even though their absolute voltage values differ. The same is, of course, true of in increase in voltage detected at different properties.

Once the comparison engine has compared the data from the first consumer with data for all consumers within a reasonable geographic distance of the first consumer's premises, the database will contain a first list of consumers that are fed from the same substation and a list of consumers that aren't. A second consumer is then selected from outside the first list and the process repeated for all of the consumers that are within a reasonable geographic distance of the second consumer and are not supplied by the first substation.

This process continues until all of the consumers have been allocated to a substation. The information may be communicated in any suitable form, such as a map. The map may be supplemented by historical performance data, particularly data collected during times of strain on the network. These could be times when the network operator struggled to comply with performance parameters and may be used to plan upgrades to the network such as a sub-station transformer upgrade or the provision of (more) battery storage.

Note that in rural areas, a transformer (typically mounted on a pole) may feed only a small number of properties. Since these properties may have different electricity suppliers, it is possible that a consumer's data does not match that of any other consumers. This simply means that there is only one consumer supplied by a particular supplier and a particular transformer.

It is also possible that there are such transformers which do not supply any consumers which are customers of a particular electricity supplier in which case the supplier will not have any information concerning that portion of the distribution network. However, this is not problematic since this represents a very small proportion of consumers on the network. Alternatively, electricity suppliers could collaborate to share data, particularly in remote areas where fewer than 20, or even 10 premises are served by a particular transformer. As a further alternative, the CADs (or other devices) could be provided and/or managed by a third party that may anonymise the data before sharing it with the suppliers.

Figure 8 shows a portion 800 of the distribution network showing data collection from a plurality of local distribution networks. A portion 802 of the distribution network supplies three substations 804, 806, 808 which each supply a respective local distribution network 810 (DN1), 812(DN2), 814(DN3). Having performed, for example, the process of Figure 7, at least one CAD 816, 818, 820 (or other device) associated with each local distribution network is selected to provide data via the internet 822 to a server 824. As stated previously, further CADs associated with each local network may be selected to improve insight into the local network and/or to provide a backup should a CAD cease to send data for some reason. Each CAD sends data to the server, for example as described with reference to Figure 4. The server is thus provided with frequent data representing the behaviour of each local network. Figure 9(a) shows the characteristics of voltage and frequency for a single local network over a period of time showing a single instance when the characteristics were out-of-specification. Figure 9(b) shows the instances when either characteristic was out-of-specification for the three local networks. The thickness of the vertical line corresponds to the degree to which the characteristic was out of specification. It can be seen that DN2 suffered from more out-of-specification incidents than DN1 or DN3. Some medium to long term remediation is required to DN2 such as providing battery storage or adding transformer capacity at the relevant substation 806 (Figure 8). Other remediation techniques will be apparent to the skilled reader, including the use of market levers to influence demand. A graph or graphs as shown in Figure 9 may be provided in real time, for example on a network engineer's computer. The engineer is able to zoom in on any of the incidents to determine the nature of the problem and its duration. The graphs may be enhanced to also indicate what remediation (such as those shown in Figure 6) was applied at the time.

Figure 10 shows a flow chart 1000 of a method for predicting the remaining life of a local distribution network and ranking local networks to determine upgrade priorities. This may be executed by the server shown in Figure 8. The process starts at step 1002 and at step 1004 historical data for the local network is recovered or downloaded. This data may be the 10-second data for a period of time such as a month or even up to several years. It may comprise data for the most difficult month from a number of consecutive years (typically January in northern hemisphere cooler climates) or all of the available data. Alternatively, it may comprise the characteristic data around the failure of the network to perform within specification. Processing continues to step 1006 at which the network performance data is extrapolated for some time into the future, typically several years, to produce an estimate of how often the network will fail to perform within specification. This can be performed using any suitable extrapolation technique consistent with the nature of the data. Processing continues to a further forecasting step 1008 (which may be optional or combined with that of step 1006). Step 1008 further processes the extrapolated data to account for expected demographic changes across the whole population. This could include the expected proportion of consumers on the local network that will buy an electric car or install a heat pump at their premises. It could also include factors that are likely to reduce net consumption such as an increase in the use of energy-efficient appliances or installation of solar panels.

At step 1010 the useful life of the network is predicted using the extrapolated data. This life could be determined as the point in time at which the frequency of out-of-specification incidents exceeds a given threshold or when predicted peak consumption exceeds the design parameters of the local network. It can also be determined whether addition of local battery storage would benefit the network, in particular whether addition of battery storage would be a cost-effective measure to defer more expensive upgrades to the network.

At step 1012 it is determined whether there are more networks to analyse and, if so, processing returns to step 1004. If all of the relevant sub-networks have been analysed processing proceeds to step 1014 at which an optional ranking is performed. This step compares the predicted life of each of the local networks that have been analysed to identify the network or networks for which an upgrade is most urgent. The process ends at step 1016.

The process described with reference to Figure 10 thus provides a network operator with responsibility for multiple sub-networks to determine maintenance and upgrade priorities for years hence. This may also assist with budgeting. Clearly this situation is dynamic to some extent and the process can be repeated at regular intervals, for example every 6 months, to take account of new data and possible changes in demographic forecasts.

Claims

1. A computer-implemented method of predicting a fault in an electricity supply network, the supply network comprising a distribution network and a plurality of sub-networks each comprising at least one step-down transformer having a primary winding connected to the distribution network and a secondary winding connected via a plurality of conductors to a plurality of consumer premises, the method comprising: receiving, from a plurality of detectors located at respective consumer premises within the supply network, a plurality of samples of at least one electricity supply parameter selected from voltage, frequency, power factor and power consumed the value of the at least one parameter being sampled at a plurality of instances over a period of time, correlating the samples of the at least one supply parameter received from different consumer premises to determine which premises are supplied by each transformer, and for at least one premises determined to be supplied by a particular step-down transformer, comparing at least one value of at least one electricity supply parameter selected from voltage, frequency, power factor and power consumed, with a predetermined threshold or thresholds to generate at least one of: an alarm, an instruction to equipment connected to the sub-network to alter its behaviour, a projected life of the sub-network, and a modified design of the sub-network.

2. A method as claimed in claim 1, wherein correlating the samples comprises comparing respective values of the at least one electricity supply parameter received from a first consumer premises over the plurality of instances in time with respective values of the supply parameter received from a plurality of other consumer premises over the plurality of instances in time.

3. A method as claimed in claim 1 or claim 2, wherein the same samples of the at least one supply parameter are used for the correlation and comparing steps.

4. A method as claimed in any one of the preceding claims, wherein the predetermined thresholds comprise thresholds for different parameters.

5. A method as claimed in any one of the preceding claims, wherein the predetermined threshold comprises a plurality of samples exceeding a predetermined value during a particular time duration.

6. A method as claimed in any one of the preceding claims, wherein the predetermined threshold or thresholds is derived from historical data.

7. A method as claimed in any one of the preceding claims, wherein the instruction to equipment comprises an instruction to at least one electric vehicle to reduce charging current.

8. A method as claimed in any one of the preceding claims, wherein the instruction to equipment comprises an instruction to at least one microgenerator to reduce feed-in to the network.

9. A method as claimed in claim 7 or claim 8, wherein the equipment is selected randomly within the sub-network.

10. A method as claimed in any one of the preceding claims, wherein the instruction to equipment is an instruction to battery storage.

11. A method as claimed in any one of the preceding claims, wherein the projected life of the subnetwork is derived by extrapolation of the electricity supply parameters using a predetermined algorithm.

12. A method as claimed in any one of the preceding claims, wherein the modified design comprises an increase in local battery storage.

13. A method as claimed in claim 12, wherein the modified design comprises the increase in local battery storage and further comprises a later transformer upgrade.

14. A method as claimed in any one of the claims 1 to 12, wherein the modified design comprises an uprated transformer.

15. A method as claimed in any one of the preceding claims, wherein the modified design comprises a cabling upgrade.

16. A method as claimed in any one of the preceding claims, wherein the comparison between at least one value of at least one electricity supply parameter selected from voltage, frequency, power factor and power consumption, with a predetermined threshold or thresholds comprises a comparison of electricity supply parameters received from a plurality of consumer premises.

17. A method as claimed in any one of the preceding claims, wherein the plurality of samples of at least one supply parameter are received from a consumer access device.

18. A method as claimed in any one of the preceding claims, wherein the plurality of samples of at least one supply parameter comprise 10-second data from a smart meter.

19. A method as claimed in any one of the preceding claims, wherein the plurality of samples of at least one supply parameter are received in combined message.

20. A method of operating a consumer access device, comprising receiving at least one sampled parameter from a smart meter, determining if the at least one sampled parameter is compliant with a predetermined threshold or thresholds, if the parameter is compliant with the threshold or thresholds, storing the value of the at least one sampled parameter, if the parameter is not compliant with the threshold or thresholds, transmitting a message to a server, and in response to a plurality of stored values of the at least one sampled parameter, transmitting the plurality of stored values to the server.

21. A method as claimed in claim 20, wherein transmitting the plurality of stored values to the server comprises transmitting at least a summary of the plurality of stored values to the server.

22. A consumer access device for a smart energy system, the consumer access device comprising: a processor, memory for storing at least one application. first communication means for receiving a plurality of samples representing at least one electricity supply parameter selected from voltage, frequency, power factor and sine wave quality, and second communication means for communicating with at least one server to upload data representing quality of electricity supply.

23. A consumer access device as claimed in claim 22, wherein the data comprising quality of electricity supply comprises a summary of the data.

24. A consumer access device as claimed in claim 23 or claim 24, wherein the processor is programmed to upload data urgently if an electricity supply parameter is not compliant with a predetermined threshold or thresholds.

25. A computer implemented method of determining the extent of an electricity distribution network, comprising: receiving a least first, second and third sets of voltage values over time from respective premises allocating the first set of voltage values to a first electrical supply region processing the first or the second set of voltage values to permit a comparison, comparing the first set of voltage readings with the second set of voltage readings in response to a match within a predetermined threshold, allocating the second set of voltage values to the first electrical supply region and in response to a failed match, comparing the second set of voltage values with at least the third set of voltage values.