new thames valley vision · knowledge and support in helping the new thames valley vision project...

56
New Thames Valley Vision SSET203 LCNF Tier 2 SDRC 9.8D Project Closedown Report

Upload: truongphuc

Post on 30-May-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

New Thames Valley Vision

SSET203

LCNF Tier 2 SDRC 9.8D Project Closedown Report

SRDC 9.8D Project Closedown Report V 0.4 SSET203 NTVV

New Thames Valley Vision

Page 2

Scottish and Southern Electricity Networks (SSEN) is the new trading name of Scottish and Southern Energy Power

Distribution (SSEPD), the parent company of Southern Electricity Power Distribution (SEPD), Scottish Hydro Electricity

Power Distribution (SHEPD) and Scottish Hydro Electricity Transmission. SEPD remains the contracted delivery body for

this LCNF Project.

Prepared By

Document Owner(s) Project/Organisation Role

Joe McNeil SSEN NTVV Project Manager

Gideon Evans SSEN NTVV Project Engineering Manager

Charlie Edwards SSEN NTVV Customer Project Manager

Mark Coulthard SSEN NTVV Project Engineer

Mark Stannard SSEN NTVV Project Engineer

Joshua Martin SSEN NTVV Data Analyst

Arun Singh SSEN NTVV Customer Project Manager

Sarah Rigby SSEN NTVV Project Engineer

Chris Jeans SSEN NTVV Project Engineer

Helen Waller SSEN NTVV Project Engineer

Matthew Windows SSEN NTVV Project Support Officer

Alex Howison SSEN Innovation Programme Team Manager South

Colin Mathieson SSEN Innovation Programme Delivery Manager

SRDC 9.8D Project Closedown Report V 0.4 SSET203 NTVV

New Thames Valley Vision

Page 3

Version Control

Version Date Authors Change Description

0.1 17/03/17 Joe McNeil First Draft for review

0.2 21/03/17 Joe McNeil Updated with initial review comments

0.3 24/03/17 Joe McNeil Further initial review comments for formal review

0.4 31/03/17 Joe McNeil Revised for external Peer review comments, update for close down events

and revised project financials.

Acknowledgement Scottish & Southern Electricity Networks would like to thank their project partners General Electric (GE) - Grid Solutions,

DNV-Kema, EA Technology, Bracknell Forest Council, University of Reading, and University of Oxford for their ongoing

knowledge and support in helping the New Thames Valley Vision project successfully achieve its goals in understanding,

anticipating and supporting changes in consumer behaviour, helping the Distribution Network Operators (DNOs) in their

aim of developing efficient networks for the low carbon economy.

This project could not have delivered the learning outcomes derived from this initiative without wide ranging support

from our customers, necessary for virtually all aspects of the project including smart metering, end point monitoring,

demand response, hot thermal, cold thermal and battery storage, either directly through the installation of equipment

within their homes or businesses or indirectly through the installation of equipment on their street.

SRDC 9.8D Project Closedown Report V 0.4 SSET203 NTVV

New Thames Valley Vision

Page 4

Contents

1 Project Background ................................................................................................................................ 6

2 Executive Summary ................................................................................................................................ 8

3 Details of the Work Carried Out ........................................................................................................... 12

4 The Outcomes of the Project ................................................................................................................ 23

5 Performance Compared to the Original Project Aims .......................................................................... 34

6 Required Modifications to the Planned Approach ............................................................................... 39

7 Significant Variances in Expected Cost ................................................................................................. 40

8 Updated Business Case and Lessons Learnt for the Method ............................................................... 41

9 Lessons Learnt for Future Innovation Projects ..................................................................................... 42

10 Project Replication................................................................................................................................ 44

11 Planned Implementation ...................................................................................................................... 46

12 Learning Dissemination ........................................................................................................................ 52

13 Key Project Learning Documents .......................................................................................................... 55

14 Contact Details ..................................................................................................................................... 55

15 References ............................................................................................................................................ 55

16 Appendices ........................................................................................................................................... 56

List of Tables

Table 1 HTS Methodology ..................................................................................................................................... 19

Table 2 ADR Capabilities ........................................................................................................................................ 31

Table 3 Project Spend Vs Budget ........................................................................................................................... 40

Table 4 Business Case Forecast Comparison ......................................................................................................... 41

Table 5 Topics of Interest ...................................................................................................................................... 53

Table 6 Feedback from Internal Dissemination Events ......................................................................................... 54

List of Figures

Figure 1: New Policy Documentation Overview .................................................................................................... 21

Figure 2: NTVV Training Packages & Modules ....................................................................................................... 22

Figure 3 Large Commercial Customer Recruitment .............................................................................................. 32

Figure 4 Dissemination Channels .......................................................................................................................... 52

Figure 5 DNO appetite for NTVV ........................................................................................................................... 53

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 5

Glossary of Terms

Abbreviation and Term Abbreviation and Term

ADDM Active Distribution Device Management kWh Kilowatt hour ADR Automated Demand Response LCCAC Low Carbon Community Advisory Centre BaU Business as Usual LCNF Low Carbon Network Fund BFC Bracknell Forest Council LCP Low Carbon Promotion BMS Building Management System LCT Low Carbon Technology CIM Common Information Model LV Low Voltage CIRED International Conference on Electricity Distribution MPAN Meter Point Administration Numbers COTS Commercial Off The Shelf MID Meter Instrument Directive CTS Cold Thermal Storage MDI Maximum Demand Indicator CMZ Constraint Managed Zone MNE Multinational Enterprise DCC Data Communications Company MoU Memorandum of Understanding

DISCERN Distributed Intelligence for Cost-Effective & Reliable Distribution Network Operation MW Megawatt

DNO Distribution Network Operation NME Network Modelling Environment DNO Distribution Network Operator NINES Northern Isles New Energy Solutions DMS Distribution Management System NPG Northern Power Grid DRAS Demand Response Automation Server NPV Net Present Value DSO Distribution Service Operator NTVV New Thames Valley Vision DSR Demand Side Response PEU Power Electronics Unit DPA Data Protection Act PoF PowerOn Fusion EDRP Energy Demand Research Project PSR Priority Services Register EMMA Energy and Micro-Generation Manager PV Photovoltaic ENW Electricity North West Q&A Question and Answer EPM End Point Monitor RAG Red Amber Green ESQCR Electricity Supply, Quality and Continuity Regulations RCPC Remaining Capacity per Customer ESMU Energy Storage and Management Unit RSL Registered Social Landlord ESU Energy Storage Unit RTS Real Time System EV Electric Vehicle RTU Remote Terminal Unit GAM Generalised Additive Model SAVE Solent Achieving Value from Efficiency GB Great Britain SCADA Supervisory Control and Data Acquisition GE General Electric SDRC Successful Delivery Reward Criteria GIS Geographic Information System SIM Subscriber Identification Module GPRS General Packet Radio Service SIMS Service Interruption Management System GSM Global System for Mobile Communications SME Small and Medium sized Enterprise HP Heat pump SMOS Smart Meter Operating System HTS Hot Thermal Storage SOAP Simple Object Access Protocol HV High Voltage SPEN Scottish Power Energy Networks I&C Industrial and Commercial SSEN Scottish and Southern Electricity Networks ICT Information and Communications Technology TNO Transmission Network Operator IET Institute of Engineering and Technology ToU Time of Use IFI Innovation Funding Initiative UKPN UK Power Networks kVA Kilovolt-amp UMTS Universal Mobile Telecommunications Service kVAr Kilovolt-amp reactive VIP Very Important Person kW Kilowatt WPD Western Power Distribution

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 6

1 Project Background

On 18th August 2011, SSEN submitted a Low Carbon Network Fund (LCNF) Tier 2 funding request to Ofgem for the New Thames Valley Vision (NTVV) project. Ofgem confirmed funding on 27th November 2011, issuing their project direction on 19th December 2011.

This five year project focused on the Low Voltage (LV) network with the aim to demonstrate how electricity distribution networks can better serve their customers by understanding, anticipating and supporting energy use as we move towards a low carbon economy.

Project objectives 1. Applying proven data analysis from the Energy Demand Research Project (EDRP) to understand the different

customer types connected to the distribution network, and their effect on network demand.

2. Understanding how the behaviour of different customer types allows informed network investment decisions to be made.

3. Demonstrating mitigation strategies, both technical and commercial, in a live environment, to understand: (a) The extent to which demand side response (DSR) (when customers change their energy usage in response to certain external triggers) can contribute to network flexibility, and identifying which customers are most likely to be early and effective adopters of DSR, and; (b) Where and how power electronics (with and without energy storage) can be used to manage power factor, thermal constraints and voltage to facilitate the connection of renewables on the LV network.

4. Undertaking dissemination and scaling activity to ensure validity and relevance to Great Britain (GB), with learning and understanding provided at two levels: (a) Provide front line training courses for the industry to embed real practical knowledge and skills, and; (b) Keeping the public informed so the intentions and benefits of the smart grid are clear.

Project Purpose Electricity demand patterns are changing as individuals, small businesses and larger companies increasingly act, either on their conscience or in response to economic stimuli, to reduce their carbon footprint. The options available to customers include: energy efficiency measures; the installation of solar thermal or photovoltaic (PV) panels and other small-scale renewable energy devices; an increased uptake of electric vehicles (EVs); and adoption of heat pumps (HPs).

This clearly poses challenges for DNOs, some of which have been demonstrated by data gathered from our LCNF Tier 1 Chalvey LV monitoring project. In order to maintain and operate a reliable and cost-effective electricity distribution system, DNOs need an understanding of the expected power flows on their networks. However, at present, DNOs have no sight of the demand of the smaller individual customers, and can only make estimates based on averaging data relating to the total number of customers fed from a distribution substation. The NTVV project is focused towards, but not exclusive to, LV networks. Data is already indicating issues on the LV network resulting from changing demand profiles, however traditional reinforcement of this asset across GB could cost up to £30.9 Billion1.

At present, LV network investment is informed by periodic measurements taken at substations. This approach is widely used, but has the disadvantage of being a relatively crude lagging indicator with no cognisance of variations over shorter timescales. With low carbon technologies now accelerating change on network, this crude monitoring leaves the network exposed to stress as load factors vary; it also hides capacity that could be otherwise utilised. Without advanced monitoring and the smart use of data, the network will require significant capital investment to support the transition to a low carbon economy, whilst ensuring security and quality of supply to customers.

1 NTVV Project bid submission http://www.thamesvalleyvision.co.uk/download/bid-submission-pro-forma/

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 7

Project Methods NTVV proposed that a better understanding of electricity consumers and their more active engagement with the energy system can minimise the investment required to maintain secure distribution networks that meet customer needs.

1. Link network reinforcement to a better understanding of electricity consumers The project aimed to demonstrate that mathematical and statistical techniques commonly used in other fields, such as consumer retail, can be applied to electricity consumers and incorporated into network planning processes. Such analysis will help to:

Target investment and the strategic placement of 'distributed LV solutions';

Facilitate scenario planning;

Minimise errors in network design; and

Reduce risk to connected customers

This sophisticated analysis will be complemented by credible alternatives to conventional network reinforcement for resolving any network issues identified.

2. Interact with demand response provided by both large and small businesses Demand response trials undertaken with both large commercial and Small and Medium sized Enterprise (SME) customers, supported by project partner Honeywell, investigate the capability of this approach in supporting operation of distribution networks. The learnings from thermal energy storage in the form of domestic water heating (as initially trialled in our Northern Isles New Energy Solutions (NINES) project on Shetland) expand the project’s investigation of demand side technologies.

3. Use mathematical techniques to reduce the need for LV monitoring Utilisation of customer profiling to inform where network monitoring should be targeted will optimise the investment needs for application of monitoring the LV networks.

4. Tactically deploy power electronics and electrical energy storage on the LV networks Deployment of power electronics and electrical energy storage on LV networks will demonstrate the extent to which these technologies can manage power factor, harmonics and voltages to provide a fast and flexible alternative to managing network issues. By providing a ‘tactical buffer’, customers have the freedom to deploy low carbon technologies without waiting for time-consuming reinforcement (or their alternatives) to take place. These technical solutions can be fully integrated into the distribution network control room.

Location - The Thames Valley area, to the west of London, takes in parts of Oxfordshire, Berkshire, Buckinghamshire, Hampshire and Surrey. It has a diverse mix of industrial, commercial and small business developments, along with a range of housing types accommodating a mix of socio-economic demographics. Project activity is centred on Bracknell, where the local authority partners, Bracknell Forest Council (BFC), are committed to developing a lower-carbon economy and are keen to promote the uptake of smart-grid technologies. At project initiation, there were no significant low carbon initiatives in this area, and Bracknell’s representative domestic and commercial demographics made this area the ideal blank canvas for the proposed trials. The project expected that there would be increased electricity demand associated with further economic development, along with demand changes linked to the anticipated uptake of EVs, solar PVs and HPs. As such technologies become increasingly mainstream the rate of adoption is expected to accelerate, placing additional strain on the network.

Customer engagement - Customers have a key role to play in the transition to a low carbon economy. Using a representative sample of large commercial, SME and domestic customers, NTVV assessed solutions applicable to different customer groups. A key element of the projects customer engagement is the Consumer Consortium programme working in collaboration with BFC and Bracknell Forest Homes.

Advisory Centre - The project established a Low Carbon Community Advisory Centre (LCCAC) to engage and inform customers and other stakeholders. This provided an interactive 'high street' presence in Bracknell, primarily aimed at keeping the public informed about our motivations and plans, and ensuring that the benefits of smart metering and smart grid programmes are understood and welcomed by energy customers.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 8

2 Executive Summary

The NTVV Project represents a significant undertaking which explores a mixture of analytic, technological and commercial solutions for network planning and operation. The £30m (£25m customer funded) project budget has been invested in developing and trialling technological solutions, commercial agreements, procedures, policies and training over five years to create new learning for DNOs and other stakeholders on understanding, anticipating and supporting the requirements of our customers through the transition to a low carbon economy.

Disruptive network connected technologies such as EVs, HPs and micro generation will change the scale, nature and variability of energy flows on the network. Historical maximum demand profiles at the heart of network planning will cease to provide the necessary decision making information as load flows become more dynamic and unpredictable.

Adapting the principles applied to a supermarket loyalty scheme2, the project has developed monitoring, modelling and forecasting capabilities to increase our understanding of future network usage. The valuable insight provided will enable informed decision to be made on future policies and investment to accommodate disruptive connected technologies. Application of the approaches trialled through NTVV will enable DNOs to avoid an estimated £900m of network reinforcement through the involvement of all customers groups and a comprehensive understanding of networks.

Whilst some of the technology trials have answered the questions posed at project initiation, others have identified additional learning and potential solutions in new areas

The project successfully achieved the following;

Installed Substation Monitors (SSM) and associated ICT infrastructure

Installed and trialled End Point Monitors (EPM) at both domestic and commercial customer premises

Developed and operated Network Modelling

Developed methods for Buddying, Aggregation and Forecasting

Deployed an LV Distribution Management System (DMS) (PowerOn Fusion (PoF))

Trialled Automated Demand Response (ADR) across a range of commercial customers

Trialled LV network street side battery storage Technology (ESMU)

Trialled Hot Thermal Storage (HTS) technology

Trialled Cold Thermal Storage (CTS) technology

Created and operated the LCCAC

Developed and trialled centralised smart system control for battery storage (Active Distribution Device Management (ADDM))

Obtained Smart Metering data from Electricity Supply Companies

Developed new Policies and Training material

Installed Substation Monitors and associated ICT infrastructure Over 300 LV secondary substation monitors were installed together with the supporting infrastructure and communication technology for streaming of real-time energy data. This was the first scaled deployment of substation monitoring by a DNO that was integrated into a DMS to provide data and alarms for immediate operational use, as well as facilitating the development of smart analytical analysis. The captured energy data allowed for substation categorisation, aggregation and relevant forecasting of future network loading. The learning and experience gained has been fed into the NIA funded low cost monitoring project (NIA_SSPED_0027) and will further support business as usual (BaU) users to become familiar with data that is inherently useful for their everyday operations, including network and system planning, investment planning and fault location.

Installed and trialled EPMs at both domestic and commercial customer premises 300 domestic monitors were installed at customer trial locations, supplying half hourly power utilisation data for over three years. GE supplied EPM and Senical smart fuse devices were deployed and assessed for ease of installation, operational performance and standardisation. Statistical modelling techniques were applied to determine the optimal level of customer monitoring required, which found that scalable modelling can be carried out using a sample of less than 1,000 monitors.

2 The use of Statistical techniques – learning from supermarkets – NTVV Project bid submission.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 9

Developed and operated Network Modelling A Network Modelling Environment (NME) was created for the NTVV project area to integrate a Geographic Information System (GIS) connectivity model with GE’s Smallworld Electric Office and the Cymdist power analysis tool. The project developed mathematical techniques for applying half hourly energy profiles within power flow analysis studies to determine the capacity headroom on the selected LV Network. This incorporated methodologies for buddying similar property types, deriving a confidence measure to address uncertainty, and clustering the uptake of Low Carbon Technologies (LCT) for forecasting purposes. The development of indices calculated from connectivity data for each feeder indicates areas of the network more susceptible to thermal or voltage constraints allowing a DNO to identify optimal locations for strategic network monitoring. The detailed modelling outputs reveal the location and extent of future capacity and voltage excursions, providing a tool for a DNO to inform future network investment decisions.

Developed methods for Buddying, Aggregation and Forecasting The project successfully developed methodologies for short term forecasting to the support smart control of energy storage. Medium term forecasting models to support investment decision making were developed, and these achieved better results than using traditional annual usage information alone. Long term forecasting to assess Low Carbon Technologies (LCT) uptake based upon actual energy profiles and accounting for social demographics, technology adoption and human behaviour were developed; and additional confidence bounding enabled these simulation outputs to form inputs to detailed network modelling.

Deployed an LV DMS (PoF) Whilst a DMS such as GE’s PoF platform is a BaU system for High Voltage (HV) network operation, its use at LV level and integration into existing systems represents a novel solution for LV network management. The development of the NTVV LV DMS resulted in the first successful integration by a DNO of a DMS with a GIS system through the NME, using the Common Information Model (CIM) standard.

The Supervisory Control and Data Acquisition (SCADA) system was used to transfer data from the substation monitoring devices and ESMUs to the centralised DMS, and allow centralised control of the ESMUs. The DMS then provided a visual representation of the LV network including monitoring data and alarms from the 300 installed substation monitors and the 25 ESMU battery storage systems in an operational environment. The developments supported by GE allowed the presentation of both streamed and half hourly data within the DMS, representing an additional functionality for the PoF system.

A further development through the implementation of secure Simple Object Access Protocol (SOAP) interfaces to the ADDM system and Honeywell’s Demand Response system enabled the control engineer to successfully schedule and perform ADR load shedding event.

Trialled ADR across a range of commercial customers The project recruited 30 commercial customers for installation and trial of ADR. Over the course of the project over 2,000 load shedding events were performed, providing more data than any other DNO run commercial DSR trial. The schedule for load shedding events was structured to ensure a balance of each of the different elements of a load shed across all customers throughout the programme, including time of day, duration, day of the week and notification period. The project has therefore accumulated a wealth of learning on load-shedding capabilities across both large and small commercial sites and the external variables that can influence the level of reduction seen; including time of day, weather and holiday periods. The learning generated from these ADR trials has supported SSEN in developing the Constraint Managed Zone (CMZ)3 schemes implemented directly into BaU.

Trialled LV network street side battery storage technology The project has developed 25 three phase street side power electronics and battery storage units and installed these around Bracknell to support LV network operation within technical standards. This was the first scaled deployment of such technology by a DNO on LV feeder circuits away from a substation environment. These street side LV battery storage units were deployed at locations which could influence feeder constraints. The trials successfully demonstrated their ability to impact voltage and thermal constraints and provide dynamic phase balancing capability, proving their value as a modular, relocatable and scalable alternative to traditional reinforcement. While the functionality was demonstrated successfully, in their current form it remains more economic to apply targeted network reinforcements rather than roll out these devices to solve network thermal or voltage limitations. In addition to reducing the price of this technology to

3 A CMZ is a geographic region served by an existing network where network requirements related to peak electrical demand under fault conditions are met through the use of DSR techniques (including demand reduction from commercial or aggregated domestic premises or export from third party owned generation and storage), provided as a managed service to SSEN by a CMZ service provider.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 10

ensure its economic viability, future technical development should look to reduce the size of the units and reduce the noise generated when the cooling fans are operating to ensure suitability for roadside installation in residential areas.

Trialled HTS technology The project installed 102 HTS Energy and Micro-Generation Manager (EMMA) units within domestic properties. These EMMA units reduce the peak export of PV micro generation on to the LV network whilst allowing customers to store and benefit from the electricity generated by their PV systems through its use in heating hot water for use in their properties. The expected reduction in output to grid was 3-6 kilowatt hours (kWh) per day per property, with an average of 500W per property being alleviated at times of peak generation. The impact of this technology was then incorporated within network modelling to understand the value at scale. Whilst SSEN supports customer’s uptake of this technology, it is not SSEN’s intention to install these devices on the customer side of the meter as BaU.

Trialled CTS technology The original intent was to recruit 50 customers who owned small-scale CTS to participate in a project trial. However, as no such installations could be identified within the project area a change request was submitted and approved for DNO installation of three CTS systems (Ice Bears) at commercial premises to provide cooled air as an alternative to their existing air conditioning plant. These were successfully trialled and demonstrated the capability for shifting peak demand, however DNO ownership of the devices meant that there was a need for ongoing interaction with customers for successful operation of the technology. As with the EMMA devices, whilst SSEN supports customer uptake of CTS technologies, at present it is not SSEN’s intention to install such devices on the customer side of the meter as BaU.

Created and operated the LCCAC The engagement of customers is key to the widespread adoption of low carbon technologies such as energy efficiency, embedded generation or demand side flexibility. The NTVV project created a DNO first in the shape of a LCCAC to trial and evaluate new and innovative means of engaging with a DNO’s customer base, and attract interest from domestic customers and educational groups. Whilst an LCCAC is unlikely to be cost-beneficial to a DNO, many of the functions of the LCCAC can be replicated by a DNO without requiring a permanent physical location. Value may be found in combining activities within the DNO (i.e. stakeholder engagement, identification of Priority Services Register (PSR) and fuel poor customers, education around demand reduction, project plans) and with partner organisations (e.g. local council, trade organisations, energy efficiency organisations and community groups). Learning outcomes from the interaction achieved through the LCCAC have been used to inform wider customer engagement activities.

Developed and trialled centralised smart system control for battery storage The development of an ADDM system allowed short term demand forecasting to be integrated with live LV network information to determine the appropriate smart control strategy for ESMU operation. The system successfully managed the street side battery storage units in phase balancing, power factor correction, voltage regulation and network support, mitigating thermal constraints. It was found that whilst the system delivered the necessary capability to support the ESMU trials, the latency of data exchange experienced highlights the value of de-centralised intelligence with local automated EMSU control based upon localised prevailing conditions.

Obtained Smart Metering data from Electricity Supply Companies Our project procurement partner prioritised the marketing and installation of smart meters within the project region, and successfully installed 895 units with consent for the use of the half hourly power data for NTVV research purposes. The project worked with the customer supply team to ensure the secure transmission and management of the usage data, classed as personal information. For ease of development, three large data extracts were created and securely delivered to the project, rather than create a more complex, regular data transfer process.

Developed new Policies and Training material To support the transfer of each of the methods trialled within NTVV into BaU, a framework of new policy documentation has been developed across the themes of LV Monitoring & Control; LV Design; LV Network Storage; Capacity Response and Customer Engagement. Further, to close the gap between the trial outputs and BaU, training modules have been created and targeted at specific audiences, including operational staff, design teams and senior management. In addition to providing a suite of project deliverables of use within the business, this material is available externally to industry stakeholders, including other DNOs and the regulator, to inform policy development across the industry and support the effective future operation of these solutions at scale.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 11

Changes to the Project Three change requests were approved by Ofgem. Two were with reference to the ESMU battery storage solution and were associated with the lack of market readiness for supporting this technology, and subsequently the suppliers own supply chain problems associated with the complexity of design. The delays in achieving an operational battery solution had consequential impact upon the integration ESMU control with the ADDM smart control system.

The final change request related to the number of participants for CTS trials and assumptions over availability of these systems in the UK. More detail of these changes can be found in section 6 “Required Modifications to the planned approach”.

Project Delivery Ambitious projects such as the New Thames Valley Vision have not been without their challenges. The five year project duration raises issue of technical viability as seen with the changes agreed with Ofgem, the volume of project information to be managed, the volume of data generated within the project (in excess of 1.6 Terabytes), resource turnover and the need for succession planning. In addition, the need to adhere to security requirements for both the operational control environment and business Information and Communications Technology (ICT) environment and interface specific aspects with the existing systems represent an additional layer to be managed above the technical trials and customer engagement aspects of such schemes.

However through overcoming these issues, all of the key project objectives have been successfully met, with all Successful Delivery Reward Criteria (SDRC) completed and delivered to the agreed schedule.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 12

3 Details of the Work Carried Out

3.1 Linking Network Reinforcement to a Better Understanding of Electricity Consumers 3.1.1 Forecasting Improved forecasting techniques are key to helping DNOs in operational, planning and investment activities. Agent Based Modelling (discussed in Section 3.3.4) was used to prepare energy usage profiles for all customers connected to selected LV feeders. This enabled short, medium and long term forecasts to be considered in line with Method 1 as described in the project background in section 1.

Short Term Forecasts are of primary benefit to the smart control of power electronics and energy storage, and in particular, for the management of peak demand. Four methods were assessed (adjusted average, permuted merge, genetic algorithm and maximum likelihood). These were compared with “average of previous weeks” and “last week as this week” as benchmarks, comparing errors and computational effort.

Medium Term Forecasts are helpful in assessing investment choices for a subsequent financial year. The data used was at the feeder level, with feeders ranging from supplying one non-domestic customer, up to a feeder supplying 119 properties (mix of domestic and non-domestic customers). Three methods were simple benchmark methods including “last year as this year”, “the average from the last two years”, and a “linear regression model”. These benchmark methods were compared to two generalised additive models (GAMs), which outperformed the benchmarks.

Long Term Forecast scenarios enable a DNO to assess potential impacts from scaled up adoption of low carbon technologies (LCTs) by customers. It was recognised that LCT uptake will cluster, based on demographics, existing LCT adoption and human behaviour (the “Jones Effect” of copying one’s neighbour). For this analysis, a clustering methodology was developed, including evaluation of the number of simulations to run to allow for uncertainty. Actual profiles for EVs, HPs and solar panels were derived from other projects (e.g. SSEN My Electric Avenue and Northern Power Grid (NPG), Customer Led Network Revolution (CLNR) or from end point monitoring data. For each category, a representative mix was established to ensure that when applied using the clustering algorithm, the impact on the network would be representative. The output of these scenario forecasts were loaded into the NME for electrical impact analysis.

Confidence Bound Generation was considered essential to allow network modelling activities to take account of the variability of actual network loads, taking account of day of the week seasonality, weather conditions, and other social factors. For this purpose, confidence bounds were calculated from the monitoring feeder data; this allowed an additional spot load to be added into the network model, the size of which was determined by the level of confidence. A technique known as bootstrapping was also tested for generating the confidence data for when monitoring data was not available; the results were compared with results derived from the techniques used in ACE49, the UK electricity industry network planning standard which outlines a statistical method for the design of LV networks.

These methods were described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

Customer engagement lies at the core of NTVV underlying the monitoring, modelling and management of the project. Within NTVV the customers engaged were reflective of each sector of society inclusive of domestic, SME and Industrial and Commercial (I&C). Prior to project inception a rigorous customer engagement plan was produced and updated throughout project delivery, providing a communication strategy for ongoing customer interaction. Project trial interaction was split across three groups, defined as recruitment, ongoing engagement and response4.

3.1.1 Local Authority Bracknell was chosen for the NTVV project because of the diverse customer demographics; with a varied mix of domestic ‘fuel poor’ customers, through to more affluent rural customers and large commercial buildings. The long engagement with BFC allowed us access to information regarding the location of these customer types and gave advice on engaging the local community – this engagement proved to be vital in understanding the demographics of both the domestic and commercial customer base.

A member of BFC was based at the LCCAC for two hours every day. This brought more customers into the LCCAC for pre-booked appointments, giving us the opportunity to inform them of the project. More information regarding housing associations and local authority input can be seen in SDRC 9.8(b)2 Evidence Report 100 Substation Monitors Installed.

4 Note SME and I&C customers are broadly categorised into ‘Commercial’

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 13

The Director of Environment and Culture was the main contact within the council. This engagement can be seen as being a success as we not only had access to senior stakeholders within the council but also additional contacts in other housing associations.

3.1.2 Low Carbon Promotions (LCPs) The diagram below illustrates the method used to develop the LCPs work. It highlights key principles which were taken into individual event planning/ execution activities.

Five LCP events were run across the course of the project, summaries of each of these can be found in SDRC 9.8b 2, Appendix A. This included the Bracknell Low Carbon Day event which involved advertisement through numerous channels including a live radio interview and resulted in approximately 1MW of load-reduction off Bracknell Primary substation.

3.1.3 Website The NTVV website was created as a central hub for project participants to view energy saving advice, take part in engagement competitions, and view achievements that were being made by the project team. This gave us the opportunity to fully engage participants with ideas and advice designed to drive a reduction in electricity use. The website also served as a project library, where we stored all documentation relating to SDRCs and project updates. During the life of the project, small extracts of data were uploaded, so that academics and other interested parties, outside of the project, could get some inkling as to the sort of information we were gathering. In the period from 22 May 2012 to 30 January 2017, the project website attracted 18,580 sessions, with 63% of those being unique. Visits were recorded from 128 countries around the world, which would not have been possible without this dissemination medium.

3.2 Interact with Demand Response Provided by Both Large and Small Businesses The project has built upon the lessons learned from SSEN’s LCNF Tier 1 project “Honeywell I &C ADR: Demonstrating the functionality of automated demand response”. NTVV demand response trials explore those customers likely to be early and effective adopters of DSR and how it can contribute to network flexibility and commercial arrangements that can facilitate the adoption and operation of DSR.

3.2.1 Demand Response The project looked to procure ADR5 on 30 commercial sites; this required a strategy of identification, engagement and installation:

Identification - Trial 1 involved a study of the network to highlight ‘at risk’ substations and their associated network structure. Commercial customer identification was through record assessment by SSEN and Honeywell with support from the local council, local chamber of commerce and facilities management groups. These were cross-referenced to create a list of available premises connected to ‘at risk’ substations. To maximize the network value of ADR, initially only sites with a load over 200kW were targeted. Trial 2 widened the target area to allow for installations at other local Primary substations. The initial 200kW limit/ threshold were removed. More detail can be found in SDRC 9.8a 2&3, section 2.

5 ADR is a subset of DSR as procured on NTVV.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 14

Engagement - The project explored targeted messages and tailored avenues of engagement for different customers, including a consumer consortium to explain the NTVV project, recruit customers and as a platform to share energy efficiency best practice. A detailed overview of the customer engagement process can be found in Appendix 16.3.2.1

Installation - The installation process adopted mirrored that applied in SSEN’s LCNF Tier 1 project “Honeywell I &C AD”. This process is detailed in Appendix 16.3.2.1.1 and looked to focus on control of flexible and inflexible load.

The project trials commenced March 2014 and completed over 2,000 load-shedding events. The schedule for load shedding events was structured to ensure a balance of each of the different elements of a load shed across all customers throughout the programme, including time of day, duration, day of the week and notification period. Customer engagement was crucial in assessing their reactions and ascertain where shed intensity could be increased. Final trials tested an average of 1.5 sheds per week with duration up to 4 hours. This has not only vastly advanced previous learning around load-shedding (Element Energy (2012)), but also tests the business cases of DSR (both sole use and shared use) for other potential procurers including suppliers and TNO’s.

3.2.2 Cold Thermal Storage (CTS) Given CTS units sit customer side of the meter, on commercial premises; the methodology for interacting with customers drew upon the learning and relationships from previous ADR trials. The identification activities already utilised for ADR trials were mirrored for CTS customers. The project selected sites based upon unique installations and applications of CTS. Relationships from ADR trials were drawn upon to engage customers, and those interested in the technology (10/10 customers) were progressed to site survey (7/10 customers). Site surveys required to confirm a room with cooling requirements around 6kW (this is the level at which Ice Bears are most efficient) within 45m (10m in elevation) of an area able to accommodate the 3.5m by 3.35m footprint of the unit. Only 1/7 customers failed site survey due to space requirements. Full details are available in SDRC 9.8a 4 Cold Thermal Storage.

The CTS units operate largely out-the-box following shipping and appropriate handling of refrigerant R410a as per European F Gas regulations. Installation including civil, mechanical and electrical works took 2-4 weeks to complete. Ice Energy (CTS manufacturers) were required to spend 2 weeks in the UK commissioning three sites and training a local contractor in their maintenance (annual checks were carried out on the units). Installation accounted for approximately 65% of total set-up costs. Operation of CTS required analysis of the network, and customer engagement in order to identify the optimal times of operation to maximise benefits for both the DNO and the host customer. Following a week’s baselining on the units it was determined that lower/less consistent cooling requirements in the UK, as opposed to California (where Ice Bears are manufactured), suggested a 1:1 ratio between cooling (9 hours) and charging (9 hours) on the Ice Bear applications as opposed the 1:2 ratio noted from US experience.

3.2.3 Commercial Drawing upon learning from SSEN’s Tier 1 Project “Demonstrating the Functionality of Automated Demand Response ADR, commercial agreements were designed to be as simple as possible. A three-way arrangement was required between the participating company, equipment supplier and the DNO; a consolidated version of this can be found in Appendix 5 of SDRC 9.1C.

The project introduced payment to its ADR trials 18 months in, in November 2015. Given previous challenges in obtaining signatories on legal agreements it was deemed this process was best simplified through the use of a Memorandum of Understanding (MoU). The MoU outlined a payment amount (£/MW), the trial period and load-shed parameters excluding the time of shed. By having no load-shed ‘windows’ the project team had complete flexibility over this parameter and could test a range of use cases for DSR.

Payment of trials utilised learning from CLNR, this calculation of Net Present Value (NPV) looked at deferred reinforcement across 4 years at a time. NTVV methods were trialled without availability payment. This looked to allow a variable level of demand reduction and incentivises a participant to deliver the maximum reduction possible when called.

3.2.4 Governance The NTVV project was able to demonstrate the use of DSR without the need for derogation against P2/6. It was recognised that no special dispensation was required and the DSR trials could operate as planned.

3.3 Use Mathematical Techniques to Reduce the Need for LV Monitoring

3.3.1 EPMs For mathematical techniques to be developed there was a fundamental requirement for half hour demand data from customers. This was predominantly obtained within the project by fitting EPMs directly at properties where customers had volunteered to participate in the project. An initial batch of 250 monitors were installed using the EDMI Mk7C meter;

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 15

this was a Meter Instrument Directive (MID) approved device, known to be compliant with operational, safety and tariff metering standards, and was configured to comply with the latest expectations of data encryption to ensure that Data Protection obligations could be met. Data was collected using GE’s Smart Meter Operating System (SMOS) and transferred to SSEN daily for storage in the Pi Process Book; this system was made accessible to the wider project team for data analysis. The installation of EPMs is described in SDRC 9.2 (a) 250 End Point Monitors Installed.

The end point data received was used primarily for the development of the mathematical techniques proposed; in addition, the opportunity was taken to evaluate the suitability and relevance of the selected monitor locations, the effectiveness of the installations, and the performance of the monitoring system, particularly with regard to operational experience, application, standardisation and installation considerations.

The output of the mathematical techniques was used to better understand the optimal balance of monitoring and statistical methods while seeking to reduce the need for monitoring, understanding the impact of monitoring availability on statistical quality, the applicability of Supplier based Smart Meter data for the future, and the balance of customer disruption against the statistical benefit. Further analysis was also carried out on non-domestic customer data to develop useful statistical techniques, and to assess the effectiveness of the techniques. These are described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

To support the analysis an additional 60 cut-out based monitors were deployed; these were developed on a preceding Innovation Funding Initiative (IFI) project (Supply Point Monitoring (SSE 2011_13) Senical) and it was hoped that these would offer an effective means for a DNO to gather customer data without depending upon Supplier based Smart Meters, and to do so with a minimum of disruption to customers. All Suppliers were approached to seek consent for access to customer data where the Suppliers had already installed early Smart Meters in the project area; any data obtained could then be used to assess how this data could be integrated into new processes, and to validate the techniques developed. End point monitoring is more fully assessed in SDRC 9.8 (a) Part 1 End Use and Network Monitoring Evaluation.

3.3.2 Substation Monitoring As for EPMs, data regarding the loading of LV feeder cables was required on the project for the development of mathematical techniques, and this was to be obtained using substation monitoring. Method 3 of the project specifically aimed to use mathematical techniques to reduce the need for LV monitoring. Method 4 of the project aimed to tactically deploy power electronics and electrical energy storage on the LV network; the control of this technology was to be linked to short term forecasts regarding the loading of the feeders, hence substation monitoring was required to be installed on all feeders where this technology was to be deployed.

To support the installation of a large number of substation monitors, learning was taken from the Tier 1 project SSET1002 Demonstrating the benefits of monitoring LV network with embedded PV panels and EV charging point, particularly the development of a procedure allowing substation monitoring to be retrofitted at substations without disrupting supplies to customers. An initial batch of 100 substation monitors were supplied by GE and installed at substations selected on the basis of having a particular combination of customers connected (a matrix of customer density (number of Meter Point Administration Numbers (MPANs)) and homogeneity was established). It was intended to reveal categories of substations that would then inform subsequent monitoring installations. This initial substation monitoring installation process was described in SDRC 9.2(b) 100 Substation Monitoring Installations Installed.

The data gathered from the initial substation monitoring installations was shared with the project academic team for substation characterisation analysis, preparation of short term feeder loading forecasts and buddying and aggregation analysis. Subsequent analysis of confidence bounds was used to support the running of power flow studies in the NME. The data was also made available to operational teams to facilitate trials of the DMS with operational data available. SDRC 9.2(d) Develop and Trial Method of Optimising Network Monitoring Based on Installation of First 100 Substation Monitors described the analysis of substation monitoring data.

A further 200 substation monitors were installed following the initial analysis. These were targeted at locations chosen to give optimal benefit based on the learning already gained, and were installed efficiently by applying learning from the initial installation (e.g. bulk preparation of templates within the DMS, use of masts to raise antennas optimising signal strength etc). Specific learning regarding the use of monitoring data in supporting the forecasting, buddying, aggregation and was described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation. SDRC 9.8(a) Part 1End Use and Network Monitoring Evaluation described the performance and analysis of substation monitoring.

3.3.3 Characterisation The grouping of customers in terms of their energy usage was recognised as a key means by which it might be possible to support the buddying of unmonitored customers to customers where an EPM had been fitted, facilitating analysis of the

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 16

LV network. Linking of LCT uptake to specific groups of customers may then allow improved forecasting of network impacts. Further, grouping substations by demand may also allow benefits to be gained from the forecasting of demands, reducing the total quantity of end point and substation monitoring required by targeting monitoring to locations of greatest value.

Having installed EPMs as described in Section 3.3.1, the energy profile data for each customer was assessed in terms of demand during four periods (breakfast, daytime, evening and overnight), and then further considered for variability, seasonality and weekend difference; this analysis allowed 10 categories of customer to be created and defined. This categorisation was described in SDRC 9.5(a) LV Customer Groups Presented.

The grouping of substations by customer count and homogeneity was already described in Section 3.3.2 as a means of targeting the monitoring. The substation and feeder loading data gathered was assessed for correlation with property size, homogeneity and density. Correlation was hard to find and this supported 194 further substation monitor installations as a means to provide statistical confidence. This analysis was described in SDRC 9.2(d) Develop and Trial Method of Optimising Network Monitoring Based on Installation of First 100 Substation Monitors.

Although Council Tax Band has been assessed in other research and found to have very limited correlation with energy usage, the use of Council Tax Band was considered a reasonable means of grouping customers for use with buddying. This was analysed as more end point data became available, and scalability became a bigger priority. The analysis of this form of characterisation was described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic Evaluation. The use of profile class was assessed and subsequently used to inform the groupings used in the buddying algorithms developed as described in Section 3.3.5.

The substation clustering method used by Western Power Distribution (WPD) was applied to 134 substations in Bracknell, thus creating an approximation to the electricity load every 30 minutes for a year. Their method generally captures weekly behaviour, albeit with a smoothed profile. In contrast, to capture key features, such as the maximum demand and peak behaviour, a GAM was developed. This model uses only a fraction of the input required for the WPD method6, and predicts these key features more accurately. Moreover, it confirmed the intuition that the number of properties on a substation, and the proportion of which are domestic (either profile class 1 – customers without electric heating or 2 – customers with electric heating) have the most predictive power. That is, including additional information, such as separating profile class 1 and 2, does not greatly improve accuracy when predicting key features.

3.3.4 Network Modelling The assessment of headroom on the LV network is ultimately dependent upon the running of power flow analysis studies. Mathematical techniques for buddying, confidence assessment and LCT clustering have been developed as described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation but the output of the techniques is a set of energy profiles, one for each endpoint on the network. These endpoint profiles needed to be loaded into a modelling tool to facilitate network analysis. For this purpose a NME was established.

The fundamental principal of the NME was that it contains the complete connectivity model of the LV network being assessed, including services. This was achieved in the project by migrating the existing BaU GIS connectivity model into a version of GE’s Smallworld Electric Office as described in SDRC 9.6 Low Voltage Network Modelling Environment Built, Installed and Commissioned. Electric Office was further integrated with a power analysis tool called Cymdist, and an Energy Profile Manager. These together form the NME. Before use the NME needed to be loaded with component types (e.g. electrical cables) and their respective electrical characteristics (e.g. resistance and reactance) for use in electrical calculations.

It was recognised that there were gaps in the available source data regarding connectivity, types of cables, customer addresses, and customers with known PV generation. To make the model and the data useable for power analysis studies required a number of functionalities to be established; these included the connection of service end points to the nearest LV feeder cable, the allocation of each service to an electrical phase to a defined methodology (L1, L2, L3, L3, L2, L1….), the mapping of services from the SSEN Service Interruption Management System (SIMS) database to the Ordnance Survey Address Layer database used in the NME and others. These functionalities achieved the objective of making the NME

6 WPD’s method generally captures weekly behaviour, albeit with a smoothed profile. In contrast, to capture key features, such as the maximum demand and peak behaviour, a GAM was developed. This model uses only a fraction of the input required for the method developed by WPD on their Templates LCNI project, and predicts these key features more accurately

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 17

usable and are replicable when scaled up for use with any DNO LV network. The learning gained from the address matching process has already been used by SSEN for the migration of address data beyond the project.

Further work to correct and improve the connectivity model was carried out after gaining familiarity and recognising inconsistencies. Primary areas of improvement included the mains cable connectivity at substations (informed from substation monitoring and site survey photographs), connectivity of individual mains cables (informed by tracing cables within the NME) and address matching (informed by SIMS address points left unmatched). This stage of “data cleansing” was crucial on circuits where actual power flow studies were to be run to ensure that the outputs were complete, correct and meaningful; remaining assumptions regarding phase allocation and cable types could be allowed for by appropriate interpretation of study outputs.

The NME provided graphical and numerical power flow study results. The graphical results provided a sample of the GIS diagram for each LV feeder included in the study coloured Red, Amber or Green (RAG) to reflect any sections operating out of limits. Numerical results were initially provided to give a minimum of data regarding the date, time and magnitude of the greatest excursion out of limits for each feeder; this was developed throughout the lifecycle of the project to give further information including counts, durations, cable lengths and other feeder specific data. This allowed the study results to inform the available headroom and also classification of each feeder for any loading scenario. These feeder classifications contributed to the conclusions regarding the minimisation of substation monitoring by identifying the feeders where monitoring is most justified. This was described more fully in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

3.3.5 Aggregation Method 1 of the project links network reinforcement to an improved understanding of customer energy usage. EPM data collected from a sample of 250 non-commercial customers was used to prepare buddies for all the other unmonitored customers on selected feeders. Various techniques for buddying were considered. The simplest method of buddying was found to be the selection of buddies of equivalent mean daily demand from a group of customers of similar council tax band and profile class. Actual consumption data for all customers was available from quarterly meter readings allowing mean daily demands to be calculated for each customer. This simple algorithm technique was computationally efficient and allowed profiles to be loaded into the modelling environment for power flow analysis.

When the aggregated load of each customer along a feeder was compared to the feeder demand measured using substation monitoring, a further assessment of the difference allowed a fresh iteration of the buddy allocation; in other words, a number of buddies were substituted with other buddies for which the profile allowed a better match of the aggregated demand with the monitored feeder demand. By repeating this process a number of times the error could be reduced and optimal buddies assigned. This optimisation problem was implemented using a genetic algorithm; the genetic algorithm was established and refined to minimise computational resources required. Further work was carried out to identify the preferred sample data in terms of seasonality and duration. This analysis was described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

The use of buddying and aggregation techniques as described can be seen as virtual monitoring. Minimal samples of end point data allow a reasonable assessment of the total feeder demand, and where a more accurate analysis is required; the provision of substation monitoring allowed the EPM data to be applied most effectively. The minimisation of monitoring data by these techniques was described in detail in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation. This was in line with Method 3 (Use mathematical techniques to reduce the need for LV monitoring) and was a precursor to all the modelling trials of the project.

3.4 Deploy Power Electronics and Electrical Energy Storage on the LV Networks

3.4.1 Energy Storage Part of the third objective of the project was to demonstrate a network mitigation strategy using power electronics (with and without energy storage) as a means of managing power factor, thermal constraints and voltage to facilitate the connection of renewables on the LV network. The fourth method of the project was to deploy such power electronics and energy storage as a tactical buffer to demonstrate the extent to which these technologies could achieve the objectives and provide a fast and flexible alternative to traditional reinforcement.

To ensure optimum learning would be achieved from the deployment of power electronics and energy storage a discussion document SDRC 9.4A “Energy Storage and Power Electronics on the Low Voltage Distribution Network” was prepared. This document was informed by the successfully delivered Tier 1 project SSET1008 – LV Connected Batteries. The discussion document allowed a detailed specification for the power electronics to be prepared.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 18

The detailed specification was used as the basis for a tendered procurement of the 25 Energy Storage and Management Units (ESMUs); Electrovaya (Toronto, Canada) was selected as ESMU supplier and a safety case was successfully prepared.

Each ESMU consisted of a 36kVA Power Electronics Unit (PEU) and a 12.5kWh Energy Storage Unit (ESU). A full description of the ESMU design and installation considerations was provided in SDRC 9.4(c) Install 25 LV Connected Batteries and SDRC 9.8 (a) Part 4 LV Network Storage – ESMU Trials.

Specific trials as proposed in the discussion document Energy Storage and Power Electronics on the Low Voltage Distribution Network were run to confirm the ability of the ESMUs to; balance load, balance peaks, provide reactive voltage support and improve power factor, provide demand reduction and demonstrate frequency response. These trials were carried out by manually instructing the importing and exporting of energy from the network; detailed descriptions and assessments were provided in SDRC 9.4d Produce learnings from energy storage and power electronic deployment and SDRC 9.8a (4) LV Network Storage - ESMU Trials.

3.4.2 Smart Control The hypothesis established in the discussion document Energy Storage and Power Electronics on the Low Voltage Distribution Network suggested that economic and flexible support for LV networks will be provided by power electronics with energy storage running smart control algorithms which make use of forecasted demand to provide a coordinated response to addresses the technical standards of voltage and thermal performance in the most efficient manner possible. In other words, the project sought to harness the short-term forecasts derived from the smart analytic development work for use in the development of smart electrical control algorithms to drive the ESMUs. Without these it would not be possible to deploy ESMUs efficiently.

Smart control algorithms have been developed and deployed in the core areas of phase balancing, peak demand management and voltage control. The early development of algorithms was described in SDRC 9.4(d) SDRC 9.4d Produce learnings from energy storage and power electronic deployment and the final work on algorithms is included in Appendix 16.3.4.2 Smart Control. To facilitate the operation of Smart Control, an ADDM system was developed.

The ADDM system allows a forecasting engine to draw on short term forecasts while the state database combines the forecast with the analogue values recorded in the Pi Process Book database. The output of the state database is directed to the ESMU Gateway which interacts with the DMS. The DMS sends the actual instruction to the ESMU via a SCADA interface.

A summary of the trials of the algorithms is presented in Appendix 16.3.4.2 Smart Control. These trials naturally resulted in refinements to the algorithms, each time seeking to achieve an optimum of improvement to the phase balance, peak demand reduction or voltage regulation of the network using forecast data. Assessments were also made regarding the performance of ESMUs using a less centralised approach, reduced access to data from forecasts and analogues, and the aggregated benefits.

3.4.3 DMS Method 2 of the project demonstrated the deployment and operation of demand response, and Method 4 has demonstrated the deployment of power electronics combined with energy storage. It was recognised that both these technologies can be operated with their own bespoke control system, but for wider co-ordinated deployment across a DNO network, the convergence of the control function into a single DMS is a natural expectation. For this reason a DMS was established early in the project using GE’s PoF system.

For the deployment of power electronics and energy storage, a traditional SCADA interface was commissioned between the DMS and each ESMU. The Smart Control system described in Section 3.4.2 also makes use of a SOAP interface with the DMS. The deployment of the DMS was described in SDRC 9.2c Install and Commission the Network Management Component of the Distributed Solutions Integrator System (DSI), and the integration of the systems was described in SDRC 9.8(a) Part 7 Integration Solution Control Evaluation.

An effective way of deploying a DMS was recognised as being the transfer of network and connectivity data from the NME. This was carried out in the project using the CIM as a means to connect disparate systems. The visualisation of GIS data in a DMS environment allowed the use of the CIM to be assessed.

The DMS was also used as a head-end system for the substation monitoring; the substation monitors were polled by the DMS for their half hourly data and constantly scanned for streamed data. This was deployed using SCADA and this allowed the handling of large data volumes to be assessed. The DMS was also set up to visualise the monitoring data to users (project team and operational teams) allowing real time operational benefits from the data to be realised. These were assessed as part of a set of trials to review the centralised control of the LV network. SDRC 9.2(d) Develop and Trial

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 19

Method of Optimising Network Monitoring Based on Installation of First 100 Substation Monitors described the uses made for the substation monitoring data in a DMS environment.

3.4.4 HTS The project looked to install 100 HTS devices within the project trial area. This was to be performed in two stages with a recruitment ratio of around 30:70. The three steps of identification, engagement and installation across stage 1 and 2 are detailed in Table 1 below:

Table 1 HTS Methodology

Recruitment Stage

Strategy Element

Detail Outcome

Stag

e 1

Identification

Local registered social landlords (RSL')s engaged to identify high density PV

No PV stock or limited interest.

Local council engaged to identify spread of PV

BFC building and planning records identified approx. 250 PV installations

Engagement

Mail-drop addressed ‘Dear Homeowner’

Limited Response.

Recruitment carried out on high-density area/streets

Challenges to recruit customers at high density due to interest and technical requirements.

Recruitment by bicycle Successful - more relatable to sustainability cause and customers more inviting/willing to listen.

Installation EMMA 3G unit EMMA 3G had an install success rate of 52.5%

reasons for failures include: no immersion heater, no space and complex wiring.

Stag

e 2

Identification

Wider RSL engagement Semi-structured interviews revealed a risk averse attitude to new technology, fabric first approaches to energy efficiency and difficulties matching PV, hot water tanks and trial areas.

Local councils across Thames Valley engaged

Of 5 councils engaged 2 were able to provide records of PV- approx. 300 records provided.

Engagement Mail-drop addressed ‘Dear Firstname Surname’

Successful - response rate of 30-35%.

High-density recruitment by bicycle Successful - replicated Stage 1 approach.

Installation EMMA 4G units- improved power quality, smaller size and slicker look

Smaller unit improved install success rate to 68%.

In total 102 EMMA units were installed across the project. EMMA units operate without DNO engagement based upon a learning algorithm. Further details can be found in SDRC 9.8a HTS section 1 and 2.

3.5 Policy & Training Material to Support Adoption into BaU

To support the transfer of the methods trialled within NTVV into BaU, the project set out to provide policy guidance to DNOs on application of these approaches and technical solutions. The scope of this work included changes to DNO operational policies and/or the creation of new policies and procedures as appropriate.

In addition, to close the gap between the trial outputs and BaU, the project plan included the development of vocational and technical training courses targeted at specific audiences, including operational staff, design teams and senior management.

EA Technology’s role as a Project Partner included support for the delivery of new policy documents and supporting strategies for DNOs, and the development of training materials.

The policy documentation and training packages subsequently created through the NTVV project have been specifically developed to inform and support the transition of the technologies trialled into BaU, as well as providing resources for use in disseminating project learning. The material delivered reflects the experience and learning generated from the project, and insight is also drawn from wider research programmes, including other recent LCNF project outputs, where appropriate.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 20

The work carried out is presented here, with detail provided in SDRC 9.8(c) Part 3 ‘DNO Training & Policies Review’7.

The approach taken to identify the requirements for new policy documentation and training packages comprised the following four stages:

Phase A - Review: Relevant documentation was reviewed to enable the new policy and training requirements to be determined;

Phase B - Hypothesise: A hypothesised list of policy documentation and training material requirements was completed based on the findings of the Phase A review process;

Phase C - Monitor & Review: During the course of the technology trials the hypothesised list of policy documentation and training material was reviewed and amended as necessary in response to learning and experience; and

Phase D - Outputs: The new policy documentation and training material was developed and delivered.

The nature of the new technologies and innovative solutions trialled as part of the NTVV project resulted in a relatively small overlap with existing policy documents, therefore, the decision was taken to develop a new set of policy documents based on the learning from the project. This new policy documentation is structured into three tiers, as follows:

Tier 1 Documents: The audience for tier one documents is senior level management, these documents also provide orientation for the overall set of new policy documents;

Tier 2 Documents: The audience for tier two documents is Planning Engineers, Design Engineers and Operational staff; and

Tier 3 Documents8: The audience for tier three documents is Field staff, this includes both Depot staff and installation contractors.

The documentation has also been developed under five categories to reflect the different aspects of the NTVV technology trials, as follows:

LV Monitoring & Control: These documents cover the LV monitoring aspects of the project. This includes the implementation of substation monitoring and end point monitoring, and the associated management of the LV Network utilising the DMS;

LV Design: These documents cover the LV design processes developed within the project. This includes the use of LV monitoring data within the NME to model the impacts of future demand on the LV network and assess the suitability of innovative solutions as alternatives to traditional reinforcement;

LV Network Storage: These documents cover the storage technologies deployed during the project. This includes HTS, CTS and battery storage devices;

Capacity Response: These documents cover the non-storage related technologies that can be used to manage network capacity. This includes DSR, the phase balancing functionality provided by the battery storage devices implemented as part of the NTVV project, and Demand Side Management (DSM); and

Customer Engagement: These documents cover the customer engagement processes required to support the introduction of new network and customer based technologies.

Figure 1 provides an overview of the new policy documentation that has been developed through the NTVV project. The full list of policy documents is provided in Appendix 16.3.5.

7 SDRC 9.8(c) Part 3 ‘DNO Training & Policies Review’ is available from NTVV project website www.thamesvalleyvision.co.uk/library/sdrc-9-8c3-dno-training-and-policies-review. 8 Note that the Tier 3 documentation was not delivered as part of SDRC 9.8(c) Part 3 ‘DNO Training & Policies Review’ as these documents relate specifically to the implementation of the devices and systems trialled within the NTVV project, whereas BaU application by other DNOs may involve the use of alternative devices or technologies from different product vendors. These documents will be made available on request.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 21

Figure 1: New Policy Documentation Overview

The policy documents have been developed in line with SSEN’s standards for documentation, and prepared such that they are suitable for formal adoption into BaU as appropriate, with minimal revision. Prior to formal issue to the business through standard BaU processes, the Policies and Technical Guides represent strategic documents to inform the transition of technologies into BaU as part of SSEN’s wider Innovation and Asset Management programme, drawing on the learning and recommendations delivered through the NTVV project.

The themes for training differ slightly from the themes for policy to reflect the requirements of the project and ensure the relevance of the material produced. The ongoing review during the course of the technology trials led to the decision to create the following four training packages:

LV Monitoring: Covers the business drivers for LV monitoring, the installation of LV monitoring devices, and the utilisation of LV monitoring data.

LV Design: Covers the need for decision support tools for LV Planners and Designers, the functionality provided by the new tools developed within the NTVV project, new ways of working as a result of the learning generated by NTVV, and how to assess the suitability of alternative, smart solutions for accommodating LV network load growth due to customer adoption of LCTs.

LV Battery Storage: Specifically covers the ESMU devices trialled within the NTVV project, including how to deploy, commission, control, manage and maintain them; and

Overall NTVV Learning: Provides an introduction to the changing environment for electricity distribution networks due to increased customer uptake of LCTs, and the alternative approaches to planning, operating and managing them, in addition to presenting the overall key learning from NTVV project.

A ‘training needs analysis’ was completed for each training package to:

Identify the staff group(s) that would be involved with the new technologies;

Identify the training needs for each of the staff groups;

Define training modules suitable for different staff groups under each of the training packages; and

Provide a matrix outlining which training modules were relevant for each staff group.

The creation of training modules targeted at different staff groups provides maximum flexibility in the use of the training resources developed, with training delivered as applicable to the staff groups involved. The suite of material can be used to embed an understanding of the practical application of new concepts, systems and procedures at a suitable level for each of the target audiences identified.

The material developed comprises sixteen individual training modules across the four training packages, as shown in Figure 2.

TIER 1 DOCUMENTS

Audience – Higher level management &

orientation for all

POLICY (PO)Supporting Customer Uptake of Low Carbon

Technologies

TIER 2 DOCUMENTS

Audience – Planning and design engineers

and operational management

TECHNICAL GUIDE (TG)Supporting Customer Uptake of Low Carbon

Technologies

POLICY (PO)LV Monitoring

POLICY (PO)LV Design

POLICY (PO)LV Network Storage

POLICY (PO)Capacity Response

POLICY (PO)Customer

Engagement

TECHNICAL GUIDE (TG) – LV Monitoring

TECHNICAL GUIDE (TG) – Network

Modelling Environment

TECHNICAL GUIDE (TG) – Capacity

Response

TECHNICAL GUIDE (TG) – Storage Technologies

TECHNICAL GUIDE (TG) – Customer

Engagement

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 22

Figure 2: NTVV Training Packages & Modules

All NTVV policy documents and training modules are available from SSEN’s Asset Management & Innovation team, and can be requested via the Contact Us page of the NTVV project website (www.thamesvalleyvision.co.uk/contact-us) or by sending an email to [email protected].

LV Monitoring Training Package

Modules

LV Design Training Package Modules

LV Battery Storage Training Package

Modules

Overall NTVV Learning Training Package Modules

Introduction and Basic Overview

Business Overview

Information for Third Parties

Storage and Handling

Scheduling and Control

Commissioning

Construction

Inspection and Maintenance

Introduction and Basic Overview

Installing End Point Monitors

Installing Substation Monitors

Active DeviceDistribution

Management (ADDM)

LV Monitoring Data

Introduction and Basic Overview

Enhanced LV Network Design to

Accommodate Load Growth from

Customer Adoption of Low Carbon Technologies

Monitoring

Modelling

Managing -Operations

Managing -Energy Storage

Managing -Automated Demand

Response

Customer Engagement

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 23

4 The Outcomes of the Project

4.1 LO-1 - Understanding - What do we need to know about customer behaviour in order to optimise network investment?

4.1.1 LO-1.1 - What is the optimum level and location of network monitoring? 4.1.1.1 Understanding Changes on the Network Monitoring was deployed successfully at over 300 distribution substations in Bracknell; these provided voltage data at the LV busbars as well as feeder data including the load current, real and reactive power, energy and voltage harmonic content of every electrical phase. No interruptions in supply to customers were caused, and this gave confidence that the new procedure developed in the SSEN Tier 1 project “Demonstrating the Benefits of Monitoring LV Networks with embedded PV Panels and EV Charging Point” was scalable for a large monitoring installation programme.

A key area of learning from the installations was linked to the resolution of the data provided; for voltage a resolution of +/- 0.1 V was found to be necessary to facilitate a true understanding of the voltage performance. Energy consumption resolution of +/- 1 Wh was found to be the preferred resolution, where the data was to be used in conjunction with mathematical techniques. The learning gained from substation monitoring is described in SDRC 9.2 (d) Develop and Trial Method of Optimising Network Monitoring Based on Installation of First 100 Substation Monitors and is further assessed in SDRC 9.8 (a) Part 1 End Use and Network Monitoring Evaluation.

4.1.1.2 Understanding the interaction between increased network Monitoring and smart meter data Suppliers were unwilling to share available smart meter data in the early stages of the project as described in SDRC 9.8 (a) Part 6 Smart Meter Performance, and only very late in the project (Q4 2016) was any actual data received; this was used to validate the data obtained from the 250 EPMs deployed as part of the project. It was found that the end point data samples (identical to data from real smart meters) allowed a buddy profile to be assigned to every unmonitored customer using only the mean daily demand from existing data flow data and the council tax band, and these buddies allowed the network to be modelled successfully for the purpose of understanding the status of the network. However, it was found that if the feeder load profile data obtained from the substation monitoring was used to support the buddying, the accuracy of the buddies could be significantly improved, allowing network studies to be relied upon more consistently at the individual feeder level. This is described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

It was not possible to assess the aggregated effect, where every service on a feeder is monitored, as a proxy for substation monitoring. This was due to the extreme difficulties encountered in persuading all customers connected to a given feeder to accept a monitor. While this was disappointing for the project, it was seen as indicative of future smart meter data availability for which the rollout will not complete until 2020, and customers have the opportunity to opt-out.

Where available, aggregated smart meter data from a Data Communications Company (DCC) can be used to enhance the buddying techniques, allowing true substation monitoring to be confined to locations where there is a dominant operational requirement. Conversely if aggregated smart meter data is not available (e.g. prior to 2020, or where more than a few percent of customers have opted out) then substation monitoring provides an effective means of improving the buddying accuracy; it would be expected that this would be targeted at substations where a feeder index has already identified the feeder and substation as being particularly vulnerable to the uptake of EVs or HPs (for load) or PV (for voltage).

GE’s SMOS was successfully used as a meter communications and data management service; this demonstrated that this role could be carried out by a DNO as a means of capturing metering data. However, it was expensive to operate and more importantly, the real difficulty remains in obtaining data from particular customers targeted by virtue of their location on the network. Running a separate system from that operated by supply companies did not solve that problem, so no evidence was found that would justify a DNO running its own meter communications and data management service. These learning points are discussed in SDRC 9.8 (a) End Use and Network Monitoring Evaluation.

4.1.1.3 Managing high volumes of data in a DNO environment Extensive use was made of the Pi Process Book as a means to store monitoring data. After initial configuration troubleshooting, this has proved reliable, and it has given wide access to the data for both the project team and operational users, and in a format already familiar for users of HV data. The verification and cleansing of data was carried out by the project network controller and such a role would have to be re-created for the management of data if a programme of monitoring installations is to be established after the project. Both IT teams and Real Time Systems teams have contributed to the establishment of the project data architecture and this can be recreated for BaU use.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 24

The real challenge has been to facilitate and maintain the communications system. In line with other LCNI projects there was a convergence in the use of cellular network technology 2.5G GSM (Global System for Mobile Communications) and 3G UMTS (Universal Mobile Telecommunications Service) for of ease of deployment. This has without question facilitated the project and is completely replicable for monitor data communications in urban areas. The key learning point here is that this convergence, combined with the volume of data involved, has put the access point (the gate way between the mobile network and the corporate computer network) under massive strain. If pushed, this has the potential to destabilise the system resulting in the inability to communicate with any mobile device whether linked to monitoring or other BaU system. This potential point of failure requires careful consideration to ensure that the capacity is sufficient.

The design of the data storage and retrieval system clearly needed to take account of security considerations, and the access to endpoint data in particular had to be very carefully controlled (end point data was deemed to be “personal data” as defined by the Data Protection Act (DPA)). The solutions found within the project to manage compliance with the DPA would be difficult to scale up and the key learning point remains that awareness of the issue is crucial prior to designing a system intended to handle large volumes of this type of data. Key learning points on this are described in SDRC 9.8 (a) End Use and Network Monitoring Evaluation and SDRC 9.8(a) Part 7 Integration Solution Control Evaluation.

4.1.2 LO-1.2 - To what extent can customers be categorised in order to better understand their behaviour? The NTVV project set out to extend the learning gained from the IFI project Smart Analytics (SSE 2011_10) carried out in conjunction with the University of Reading. The initial stage of this extension was the classification of customers into a number of types according to energy demand using the customer data gathered from EPMs. From the analysis of these households’ half hourly usage data ten behavioural clusters were found which described customers typical demands for certain periods of the day as well as their seasonal usage and general variability. The clusters were established by analysis of seven yearly variables (demand at four key times of the day, difference between summer and winter, difference between weekends and week days, and standard deviation); these allowed the categorisation of all the potential domestic electricity usage behavioural types describing their typical yearly electricity demand behaviour. This categorisation was described in SDRC 9.5(a) LV Customer Groups Presented.

Once the clusters were found the links between the clusters and other household information such as Council Tax Band, profile class and their daily mean usage (calculated from quarterly meter readings) were considered. From this analysis only broad links could be made between such information with the strongest links found between economy 7 customers (Profile Class 2) and the most seasonal customer clusters (ie clusters with customers that had the greatest difference between winter and summer demand).

The learning gained from this initial clustering methodology informed the development of the buddying methodology. Full details are described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

4.2 LO-2 - Anticipating - How can improved modelling enhance network operational, planning and investment management systems?

4.2.1 LO-2.1 - How could network headroom change as customers react to low carbon stimuli? As described in Section 4.1.1, large volumes of feeder data was gathered from over 300 distribution substations and customer energy usage data gathered from 250 EPMs over a three year period. This data was used to assess the performance of a number of smart analytical techniques developed on the project. These include agent based modelling (buddying) as a means to assign un-monitored customers to monitored customers, probabilistic forecasts to establish confidence levels, and clustering techniques to assign LCTs to individual customers in a realistic manner. Full details are described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

It was recognised that the modelling of network impacts from LCTs would require the use of energy profiles as the new loads (e.g. EV charging and HPs) would be large compared with existing customer demands and likely to be non-diverse. The modelling challenge was therefore to find a technique to assign relevant profiles to unmonitored customers from a sample set of monitored customers. This was initially achieved using a simple algorithm for buddying; this grouped the customers by Council Tax Band, and then calculated the mean daily demand for each customer from quarterly read data. This allowed a buddy of equivalent mean daily demand and in the same Council Tax Band to be selected; the energy usage profile for the buddy could then be deemed to apply to that unmonitored customer. This technique was found to be a computationally efficient means of assigning buddies and scalable for larger network.

A further refinement of this technique was the aggregation of buddied profiles to allow comparison with the feeder load profile as measured using the substation monitoring; buddies could then be re-assigned until such time as the error (difference) between the aggregated buddies and the feeder loading was minimised. This became known as the genetic

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 25

algorithm, and by definition this resulted in much more accurate buddying in terms of overall feeder load. Scaled up usage of this technique is dependent upon the roll out of substation monitoring, but this may be justified on a targeted basis.

Buddying provides the basis for loading a power flow analysis tool with relevant loading data, but actual loading is volatile and consideration needs to be given to the actual level of volatility. Probabilistic forecasts allowed a confidence factor to be calculated for each feeder; this was then converted into a spot load that could be added at the end of the feeder, allowing the power flow analysis to be re-run with the additional load. A non-compliant outcome of this confidence study was defined as an amber outcome; in other words, although the basic feeder loading is within limits, the combination of available headroom and loading volatility suggest that the feeder is vulnerable to increased LCT loading and could be targeted for deployment of substation monitoring.

To assess the network headroom as customers adopt LCTs, it was first necessary to identify a set of energy profiles that represent the individual LCTs, and then to establish a methodology for applying these to customers. EV profiles were sourced from the SSEN LCNF Tier 2 My Electric Avenue project, PV profiles were derived from EPM data where PV was known to be present, and HP profiles were derived from the LCNF funded NPG Customer Led Network Revolution project. An algorithm was developed that assigned these LCTs to customers based on an initial seed (e.g. known existing PV or EV present). The algorithm then added LCTs based on Council Tax Band and proximity to neighbours with LCTs. Multiple simulations were run to identify the level of “application uncertainty” which was used to apply a percentage of the total LCT load to each property. The output was a set of profiles that could be run in a power flow analysis study to assess the impact of a cluster LCT uptake scenario.

Studies were run for winter, spring and summer for scenarios of 30% EV, 50% EV, 10% PV, 30% PV, 10% HP and 30% HP. The remaining headroom for each feeder was quantified and the percentage of overloaded feeders identified. Comparison between the scenarios reveals the sensitivity of the network to those scenarios. While it is theoretically possible for all feeders to be assessed in this manner the data preparation and processing required is significant. For this reason a number of feeder indices were developed and trialled. In particular, the Remaining Capacity Per Customer (RCPC) was found to correlate well with the feeder scenario studies. In other words, where the RCPC was below a particular threshold, it was observed that those feeders were particularly vulnerable to overload for the LCT scenarios. This RCPC index is therefore proposed as an efficient means of identifying the feeders that justify both substation monitoring and more detailed modelling and power flow analysis. The scenarios assessed and results of studies can be found in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

4.2.2 LO-2.2 - How can modelling outputs be fed into operational systems and processes in a meaningful manner? GE have implemented a working PoF system set up as a shadow control environment. This was integrated with the NME in the sense that it was populated with the same network connectivity model and can interact directly with monitoring, energy storage and demand response resources. Learning from this integration is described in SDRC 9.8(a) Part 7 Integration Solution Control Evaluation. The primary modelling output leading into the operational environment is the output of power flow studies; where these studies reveal that the network is likely to be overloaded in a particular set of circumstances, the DMS is the tool from which to visualise the more detailed loading information from the substation monitoring, and then to initiate switching (by SCADA if remotely controllable devices are fitted) or manually by instruction to site operatives as appropriate. Where established, other smart solutions can be activated from the DMS.

The locations justifying smart solutions such as network connected energy storage were revealed by analysis of power flow study results as described in in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation. It was deemed that a feeder that is overloaded by less than 10%, for one season only and for less than two hours would be within the capability of a network connected energy storage device. 25 such storage devices (ESMUs) were deployed in the project area and controlled from the DMS as described in SDRC 9.8(a) Part 4 LV Network Storage – ESMU Trials. Direct manual control from the DMS was achieved using SCADA. A Smart Control system known as ADDM was also deployed to allow the storage devices to be managed on an automated and optimised basis as described in SDRC 9.4(d) Produce Learnings from Energy Storage and Power Electronic Deployment. This included a Smart Control Agent where instructions were set, a State Database where forecast data operational values were consolidated, and a Gateway to the PoF SCADA system delivered through a SOAP interface. The forecast data is the secondary feed of modelling outputs into the operational environment; the short term forecasts derived from substation monitor feeder data allow the Smart Control algorithm to anticipate when the peaks of demand or generation will occur on the feeder and ensure that the available storage capacity is used optimally over a 24 hour period. This aspect of Smart Control was challenging to achieve and was demonstrated to be effective, but is an area where future deployment will depend upon refinement and simplification.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 26

The dispatch of demand response was also demonstrated using a SOAP interface connected to Honeywell’s Demand Response Automation Server (DRAS) linked to 30 participating customers. For the project, numerous demand reduction trials were carried out to prove the functionality and responsiveness of the technique. In this context, and to provide a tangible benefit to the network, it was the ability to group a batch (e.g. by HV feeder connection) that was most relevant due to the limited confidence level in any one customer being able to provide the full reduction required.

4.2.3 LO-2.3 - How can modelling outputs be fed into planning systems and processes in a meaningful manner? Section 4.2.1 described the key learning regarding the analytical methods underpinning the NME. The NME was set up to run power flow studies based on energy profiles as described in SDRC 9.6 Low Voltage Network Modelling Environment Built, Installed and Commissioned, and to provide geographic and numerical outputs. To determine the criteria that represent “out-of-limits”, for loading the individual cable ratings were loaded for each season, and for voltage, the statutory voltage limits at the customer’s meter (230V +10% and -6%) were assigned. These ensure that any studies would reveal where and when the network was found to be out of limits in line with existing planning standards (ER P2/6, Electricity Supply, Quality and Continuity Regulations (ESQCR) and internal planning documents). RAG definitions were assigned as follows:

Red Out of limits

Amber Out of limits only with confidence data assigned

Green When neither Red or Amber

These colours were applied to the geographical outputs to reveal the location of out-of-limit sections of the network. This proved invaluable for understanding the impact of LCT loading on individual feeders, as the extent of the issue was very visual. The outputs were also largely intuitive (i.e. for overloading, the section of cable nearest the substation was always the first to be out-of-limits), but this was valuable in building confidence in the modelling. For larger (bulk) studies, representing large counts of substations (e.g. up to a Primary Substation) numerical results were far more valuable. Counts, durations, percentages of overload, cable lengths and timings were all recorded allowing the post study analysis to concentrate on the area of interest. A set of rules were established for load and voltage that use the bulk study results to categorise individual feeders.

Analysis of the study results informed relevance of running studies on different days representing different seasons, special days (e.g. Easter) which can be particularly demanding for the network, and the impact of the source data quality, in particular, the phase allocation of services. The actual phase allocation of the majority of services was unknown, so the model was built up with an assumed phase allocation; this was intentionally chosen to be set up as balanced (L1, L2, L3, L3, L2, L1…). For a sample of substations, real phase allocation was checked on site. This allowed a comparison study to be run where the phase allocation was corrected. This revealed that the actual phase allocation was up to 25 % more unbalanced than the assumed allocation, and this made the study results significantly worse; further, the system unbalance does not improve as the number of customers on the feeder increased. The key learning point from this was that obtaining corrected phase allocation data is justified where more detailed analysis is required. Full details are described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

The use of the indices has already been discussed in Section 4.2.1. This provides a meaningful way of prioritising the assessment of LV networks in a planning system. The sequence can be considered as:

1) Run RCPC index for all feeders in area of interest. 2) Classify feeders based on RCPC index value. 3) For the small % of feeders of concern, consider fitting of substation monitoring, and prepare data for analysis in NME 4) Run power flow studies on targeted substations (use genetic algorithm if substation monitoring fitted) 5) Classify feeders in line with thermal and voltage rules 6) Review fitting of substation monitoring in line with categories (add more if necessary) 7) For selected problematic feeders, survey phase allocation and re-run studies. 8) Design solution for individual feeders (Smart Technology deployment where appropriate, otherwise traditional

reinforcement)

For the analysis of new connections, the same principles can be applied. For every feeder where the RCPC is above a defined value, a next single phase new service connection can be made without further analysis. For feeders where the RCPC is too low, substation monitoring may already have been fitted and this can inform a connection decision; otherwise a more detailed study can be run in the NME to assess performance after connection of the new service and to inform any reinforcement work required.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 27

4.2.4 LO-2.4 - How can modelling outputs be fed into investment systems and processes in a meaningful manner? It has been demonstrated in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation that the power flow analysis studies built on monitoring data and buddying techniques, when combined with forecast scenarios of LCT uptake, deliver numerical and objective outputs for performance of each feeder relating to both load and voltage. A methodology for the classification of the feeders based on the outputs has been established. This classification defines each feeder as requiring:

a) no action (80% of feeders in sample for 30% EV uptake scenario); b) Would benefit from closer monitoring (2%); c) Problematic but able to be resolved with a smart solution such as energy storage (4%); d) Problematic and requires traditional reinforcement (14%).

The count of feeders in each category forms the primary basis for a load related capex calculation; in other words a unit cost can be applied to each category and multiplied by the number of feeders affected. For the feeders requiring traditional reinforcement the modelling outputs include the length of each feeder that is out of limits for that scenario; the cumulative feeder length requiring reinforcement can then be established giving a potentially more accurate indication of the aggregated reinforcement cost rather than working with an average per feeder.

Clearly the methodology described above requires an assumption to be made regarding which scenario is to be applied (e.g. 30% or 50% EV uptake), although it is feasible to run studies for multiple scenarios and carry out a sensitivity analysis to identify if a small change in a scenario results in a large change in network impact.

To be able to carry out a large scale analysis of this type would require the provision of a scaled up NME and a vast amount of preparatory work to prepare network data and customer load data. It was recognised that this would not be appropriate in the short to medium term and hence a number of feeder indices were created. In particular, the RCPC has been identified and shown to correlate reasonably well with impact of EV uptake. This can be calculated easily for all feeders in the modelling environment (and potentially with raw feeder data in a spreadsheet). All feeders with an RCPC above a threshold relating the chosen uptake scenario can be expected to operate within limits for that scenario, and hence there is no need for further analysis of these feeders; this allows the DNOs systems and resources to be targeted on the remaining feeders which may require the load related capex.

4.2.5 LO-2.5 - How can network modelling outputs be fed into town planning systems and processes and vice-versa?

It was suggested that local government is a key stakeholder to a DNO both in terms of understanding where reinforcement may be required and also the possible suitability in providing demand response support. A number of workshops were held with the local authority BFC and representatives from commercial developers. The starting point of the discussions was recognition of the mechanistic approach that prevails which is based on the historic responsibilities of each party in line with planning, political and regulatory interventions. The interactions with the local authority and developers drew attention to potential opportunities including:

As the modelling environment develops and the availability and quality of its input data is verified it becomes increasingly important to review how SSEN’s network modelling can be tailored to inform an integrated town planning approach. Emerging network modelling simulations are likely to play a key role.

Informed and coordinated planning processes are important for effectively managing the impact of a rapid transition to a low carbon economy. A common understanding of this evolving impact between SSEN and BFC will also act to improve stakeholder confidence through coordinated and informed town planning initiatives and decisions.

As the demand for network capacity increases and constraints emerge the refurbishment decisions of existing large network customers (e.g. housing associations) will become more important for maintaining a reliable network. SSEN will need to maintain its focus on new connections and to increasingly engage with existing connection customers.

Regulatory policy will need to be sufficiently flexible and adaptable to accommodate uncertainty and potentially radical change. SSEN may need to react in relatively short timeframes to effectively manage network constraints. The ability to do this in a cost effective manner will be key.

The underlying theme of each of these opportunities was found to be one of on-going liaison and sharing of information with the local authority. Larger developers were more focussed on a DNO’s high level capacity constraints and the mechanism for reserving the available capacity for their own commercial benefit. The local housing association was more focused on delivering energy efficiency solutions for its tenants (e.g. improving insulation) and less interested in deployment of new technology (e.g. solar panels). Full details of this analysis are in SDRC 9.8(a) Part 8 Overall Proven Benefits (Both Financial and Customer Service).

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 28

4.2.6 LO-2.6 - What changes are required to industry governance and documentation to facilitate a modelling based approach to network monitoring?

Security of Supply standard ER P2/6 sets out the minimum standard of supply security based on the maximum forecast demand. Distribution transformers are typically rated and loaded to less than 1000kVA which is the threshold for Class A demand groups which is the lowest class. For Class A, the Group Demand “will be restored in repair time hence minimal or no security/redundancy design requirements”. An indication of the recent maximum demand at each distribution transformer is provided by means of Maximum Demand Indicators (MDIs) and hence there is no dependency upon more modern granular substation monitoring for this purpose. Network modelling has been used in the project as a means of predicting the load on LV feeder circuits; the benefits of the techniques applied are that the non-diverse loads can be assessed in a relevant statistical manner, allowing decisions to be made regarding optimal solutions at the individual feeder level. In practise, the techniques can be extended to larger groupings such as the distribution transformer level, but the larger the grouping the more existing techniques remain valid. No change is proposed to ER P2/6 based on the studies carried out in the project.

P2/6 or ETR130 does not account for non-firm demand and suppressed / despatchable demand (interruptible load or generators contracted to generate during peak periods), but this is currently under review in the P2 workshop and is taken into consideration in RIIO-ED1 reinforcement reporting for 132 kV and EHV substations. ETR130 could be expanded to provide guidance on:

group demand calculation (e.g. an interruptible load will decrease the group demand while demand reduced using on-site generation will not), and possibly on

generation security contribution (e.g. the criteria when a generation side response is deemed to be providing a contribution to security of supply).

Although there is no security/redundancy requirements for Class A demand groups (less than 1000 kVA), there is merit for ETR130 (Application Guide for Assessing the Capacity of Networks Containing Distributed Generation) to provide more guidance and clarity on security assessment for HV networks, or even higher voltage levels. Therefore, we would fully support the discussion in the P2 workshop (and any other relevant industry working groups) in amending/expanding the guidance on treatment of despatchable demand.

The need for Load Managed Areas may have been reduced but would still be necessary, for example, due to unforeseen demand growth and intervention or solution (including those identified in this project) could not be implemented in time. No change is proposed to Section 8 of the DCUSA Document.

4.3 LO-3 - Optimising - To what extent can modelling reduce the need for monitoring and enhance the information provided by monitoring?

4.3.1 LO-3.1 - To what extent can modelling be used in place of full network monitoring? The use of agent based modelling as a tool to support the modelling of LV networks has been discussed in Section 4.2.1 and in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation. It has been shown that power flow studies successfully indicate the network impact of particular scenarios and the study results can be interpreted to allow each feeder to be classified as described in Section 4.2.4. The first category (where the study results reveal that the loading is less than 90% of the feeder rating) for the scenario being analysed do not need to be monitored. From the analysis of a sample of 44 feeders, it was shown that nearly 85% of feeders were loaded below 90% of their rating for a 30% EV uptake scenario; when the EV uptake scenario increased to 50%, the percentage of feeders below the 90% loading threshold dropped to around 75%. In other words, three quarters of feeders do not require monitoring based on network modelling derived from agent based modelling combined with clustering.

For feeders loaded above 90% of their rating a more detailed analysis of the results is helpful before committing to the installation of monitoring. For the most extreme cases of overloading, assessment of the feeder configuration (e.g. length, number of spurs and total count of connected customers) may suggest that the overloading is inevitable in the event of even minor LCT uptake scenarios, and a reconfiguration of the circuit would be advisable, possibly involving traditional reinforcement activities. Provision of enduring monitoring data for such a feeder will not add value to the investment decision to be made. It is the feeders that are just under or just over the threshold for overload where the provision of monitoring data can support the modelling output data and inform investment decisions regarding whether to invest and how. Substation monitoring can reveal the actual level of unbalance between the electrical phases; this helps to understand the likelihood of a real problem arising, and if it does, it also points to a potential solution (e.g. use of power electronics to balance the circuit loading).

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 29

Reference has been made to network loading for ease of description; the principles can be equally applied to the voltage study outputs from power flow studies.

As described in Section 4.2.4, the use of feeder indices was found to be an appropriate pre-cursor to the running of full power flow studies. The indices are calculated using connectivity data held within the connectivity model (e.g. number of customers connected to the feeder, rating of first section of cable on each feeder, distance of each customer to the source etc). The key learning on this point was that the indices can be established from basic connectivity information and are easy to run when the connectivity model is available; the RCPC can also be run in a spreadsheet as a pre-cursor to building the full NME. This will not only provide the earliest indication of where to fit monitoring, but will also guide the prioritisation of data preparation for the building of a full scale NME.

4.3.2 LO-3.2 - How might modelling assumptions change over time? Tracking actual changes of customers was virtually impossible from a modelling perspective. Non-domestic customers tend to have more predictable demand profiles than domestic customers, and if one of these changes from one type to another the impact on the network from increased loading can be significant. However, this is a pre-existing modelling problem, and the use of buddying was not able to improve on this; buddies for non-domestic customers were manually assigned from knowledge of the actual non-domestic customer. It was easy to substitute a different buddy where it was known that a premise had changed trading activity. These network endpoints were excluded from the clustering algorithm for the studies run within the project, but it was agreed that the algorithm could refer to geographic code for each property, so a non-domestic property could be specifically included or excluded in the clustering based on the code. This was described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation.

The clustering algorithm developed allows any percentage of any LCT to be assigned in the model; in other words, as LCT adoption assumptions change, the modelling analysis can be re-run with current data and an updated clustering percentage. At the time of running, it was found to be useful to run multiple different uptake scenarios. There is value in performing a sensitivity analysis on individual feeders; where a small change in LCT uptake results in a large net impact on a particular feeder this needs to inform any investment decision made, either to justify the deferment of an investment, or to justify a larger investment to avoid having to invest twice in the same feeder as the LCT uptake rises.

4.4 LO-4 - Supporting Change (technologically) - How might a DNO implement technologies to support the transition to a Low Carbon Economy?

4.4.1 LO-4.1 - How could distributed solutions be configured into the DNO environment A number of distributed technology solutions were successfully deployed on the project including power electronics and storage (network side solution), Building Management Systems (BMS’s) to facilitate demand response, and thermal storage (as customer side solutions).

The use of power electronics with and without energy storage was trialled using ESMUs. LV feeders are vulnerable to peak demands causing thermal and voltage problems. The feeder locations with such problems were revealed by analysis of power flow study results as described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation. It was deemed that a feeder that is overloaded by less than 10%, for one season only and for less than two hours would be within the capability of a network connected ESMU. It was demonstrated that a network demand reduction of up to 12 kW per phase was possible using an ESMU as described in SDRC 9.8(a) Part 4 LV Network Storage – ESMU Trials. The use of power electronics alone allowed energy to be transferred from one phase to another, effectively allowing an improvement in the balance of load on each phase.

The voltage profile of a feeder was successfully manipulated by up to 4 V per 10 kVAr of reactive power and up to 6 V per 10 kW of active power by injecting reactive power using an ESMU. This is sufficient to mitigate the impact of micro-generation by a modest number of customers on a lightly loaded feeder, keeping the voltage within the statutory limits. The use of energy storage (lithium ion batteries) in addition to the power electronics effectively allowed the ESMUs to provide network support for more extended time periods and with reduced dependency between the phases. Both the demand reduction and the balancing of loading between the phases effectively reduced the losses on the network; taking account of the energy losses associated with the operation of the power electronics and energy storage, mostly as heat, the overall energy benefit was at best negligible, and it was recognised based on the current technology, loss reduction would not justify the installation of ESMUs.

To enable the above benefits to be achieved using multiple ESMUs across the network a form of Smart Control system known as ADDM was also deployed. This allowed the storage devices to be managed on an automated and optimised basis drawing on short term feeder load forecasts as described in SDRC 9.4(d) Produce Learnings from Energy Storage and Power Electronic Deployment and Appendix 16.3.4.2 Smart Control. The system seeks to anticipate when the peaks

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 30

of demand or generation will occur on the feeder and ensure that the available storage capacity is used optimally over a 24 hour period. It was demonstrated that this could be achieved, but the technology readiness level of the ESMUs and control system were recognised to be lower than required for widespread deployment beyond the project.

The primary objective of demand reduction negating the need for network reinforcement was demonstrated. A theoretical limit of 12kW per ESU or 36 kVA per ESMU can be aggregated to give the theoretical maximum demand reduction on an HV feeder or HV primary substation. The actual ESMU demand reduction capability was more limited due to battery performance and operational limitations, resulting in a more practical limit of 9 kW per ESU or 27 kVA per ESMU; when combined with the diversity of timing for peak demands between different LV feeders and distribution substations the aggregated effect on the HV network was observed to be negligible based on the capabilities of the supplied ESMUs.

The capacity of an ESMU is equivalent to three 4kW PV installations and three 3.5kW EV chargers; hence the functionality demonstrated confirmed that an ESMU could enable those quantities of low carbon technologies to be connected and operated on a feeder that was already operating close to its thermal or voltage limits. Based on the current technology readiness level and equipment costs experienced within the project, it was clear that this smart solution was uneconomic at the time of writing. Further work to put the technology into production and reduce battery costs is required before an economic case can be made. A large array of inter-connected ESUs requires reliable equipment and a robust decentralised control system with access to feeder data and forecasts to achieve the optimum benefit from this type of investment.

With regard to customer side solutions, ADR was successfully deployed on 30 customer BMSs in the project area. The largest forms of demand response were seen to arise from heating or cooling technologies in buildings. From a network perspective, it is the aggregated effect from ADR at several customers that are connected to the same feeder that will give the greatest benefit. It was observed that the reliability of individual customers (mixture of communications issues, opting out, technical issues, and parameters within a site outside a DNO’s control) would mean that interest would be greatest in the aggregated use of ADR as part of a programme, rather than seeking out individual customers. This was described in SDRC 9.8 (a) Part 2 Part 3. The use of standby generation in buildings as a form of demand reduction was assessed by liaising with existing ADR customers. Four customers revealed their back up generation capability and these ranged from 200kVA to 2.25MVA with an aggregate of 5.3MVA sustainable for 24 hours; for UPS support the range was 115kVA to 710kVA with an aggregate of 1.2MVA sustainable for 20 minutes, up to an hour in some cases. These are large values in comparison with some of the demand reductions observed, but do have limitations and costs which would need to be factored in by the DNO.

As for the demand response, the use of hot and CTS technologies was demonstrated to work effectively, but these are only of benefit to the DNO if the point of connection for the customer is on a feeder where there is a constraint, or if the upstream primary substation is close to its firm capacity limit.

4.4.2 LO-4.2 - How could a network management solution integrate with BMS’s During the project, GE implemented a SOAP messaging system, and Demand-Side Response (DSR) interface within their DMS. We used this to send SOAP messages from our DMS, to our ADR project participant’s BMS's, in order to enact a DSR event. This method gives a Control Engineer, the ability to remotely reduce electricity demand on a part of the network, with ease and simplicity. Future applications of this would see all ADR sites on a feeder linked together, to enable larger shedding on overloaded parts of the network.

4.4.3 LO-4.3 - How can the DNO best engage with customers to encourage demand reduction, and where on the network is each most effective

The project trialled a range of different approaches and messages for recruiting customers. A full summary of the processes trialled and learning outcomes can be found Appendix 16.3.2.1. Key engagement opportunities include:

Focus on the local nature of the project, support from BFC and the impact on the local network aided sign-up.

Events and focus groups may be a quicker and more effective means of introducing ADR to customers than one-to-one meetings.

It is important to identify at as early a stage possible who the business sponsor will be, local and senior interest will not always be enough.

On-going engagement through varying channels is key to unlocking full ADR potential.

NTVV project trials recruited 30 customers from varying sub-sectors of society representing around half of the types of building that make-up I&C building load during SSEN’s ‘red zone’ (4:30pm) (note NTVV recruited no retailers who account around 30% of total load) (Element Energy 2012). The size and type of these buildings vary and include: multinational

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 31

enterprise (MNE), local government and SME. For clarity customers are categorised by the size of their site as opposed a specific industry sector. The Project bid clarifies large commercial as customers with a max demand of >200kW, it is therefore assumed customers with a max demand of <200kW are light commercial SME.

Table 2 ADR Capabilities

You can see from Table 2 above that those NTVV customers within the largest categories are most likely to have an on-site BMS and resultantly the average load-shed is significantly higher. When looking at the load-shed between those customers with a BMS and those without it is apparent there is a perfect correlation between BMS and high load-shed. Where a BMS was not available (largely educational institutes and small offices), appliances with a safety interlock circuit, timer circuit or remote enabling can be utilised for remote load-shedding. Given the low-level of shed available from these sites automation of demand response is unlikely to be cost-effective.

ADR has also been applied to the modelling work carried out on NTVV. Through modelling, the project can anticipate at what level of uptake of LCT the network is likely to be under-constraint; in turn by understanding the breakdown of commercial premises on the case-study network area, ADR analysis can reveal theoretical load-shed levels and henceforth the related level of load-reduction possible. The outcome highlights which percentage uptake of LCT’s could be managed by ADR and at which point the threshold becomes too significant for ADR alone to manage the increased load on the network.

4.4.4 LO-4.4 - How would network storage be used in conjunction with Demand Response LV networks have traditionally been managed in a reactive manner, responding to concerns as they arise. The project has developed a NME and techniques which allow large parts of the network to be assessed for both thermal and voltage performance. This can be carried out using current loading information or projected loading based on the uptake of low carbon technologies. As described in Section 4.4.1, study results reveal which feeders require support. Where that feeder is dominated by connections to domestic customers, the use of network connected energy storage was found to be effective for the purpose of managing thermal and voltage constraints. The use of HTS by domestic customers was recognised to be useful to both the customer and the DNO, but the benefits to a DNO are too small to justify incentivisation.

For LV feeders where non-domestic customers dominate the connections to the feeder, it may be possible to use demand response. It was found that the smaller non-domestic customers typically have a very limited demand that can be shed, and the additional considerations of reliability (opt-out, technical and communications issues) and costs make this approach less attractive for both the DNO and the customer. The possibility of using network connected energy storage remains an option subject to the actual constraint parameters.

For larger non-domestic customers, typically connected at HV or with a single consumer distribution substation connection, the use of LV connected storage as trialled on the project has no role, but it is much more likely that such customers would be able to participate in demand response. For customers with an air-conditioning dominated load, the use of thermal storage as demonstrated using the Ice Bear technology is possible. For a DNO to incentivise these solutions there would have to be a corresponding HV feeder or primary substation loading constraint.

In any given area of the network it may be that a combination of network management technologies provides the optimal solution for constraint management. Through smart control a DNO may be able to optimise the use of an energy storage technology by ensuring a full charge/discharge, as opposed partial usage, utilising the flexibility provided by ADR.

To summarise, each storage solution has an optimal application situation, and the solutions complement each other. Customer side storage may be further influenced by Supplier based incentives; this could serve to benefit the DNO indirectly, or potentially to create a new network peak demand requiring an intervention.

Site peak load Quantity BMS (%) Average load-shed Average shed (with BMS) Average shed (without BMS)

Very Large (>500kW) 4 100 70.8 70.8 N/A

Large (>200kW) 8 87.5 31.4 34.3 14

Medium (>100kW) 8 75 20.3 23.0 4

Small (>50kW) 5 60 12.0 12.7 10

Very Small (>20kW) 5 60 5.2 10.7 3

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 32

4.5 LO-5 - Supporting Change (commercially) - Which commercial models attract which customers and how will they be delivered?

As described, customers were signed on to the ADR trial without financial incentive. The sections below explain the attractiveness of the commercial models adopted to each sector.

4.5.1.1 LO-5.1 - Large commercial At initial sign-up the project identified 345 customers with load varying between 10kW and 3.7MW. Given limitations around the contact details and loading of different sites this list was narrowed down to 110 customers, assumed or proven to be eligible for a min shed (load of >200kW). Of these 110 customers Figure 3 below summarises recruitment progress, of which 16% were recruited onto the project.

Figure 3 Large Commercial Customer Recruitment

There were a wide range of reasons for which other customers could not participate, notably at 19% of sites no contact could be established despite numerous and varied attempts to establish communication. Despite the project teams best efforts, just over 20% of sites did not wish to progress discussions past initial stages; with a further just under 20% of sites losing momentum at a later stage in the sign-up process.

It is important to note here how NTVV trials were completely un-incentivised up until November 2015 at which point trials were introduced to test commercial mechanisms and detail how payment affected participation rates. As a result whilst it might be assumed that approximately 20% of sites could still be challenging to establish contact with; a greater proportion of sites may register interest in ADR given a commercial offering shifting stats from the ‘lost momentum’ and ‘not interested’ categories to the install category.

Following the introduction of incentivised trials in November 2015 a MoU was issued to each customer in order to offer payment over the summer and winter periods of 2015/16. Meanwhile Spring and Autumn trials remained un-incentivised. It was intended that by running trials in this manner the project could understand how an incentive may impact a participant’s willingness to respond to more intensive load-shedding events, including no notice load-sheds and sheds up to 4 hours in duration.

Given the close customer engagement adopted on the project opt-outs up to, and indeed throughout the incentivised trials, were minimal, with just 39 opt outs in total across the project out of over 2000 load-shed events (<2%). As a result the introduction of an incentive, even despite more intensive trials, had no definable impact on customers willingness to participate in events. Two hypothesises can be drawn from this: 1) personnel opting out of the load-shedding events are those driven by a need to keep a premise comfortable, not where the financial benefits of ADR participation are missing; and 2) without close customer engagement NTVV trial opt-outs will have been significantly higher.

NTVV ADR trials were closed in November 2016, at which point the project team initiated discussion with a third party aggregator in order to cost-effectively transfer operation from a project domain to BaU. It was agreed that all customers with a load-shed of 20kW + would be offered this opportunity. Given large commercial( as defined in the NTVV project bid

No. of installs

16%No. of lost

momentum16%

Unable to establish contact

19%

Not interested22%

Other27%

Customer Recruitment

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 33

document) relates to those customers with loading over 200kW; all large commercial customers were applicable to progress discussions, of which 70% obliged. Meanwhile four light commercial customers qualified to commercialise the value of their ADR, of these customers 100% were interested in taking discussions forwards.

4.5.2 LO-5.2 - Light commercial SMEs Following recruitment attempts of first priority large commercial sites achieving approximately two-thirds of target installs, the project looked to further its bid commitment to “explore the extent to which… [ADR] could be applied to SME customers” to actually trialling ADR on these sites. Some of the key motivations for recruitment of SME and light commercial customers included:

The benefits of being seen to be part of a local project supported by the local council.

A fully funded site audit & installation was necessary for SME customers to benefit from non-incentivised ADR trials.

Potential challenges may occur in connecting to a corporate network. To overcome this hurdle switching to General

Packet Radio Service (GPRS) as soon possible Is advisable. (3 customers required GPRS as opposed fixed

communication to allow communication of the DRAS portal, all of these were light commercial).

Drawing upon experience of industry aggregators using the Honeywell ADR equipment, signing up and installation of ADR on large commercial sites (>200kW) may typically be paid for by a third party; however on smaller customers (<200kW) ADR capabilities are usually only cost-effective when funded by the site itself. This operation, through an aggregator, however allows a customer to participate in multiple markets stacking potential benefits from ADR. In a DNO led model whilst direct recruitment of larger customers may be cost effective it may only be viable to recruit light commercial customers through an aggregator who can facilitate stacked benefits for said customer. Whilst DNO’s entering the DSR market, if not managed closely, has the potential to add additional complication to an already complex market, DNO’s will provide an additional layer of benefits to entry for light (and large) commercial customers.

4.5.3 LO-5.3 - Domestic Whilst the NTVV project did not actively procure DSR on domestic customers, five domestic consumer consortium events held with project participants to the monitoring trials were held across the course of the project. Aims and objectives varied and included: retaining customer interaction, improving understanding of how customers perceive their energy usage, dissemination and appetite for time of use (ToU) tariffs. Focus groups showed customers categorise themselves as either ‘interested’ or ‘very interested’ in RAG pricing, with an even split between those preferring fixed or variable time band. Most customers noted they would rather have control of this themselves (57%) as opposed an automated system or fixed timers.

Through the projects LCP work events were held to discover appetite for different low carbon initiatives/models. This included ‘Bracknell Low Carbon Day’ an un-incentivised marketing event which asked residents of Bracknell to reduce energy usage between 5-6pm. This event resulted in an approx. 1 MW reduction on Bracknell Primary Substation, demonstrating that with the correct messaging, customers are prepared to alter their consumption habits. Full details can be found in Annex A of SDRC 9.8b 2.

NTVV has also demonstrated how a DNO with accurate modelling of the network could reveal feeders likely to experience thermal or voltage constraints under particular scenarios. This may then be managed through a variety of means, whether a management signal such as tariff incentives (Solent Achieving Value from Efficiency (SAVE) project, SDRC 4- June 2017), or a technological solution such as EMMA (SDRC 9.8a 4) for PV load, or smart charging (My Electric Avenue, Project Closedown Report) to manage EV load.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 34

5 Performance Compared to the Original Project Aims 5.1 Objective 1: Applying proven data analysis from the EDRP to understand the different customer

types connected to the distribution network, and their effect on network demand

Proven data analysis from the EDRP was used successfully to categorise customers based on their energy usage using a finite mixture method; this was applied to the customer data derived from end-point monitors within the project. As discussed in Section 3.3.3, ten categories of customer were defined and this was described in SDRC 9.5(a) LV Customer Groups Presented. These groups were clearly defined based on the levels of demand in the key periods of breakfast time, day time, evening and overnight. Identifying the number of customers in each category connected to a feeder would allow immediate impacts to be assessed; if the customers in each category show similar trends regarding the uptake of low carbon technologies, then future demands can also be assessed. The EDRP data was also used to, generate and test the first household level short term load forecasts developed.

The objective was met with regard to the understanding of the customer types. The clustering into weekly behavioural demand groups helped improve the understanding of the dynamics and instability of energy behavioural groups. This lead to the main clustering that was performed with the yearly attributes. The clustering also provided the first evidence of the poor link between energy demand groups and non-energy characteristics, in this case the MOSAIC (http://www.experian.co.uk/marketing-services/products/mosaic-uk.html) classifications. A further benefit of the clustering was the foundation for the research into understanding domestic energy usage behaviour and properties and informing first development of the buddying algorithm that links monitored customers to unmonitored customers; this fundamentally enabled the modelling of the network. Hence the effect of customer types on network demand could be assessed, fulfilling the second part of this objective.

Short term load forecasts were developed on the EDRP data which was instrumental in the development of the household level forecast error measure.

5.2 Objective 2: Understanding How the Behaviour of Different Customer Types Allows Informed Network Investment Decisions to be Made

As for Section 5.1 / 5.2 above, the analysis of customer types led to the process for buddying un-monitored customers to monitored customers; this allowed the connectivity model established in the NME (see SDRC 9.6 Low Voltage Network Modelling Environment Built, Installed and Commissioned) to be populated with representative customer loads. Power flow analysis studies were run in the NME on the base load profiles as assigned, and following the addition of various combinations of low carbon technologies (long term scenario forecasts). The outputs of the studies were analysed in line with electrical rules to allow each feeder to be classified as no problem, requiring further consideration or monitoring, slightly out of limits justifying support from a smart solution, or significantly overload justifying traditional reinforcement.

For the clustering, it was found that one of the main clusters of the behavioural groups was linked strongly to the profile class 2 (Economy 7 tariff) customers. Thus when determining the grouping for the buddying it was necessary to ensure that domestic customers were grouped according to profile class. In addition, the analysis determined that although the distribution of daily demand was limited to 10 groups the daily mean demands were much more diverse. Hence for the buddying algorithm sufficient customers, with varying demands had to be allocated to each group. This determined which council tax brackets would be grouped together. The energy usage behaviour of customers was then modified by assigning them low carbon technology profiles in the scenario driven forecast algorithm. They were assigned based on predetermined or user defined specifications linked to clustering behaviour.

The detailed analysis of customer types and process for using these to run power flow studies, including study results are described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation and this provides evidence that this objective has been met.

The success of the grouping was determined by the accuracy of the buddying algorithms. Two algorithms were developed which used this grouping, the simple and the genetic algorithm. Analysis and results of the algorithms for both real and pseudo feeders were presented which showed the effectiveness of the methods. In particular, it was shown that both methods were accurate at the feeder and individual level compared to a standard Monte Carlo approach, with feeder loading information from substation monitoring helping to improve the overall accuracy of the buddying.

In addition from the experiments clear links were found between the accuracy of the modelling and the number of customers connected to a particular network. This can be used by a DNO as a first indicator of which LV feeder circuits should be monitored (i.e. those circuits where the modelling is known to be least accurate). Power flow studies were also run on the networks where low carbon technologies were included. This demonstrated that simulations of long term

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 35

forecast scenarios could be run effectively, and the outputs interpreted to understand the performance of the network. The quantity and location of thermal and voltage excursions were revealed, allowing solutions to be proposed, and hence informing network investment decisions.

5.3 Objective 3: Demonstrating Mitigation Strategies, Both Technical and Commercial, in a Live Environment, to Understand:

5.3.1 Flexibility of DSR Procured by the project, Imperial College London published their 2015 report “Role and benefits of ADR in distribution network planning uncertainty” (See Appendix 16.5.3.1 for more information). Using project data and outcomes this report discusses a minimum-maximum (min-max) regret decision making approach to assess how a network planner can understand the flexibility that DSR can offer to a specific network scenario. This works by minimising the cost of uncertainty under the most unfavourable future realisation.

As would be expected this analysis demonstrates the value case largely depends upon the cost of ADR. Specifically noting that relatively low cost ADR, deployed at the beginning of the planning horizon (when uncertainty is faced around future demand growth) leads to very significant maximum regret reductions by postponing network reinforcement decisions. Higher cost ADR however is likely to have a more competitive business case at later stages under some scenarios and reduces regret associated with stranded capacity and/or reinforcing network assets multiple times.

As section 4.4.3 and LO-5 above have shown those customers most likely to be early and effective adopters of DSR should have a BMS and include but are not limited to larger commercial customers.

5.3.2 Use of Power Electronics and Storage As described in Section 4.4.1, a number of distributed technology solutions were successfully deployed on the project including the use of power electronics with and without storage. 25 ESMUs were installed and connected to LV feeders in the project area allowing each of the functionalities to be trialled. Analysis of data from the network monitoring and power flow simulations of several LV feeders have shown that the loading between phases is mostly unbalanced and exhibiting distinctive peak demand periods. The unbalanced phase loading causes voltage unbalance and high neutral current.

The modelling methodology adopted by the project (buddying combined with clustered allocation of LCTs to the connectivity model in the NME) allowed the feeders that exceed thermal or voltage limits to be identified. This was described in SDRC 9.8(c) Part 1 University of Reading Smart Analytic and Forecasting Evaluation. It was shown that a feeder that is overloaded by less than 10%, for one season only and for less than two hours would be within the capability of a network connected ESMU. The identification of the optimal location along a feeder for an ESMU to be installed was established using the ESMU Location Tool. This was developed in the project and described in SDRC 9.4d Produce learnings from energy storage and power electronic deployment, Appendix K.

It was demonstrated that a network demand reduction of up to 12 kW per phase was possible using an ESMU as described in SDRC 9.8(a) Part 4 LV Network Storage – ESMU Trials. The use of power electronics alone allowed energy to be transferred from one phase to another, effectively allowing an improvement in the balance of load on each phase. The voltage profile of a feeder was also successfully manipulated by up to 4 V per 10 kVAr of reactive power and up to 6 V per 10 kW of active power by injecting both active and reactive power using an ESMU. This is sufficient to mitigate the impact of micro-generation by a modest number of customers on a lightly loaded feeder, keeping the voltage within the statutory limits. The power factor was fundamentally manipulated by injecting reactive power, and hence it was shown that an ESMU could be used to manage the power factor to near unity (between 0.95 and 1).

The use of energy storage (lithium ion batteries) in addition to the power electronics effectively allowed the ESMUs to provide network support for more extended time periods and with reduced dependency between the phases. To enable the above benefits to be achieved using multiple ESMUs across the network a form of Smart Control system known as ADDM was also deployed. This allowed the storage devices to be managed on an automated and optimised basis drawing on short term feeder load forecasts as described in SDRC 9.4(d) Produce Learnings from Energy Storage and Power Electronic Deployment and Appendix 16.3.4.2 Smart Control. Four quadrant voltage support algorithms utilised both active and reactive power to maintain phase-to-neutral voltages measured at the ESMU within the predefined target range.

Peak-reduction algorithms, using short-term forecasts, generated schedules for the ESMU to charge and discharge the energy storage to reduce peak loading across the phases. Two versions were deployed: fixed half-hourly schedules per phase and an aggregated schedule for shifting energy whilst performing phase-balancing at 3 minute intervals. Fixed

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 36

schedules were not adjusted to compensate for forecast errors and applied constant power over half-hourly periods. The schedules were calculated for each phase and aimed to perform phase-balancing and charging/discharge energy storage for peak-reduction across all phases. The results obtained showed some improvement on thermal constraints and voltage improvement. Due to inevitable errors in demand forecasts and the half-hourly control interval, the control system could not guarantee peak-reduction compared to historical demand data as observed at a higher frequency.

For the phase balancing operation, it was possible to specify an offset on power flow in and out of the energy storage which increases and decreases loading across three phases. Aggregation of phase forecasts was used to calculate the schedule to charge and discharge energy storage equally across three phases at fixed rates during a half-hourly period. Phase-balancing then operated at 3 to 15 minute periods based on substation monitoring data whilst using energy storage at specified rates for each half-hour across three phases. The full capability of the battery could not be utilised due to hardware limitations.

The capacity of an ESMU is equivalent to three 4kW PV installations and three 3.5kW EV chargers; hence the functionality demonstrated confirmed that an ESMU could enable those quantities of renewable low carbon technologies to be connected and operated on a feeder that was already operating close to its thermal or voltage limits.

Hence it is confirmed and demonstrated that this objective has been met, both in terms of where and how power electronics (with and without energy storage) can be used to manage power factor, thermal constraints and voltage to facilitate the connection of renewables on the LV network.

5.4 Objective 4: Undertaking Dissemination and Scaling Activity to Ensure Validity and Relevance to GB, with Learning and Understanding Provided at Two Levels:

Across the course of the project the NTVV project team and its partners have delivered over 220 individual dissemination events (around 20% internal, 80% external), averaging approximately 1 event per week. Key stakeholders have included DNO’s, Government, National grid, electricity suppliers, generators, equipment suppliers, local councils, academics, housing associations, local businesses, local chamber of commerce, aggregators, media and international experts.

Headline events have included roadshows to each of the other DNOs, offering a 1 day presentation with questions and answers (Q&A), the project team asked each DNO which areas of the project were of most relevance to them and tailored each event to these needs. NTVV has also carried out 6 internal dissemination events with audiences including network planners, connections team and stakeholder engagement teams. These events utilised material produced within the NTVV policy and training package of work (SDRC 9.8C3) the project team has delivered with the support of EA Technology. Key areas covered include: LV Monitoring, LV Design and LV Network Storage. A formal project closedown event has been held with key stakeholders covering the entire NTVV project with 213 registrations

A key part of NTVV was to ensure on-going public engagement to test different techniques and interact with differing areas of society around the project. This was largely accomplished through the operation of the LCCAC, LCP’s, Very Important Person (VIP) project participant events school engagement and the NTVV project website. It is important to keep project participant events interesting and fresh, in order to achieve this NTVV tied it’s events with tours of relevant attractions within the local area including: The LCCAC, Waitrose Eco Tour, GE Grid IQ Centre and National Grids Control Centre. More information on dissemination and on-going engagement can be found in SDRC’s 9.3c and 9.8b2.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 37

5.5 Replicate Project Direction SDRC table with additional columns for submitted evidence and links to evidence

SDRC Delivery Details Date Delivered

Delivered to Schedule?

URL

9.1A First Automatic Demand Response (ADR) Agreement negotiated and signed with Commercial Customer.

31/05/2012

9.1B

Install the Honeywell/SSEPD interface equipment, programme the Building Management System (BMS) and implement a manual Peak Load Shedding event, via the DRAS and track the actual kW shift in Peak Load;

31/07/2012

9.1C

Demand Side Response Evidence Report, 30 Customers signed up to Automatic Demand Response (ADR) programme and host customer event-renew new arrangements.

30/04/2015 tinyurl.com/hura54q

9.2(a) Evidence Report: 250 In House End Point Monitors Installed 31/01/2013 tinyurl.com/z894l7k

9.2(b) Evidence Report 100 Substation Monitors Installed 30/04/2013 tinyurl.com/j85y8kq

9.2(c) Evidence Report: Install and Commission the Network Management Component of the Distributed Solutions Integrator System (DSI)

31/01/2014 tinyurl.com/z587gr6

9.2(d) Evidence Report: Develop and Trial Method of Optimising Network Monitoring Based on Installation of First 100 Substation Monitors

30/04/2014 tinyurl.com/gq2g3l3

9.3(a) Learning Report: Customer Engagement

29/02/2012 tinyurl.com/jxs36uy

9.3(b) 29/02/2012

9.3(c) Public Engagement 28/02/2013 tinyurl.com/zuhf3r9

9.4a Discussion Document: Energy Storage and Power Electronics on the Low Voltage Distribution Network

31/07/2012 tinyurl.com/hjvpdex

9.4b Evidence Report: LV Network Storage – ESMU Trials 31/03/2014 tinyurl.com/zxzuuad

9.4c Evidence Report: Install 25 LV Connected Batteries 31/07/2015 tinyurl.com/zdo2d89

9.4d Evidence Report: Produce learnings from energy storage and power electronic deployment

30/11/2015 tinyurl.com/gt88nt6

9.5(a) Evidence Report: LV Customer Groups Presented 30/11/2013 tinyurl.com/z3cpkmj

9.5(b) Evidence Report: Testing Methods to Produce Accurate Half Hour Forecasts of Domestic Energy

30/04/2014 tinyurl.com/zeorl78

9.5(c) Evidence Report: Aggregate and integrate forecasts to produce first report on the modelling of LV load profiles.

30/04/2014 tinyurl.com/jv8qtu8

9.6 Evidence Report: Low Voltage Network Modelling Environment Built, Installed and Commissioned

31/12/2013 tinyurl.com/zwlflsg

9.7 Public Engagement 28/02/2013 tinyurl.com/zceqazl

9.8a1 Knowledge Sharing Report - Monitoring Evaluation 30/11/2014 tinyurl.com/hdda7ed

9.8a2 Demand Side Response Evaluation & Network Controlled Demand Side Response & Energy Efficiency

30/11/2014

tinyurl.com/hkwhlcw 9.8a3

9.8A4.1 Knowledge Sharing Report – Hot Thermal Storage 30/11/2014 tinyurl.com/gvc98ud

9.8A4.2 Knowledge Sharing Report – Cold Thermal Storage 30/09/2015 tinyurl.com/gl8yucv

9.8A4.3 Knowledge Sharing Report – Battery Storage 31/10/2015

9.8A5 Knowledge Sharing Report - EV Chargers Usage Evaluation and Issues

30/11/2014 tinyurl.com/htus573

9.8A6 Knowledge Sharing Report - Smart Meter Performance 30/11/2014 tinyurl.com/hf4pqkc

9.8A7 Knowledge Sharing Report - Integration Solution Control Evaluation

30/11/2014 tinyurl.com/glcud7k

9.8A8 Knowledge Sharing Report - Overall Proven Benefits 30/11/2014 tinyurl.com/gpbppe3

9.8B1 Part 1: Low Carbon Fuel Poor Evaluation 30/11/2015 tinyurl.com/jo8jqsr

9.8B2 Part 2: Housing Associations and Low Carbon Promotions 30/11/2015 tinyurl.com/hj6o9tf

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 38

9.8b3 Technical Impact Evaluation: Impact on DNO Network from Low Carbon Promotions

30/11/2015 tinyurl.com/zajudef

9.8c1 Part 1 Smart Analytic and Forecasting Evaluation 30/11/2016 tinyurl.com/hauh823

9.8c2 Knowledge Sharing – Low Carbon Community Advisory Centre Final Version

30/11/2016 tinyurl.com/gtf2yxo

9.8C3 DNO Training and Policies Review 30/11/2016 tinyurl.com/j9ng26c

9.8D1 Review seminar, discussing project learning from the project, with attendees from Ofgem, DNO’s, product suppliers, customers and other stakeholders.

28/03/2017 On target

9.8D Project Close Down Report 30/03/2017 On target

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 39

6 Required Modifications to the Planned Approach

Battery Storage - Whilst the specification for ESMU battery storage devices has been well defined, the selection process initially found no vendors with a readymade system suitable for the project trials, demonstrating the lack of market readiness for this technology. As a consequence a further iteration of vendor selection was required, resulting in a four month delay in the procurement process, impacting the initial battery installation scheduled for March 2014. Change request CR001 was raised as a consequence and approved in February 2014. It was identified that the product would be assembled and integrated from separate components, and this added five months to the production period. Further, the desired 15 off single-phase devices proposed were not readily available, and a revised methodology for using three phase devices in single phase mode was considered expedient. The number of three phase devices was increased by 9 to reasonably match the cumulative energy storage capacity originally proposed. These changes to the delivery schedule and equipment configuration had no consequential effect on other project elements and allowed the proposed project learning to be gained; the change request was approved based on a revised delivery date of November 2014.

A number of design iterations to the cabinets housing the battery storage unit, necessitating multiple prototypes were required before the design could be finalised. The vendor incurred supply chain issues, where one could not supply the design either to time or cost. Further complexity in the design of the cabinet transformer and electronics also impacted their ability to deliver to the required specification, resulting in further delays to the ESMU installation and trials. Change Request CR003 was approved in October 2014. Consequentially the installation date was amended to July 2015. The associated learning reports SDRC 9.4(a) and SDRC 9.4(d) were amended to October 2015 and November 2015 respectively. This change request was also approved on the basis that there was no net increase in cost and the learning outcomes were unaffected.

The delays in achieving operational battery storage therefore had a consequential impact upon the integration ESMU control with the ADDM smart control system.

CTS - The project’s original intent was to incentivise 50 customers who owned small scale CTS to participate in trials. After an extensive assessment, the project determined that there were no CTS devices installed, nor were there plans for any to be installed within the trial area of the project. Change Request CR002 approved in February 2015 revised this trial to three large scale CTS systems, successfully demonstrating the shifting of peak demand.

Smart Meter Data Collection - All energy supply companies were approached during the project for the provision of smart meter data as described in SDRC9.8 (a) Part 6 Smart Meter performance. Unfortunately, the SSE supply business was the sole respondent with the capability to support volume smart metering, and committed to the provision of 1000 smart meter data sets throughout. SSE Supply assessed their customer base and estimated from their current installation rate that approximately 300 would sign up to the trial and that it would require a customer engagement manager to campaign in the project area to achieve the goal of 1000. By the end of the project 895 smart meters had been deployed in the Bracknell area.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 40

7 Significant Variances in Expected Cost

7.1 Produce Final Table of Costs Table 3 below provides a summary of the spend variance at project closure against the original project budget of £25.520.04k as defined in the project direction.

Table 3 Project Spend Vs Budget

Ofgem Category Budget £k

Cost at Completion £K

% Spend Note

LABOUR 5,932.76 6,333.96 107% 1

CONTRACTORS 8,710.71 8,810.32 101%

EQUIPMENT 4,526.44 4,216.42 93% 2

IT 4,043.53 4,230.95 105%

TRAVEL & EXPENSES 335.22 106.64 32% 3

PAYMENTS TO USERS 591.00 109.68 19% 4

DECOMMISSIONING 392.00 290.70 74% 5

OTHER 988.38 618.09 63% 6

TOTALS 25,520.04 24,716.76 97%

The financial statement in this report is subject to accruals and financial year end processing as a result very minor alterations to the final outturn costs should be anticipated.

The following notes provide the summary justification for project variances over 10% of budget.

Note Spend Variance Justification

1 Project Engineering spend higher than expected, primarily due to protracted development and delivery of the ESMU battery storage system, necessitated additional support, interaction and testing as well as incurring the need for extended on-site support. This relates to Change Requests 001 and 003. A projected overspend due to the extended completion date of this element of the project, necessitating additional project management resource as well as increased costs due to the requirement for contract project management resource, as an interim measure due to loss of internal resource due to ill health.

2 Extensive and successful LV monitoring and access to industry half-hourly data flows mean that HV monitoring is not required to fill gaps. Learning Dissemination & website increased costs due to additional work associated with the website animation material and planned changes to improve the usability of the website. Also additional planned ADR learning. Anticipated underspend due to reduced communication equipment costs. The late availability of battery storage has reduced the window for operation and hence reduced data communication costs such as Subscriber Identification Module (SIM) cards. Also due to a need for protracted ESMU support, requirement for additional ESMU hardware to enable ground fault protection and other modifications for 25 PEUs and 35 ESUs in line with council and local community requirements and additional costs for further approval testing.

3 Movement of cost allocations within the activity “Integration of monitoring, modelling and management” to better reflect the nature of project costs/milestone payments. Travel & Expenses not treated as exceptional items within the performance of this activity. No substantive change in overall in cost of activity.

4 Underspend for payments to users through tight control of incentivised ADR trial costs.

5 Lower than expected decommissioning costs as not all substation monitoring will now be removed.

6 The project anticipated the need to purchase land for the location of the ESMU battery Storage solution. Through negotiation, it was agreed that the project would only rent the land for the duration of the project and that they would be subsequently returned to green field. Detailed design has identified savings in some licensing costs. Budget reallocated to enhance customer experience through full-time staffing at high street outlet. No substantive change in combined cost of activities. Anticipated underspend for ICT field resource costs. Fewer system issues identified than originally anticipated.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 41

8 Updated Business Case and Lessons Learnt for the Method

The original NTVV business case was benefit developed in 2011 and was largely predicated on the levels of LCT uptake which were prevalent at that time. It also used the best modelling tools and information available at the time. The model used at this initial stage had a much lower level of granularity, and was based on three representative LV feeders, rather than the approximately 100 variants that exist current modelling tools. The other key difference between what we know now, compared to then, is the uptake of technologies by customers. There have been faster than expected uptakes of Solar PV, lower than anticipated uptake of EVs, with significantly a slower than expected uptake of HPs. The business case was then reviewed at the end of the project using the best industry view on the update of LCTs currently available, and also utilised the industry accepted Transform model to assess the benefits. The Transform model provides a much higher level of granularity in its representation of the LV network and therefore, gives a far higher degree of accuracy. After revision, the forecasted benefit is £900m, compared to the earlier assessment of £4,952m as shown in Table 4.

The use of substation monitoring and ADR gives a net benefit in discounted totex terms of £658m over the period to 2050. If the low carbon technology uptake is higher than currently expected, this could rise to over £2bn. However if uptake is lower, then the benefits are somewhat resilient, and are anticipated to reach £643m. More detail of the revised project business case can be found in Appendix 16.8 Quantification of NTVV Benefits. Table 4 Business Case Forecast Comparison

Item 2011 forecasted Benefit £m

2016 forecasted Benefit £m

Method 1 Better Understanding of Customers Method 3 Reduced Monitoring of LV feeders 4,713 800

Total deferred & avoided LV reinforcement 3,684 625

Deferred & Avoided Primary Substation Replacement 861 146

Monitoring of LV feeders 168 29

Method 2 Demand Management of Commercial Customers

Deferred & Avoided Primary Substation Replacement 237 100

Total 4,952 900

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 42

9 Lessons Learnt for Future Innovation Projects

9.1.1 Forecasting Medium term forecasting of several years in the future required three years’ worth of historical data. For future projects data should be planned to be collected over many years so that the annual trends can be fully examined.

9.1.2 Customer Engagement Customer letters - Recruitment for the Hot Thermal EMMA unit trials was slow and resource intensive, altering the customer letters to ‘Dear Firstname Surname’ saw a rapid increase in response rate to approximately one-third. Use of Project Partners – Those which are respected and trusted within the local community provide significant leverage in trial recruitment. Mention of universities and the local council were especially valuable in recruiting customers. Commercial Customer Engagement - supplying multiple documents such as installation and collaborative agreements was perceived as doubling the legal review and sign-off activities. Combining these into a single document simplifies the information delivered to the customer, minimising any confusion. For engagement with commercial customers it is important to develop working relationships with the local business community and the local authority. 31% of ADR contacts came through consortium events and local bodies which give the DNO a platform and relationship to fully understand the potential for demand response.

9.1.3 Local Authority Creating a mutually beneficial relationship, engagement and collaboration with local councils such as BFC is an invaluable route to personnel, local knowledge and relevant data sources not available elsewhere.

9.1.4 Website Significant project information including SDRC’s and progress reports were produced and available on the website. Feedback indicated the website was difficult to navigate and find relevant information. As a result, the website was refreshed and the library was upgraded to provide a search function with more detailed document descriptions; easing the search for relevant sources of information. On subsequent projects, the requirements of the website should be considered in more detail, with assessments of likely document storage and search facilities.

9.1.5 Demand Response System Profile Access - The DRAS system does not currently allow for different role types, allowing anyone with system access to schedule a load shedding event across the entire customer base. This did result in an unauthorised load shed being scheduled, but identified prior to initiation. The project created a dummy customer, who would therefore only be requested to perform a load shed if such an error re-occurred. Consideration for future systems deployment is the need for different profile login and access control.

9.1.6 EPMs SSE IT revised their security protocols and revised their encryption standard for information reception. This resulted in a loss of communication and stopped the delivery of data. This necessitated development of new firmware, delivered remotely. Consideration for encryption standards early in the project cycle will minimise the need for costly upgrades. Firmware delivery over the air should be encouraged as this minimises site visits.

9.1.7 Substation Monitoring Equipment failures occurred, necessitating site visits and unit replacement. Consideration should be given to the level of spares based upon product maturity and expected failure rates. Data streaming in near real time is valuable for operational decision making, but is data intensive. For volume system deployment data volumes and update frequency need to be considered.

9.1.8 Energy Storage The remote relationship with the Canadian ESMU manufacturer did not breed the close collaborative working relationship that was originally anticipated. Maintaining good communication with them was difficult due to their geographical and time differences. While the commitment of the manufacturer remained in place it was recognised that the use of a European or UK based supplier would have been advantageous. Ensuring that the ESMU product was compliant with EU CE Marking criteria became problematic as the manufacturer was geared up for testing to American standards. Compliance with the Low Voltage Directive and Electromagnetic Compatibility requirements was therefore achieved later than desired.

The commissioning of ESMUs revealed some short comings in equipment firmware and software. To support this commissioning activity the manufacturer dispatched an appropriate development engineer to support the SSEN team in the UK. Numerous refinements and upgrades were made and then deployed to all ESMUs; this process had not been

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 43

allowed for on the scale required for either the manufacturer or SSEN. This partly reflected an optimistic view of the technology readiness level, and partly the lack of project anticipation in having to deal with supply chain issues from outside of Europe.

9.1.9 Smart Control Data Latency - Due to the data transmission delays from the ESMU’s and substation monitoring to the control system; Smart control of ESMU’s ability to control in real time was impaired and at best the control system could effect change every three minutes. Placement of the control system within the ESMU would open a possibility to apply instructions in a near “true” real-time increasing responsiveness of the control algorithms to the changes in loading of the network. This approach would still require a communications link maintained with the ESMUs to convey control configuration parameters, status and demand forecasts.

Development Strategy - The development of the algorithms has been closely linked to the live system; allocation of dedicated ESMU, connected to a real-time network simulator or a test-bed, for testing and development of the control system would significantly reduce the development time and allow improvements in the performance of the algorithms.

9.1.10 Policy & Training Material to Support Adoption into BaU In addition to creating useful, practical outputs from the project, the process of developing the policy and training material was valuable as a means of building and consolidating the project team’s thinking on applying the technologies deployed within the NTVV project as BaU; taking into account the current industry environment (regulatory framework, market development, etc.). The consideration of this context adds more to the learning than simply demonstrating whether a solution works technically and provides the required functionality. It is recommended that future projects consider some element of BaU 'strategy development' as a deliverable. This may take the form of the production of BaU style 'Policy' & 'Technical Guide' documents as with NTVV. Alternatively the output may suit the creation of a strategy document that concisely summarises the project's conclusions on technology implementation, and identifies how future industry or market changes may influence those conclusions.

9.1.11 Project Management Planning and managing multiple activities with partners - Some bottlenecks in response time from one partner were seen during this period; the lesson learned from this was that initial programme planning had underestimated the impact of milestones from multiple activities on a single team in the partner organisation. Future projects should consider the possible impacts of such milestones on the partner organisations through early engagement and planning reviews.

Planning and Estimation - The project underestimated the time required to capture, document and review requirements. Consideration of confidence factors, accounting for the level of uncertainty in the project deliverable would have minimised these overruns.

Test Planning - User Acceptance Testing for the LV NME was planned to occur shortly after requirements capture. However this activity should have been scheduled later when the software had been developed. When planning development based activities, a walk-though of the activities with the development and test teams would have highlighted these inconstancies. Consequently, it was not possible to produce training materials or conduct training until development had been completed.

Managing Requirements - The ADDM development encountered significant delays due to issues of security. This required an extensive review of the architecture and location of the firewalls, even those which were internal to SSE, crossing the boundaries of SSE IT and SSE Real Time Systems (RTS). Clear understanding of the differing security options and earlier engagement with the IT security team could have resolved the requirements and development and integration could have been planned better.

Technical Language - The project noted an instance where project specific language could have caused confusion had the recipient not checked the context and asked to clarify. In this case, a server system was termed the ‘test environment’ but was being reconfigured such that it would ultimately become a ‘production environment’ – care is needed to ensure specific meanings are understood by all.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 44

10 Project Replication

The table below summarises the information required to replicate the different aspects of the NTVV trials, including reference to relevant project documentation. If further detail is required, please contact [email protected].

Trial Details

Short term Forecasting

Short term Household forecasts: executable files have been created; methods have been published in the article ‘A refined parametric model for short term load forecasting’ in the ‘International Journal of Forecasting’ and this is freely available as a preprint. LV point and probabilistic forecasts for storage control have been integrated into the smart control algorithms as

per section 4.2. Smart Control; these have been documented and submitted to International Conference on Electricity Distribution (CIRED).

Medium term Forecasting

To replicate the simple year ahead forecasting, one to two years’ worth of historical data is required together with a simple calculation tool such as an Excel spreadsheet. To replicate the GAM, the free ‘R’ software for statistical analysis with the GAM package is required.

Forecasting LCT Uptake

The code (in the form of executables) used to forecast LCT uptake and to calculate uncertainty were made available to GE for incorporation into the NME, allowing simulations to be run. See Hattam & Greetham (Jan 2017) for a detailed description of the clustering and uncertainty algorithms applied by these executables.

Customer Engagement

The individual techniques used to engage customers (domestic and commercial) can be selected and replicated to tailor future engagement strategies to the aims of other projects or schemes. A LCCAC is unlikely to be cost-beneficial to a DNO, however many of the functions of the LCCAC could be replicated by a DNO, without requiring a permanent physical location. Value may be found in combining activities both within the DNO (i.e. stakeholder engagement, education around demand reduction, project plans) and with partner organisations (e.g. local council, trade organisations, energy efficiency organisations, community groups, etc.), for example through ‘event days’.

Local Authority

Replication of this approach in a BaU environment would require labour resources as it will take time to effectively engage with the local authority. As with NTVV, it is recommended that the objectives of any project are linked with the low carbon initiatives of the Local Authority, allowing both parties to share knowledge without formal contracts or financial agreements, to minimise procurement or legal costs.

LCP

The LCP activities detailed in SDRC 9.8b2 could be replicated, however these are unlikely to prove cost effective in a BaU context. Alternatively, participating in events run by third parties would replicate the benefit of these activities whilst minimising the cost.

Website

Creation of a project website could be done simply and at relatively low cost, and brings arrange of benefits to a project for communication plans, disseminating information and providing a point of contact.

Demand Response

DSR schemes may have many forms, however to replicate NTVV’s ADR the following technical aspects are required, as detailed in SDRC 9.1c:

Honeywell’s DRAS with licensing

DR Gateway - the interface between the premises BMS and DRAS

Main Electrical meter interface

CTS Ice Bears provide an off-the-shelf solution to smoothing air conditioning load, however the space requirements for installation of the Ice Bear are likely to inhibit future adoption at some premises.

Commercial

Replication of a single contractual agreement is recommended, even where there are several parties, e.g. NTVV’s ADR at commercial premises. The full learning around the development of this can be seen in SDRC 9.8a2&3.

EPMs EDMI Mk7C endpoint monitors were installed between the cut-out fuse and the Supplier’s tariff meter as described in SDRC 9.2a (product brochure included in Appendix 16.10.3.1a). A Vodafone VPN transmitted the usage data overnight to GE’s SMOS, and data was subsequently stored in Pi Process Book for access.

Senical EPMs Senical Smart Fuse monitors were typically installed in place of the cut-out fuse, but where not compatible with the existing cut-out fuse connection an additional cut-out base was installed between the cut-out fuse and the tariff meter.

Smart Meters For NTVV these were fitted with three individual data extracts.

Substation Monitoring

GE supplied Multilin DGCM Field Remote Terminal Units (RTUs) installed under work instruction WI-PS-912 from

the Tier 1 project SSET1002 Demonstrating the Benefits of Monitoring LV Network with Embedded PV Panels and EV Charging Point (see Appendix 16.10.3.2a for the Instruction Manual). JRF-1 split-core flexible Rogowski coil CTs were used as current sensors (see Appendix 16.10.3.2b for the product brochure). Voltage connections were made using Drummond G Clamps (Drump25) (see Appendix 16.10.3.2c for the product brochure). Fused leads were used throughout for the voltage connecting cables. The installation of the first 100 monitors is described in SDRC 9.2b.

Characterisation The NTVV buddying algorithm uses Profile Class, mean daily usage (from quarterly meter readings), knowledge of generation at the property (available through G83 notices), and Council Tax Band (CTB), making this easy to replicate. Further information is available as an appendix to SDRC 9.8c1.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 45

Trial Details

Network Modelling

The NTVV NME integrated GIS Smallworld Electric Office Interface, CYME CYMDIST 5.04 Power Analysis Tool, and an ESMU location Interface. The preparation of network data for use in the NME is described in SDRC 9.6. The buddying approach and aggregation executable code incorporated in the NME is described in Appendix 16.10.3.4a and SDRC 9.8c1.

Aggregation The aggregation algorithms are described in SDRC 9.8c (see G. Giasemidis, et al. (Dec 2016) for further information).

Energy Storage To replicate the NTVV approach LV network modelling should be used to identify the LV feeders that would benefit from power electronics, and identify the optimal location along the feeder where the ESMU should be deployed. This ESMU location tool is described in detail in Appendix 16.10.4.1a. The original specification for the ESMUs is provided in Appendix 16.10.4.1b. The allocation schedule of DNP3 points is shown in Appendix 16.10.4.1c. Work instruction WI-PS-1184 created for the installation, maintenance and de-commissioning of ESMUs is given in Appendix 16.10.4.1d. The experience gained during installation and deployment of the ESMUs is presented in SDRC 9.4d.

Smart Control

The ESMU operational management high level architecture is described in Section 3.4.2, and a more detailed view of the ADDM system is given in Appendix 10.4.2. This is replicable by any DNO operating a SCADA based DMS with access to DNP3 and Simple Object Access Protocol (SOAP) interfaces. This arrangement was designed as a project solution; learning regarding the preferred solutions for BaU implementation is described in Appendix 16.3.4.2.

DMS The NTVV project used GE Digital Energy PoF system Release 5.2.2.1.2. The requirements for the CIM integration, SCADA support and SOAP interface are documented in SDRC 9.2c. Reports that may support future implementation of an LV DMS are included in Appendix 16.10.4.3a, Appendix 16.10.4.3b, and Appendix 16.10.4.3c.

HTS EMMA SP40v2.1 controller supplied by Cool Power Products Limited. The installation of the EMMA devices is reported in SDRC 9.4b.

The Policy and Training material delivered through the NTVV project also supports the transition of new technologies to BaU at scale and the replication of the technologies on other DNOs’ networks. This has been created in such a way that it may be adopted where appropriate, adapted in light of further research or wider external factors, or used as a basis to incorporate and develop policy for other new, innovative solutions as technologies are transitioned from innovation trials to BaU application.

During the development of this material consideration was given to a list of key factors which reflect circumstances, characteristics or considerations that will have an influence on the scalability or replicability of a solution, as identified through SSEN’s participation9 in the European collaborative project DISCERN (Distributed Intelligence for Cost-Effective and Reliable Distribution Network Operation)10. These factors provide an assessment framework to highlight where detailed consideration would be of value to ensure the feasibility of scaling or replicating a solution in a different network environment11.

In addition to providing a suite of project deliverables of use within the business, this material is available externally to industry stakeholders, including other DNOs and the regulator, to inform policy development across the industry and support the effective future operation of these solutions. The NTVV policy documents and training modules are available from SSEN’s Asset Management & Innovation team, and can be requested via the Contact Us page of the NTVV project website (www.thamesvalleyvision.co.uk/contact-us) or by sending an email to [email protected].

9 To leverage the work undertaken in the NTVV project, SSEN participated in the European Seventh Framework Programme (FP7) project DISCERN. Through the DISCERN NIA project , SSEN shared experience and information drawn from NTVV to access the range of additional knowledge and expertise represented by the DISCERN project participants themselves, together with information on the solutions implemented by the DNO/DSOs partners in their smart grid demonstration site projects. The DISCERN NIA Close Down Report is available on the ENA’s Smarter Networks Portal www.smarternetworks.org/NIA_PEA_Docs/NIA_Project_Closed_Down_Report_-_DISCERN_-_final_p_160729134722.pdf 10 The DISCERN EU FP7 project ran from February 2013 to April 2016, and the project consortium comprised eleven project partners drawn from DSO, technology provider, research and consultancy fields across Germany, Great Britain, Spain and Sweden. The DISCERN Final Report and accompanying deliverables can found on the project’s website www.discern.eu/project_output/finalreport.html 11 Detailed information on this is provided in SDRC 9.8(c) Part 3 ‘DNO Training & Policies Review’, available from NTVV project website www.thamesvalleyvision.co.uk/library/sdrc-9-8c3-dno-training-and-policies-review

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 46

11 Planned Implementation

The following subsections outline the current status regarding the adoption of the NTVV technologies into BaU.

11.1 LV Monitoring The NTVV project implemented both secondary substation and end point monitoring on the LV networks.

Plans for whether, and how, our Distribution System will be modified in light of project learning

Further actions required to support implementation of beneficial methods

Likelihood that the methods will be deployed at scale in future

Recommendations for how the outcomes of the project could be exploited further (inc. further trials necessary to move to the next TRL)

DNO Actions Non-DNO Actions

The strategic use of LV substation monitoring devices has been accepted as necessary for adoption into BaU by SSEN.

Both SSM and EPM monitoring devices are available as Commercial Off The Shelf (COTS) products and are cost effective for LV monitoring in line with the LV Monitoring policy developed through NTVV.

Cellular network services are widely available, and radio communication can offer an alternative for sites with low mobile coverage.

A number of the NTVV SSM devices are being retained in situ to support SSEN’s strategy for early EV cluster management.

Project learning has been shared with SSEN’s ‘Low Cost LV Substation Monitoring’ NIA project12.

The implementation of centralised systems falls under SSEN’s wider IT Transformation programme.

The project has enabled SSEN to collect and store LV monitoring data from substation monitoring devices, EPM devices and smart meters in a single database. This database (Pi Historian) is already in use as a BaU system. As a result, SSEN has learnt how to upgrade existing systems, and how to specify functionality requirements for new systems, to enable this data to be collected, stored and made available to other systems to successfully integrate LV monitoring data into the BaU environment.

Information needed to support DNO adoption of LV monitoring devices is provided in the LV Monitoring Policy (T2-LVMON-PO-001, Nov 2016) and Technical Guide (T2-LVMON-TG-002, Nov 2016) documents developed through the NTVV project.

The envisaged use of smart meter data in place of EPM data is dependent on the national roll-out of smart meters and finalised arrangements regarding the data available to DNOs, and whether this represents a cost effective alternative to EPM data.

Manufacturers of substation monitors could increase market share by finding ways of reducing device costs or increasing their functionality and value (e.g. by incorporating capabilities such as distance to fault calculations).

NTVV has demonstrated that the strategic use of LV substation monitoring devices is of value to DNOs in providing data that can be used for both operational and planning purposes.

The strategic deployment of EPM devices by SSEN is supported in the short term (e.g. where relevant to address specific customer issues), however in the longer term it is expected that the national roll-out of smart meters will provide data that can be used by DNOs for planning purposes.

The availability of EPM data through the NTVV project has allowed SSEN to develop approaches for incorporating half hourly customer demand patterns into the network planning and design tools created, and the analysis approaches have been designed such that smart meter data can be utilised in place of EPM data.

To ensure the optimised application of LV substation monitoring at scale, the project outputs set out where retrofit application should be targeted to support planning and operational use cases, and bring benefit to customers and/or the business. New secondary substations should be designed to include LV monitoring equipment to minimise costs, and new equipment should be designed to incorporate this functionality.

12 SSEN’s ‘Low Cost LV Substation Monitoring’ Network Innovation Allowance project runs from March 2016 to March 2018

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 47

11.2 LV Design

The NTVV project created a NME which combines a geographical information system, equipment database and power flow analysis tool to allow network models to be created and used for detailed LV network studies.

Plans for whether, and how, our Distribution System will be modified in light of project learning

Further actions required to support implementation of beneficial methods

Likelihood that the methods will be deployed at scale in future

Recommendations for how the outcomes of the project could be exploited further (inc. further trials necessary to move to the next TRL)

DNO Actions Non-DNO Actions

The NTVV project has demonstrated that the functionality provided by the NME has the potential to provide significant benefit and value to a DNO.

Whilst the NTVV NME is not yet suitable for immediate roll-out at scale across the business, it is the intention to bring this functionality into BaU. In line with business practices this will fall under the remit of the IT Transformation programme, and the NTVV project team are actively engaged in the specification and tendering of a replacement GIS system capable of providing the functionality developed and demonstrated in the NME. The LV Design policy material developed through the NTVV project reflects the principles to be applied.

In keeping with the introduction of any enterprise level system, detailed systems analysis will be required prior to the commissioning of a system such as the NTVV NME within BaU. This would include amendments to processes and working practices as well as the systems hardware.

DNOs must to learn to extract value from aggregated smart meter data, as provided on request from the DCC. In addition, the findings from research projects, such as NTVV, should be assessed for evidence which may justify the provision of selected aggregated smart meter data to DNOs in real time, or the provision of selected non-aggregated smart meter data, where this would support secure and efficient network operation and planning.

The detailed principals relating to the functionality provided by the NME and the use of LV decision support tools are provided in the NTVV LV Design Policy (T2-LVDES-PO-001, Nov 2016) and Technical Guide (T2-LVMON-TG-002, Nov 2016) documents.

Systems providers should look to develop the functionality provided by the NTVV NME into COTS products. This should include consideration and refinement of the data input methods, such as the process for incorporating and updating the latest and most appropriate buddying and clustering algorithms, in addition to the development of a suitable graphical user interface (GUI).

The application of half hourly demand profiles to individual customers allows the impact of potential future LCT uptake scenarios to be investigated. This provides a significant enhancement to LV network design methodologies for DNOs in adapting to changing energy usage patterns.

The decision support methodologies developed through NTVV provide a staged approach for assessing networks, to ensure the efficient use of resources:

the NME can derive Thermal and Voltage Stress Indicators to identify networks at risk from demand growth due to future adoption of LCTs, meriting detailed study

the more complex NME modelling functionality then applies half hourly load profiles for each property to determine the level of risk

the NME can then be used to assess whether new, innovative technologies can be used to manage the network or whether traditional reinforcement is required

These capabilities bring significant benefits to DNOs for network planning & new connections assessments, and merit adoption into BaU at scale.

Within NTVV, Council Tax Band and mean daily demand calculated from quarterly meter reading data were used to ‘buddy’ EPM households with non-monitored households. This has proven the concept of a ‘buddying’ algorithm to assign more refined demand profiles to properties for planning purposes, however further work to investigate alternative matching characteristics would be valuable (possible alternatives for consideration include Elexon or MOSAIC classifications).

The clustering algorithms developed to model the uptake of EVs may be adopted for use in future research or planning projects.

Further research trialling the use of smart meter data aggregated from very small numbers of customers would establish whether this provides granularity of data of value to DNOs and is practicable to utilise at scale.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 48

11.3 LV Network Storage

This section firstly address the implementation of LV network connected electrical energy storage devices, and then considers the thermal energy storage technologies trialled through installation within customers’ premises.

11.3.1 LV Network Connected Electrical Energy Storage Devices The ESMU devices trialled within the NTVV project contain battery modules which can be used to import and export electricity as required to support the LV network, in addition to a PEU (see section 11.4) that is able to provide dynamic phase load balancing, feeder voltage regulation and power factor correction.

Plans for whether, and how, our Distribution System will be modified in light of project learning

Further actions required to support implementation of beneficial methods

Likelihood that the methods will be deployed at scale in future

Recommendations for how the outcomes of the project could be exploited further (inc. further trials necessary to move to the next TRL)

DNO Actions Non-DNO Actions

The learning and experience gained from the deployment of ESMUs within the NTVV project demonstrates the value in DNOs owning and operating energy storage assets to provide flexibility in network management. LV network connected energy storage devices represent a technically feasible solution both for managing intermittent network constraints, and for allowing networks to be operated safely and securely where there is uncertainty over how energy usage patterns will change through time.

At present, deployment is primarily constrained by technology readiness levels and relative economics.

Maintain awareness of developments in the field of LV network connected energy storage, and actively engage in debate and further research in this area.

Such considerations may include DNO/TSO interactions, particularly through the DNO transition to Distribution Service Operator (DSO).

The NTVV LV Network Storage Policy (T2-LVSTOR-PO-001, Nov 2016) and Technical Guide (T2-LVSTOR-TG-002, Nov 2016) documents support the transition of battery storage technologies into BaU.

Three key non-DNO actions are required for the BaU deployment of such technologies by DNOs:

Technological maturity - manufacturers have a role in developing and increasing the availability of a range of COTS devices which apply defined standards and protocols for safety, security, communications and data exchange;

Economics - manufacturers can access and promote development of this market by reducing the cost of such devices; and

Regulatory issues - clarity is required over the possible future regulatory framework around DNO ownership and operation of battery storage, including licence considerations, to remove the currently perceived uncertainty around the future DNO ownership of storage. Regulatory barriers to the uptake of this technology as an efficient and effective solution for network management should be avoided.

As the ESMUs are not yet a COTS product, they are not suitable for deployment in a BaU environment at this point in time. However, energy storage is a growing market, with a number of manufacturers and product vendors developing electrical energy storage devices. Similarly, the costs for lithium ion batteries (the technology used within the ESMUs as well as some EVs), is falling. Within the RIIO-ED1 period it is therefore feasible that the availability of suitable, cost effective COTS products will increase.

The three factors cited under ‘Non-DNO Actions’ are currently under consideration within the wider industry, and the outcomes may influence DNO use of LV network storage technologies.

The ESMUs were developed specifically for the project and the Technology Readiness Level (TRL) of these devices has been taken from TRL2 (concept) to TRL7 (prototype & demonstration). Further work to advance these specific devices into COTS products should include reducing the size of the units and reducing the noise generated when the cooling fans are operating, to ensure suitability for roadside installation in residential areas.

The forward scheduling of ESMU activity based on historical monitoring data minimises the need for real-time monitoring and closed loop control. This approach could be further developed for the deployment of LV connected energy storage solutions.

The Commercial WP1 & WP3 reports commissioned for NTVV provide useful insight regarding regulatory aspects associated with LV network connected storage (see Appendices 16.11a & 16.11c).

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 49

11.3.2 Thermal Storage Devices The NTVV project demonstrated two thermal storage solutions to meet different network requirements, both of which are installed within the customers’ premises:

EMMA units - HTS devices for residential customers with solar PV generation which reduce the electricity exported to grid by diverting PV generation to heat water stored within a domestic hot water cylinder

Ice Bears - CTS units for commercial customers with air conditioning load that delivers a reduction in summer peak demand by creating ice at off-peak times for use to provide cooling during peak times

Plans for whether, and how, our Distribution System will be modified in light of project learning

Further actions required to support implementation of beneficial methods

Likelihood that the methods will be deployed at scale in future

Recommendations for how the outcomes of the project could be exploited further (inc. further trials necessary to move to the next TRL)

DNO Actions Non-DNO Actions

SSEN supports customer uptake of thermal energy storage technologies, however at present it is not SSEN’s intention to install such devices on the customer side of the meter as BaU. This policy similarly applies to domestic or commercial electricity storage technologies.

These technologies have demonstrated benefits for the local LV network, however as not all premises are suitable for installation, some customers receive direct financial benefit from the technologies where others do not. In addition, significant resources are required to manage the relationships associated with the installation and ongoing maintenance of devices owned within customers’ premises.

SSEN accept the use of these technologies where they form part of a service provided by a third party, such as a DSR aggregator, procured by SSEN to manage network constraint(s) and meet Security of Supply requirements, in keeping with the NTVV Capacity Response policy.

Maintain awareness of developments in the field of customer energy storage, and actively engage in debate and further research in this area.

Product manufacturers, flexibility service providers and customers each have roles to play in promoting the development of this market.

Standards may need to be enhanced or created to ensure the safe and efficient application of such technologies.

Regulation and legislation may need to adapt to ensure that there are no artificial barriers to the uptake of such technologies where these bring benefits to end customers or other parties.

Though it is unlikely that DNOs will choose to install, maintain or operate devices beyond the customer meter at scale under BaU, the use of such energy storage technologies by third party flexibility service providers, such as DSR aggregators represents a viable future scenario.

Similarly, these technologies have the potential to provide direct financial and energy efficiency benefits to the customers who have them installed, and so the market for such devices is likely to increase.

Cost reductions due to economies of scale or increased numbers of product vendors entering the market will also increase the financial attractiveness and customer acceptability of such devices.

The findings and data from the NTVV project may be of interest to DNOs, flexibility service providers, product manufacturers, customers or other relevant parties in allowing the potential impact and benefit of such technologies, installed within customers’ premises, to be assessed. These outputs contribute to broader industry knowledge in this area.

The Commercial WP1 & WP3 reports commissioned for NTVV provide useful insight regarding regulatory aspects associated with customer owned thermal storage technologies (see Appendices 16.11a & 16.11b).

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 50

11.4 Capacity Response

The term Capacity Response comprises a range of technologies or solutions that can be applied to cost effectively optimise the use of existing LV network assets. Such approaches enable the distribution network to accommodate additional load without the requirement for traditional network reinforcement.

Plans for whether, and how, our Distribution System will be modified in light of project learning

Further actions required to support implementation of beneficial methods

Likelihood that the methods will be deployed at scale in future

Recommendations for how the outcomes of the project could be exploited further (inc. further trials necessary to move to the next TRL)

DNO Actions Non-DNO Actions

SSEN supports the adoption of Capacity Response techniques into BaU.

The learning generated from the ADR trials undertaken within the NTVV project has supported SSEN in developing the CMZ schemes implemented directly into BaU. A CMZ is a geographic region served by an existing network where network requirements related to peak electrical demand under fault conditions are met through the use of DSR techniques (including demand reduction from commercial or aggregated domestic premises or export from third party owned generation and storage), provided as a managed service to SSEN by a CMZ service provider.

The PEU incorporated within the ESMU devices is able to provide dynamic phase load balancing, feeder voltage regulation and power factor correction. The NTVV project has demonstrated the value of this collective functionality as a new solution for LV network operation, and for addressing issues identified through the project’s installation of LV substation monitoring. Whilst the

Maintain awareness of developments in the field of capacity response technologies and services which will support flexibility in network operations, and actively engage in debate and further research in this area.

Such considerations should include DNO / TSO interactions, particularly through the DNO transition to DSO.

The learning captured within the NTVV Capacity Response Policy (T2-CAPRES-PO-001, Nov 2016) and Technical Guide (T2-CAPRES-TG-002, Nov 2016) documents support the transition of Capacity Response techniques into BaU.

Product manufacturers, flexibility service providers and customers each have roles to play in promoting the development of the DSR market.

Standards may need to be enhanced or created to ensure the safe and efficient application of such technologies.

Regulation and legislation may need to adapt to ensure that there are no artificial barriers to the application of approaches such as DSR.

The industry will need to give consideration to the implications of DSR for Supplier balancing and settlement. This may fall under the broader scope of changes to Supplier planning and forecasting approaches as the industry adapts to smart meters and domestic billing based on half hourly settlement periods.

Consideration will also need to be given to the charging methodology and how potential ToU tariffs made possible through the national roll-out of smart meters may sit alongside and/or interact with the market for DSR.

The Commercial and ADR reports

All cost effective techniques for maximising the use of existing network capacity are valuable.

Accordingly, there is increasing interest across the industry in the potential for enhancing flexibility in network operations through the use of capacity response technologies such as DSR and phase balancing.

The ADR and power electronics technologies trialled within NTVV each utilised commercially available systems, specially configured for the project.

At present, ADR is particularly suited to commercial premises with an existing BMS and significant building demand. The wider DSR market may allow participation by a range of property types, for example through aggregation.

Cost reductions due to economies of scale or increased numbers of product vendors entering the market will also increase the financial attractiveness and customer acceptability of such devices.

The reports, findings and data from the NTVV project may be of interest to DNOs, flexibility service providers, aggregators, product manufacturers, academics & researchers, customers or other relevant parties in allowing the potential application and benefit of technologies such as LV level DSR or phase balancing to be assessed. These outputs contribute to broader industry knowledge in this area.

The demand reduction levels achieved across the >2000 ADR load shed events run during the project may be of value to DSR aggregators with regard to the potential consistency or variability in response levels from different premises under different load shed scenarios.

The methodology for assessing the impact of demand reduction events developed for inclusion in the contractual arrangements for CMZs can be applied by a range of other DSR stakeholders. This provides a practicable and impartial methodology for calculating the baseline against which a load profile can be compared to ascertain the reduction seen during a demand reduction event. This performance

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 51

ESMU devices cannot be adopted into BaU at the current time, SSEN support the application of power electronics to provide real time phase balancing in addition to voltage regulation and power factor correction, thereby making efficient use of existing network capacity and reducing losses on the LV network.

commissioned for NTVV provide useful insight relating to potential market developments for DSR and other NTVV technologies (see Appendices 16.11a to 16.11f and Appendix 16.5.3.1).

assessment methodology is set out in the NTVV Capacity Response Technical Guide (T2-CAPRES-TG-002, Nov 2016).

The development & availability of cost effective devices which combine dynamic phase balancing, voltage regulation and power factor correction capabilities would support the exploitation of these benefits by DNOs.

11.5 Customer Engagement Customer Engagement ran through all aspects of the NTVV project. The Customer Engagement policy material developed within the project captures the learning gained from engaging with a wide range of stakeholder groups throughout the project.

Plans for whether, and how, our Distribution System will be modified in light of project learning

Further actions required to support implementation of beneficial methods

Likelihood that the methods will be deployed at scale in future

Recommendations for how the outcomes of the project could be exploited further (inc. further trials necessary to move to the next TRL)

DNO Actions Non-DNO Actions

The NTVV Customer Engagement policy documents set out the key considerations when engaging with customers both to recruit participants to schemes or trials, and to inform stakeholders of business plans and activities.

These documents will be utilised by SSEN as key reference material for the development of Customer Engagement Strategies when implementing new technologies or network management strategies (e.g.

SAVE13). Further, the experience captured in these documents will inform broader aspects of SSEN’s BaU customer engagement activities.

The recommendations and guidelines relating to DNO Customer Engagement as drawn from the NTVV project are provided in the Customer Engagement Policy (T2-CUSENG-PO-001, Nov 2016) and Technical Guide (T2-CUSENG-TG-002, Nov 2016) documents.

Trusted third party organisations can provide a valuable role in supporting a DNO with customer or stakeholder engagement activities through using their contacts to engage with relevant individuals or businesses, or by demonstrating their support for a specific project or scheme.

Customers are at the heart of activities for all DNOs, and high standards of customer engagement must be maintained at all times.

The recommendations and guidelines captured in the Customer Engagement policy documentation can be used by DNOs, aggregators and others to inform the development of customer and stakeholder engagement strategies for future research and innovation projects.

Third party organisations such as Local Councils, trade bodies and community groups may find the outcomes and recommendations of interest with regard to developing future partnering opportunities.

13 The SAVE Low Carbon Networks Fund project runs from January 2014 to June 2018 and assesses energy efficiency measures for network management

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 52

12 Learning Dissemination

12.1 Provide details of the methods used for sharing info/learning, and what the results In order to maximise project exposure and shared learning, NTVV dissemination activities have gleamed success through asking the Kipling / 5W1H (who, what, why, where, when, how) methodology. In total the project and its partners have averaged approximately 1 dissemination activity for every week of operation since project inception. Appendix 16.13.4 provides a snapshot of these.

Provide details of the methods used for sharing info/learning, and what the results were (The DNO Roadshow Events). Dissemination for the NTVV project can be broadly grouped into 6 main themes as displayed in Figure 4 below, where each bridge shows the dissemination avenue taken, with the respective cars coloured on relative level of dissemination success. These themes were utilised and trialled to engage a wide range of different stakeholders through the most appropriate options as detailed below.

Figure 4 Dissemination Channels

Website Provides a platform for regular, low cost updates and a credible link of authenticity to customers.

Presentation and Conferences

Tailored to be engaging and to the right audience a successful mechanism.

Roadshows and Training

Providing key learning outcomes and practical material for BaU operation.

LCCAC A hub within the community, central basis for events and DNO/energy exhibition.

Events Require more planning and organisation, when tailored to correct audience work well especially for obtaining feedback.

Publications Valuable for wide-spread engagement and to log learning for those most interested in the project (DNO’s).

In summary NTVV has organised over 220 dissemination events, of which approximately 42% were SSEN led and 44% project partner/other led. A sample of these;

4 x DNO roadshows to 5 DNO’s 5 consecutive utility week publications

Annual attendance at LCNI Conference Bi-annual project participant events

6 Internal training events LCCAC opening event

A project closedown event with 203 attendees across two days

Project participant closedown event at National Grid

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 53

Monitoring of

customer

end points

Monitoring of

low voltage

feeder

circuits

Modelling of

the low

voltage

network

using smart

analytics

Forecasting

and

clustering to

assess

impacts on

the low

voltage

network

Integration

of a low

voltage

distribution

management

system

Customer

engagement

Automated

Demand

Response

(ADR)

Battery

storage as a

means to

support the

network

Thermal

storage as a

means to

support the

network

NPG 1 2 3 3 3 2 2 1 1

UKPN 2 3 2 2 2 2 3 3 1

SPEN 1 2 2 3 2 2 3 3 1

ENWL 2 2 2 3 1 3 3 1 2

WPD 1 3 1 2 1 1 0 0 0

Total

Location

Preferences

12.2 State what other DNOs required/requested info. on, and how we sought their views (e.g. via DNO roadshow questionnaires, workshops, dissemination events)

Dissemination events were run for a variety of reasons whether sharing project knowledge, providing a platform to obtain information from an audience or providing an update and next steps for those participating within a project. Whilst objectives may vary, the need for review of dissemination activities remains constant. This allows for ever-improving and tailored events, maximising benefits to both host and audience.

DNO Roadshows The NTVV project team held 4 DNO Roadshows across the country between November and December 2016. These included presentations to 5 Electricity North West (ENW) and 1 WPD colleague in Preston, 29 NPG colleagues in Leeds, 4 UK Power Networks (UKPN) colleagues in Crawley and 9 Scottish Power Energy Networks (SPEN) colleagues in Cumbernauld. These presentations offered a full day in which to hear about key learning outcomes from NTVV most relevant to said audience. To maximise value for the audience NTVV provided a ‘menu’ of project activities that could be prioritised based upon relevance to each audience. The results of this can be seen below in Table 5 where a scale of 0-3 is used to indicate a DNO’s level of interest.

Table 5 Topics of Interest

Each Roadshow event was minuted, questions answered/recorded and then questionnaires sent out to each participant to uncover, where expectations were met, relevance to BaU and quality of the event. Responses indicate 99% success in covering those topics the audience expected. Of those topics presented on at the events, Figure 5 below shows relevance to BaU, with: ADR, Monitoring of LV feeder circuits, Modelling and Forecasting being found of most value.

Internal training Six internal training events were held, with 60 members of staff, to disseminate key project learnings to departments including planning, connections, stakeholder engagement, commercial and field teams. These events were tailored to outline the key areas likely to be of value to the objectives of each business unit and held across SSEN’s patches (Central Southern England and Northern Scotland). Prior to sending invites to these events senior management from across SSEN were engaged in order to highlight the importance and value of the training sessions and to identify the appropriate attendees.

Training sessions were divided to give an overview of the whole NTVV project within 2.5 hours and weighted to those areas deemed to be of greatest value to the attendees. Feedback from these events was captured in a questionnaire. The results of this can be seen below in figure xx. In summary 85% of attendees felt the information was of relevance

Figure 5 DNO appetite for NTVV

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 54

(‘definitely relevant’ or ‘might be relevant’ categories); mirroring the projects DNO Roadshows, areas of monitoring, modelling, DMS integration and ADR were sought of most relevance.

Table 6 Feedback from Internal Dissemination Events

Project Closedown Event Taking on-board feedback from the DNO Roadshows, NTVV closedown presentations, where relevant, focused on areas that could or were being taken to transition project learning into BaU.

The project closedown event was held in collaboration with the SEEN run, NINES Project, and was located at the Institute of Engineering and Technology (IET) in London. Running across two days (28th-29th March), the event split sessions across three rooms. Efforts were made to ensure similar themed topics were presented in differing time slots to avoid timetable clashes. An agenda for the closedown event can be found in Appendix 16.12.1.

In inviting people to this event detailed stakeholder mapping was carried out by both the NTVV and the NINES team, identifying customer based on interest and influence. In total over 200 people registered for the event with 203 people attending across the two days. Amongst others, attendees included 10 DNO representatives (excluding SSEN), 4 government representatives and 11 academics.

12.3 Explain how this feedback from DNOs has been taken into account when writing the closedown report

Project learning has been formally reported on throughout the project in 34 SDRCs, and 16 academic publications. Taking on-board learning from project dissemination events the project has where necessary re-visited trials and shaped the writing of this report with most weighting towards those areas of most interest and value to a DNO audience..

The report has been peer reviewed by UKPN, with comments gathered and accounted for in shaping this document and is included in Appendix 16.12.2. We recognise from our interaction that different network operators that different organisations (and individuals within them) have differing priorities and viewpoints. As a result, the project understands that it cannot be an expert in each arm of a DNO’s hierarchy; in order to get a BaU angle, the closedown report has been reviewed and directed by experts across SSEN, including: Network Planning, Connections, Stakeholder engagement and Depot Management.

12.4 Submit project learning to the Transform model and governance process and reference briefly in closedown report, confirming learning submitted to ensure Transform model assumptions are informed by latest evidence

Project learning has been submitted to EA Technology in order to populate their Transform model. This revised model has been used in the development of the revised NTVV business case described in section 8.

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 55

13 Key Project Learning Documents

13.1 Six Monthly Progress Reports

All published 6 month Project Progress Reports can be found on the NTVV website by clicking here, or via the following link:

http://www.thamesvalleyvision.co.uk/project-library/published-documents/project-progress-reports/

Individual 6 month Project Progress Reports can be accessed via the below links:

Report Date and Link URL

June 2012 tinyurl.com/jnz8evu

December 2012 tinyurl.com/j32oasp

June 2013 tinyurl.com/goebwfv

December 2013 tinyurl.com/h8mpoxf

June 2014 tinyurl.com/zdgc497

December 2014 tinyurl.com/gn4y8cf

June 2015 tinyurl.com/zzomyn5

December 2015 tinyurl.com/zscw6d2

June 2016 tinyurl.com/z6c5o95

December 2016 tinyurl.com/hjsgxs3

13.2 Academic Learnings

Appendix 16.13.3 lists published academic papers, and conferences and workshops either hosted or presented at.

14 Contact Details

To obtain further details on the learning from the project please visit the project website at: http://www.thamesvalleyvision.co.uk/

You can also contact us at: [email protected]

Alternatively, please write to us at: Future Networks, Scottish and Southern Electricity Networks, Inveralmond House, 200 Dunkeld Road, Perth, PH1 3AQ

15 References

Element Energy, 2012, Demand Side Response in the non-domestic sector, Cambridge: Element Energy

End of Document

SRDC 9.8 D Project Close Down Report SSET203 NTVV

New Thames Valley Vision

Page 56

16 Appendices

Appendix 16.3.2.1 Customer Engagement Lessons Learned

Appendix 16.3.2.1.1 ADR install

Appendix 16.3.4.2 Smart Control

Appendix 16.3.5 NTVV Policy & Technical Guide Documents

Appendix 16.4 Trial Interaction Groups

Appendix 16.5 Use Case Analysis

Appendix 16.5.3.1 Role and benefits of ADR in distribution network planning uncertainty

Appendix 16.8 Quantification of NTVV Benefits

Appendix 16.10.3.1a EDMI Mk7C Brochure

Appendix 16.10.3.1b Supply Point Monitor Brochure v2.3

Appendix 16.10.3.2a Multilin DGCM Field RTU Instruction Manual

Appendix 16.10.3.2b Current Sensor - Rogowski Coil JRF1

Appendix 16.10.3.2c G Clamp

Appendix 16.10.3.4a NME - Application of Buddying

Appendix 16.10.3.4b NME - Application of Clustering

Appendix 16.10.4.1a EESMU Location Tool

Appendix 16.10.4.1b ESMU specification v2

Appendix 16.10.4.1c ESMU DNP3 Allocation Schedule

Appendix 16.10.4.1d Work Instruction WI-PS-1184-v1.0 ESMU

Appendix 16.10.4.2 ADDM Architecture

Appendix 16.10.4.3a DMS LV Network Operations

Appendix 16.10.4.3b DMS CIM Implementation Guide

Appendix 16.10.4.3c DMS ADR Implementation Guide

Appendix 16.11a NTVV Commercial Assessment - WP1 - Regulatory Impact of Trials

Appendix 16.11b NTVV Commercial Assessment - WP2 - Review of Distribution Charging Arrangements

Appendix 16.11c NTVV Commercial Assessment - WP3 - Interaction with Industry Governance Documentation

Appendix 16.11d NTVV Routes to Market for ADR

Appendix 16.11e NTVV Opportunities for ADR in National System Balancing

Appendix 16.11f NTVV Forecasted UPS and Standby Generation DSR Potential

Appendix 16.12.1 NINES.NTVV.Agenda

Appendix 16.12.2 UKPN_NTTV Letter on closedown report_March 2017_300317rev1

Appendix 16.13.3 Academic Learning

Appendix 16.13.4 Dissemination activities