A review of the Building 16 project at BRE by Andrew Williams (BRE), Al-Azhar Lalani (University of Hertfordshire), Emilio Mistretta (University of Hertfordshire) and Johann Siau (University of Hertfordshire)
- Background – The future of electricity
- The B16 project
- System control
This paper provides a brief overview of an experimental electricity network installed in Building 16 (B16) at BRE‚Äôs site at Watford. The objective of the installation is to act as a research and development platform to explore how an electricity network in a building equipped with various generations sources, storage and demand management can be optimised for a number of outcomes including electricity cost, carbon emissions and availability (lifestyle need). The project particularly focuses on mobile device (laptop, tablet, smartphone) interfacing and explores how their internal power management functions can be externalising to engage with a broader building network strategy. It also focusses on the aggregation of product and network performance data to support the chosen control regime.
The work outlined in this report is part of a broader package of research and development initiatives aimed at systemising passive and active components in a building. The plan is to be able to integrate various components by way of a lowest common denominator and focus the resultant system on the value provided to the occupant. In terms of developing control systems, electricity is an obvious place to start and the work outlined in this paper develops on from a report reviewing the future of electricity in domestic buildings .
Peak demand for electricity across all sectors on a cold day in Great Britain is approximately 61 Gigawatts (GW). In a year, approximately 330 TWh of electricity is generated and consumed the majority of which is produced by burning coal and gas, and by nuclear power stations. By 2050 electricity consumption is expected to be over 550 TWh driven by significant increases in the use of electricity for residential heating via heat pumps and domestic transport. To meet this demand the installed capacity will almost double from approximately 80 GW today to approximately 160 GW in 2035 . However, to meet carbon dioxide targets much of the new capacity will be more intermittent as a result of renewable generation and less flexible as a result of nuclear generation.
Over the next 30 years electricity supply will go through a revolution as the grid essentially becomes decarbonised and local micro-generation becomes common place. Under the National Grids ‚ÄėGone Green‚Äô scenario  the plan is that by 2025 emissions produced by the use of electricity will fall to approximately 70gCO2/kWh from that of over 400gCO2/kWh today. By 2035 this figure is projected to be well below 50gCO2/kWh, however, to achieve this requires significant investment in low carbon generation capacity and challenging energy efficiency targets must be met.
The demand is also changing. With the introduction of smart meters, domestic users will be subject to Time-of-Use Tariffs and will be paying for electricity in a broadly similar way to many commercial and industrial sites that pay for electricity through half-hourly settlement periods. Local generation will become much more commonplace and each of the generating technologies will have their own particular characteristics related to the cost of generating electricity, efficiency, carbon emissions and availability. For example, if the generation technology is a renewable technology it can be considered as producing no carbon emissions in-use (not taking in to account production) or if it uses primary fuels to produce electricity locally it may avoid grid distribution losses and hence be more efficient overall. These differing characteristics will open up a whole new range of opportunities for managing and optimising the local electricity network in relation to user need, grid supply, local generation characteristics and storage capabilities.
One of the key approaches to balancing supply with demand is Demand-Side Management or Demand Response. In terms of balancing supply and demand, reducing demand is as creditable as increasing supply so long as it does not lead to unacceptable disruption to the consumer. Demand response is a key tool in any future electricity strategy however, at a certain stage of system development, the direct relationship between consumer demand and electricity supply places limitations on the control scenarios that can be adopted. To expand the control scenarios further it becomes necessary to decouple the time-of-use of electricity from its generation and to do this requires the addition of storage, typically batteries, to the system. While these bring additional complexity, cost and maintenance issues to the network they do allow for electrical energy to be stored and hence consumer demand to be more-or-less independent of electricity supply, either nationally or locally. In the end, the success of any control strategy for future building electricity networks depends on their ability to manage all these variables and provide a seamless electricity supply in relation to what service the electricity is providing to the consumer and the cost and carbon implications of its generation and supply.
Currently battery storage is not typically economic (depending on payback period, battery technology and cost) based just on electricity price variations (demand charges) throughout a typical day or the difference in price between import and export of locally generated electricity. Having local storage can provide backup should the grid supply fail but in the UK at least this is a fairly rare occurrence for most people. However, daily electricity charges may become more dynamic in the future and new battery technologies and advanced, high volume production processes are bringing down costs part driven by the need to satisfy the mobile, renewable energy and transport agenda. Already in parts of the US, New York City for example, high electricity demand charges mean that current battery prices are only approximately 25% greater than breakeven levels for peak shaving applications and the cost of lithium-ion batteries has fallen by nearly 50 per cent over the past 5 years. Alternative technologies to lead-acid and lithium-ion are also receiving more attention including sodium-ion batteries, lithium-air batteries, lithium capacitors, lithium-sulphur batteries, solid state batteries and many more. These technologies may further reduce the cost of storage as lithium-ion battery technology and pricing fully matures in the next 10 years.
Having storage on the local electricity network can have a number of benefits including maximising the use of locally generated electricity, reducing export to the grid and smoothing the daily profile. Electricity storage can also help with power quality issues such as frequency stabilisation and end-of-line network reinforcement although this is possibly larger scale and more likely to be associated with the supply-side rather than the demand-side.
While many of these are very important in electrical engineering terms, having electricity storage¬†¬† greatly expands the control options for the network and opens up the possibility to focus on the ‚Äėavailability‚Äô part of the energy trilemma in its own right or in relation to the cost of electricity and/or its carbon emissions. In terms of the social impacts of electricity, storage potentially allows a function to be performed by a consumer at a time when the electricity supply environment is unfavourable by using stored electricity from when it was favourable. As a result, storage can be considered not just in electrical terms but as providing a ‚Äėsocial‚Äô buffer as consumers go through a process of behavioural change driven by the new electricity horizon.
Even without the addition of batteries on the electricity network either as part of, for example, photovoltaic inverters or stand alone, the mobile revolution has resulted in a small but meaningful amount of storage distributed across most domestic and commercial electricity networks. This may not represent a significant level of storage compared to the overall demand and/or perhaps the level of local renewable generation but their numbers are growing rapidly and they are a new category of device that needs to be managed in a future low carbon electricity strategy.
For the purposes of producing a development platform, unlike many other current electricity consuming appliances and devices they do, however, come with highly advanced communication capabilities and often moderately good battery interfacing and control functions. As a result, they provide an ideal platform for exploring and developing future network management approaches. Also, interestingly, in terms of controlling the overall network a surprising amount of flexibility can be achieved in relation to optimising the system with a relatively small amount of storage compared to the overall daily consumption. Optimising demand with storage capacity (inside a device or stand alone) correctly can make a meaningful difference to the performance of a local electricity network and the cost effectiveness of storage when fitted. Even small reductions in consumption or shift in daily profile can, when aggregated up, make a significant difference to the supply of electricity nationally.
Building 16 is BRE‚Äôs Environmental Building. It was completed in 1997 with a ‚Äėpassive‚Äô design brief including building orientation, passive stack ventilation, high thermal mass, ground sourced heating/cooling etc. It was chosen as a suitable building for the research project based on the ease with which the bespoke electricity supply network could be installed, space adjacent to the building for photovoltaic installation and an office based environment with many desktop (230 V) and mobile devices.
The primary objective of the B16 project is to:
Investigate how a local electricity network with local generation and storage can be managed and instantaneously optimised for various outcomes including least cost, least carbon emissions and availability.
There are a number of other underpinning requirements, to;
- Explore the opportunities for greater interaction between the power management strategies of ICT devices (with and without internal batteries) and those of the broader building electricity network; essentially externalising the ICT power management strategies
- Explore what learning can be applied from ICT local and wide area networks to power networks
- Consider how data convergence can be used to optimise electricity consumption in line with building performance and lifestyle need
- Develop future control scenarios for a local electricity system
- Develop hardware to enable the overarching control strategy to be implemented on a conventional (non-smart) 230 V AC network.
To undertake research and product development the following system was installed. In many ways the system is technically relatively simple, however the key is the programming necessary to interrogate and control the devices together with the overarching intelligence necessary to manage the network to meet the required outcome.
To simplify the installation of a dedicated and controllable electricity network the installation in Building 16 is based on the Power over Ethernet (PoE) protocol. This protocol allows a standard Ethernet cable (cat5e and above) to distribute approximately 30 W (IEEE, 802.3at) at approximately 50 V DC to end devices. As with most office environments, B16 is fully equipped with an Ethernet network as part of its ICT infrastructure with often one or more connections at each workstation.
The hard wired Ethernet LAN (Local Area Network) is able to power almost any product that falls within the power limits whether it has the inbuilt capability or via a suitable splitter (DC to DC converter) at the point of load. This includes devices such as VOIP phones, access cameras, terminals, VDU‚Äôs, thin clients (computers) and much more. While PoE provides an end-to-end power and data capability the B16 project at BRE primarily focuses on device interfacing and functional/power control in relation to a broader network control strategy. The use of PoE provides a quick and simple way to produce a dedicated electricity distribution system that is controllable (ICT interfaced) and provides power to every work station.
The network is supplied with renewable energy from a 3 kWp ground mounted photovoltaic (PV) array located outside the front of B16 which is fed to the network via a conventional inverter. The star connected supply ensures that electricity is always available to keep business operations running as usual but that it is only taken from the grid supply as a last resort and then only ideally when the demand is low and/or grid carbon is low (high renewable generation). PV was chosen as it is a widely adopted renewable source however, the lessons learnt can be applied to other types of local generation. For the purposes of network management the electricity generated by the photovoltaics represents a zero carbon, time-dependent source. This can be changed to accommodate other parameters such as embodied carbon, tariffs etc. if required.
In the future, central storage will become more common place either via plug-in vehicles importing/exporting electricity or associated with PV inverters or stand-alone battery installations. The level of network storage is important. Having batteries at all brings increased complexity and over sizing or under sizing them has a big impact on battery life and cost effectiveness. As a result, if batteries are to be used one of the key questions is what is the minimum total network storage capacity (distributed (mobile devices) or centralised) required to meet the desired outcome?
Convergence is a common theme for the future and it applies to technologies, service offerings, content delivery and business models. Already in the commercial sector cloud based energy management systems are allowing those with significant property portfolios to collect, analyse and control energy across all their sites. In the domestic sector, while each building is different, shared data and cloud operations could provide new insights as to how buildings perform and how physically and managerially they can be improved. Cloud based operations will also ensure scalability and cost effectiveness by providing significant computing power and data storage that is well beyond the average small and medium enterprise or domestic consumer.
Convergence can also usefully describe a localised process by which the power management strategy of each individual electrical and electronic device interacts with their neighbouring device and/or the electricity supply. Electricity convergence will mean that each device will move away from demanding electricity to one where it will negotiate for electricity with others on the network depending on its device characteristics, status and priority. While it is easy to expect too much from the smart agenda, smart devices in the future will not only fulfil their design function but they will do this in relation to user lifestyle and how this interacts with a distributed and yet integrated network of generation, storage and demand. As the functional performance of the device becomes more user configurable and the network becomes smarter and self-learning then essentially electricity consumption will naturally become much more related to need or even, due to the responsiveness of many electronic devices, the location of a person within a building.
The work being undertaken by the B16 project is also exploring convergence between ICT networks and electricity networks. Clearly there is convergence by way of the Ethernet network supplying power to end devices but in the B16 project this is largely for convenience. The research being undertaken is considering what lessons can be learnt from local and wide area ICT networks (LAN‚Äôs and WAN‚Äôs) and how these lessons might be applied to managing local power networks.
Resulting from the work the concept of a VeLAN (Virtual electricity Local Area Network) is being explored:
A VeLAN (Virtual electricity Local Area Network) is a local electricity network where end devices with common requirements or characteristics can be managed independent of their physical location across the VeLAN. VeLAN also refers to a set of protocols and rules that define and bound electricity allocation and network control intelligence to optimise electricity in accordance with building performance data and occupant need.
In some ways a VeLAN can be seen as a buildings equivalent of what is happening in the supply-side by way of smart grids and it takes some of its learning from existing data network management understanding and techniques.
In terms of B16, the control of system devices and their power use is provided via the Internet for IP addressable devices and where the power management function is already embedded as part of the PoE protocol. For those devices with no power management or addressable capability an intermediate switching and storage device (a charging station) specifically designed and built for the project is included. These provide the system capabilities necessary to allow the network control algorithm to function correctly. In terms of the algorithm intelligence itself, this interacts with each component in the electrical system and, via cloud operations, gathers the necessary data to perform the appropriate functions to meet the required outcome. The cloud data can consist of grid data, weather data, generic building performance data, lifestyle data etc. By doing so the control algorithm and the service aggregation platform are able to provide the data and computing power necessary to maximise the benefit from many bits of fragmented and disparate data taken from different sites, buildings, devices, databases, equipment and sensors.
In very simple terms there are a number of basic decisions and actions that need to be made associated with the availability of PV, the status of the mains supply, the battery status and the end use. For example, if a battery is continued to be charged when it is fully charged this will result in a waste of electricity no matter whether its grid or locally generated. If the battery remains fully charged as a result of the previous days control regime then it is unable to make a contribution to storing today‚Äôs renewable electricity. Some appliances, for example a washing machine, can be run during the day or at night when electricity conditions might be more favourable whereas others, for example a TV, has to be on and consuming electricity when required by the viewer. Often one of the key aspects in developing the control regime is in relation to when and how much electricity is used by specific consumers at specific times. By knowing this, the supply and storage capabilities can be managed to do the best for today while leaving the system in a state that allows it to best optimise itself, no matter what impact the weather is having on renewable generation, tomorrow.
For the PoE devices in B16 the distribution of electricity is controlled by enabling or disabling power to the appropriate Midspan port using the protocol SNMP (Simple Network Management Protocol). By using this method remote control can be gained over charge times depending on the system criteria at the time and the desired outcome while at the same time providing a certain battery charge level to guarantee business activity carries on as normal.
The system is programmed to only apply the above control statements to ports which have a port type defined as ‚Äėbattery‚Äô which stops devices such as desktop PC‚Äôs being turned off by accident. For devices being powered via the bespoke charging stations (see 4.4 later) the precise level of charge and other details are fed back to the control algorithm via a ZigBee link. Other devices and ports are configured according to their own load characteristics.
The overall control strategy implemented on the B16 system is based on a series of interacting subroutines the simplest of which is illustrated in the flowchart shown in Figure 3. The first decision is whether the amount of solar being generated is greater than the demand. If this is the case then the system will always enable power to the PoE ports to run the devices. If solar generated electricity is not greater than the demand then a check is carried out to see if the current time is within working hours (for example 8am ‚Äď 7:00pm). If this is true then the battery level of each device is checked to see if it is equal to or greater than 50 per cent and if so power is disabled to that particular PoE port and only battery power is used. If the battery level is below 50 per cent then the system will detect whether or not the device is currently being used and if it is then power is provided; if not power is stopped.
The scenario outlined above is a simple indicative example of a subroutine that, along with many other subroutines, define a series of decisions associated with specific aspects of system control. They could be related to the nature of the electricity supplies, network capabilities, the type of end device being powered, and how and when the device is being used. Some of these can be very specific, for example avoiding charge/discharge ripple or hysteresis (small charge/discharge fluctuations) that can shorten battery life, or more simply the need to ensure that say the battery remains 50 per cent charged to meet the immediate need. The network will also have a number of set points which are related to hardware or in-use parameters and the control strategy must be able to deal with these without the whole system becoming unstable as a result of hitting an ‚Äėend-stop‚Äô. The control approach could be based on a predictive model i.e. predict the need and then try to match the supply to meet it or a more open and instantaneous model; balancing or selecting the outcome at any point in time. While business continuity is a priority the energy trilemma of cost, carbon emissions and availability (security of supply) may change on a more or less instantaneous basis depending on the system needs and capabilities.
No matter which approach is taken, for the overall network to be managed properly there is a need for a lot of data surrounding hardware, electricity supply environments and use. Some of this is already available but it needs an aggregation platform to pull it together in a way that allows the control intelligence to use it to manage the local network.
¬†4.2¬†¬†¬†¬†¬† Web service/database structure
Each application (subroutine) created requires access to a database so that data can be retrieved and stored. To allow these applications to communicate with a database a web service has been created to extend the SASH (Service Aggregation for Smart Homes) service. This service allows communication through a series of ‚Äėget methods‚Äô to retrieve data and ‚Äėset methods‚Äô to send data to various tables within the database. There are eleven tables just to store information about different parts of the system, these include:
- Device table: this table contains information regarding the end devices being powered by PoE.
- Injector table: this table contains data about the PoE injector unit including details such as injector IP address, number of ports, maximum power and injector username and password. Multiple entries can be added if there is more than one injector being monitored or controlled.
- Injector data: this table is used to store the operational values produced by the injector and it typically consists of output values such as voltage, power, current and temperature.
- Meter table: this table stores information about the primary meters (VMU 1, 2, 3; see Figure 2) such as meter address, system configuration, application mode and meter ID.
- Meter data: this table stores the operational values of the meters. These are the output values which are sent from the meters to the web service.
- Port data: port data contains information about the status and performance of each individual port.
- User table: this table contains login information for the users who will be using the applications.
- Port Control: the port control table holds values that are used when ports are enabled and disabled which includes a time stamp enabling specific events to be identified.
- Remote Power Management Hardware: this table is used to store values from each of the bespoke charging stations such as ID, battery voltage, load current, battery charge level and device address. It also ensures that the signal strength from each ZigBee link is in communication range with the main control PC.
- Remote Power Management Hardware History: this table stores meter data history, port data history and injector data history which can be used to, for example, plot graphs of previous charge cycles, photovoltaic generation etc.
- Devices: the device table is used to store performance and use data from different end devices including battery level, brightness level and device current state i.e. in-use or idl
Demand management or demand response is a key approach to managing electricity in the future and being able to manage a device either in terms of its performance, time-of-use, or if fitted, battery storage level can open up many possibilities when it comes to controlling the overall network. Microsoft Windows based mobile laptops and tablets read the state of charge of their internal battery typically in terms of a percentage value. By using primarily the system information library found in Visual Studio access can be gained to key information such as battery level and other device settings and features that influence power consumption and battery life such as display brightness.
By designing an application that runs in the background whilst a user is using, or not using, a laptop the device can play its part in a broader network control strategy. The role of the web service is essential in this application as all the data is stored in the ‚Äėdevice table‚Äô an example of which is shown in Figure 5 (power status) and Figure 6 (display and application) below.
Once the application is running the first step is to check whether the device exists in the database. A device is identified by an ID and if the application doesn‚Äôt recognise the ID in the database table it will immediately allocate one for the device (see Figure 7 below)
Figure 7 illustrates the data used in the application. The last item in the coding is known as ‚ÄėState‚Äô and it is a control variable. There are 6 system States in total:
- State 0: Disable = application is running but there are no checks taking place
- State 1: PC is on = PC is powered on
- State 2: Application running = there is currently an application running on the PC
- State 3: System is powered off
- State 4: Execute Shutdown
- State 5: Execute hibernate
States can be used for two purposes. Firstly States 0 to 3 tell the application what the status of the device currently is and secondly States 4 and 5 can be used to execute commands.
States 4 and 5 allow power management to be implemented, however, it is not acceptable to simply shut down a laptop to save power if a user has applications running and potentially unsaved work open. With this in mind State 2 comes into play. This is used to check the laptop or PC to see if any applications or processes are currently running. If the answer is no the user will be prompted that the PC will shut down (or hibernate) within 30 seconds although the user can aborted the process if required.
The second aspect of power management that can be applied using this application is to control display brightness. Display brightness has a significant impact on the battery life of a laptop (or consumption of a desktop) and by implication the electricity consumed across the local network.
To arrive at a fully integrated network turning devices and appliances on or off is only part of the solution. To do it thoroughly, the internal consumption and control capabilities, to some extent, need to be externalised to engage with the local network and their control must be closely linked to electricity consumption and the benefits the user derives from the device.
To be able to manage a local electricity network, or any other system for that matter, each of the components across the network must have the appropriate electrical performance, control capabilities and communication interfaces. Even for the most connected and smart devices such as ICT for example, the ability to influence their power consumption is often limited by one or more of these categories and for other non-smart devices the control options are simply manual, ‚Äėon‚Äô or ‚Äėoff‚Äô. Potentially the Internet-of-Things (IoT) and Machine-to-Machine (M2M) communications will help with this but to facilitate the research undertaken in B16 a custom made charging station was installed. The charging station contains a battery which, if there is no device plugged in charging, stores the power provided by the PoE connection. The charging station also has a number of sensors to monitor electricity supply, the state of charge of the battery etc. It has two charging ports, one for universal devices and the other for Apple products.
A key aim of the custom charging station was to have full control over what it does and what it can monitor and report. By introducing charging stations across the network it enables network management strategies to be researched despite many of the devices connected to the network having limited or no control options. To make this easy, and provide a stepping stone for the development of a new commercial product, the charging station uses ZigBee enabled wireless modules. ZigBee is a protocol that is used with the module XBee to provide the ability in the future to exchange data from various devices and then eventually, via a microcontroller, to the cloud and the aggregation platform.
Figure 10 below shows the completed charging station PCB and components housed in a laser cut acrylic enclosure.
To help visualise and report the network performance, energy analysis software has been produced. The software was developed to display a breakdown of total system utilisation and generation either by daily, weekly or monthly periods. A simple indicative display is shown in Figure 11 below.
The overall performance of the installation in B16 is graphically represented as in Figure 12 and 13 and this provides a visual representation of the storage levels and real-time energy readings.
By using the charging stations in association with the inherent control of normal ICT devices (with and without storage) a number of interesting scenarios have been explored in demand response and network power management. These control scenarios increase in both breadth, depth and responsiveness as the control envelopes and monitoring capabilities of each device on the network improves.
The B16 project is in its early stages but it has already identified a number of novel control scenarios for optimising a local electricity network which when fully developed will be further enabled by intelligent analytics within the cloud. This will incorporate learning and adaptive control methods based on the comparison of analysed data to BRE‚Äôs baseline building performance data.
As the research continues new intermediate hardware is being developed to support a wider role out of certain control strategies to non-IP addressable devices and appliances. The work has also led to further research areas associated with externalising device power management strategies, novel in-building occupant location systems and advanced device characterisation. These projects are supported by the BRE Trust and a summary report will be published shortly.
As ICT, and electronics more generally, becomes more imbedded throughout appliances, devices and even the fabric and fixtures of a building whole new control opportunities will open up. Already greater connection has had significant impact on the way social networks and communities are managed, the way business is done and many other aspects of daily life. As building systems become more comprehensive the potential benefits offered to occupants and how they use resources and engage with the building will explode. The link between smartness (and connectivity) and potential occupant benefit is not linear but that of a watershed. Buildings, while still fulfilling their historic functions of weather protection, safety and comfort will become an active part in a service delivery supply chain focused on occupant lifestyle and the resource implications.
In many ways electricity is an ideal candidate to become more ‚Äėconnected‚Äô and ultimately much more in tune with daily life. It has significant potential to reduce carbon emissions, it can be monitored and controlled often with tried and tested existing technologies, there are existing distribution networks and communication channels (nationally and within buildings) and many of the services it provides to the consumer are more-or-less instantaneous (perhaps excluding thermal use) allowing for a much more responsive consumer orientated control strategy.
The Building 16 (B16) project is being undertaken in association with the University of Hertfordshire. Some aspects of the work are being completed as doctoral studies and the authors are grateful to BRE Trust for funding. The authors would like to thank John Counsell for some early in-depth discussions and initial project support.
 Williams A. The Future of Electricity in Domestic Buildings ‚Äď A Review. December 2014. Designing Buildings Wiki. Downloaded from http://www.designingbuildings.co.uk/wiki/File:The_Future_of_Electricity_in_Domestic_Buildings.pdf
 National Grid. UK Future Energy Scenarios, UK gas and electricity transmission 2014 (updated 2015). Accessed 4 July 2015. Downloaded from http://www2.nationalgrid.com/uk/industry-information/future-of-energy/future-energy-scenarios/