DFES Customer Behaviour Report

Distribution
Future Energy
Scenarios 2022
Customer behaviour profiles

and assumptions report
Produced: February 2022

Version: 1
Table of Contents
Foreword ................................................................................................................... 2
Introduction .............................................................................................................. 3
Part 1: Volumes ................................................................................................................................... 3
Part 2: Customer Behaviour ................................................................................................................ 3
Part 3: MW growth data and application of growth rates .................................................................... 3
Customer Behaviour Modelling .............................................................................. 4
Context ................................................................................................................................................ 4
Intended use of profile assumptions ................................................................................................... 4
Representative Days ........................................................................................................................... 8
Changing Nature of Demand ............................................................................................................ 10
Application of customer behaviour profiles ....................................................................................... 11
Structure of profile sections............................................................................................................... 13
Generation and storage technologies .................................................................. 14
Solar Generation ............................................................................................................................... 15
Onshore Wind Generation ................................................................................................................ 19
Offshore wind Generation ................................................................................................................. 23
Non-weather Dependent Generation ................................................................................................ 25
Battery storage .................................................................................................................................. 28
Demand Technologies ........................................................................................... 31
Underlying Demand .......................................................................................................................... 32
Domestic ........................................................................................................................................... 42
Non Domestic .................................................................................................................................... 45
Electric Vehicles ................................................................................................................................ 55
Resistive Electric Heating ................................................................................................................. 64
Heat Pumps ...................................................................................................................................... 69
Air Conditioning ................................................................................................................................. 77
Hydrogen Electrolysis ....................................................................................................................... 79
Summary of future developments ........................................................................ 81
Appendix A: Technology comparison to Open Networks building blocks ....... 82
Appendix B: Primary substation clustering ......................................................... 85
Customer Behaviour assumptions ....................................................................... 88
Glossary .................................................................................................................. 88
Table of Units .................................................................................................................................... 92
National Grid | Distribution Future Energy Scenarios 2022

Foreword
This document outlines the assumptions that National Grid (NG) use for customer behaviour as part of
the strategic network analysis activities. These assumptions are to be used in conjunction with the
Distribution Future Energy Scenarios (DFES) studies which are run on an annual basis. The DFES
studies and subsequent analysis aim to understand how customer requirements for energy will change
the development and operation of the distribution network as the UK transitions to a Net Zero future.
This document sits within a suite of current and planned documents that NG aim to publish on a regular
basis as part of the investment planning process.
Forecasting Network Impact Assessment Optioneering
The Distribution Future Energy The Distribution Network Options

The Network Development Plan
Scenarios (DFES) identify how Assessment (DNOA) outlines how
(NDP) uses forecasts to analyse and
customers will use our network in we plan to invest in our network to
identify future network constraints.
future. solve constraints.
When the customer behaviour assumption in this document are applied to the DFES projections a load
set of MW/MVAr values can be generated. This load set is a key input to the network analysis
processes, whereby the DFES projections are assessed to identify any network constraints. The output
of the network analysis is used to determine where NG-led flexibility services could be used to help the
operation of the distribution network, or where conventional reinforcement may be required. The output
from the network analysis is published every two years in the Network Development Plan (NDP).
It is worth noting that any customer behaviour assumptions must be made with reference to the purpose
and level of network analysis that is being undertaken. The profiles that are presented in this document
have been created with the purpose of assessing the network’s capability and compliance of the NG 33
kV networks. These can also be applied to relevant studies on the 66 kV and 132 kV networks if
appropriate levels of diversity are applied to the projected volumes.
This document utilises a variety of data sources, including metering data from connected customers as
well as outputs from innovation projects across the industry. The profiles presented in this document
cover some emerging low carbon technologies. The impact of how large scale electrification of heat
and transport will affect the operation of distribution networks is not fully understood, these profiles will
be reviewed regularly and improved where new data sources are available.

Introduction
National Grid (NG) deliver the Distribution Future Energy Scenarios (DFES) project on an annual basis
for all four licence areas covered by the NG distribution network. This provides granular scenario
projections for the growth (or reduction) of generation, demand and storage technologies which are
expected to connect to the GB electricity distribution network. NG use a scenario framework that is
consistent with National Grid ESO (NGESO), along with all other Distribution Network Operators
(DNOs) in Great Britain.
The DFES process can be split into three parts, each of which is summarised below.
Part 1: Volumes
The first part of the DFES process aims to provide granular scenario projections for the number of
demand customers and MW of installed generation that are expected to connect to the GB electricity
distribution network. The projections are informed by stakeholder engagement to understand the
specific needs and plans of local stakeholders in each licence area. NG undertake this process in
conjunction with Regen. The DFES volumes are provided at an Electricity Supply Area (ESA) level,
which represents the geographic area supplied by a Primary substation (which contains NG-owned
distribution substations) providing supplies at a voltage below 33 kV, or a customer directly supplied at
132, 66 or 33 kV or by a dedicated Primary substation’. This allows the volumes for each technology to
be spatially allocated to where it would be most likely to connect to the distribution network.
The output of this process is a large dataset of granular projections for different technologies, years,
scenarios and areas of the NG distribution network, along with a suite of reports. The data is available
on the interactive DFES map on the NG website. It provides a key data resource and evidence base to
enable NG to appraise different investment options and develop the business case necessary to
support future investment, including regulated business plans.
In addition, an extra scenario is developed as a hybrid of the four DFES scenarios. This is called the
NG Best View, and covers the most likely growth pathway that NG expect to materialise in the next 10
years. This view is curated through extensive stakeholder engagement, to allocate one of the four DFES
scenarios that is most applicable to the medium term ambition and delivery for different local
stakeholders.
Part 2: Customer Behaviour

For the DFES volume projections to be used by NG in strategic network analysis, customer behaviour
assumptions are allocated to the projected volumes. This accounts for the expected demand and
generation profiles of new and existing customers connected to the distribution network. The output of
this process is a dataset of load profiles suitable for strategic analysis of the distribution network.
Part 3: MW growth data and application of growth rates

Once the expected customer behaviour assumptions are applied to the DFES volume projections, a
demand set of the expected loads on the NG distribution network can be generated. This data is then
mapped to a network model in power system analysis software to undertake detailed network analysis.
This data will be published on the DFES map to allow the customer to investigate future network
loadings.
The MW growth dataset for the NG Best View scenario is used to generate percentage growth rates,
which NG use in a range of regulatory submissions, such as the Long Term Development Statement
(LTDS), Regulatory Reporting Packs and annual data exchange with National Grid Electricity
Transmission.

Customer Behaviour Modelling
Context
The primary purpose of the DFES projects for NG is to inform strategic network analysis to understand
how the projected change in customer numbers will affect the operation of the distribution network as
the UK transitions to a Net Zero future. Table 1 shows a matrix which describes the different
components required in order to undertake detailed electrical analysis of any electricity distribution
network.
Table 1: Summary table of the aspects required for detailed electrical analysis of the distribution network
Network Customers
Assets Network topology and connectivity Customers connected to the distribution

information, including impedance and ‘nuts network, including the type of demand
and bolts’ data about the assets connected or generation connected. This also
to the NG distribution network. Normally this includes information on the machines or
is captured in a network model in power assets that customers have connected
system analysis software. to the network (such as Electric
Vehicles or Heat Pumps).
Behaviour Actions taken by the DNO to actively Expected behaviour of customers
manage the network. This can be in the form connected to the distribution network,
of updated running arrangements once an with reference to the focus and purpose
arranged outage is taken, or load of the network analysis to be
management schemes in place to manage undertaken.
network flows. This information is vital if
contingency analysis is required.
The DFES volumes project provides projections for the number of customer assets that are expected
to connect to the NG distribution network in the next 30 years. It is important to note that the units used
in the DFES volumes project are all quantifiable (i.e. they can be counted).
This document outlines the customer behaviour assumptions that NG use for the purpose of strategic
network analysis. These customer behaviour assumptions must be relevant to the purpose of the
network analysis that is undertaken. Different factors that may impact the customer behaviour
assumptions that are required could include:
 The voltage level used as the focus of the study – different customer behaviour assumptions
are used depending on the voltage level that is the focus of the network analysis.
 The aim of the analysis – different customer behaviour assumptions are applicable depending
on what the study aims to deliver. For example, network capability and compliance edge case
assessments will require a different set of customer behaviour assumptions to a study which
aims to calculate average asset utilisation over a year.
 The level of risk – this theme is discussed throughout the document, as there are external
factors to the customer behaviour assumptions not directly in the control of a DNO. The balance
between studying credible edge-case network conditions to achieve network compliance and
designing a network that is operated efficiently will be different for each network company.
Intended use of profile assumptions

The profiles that are presented in this document have been created with the purpose of assessing the
network’s capability and compliance of the NG 33 kV networks. These can also be applied to relevant
studies on the 66 kV and 132 kV networks if appropriate levels of diversity are applied to the projected
volumes.

Case study 1: Impact of diversity on Customer Behaviour
Background
When assessing the network impact, careful consideration needs to be given to ensure the level of
diversity applied to customer behaviour is appropriate for the study being undertaken. In the context of
customer behaviour profiling, diversity is the assessment of the coincident behaviour of a group of
customers or technology types. Not accounting for diversity would result in overly onerous expected
loading in network analysis. This could lead to constraints triggered by network analysis that are not
observed in network operation, and unnecessary reinforcement schemes started as a result.
Conversely, applying too much diversity could lead to credible network constraints being missed.
There are two methods of assessing demand diversity on the Extra High Voltage (EHV) distribution
network:
1. Aggregated demand as substation level – A load survey is undertaken to determine the
substation demand at the cardinal network edge-cases. This is with all generation unmasked
to give the true underlying demand.
2. Profiling technologies – Each technology category that is connected, or projected to connect
is assigned an explicit profile that is appropriate for the study being undertaken.
An aggregated Primary substation demand profile has traditionally been used for EHV network design,
due to the challenge of building up a representative Primary level profile from the constituent volumes
and other factors that can influence network loading. This inherently captures the level of diversity of all
demand connected downstream and is suitable for assessing existing network demand and for short-
term network assessments. The aggregated approach does not provide any information on what
customers are contributing to the edge-case peaks. This makes modelling changing demand behaviour
challenging.
Diversified Profiles
The ACE49 methodology1 used for LV design is an example of how the average demand assumed per
customer reduces as the size of the group increases. The kW per customer of domestic unrestricted
profile class 1 reduces as the group size increases. This tends towards a constant value as the number
of customers increases; this is sometimes referred to as a fully diversified profile.
When profiling specific technologies at EHV, consideration of an appropriate level of diversity is
important. At EHV, the group size of a given technology (e.g. heat pumps) may be sufficiently large that
the profile can be considered full diversified. It is worth noting that the ACE49 methodology aims to
provide a suitable level of risk such that LV networks are designed adequately, but not over-generously
for the demands on each part of the network. The profiles used in this document for application to the
NG EHV networks do also account for a level of risk; however this risk is not quantified in this document.
Coincident Peaks of different technologies
The diversity of a single technology is only one part of determining the worst credible network
conditions. All demand needs to be considered when analysing network edge-cases, as not all demand
is coincident.
Figure 1 shows a few of the DFES technologies winter peak profiles described in this report with
different profile shapes. It highlights that technologies can have notably different profile shapes and
peaks. Taking the peak of each of these profiles without accounting for the non-coincident nature would
not accurately represent the demand on the network. To account for this, the DFES profiles are full half-
hourly profiles for all representative days.
1ACE Report no.49, Report on Statistical Method for Calculating Demands and Voltage Regulations
on LV Radial Distribution Systems, Energy Networks Association, 1981;

Figure 1: Example of DFES technology winter peak profiles differences
The magnitude and time of peak of a substation will depend on the constituent parts that make up the
overall aggregated substation demand. The aggregated peak may change over time as the volumes
and individual technology profile changes.
Diversified Primary Demand
This example highlights the impact of Primary level diversity using real-world network loadings from
Portishead Bulk Supply Point (BSP) in South West NG distribution licence area. This aggregated
substation demand inherently captures the diversity observed between everything connected
downstream of the Primary substation, but does not capture the diversity between Primaries.
Substations dominated by unrestricted domestic customers will typically peak in winter around 18:00;
substations dominated by non-domestic customers often peak in the middle of the day. Not all
substations peak in winter, an example of this is a substation that feeds demand in a holiday destination,
where the peak is typically over the holiday periods.

Figure 2: Portishead 33 kV Network
Table 2 summarises the Primary peak and Primary at BSP peak demand for all substations connected
downstream of Portishead BSP. The time of the Primary peak varies notably between 09:00 and 19:30,
with the majority peaking between 17:30 and 19:00. All Primaries except British Gas Hallen 11 kV peak
in the winter season, noting that all have a different date of peak. The sum of the Primary Peaks is
51.43 MVA.
Table 2 gives a breakdown of the Primary demand at the time of BSP peak; the total is 45.12 MVA.
This is 12.3% lower than the sum of the Primary peaks. At the time of the upstream Seabank Grid
Supply Point (GSP) peak, Portishead BSP total demand is 42.64 MVA, which is 17.1% lower than the
sum of Primary peaks.
Table 2: Portishead substation peaks vs BSP peak
Substation Primary Time of Peak Date of Peak Primary @

Peak (MVA) BSP Peak
(MVA) *
Portishead Ashlands 11kV 6.17 19:30:00 09/01/2020 5.92

British Gas Hallen 11kV 0.71 15:00:00 07/08/2019 0.37
Clevedon 11kV 15.39 18:30:00 29/01/2020 12.74
Easton In Gordano 11kV 3.82 17:30:00 05/01/2020 3.47
Gas Lane 11kV 9.63 18:30:00 17/12/2019 9.22
Royal Portbury Dock 11kV 11.16 09:00:00 31/01/2020 8.40
Weston in Gordano 11kV 4.55 18:00:00 12/01/2020 5.01
Total 51.43 45.12

*BSP peak was at 19:00 on the 09/01/2020

This highlights the importance of using the appropriate diversified Primary demand set when analysing
the EHV network. Using a Primary peak demand set to assess GSP loadings would give loadings that
are 10-20% higher than are actually observed by network monitoring. Conversely, using a Primary at
GSP peak demand set to assess 33 kV constraints means genuine network issues could be missed.
Representative Days
Traditionally, distribution networks are assessed using edge-case modelling, where only a snapshot of
the network condition that is deemed most onerous is analysed. As the installed capacity and behaviour
of demand, generation and storage is rapidly changing, it has become difficult to predict what network
condition will be most onerous.
To cover a range of likely onerous cases, NG consider a range of different potential representative
days, which are used to assess network capability for the analysis purpose identified in the
aforementioned Intended use of profile assumptions section:
 Winter Peak Demand, with minimum coincident generation – an assessment of the network’s
capability to meet peak demand conditions and determine group demand as per Engineering
Recommendation P2/72;
 Summer, Intermediate Warm Intermediate Cool Peak Demand, with minimum coincident
generation – an assessment of the network’s capability to meet access window demand
conditions;
 Summer Peak Generation, with minimum coincident demand – an assessment of the
network’s capability to handle generation output.
The definition of seasons is taken from Engineering Recommendation P27/2 (Current rating guide for
high voltage overhead lines operating in the GB distribution system) 3:
 Winter: January, February and December
 Intermediate Cool: March, April and November
 Intermediate Warm: May, September and October
 Summer: June, July and August
The DFES forecasts project high customer uptake of low carbon technologies, such as Electric Vehicles
(EVs) and Heat Pumps (HPs). Low carbon technologies are expected to allow customers to manage
their demand at an individual level, such as shifting energy usage away from traditional times of peak
network loading. As a result, the five representative days studied include a half-hourly power profile.
This aims to help us to understand how the shape of the demand profile will change over time, which
could inadvertently introduce local network peaks at times which are away from the existing time of
network peak.
2 Engineering Recommendation P2 Issue 7, Security of Supply, Energy Networks Association, 2019;

3Engineering Recommendation P27 Issue 2, Current rating guide for high voltage overhead lines
operating in the UK distribution system, Energy Networks Association, 2020;

Case Study 2: Importance of studying multiple seasons
This case study highlights the potential impact of only considering a representative day for one season
only, traditionally being the Winter Peak Demand, when assessing networks. Winter Peak Demand can
often represent the maximum demand but it is not always the most onerous network condition. The
demand profile for many areas of the network show that although the peak demand may appear in the
cooler months, the reduction of the network’s asset ratings in the subsequent warmer seasons can be
greater than the corresponding reduction in demand. This can shift the network’s most onerous
condition away from the traditional winter Peak and onto intermediate cool, intermediate warm, or
summer seasons.
Consider the demand profile below which has been observed on Halesfield 33/11 kV Primary substation
in Telford in Figure 3.
Figure 3: Output graph to show total 'underlying' demand over a yearly period at a Primary substation
Prior to reinforcement, the site was fed via two 33 kV circuits each rated 31 MVA, 29 MVA, 29 MVA,
and 25 MVA for winter, intermediate cool, intermediate warm, and summer respectively. The
substation’s maximum demand across all four seasons was 27.4 MVA, falling within the intermediate
cool season in April. This makes the site Class C under Engineering Recommendation P2/7 4, requiring
Group Demand to be restored within 3 hours following a First Circuit Outage (FCO). Under a fault
condition on one of the incoming circuits, the following demand would be observed on the second
remaining circuit:
Table 3: Substation seasonal peak demand and rating comparison
Demand Edge-case Month Demand (MVA) Asset Rating (MVA) Utilisation
Winter Peak February 25.4 31 81.9%

Intermediate Cool Peak April 27.4 29 94.5%
Intermediate Warm Peak October 26.2 29 90.3%

Summer Peak July 26.3 25 105.2%

Table 3 highlights that although the site’s maximum demand appeared in the intermediate cool season
(outside of the traditionally assessed winter season), the most onerous representative day was in fact
the Summer Peak Demand when taking into account the circuit’s seasonal ratings.
These scenarios, typically seen in industrial and commercially dominated areas, are becoming more
common as customers move towards increasingly efficient and smarter technologies, and to utilise
more economic variable tariffs.
The case study shows a need to study networks across multiple seasons to identify the most onerous
conditions. It further underlines the requirement to carry out these studies periodically as it can be seen
from the case above that the asset utilisation per season can be relatively close to each other so the
most onerous season can vary from one year to another.
Changing Nature of Demand

On 31st December 2020, changes to the Electricity Distribution Licence were introduced as part of the
EU Electricity Directive 2019/944, part of the Clean Energy of all Europeans Package 5. As part of these
Licence Condition changes, DNOs are required to develop a Network Development Plan (NDP) to
outline the expected development of the distribution system over a 5 to 10 year period. As part of this
requirement, DNOs are required to justify any assumptions on the expected uptake of demand-side
response (DSR) and energy efficiency, as a means to alter existing demand observed on the network.
In order to understand the changing nature of demand, these assumptions can be split into two separate
categories of flexibility and energy efficiency services:
 Passive/Customer led: these are actions that customers actively take to manage their
electricity demand. This can be in response to behind the meter assets installed, or as part of
measures to increase energy efficiency. It is important to note that the DNO has no active part
in this process and currently would have no mechanism to determine if an individual customer
has changed their consumption behaviour.
 Active/Network led: this accounts for flexibility services that are procured and dispatched by
a network operator to alleviate a particular network constraint or to defer network reinforcement.
These are currently delivered by NG through the Flexible Power brand.
4Energy Networks Association, Engineering Recommendation P2 Issue 7, p9

5Ofgem, Open letter on changes to licence conditions as a result of the transposition of the Clean
Energy Package, 8th December 2020;
https://www.ofgem.gov.uk/system/files/docs/2020/12/electricity_directive_open_letter_0.pdf

The DFES customer behaviour assumptions aim to use credible, evidence based assumptions for how
passive/customer led flexibility and energy efficiency services can alter the existing levels of demand
observed on the network. The network led flexibility services should be employed following detailed
analysis to identify areas where flexibility services are required on the network. NG published an article
as part of the Distribution System Operability Framework (DSOF) in 2018 which investigated how
changing load profiles could impact network design6.
Whilst DNOs must account for the credible expected use of customer led energy efficiency and flexibility
measures, these must also be balanced with the ongoing requirement to operate and maintain an
economic, efficient and coordinated network. The risk associated with assumptions on the expected
level of customer led energy efficiency and flexibility measures are that if the assumptions do not
materialise DNOs could be investing in a network that is inoperable at times of peak network loading.
Consider the condition where the peak demand observed on a local network is not coincident with the
time and date of the GB electricity system peak. For this condition, the expected availability of customer
led flexibility services at GB electricity system peak demand (which if not directly contracting with a
DNO are largely driven by the electricity wholesale price signals) may not be available for the local
network peak. As a result, it is prudent for DNOs to use more conservative assumptions on the expected
use of customer led flexibility services, if the local network constraint is not coincident with periods of
high network loading on the GB electricity system.
Application of customer behaviour profiles

To capture the complex nature of the customer behaviour modelling in the DFES studies, two different
types of profiles are used:
 Unabated profiles: all technologies use an ‘unabated’ profile, which captures that credible
edge-case demand profile for each of the representative days considered.
 Flexed profiles: Some technologies may also utilise a ‘flexed’ profile, which is one where a
customer would respond to an external driver to flex their demand, not as part of a DNO
contracted and procured service.
A flexed profile in itself is not necessarily representative of the expected behaviour of a group of
customers. Instead, a split of the proportion of customers expected to be consuming energy using an
unabated and flexed profile is applied. The combination of the unabated and flexed profiles together
represent the average expected impact per customer on the distribution network. This profile split factor
is year and scenario dependent, to account for the differing levels of consumer engagement in the
DFES scenario framework.
In addition to the option of altering expected customer behaviour to account for customer led flexibility
services, a profile scaling factor is also applied to the profiles. This scaling factor accounts for any
expected energy efficiency increases made by customers connected to the NG distribution network.
The yearly scaling factor is used to linearly scale each demand profile from the baseline year. It is year
and scenario dependent, to account for different expected energy efficiency measures in the DFES
scenario framework.
To calculate the expected aggregate customer behaviour profile, the profile components are combined
with the profile scaling and expected split between unabated usage and flexed usage.
6Distribution System Operability Framework, Changing Load Profiles, National Grid, 2018.
https://www.nationalgrid.co.uk/downloads/18622

Worked Example
A visual representation of how the customer behaviour profiles are assigned to the DFES volumes
projections is included in Figure 4 for the following criteria.
 Technology: Domestic
 Units: Number of new domestic dwellings
 Licence area: East Midlands
 Scenario: Consumer Transformation
 Year: 2050
Figure 4: Representation of how customer behavioural changes are applied to the profiles for DFES
analysis
Profile Split
Proportion of customer demand at each

Electricity Supply Area apportioned to both
unabated and flexed profiles
Scenario and year dependent
Unabated Profile Flexed Profile
Based on existing customer data representing Representative profile to simulate customer

the credible edge case profile for current response to market-led price signals for
network analysis. Demand Side Response
No scenario or year dependency No scenario or year dependency
Unabated Profile Scaling Flexed Profile Scaling
Linear scaling of unabated profile to account Linear scaling of flexed profile to account for
for energy efficiency measures energy efficiency measures
Scenario, year and licence area dependent Scenario, year and licence area dependent
Final Profile
When all profile elements are combined

together, the equivalent final profile per unit at
each Electricity Supply Area can be generated.

Structure of profile sections
In the following sections of this report, each of the technology types studied as part of the DFES study
is analysed with the customer behaviour profiles used by NG. Each technology type section follows the
structure outlined below:
 Methodology – a detailed description of the methodology used to obtain representative profiles
with any data sources used.
 Representative Day Profiles – a graphical representation of the profiles used.
 How these profiles will change over time – how NG expect these profiles will change over
time, due to customer-led actions that the DNO is not able to influence directly.
 Energy modelling – outline of the assumption NG use to benchmark the expected energy
consumption of demand technologies across the NG distribution network against other scenario
based forecasting, such as those used in the Committee for Climate Change 6 th Carbon
Budget7.
 Known limitations – a description of the areas where the profiles used could be improved to
better align with expected customer behaviour
 Future developments – how NG plan to improve the profiles to address some of the known
limitations.
7Committee for Climate Change, The Sixth Carbon Budget: The UK’s path to Net Zero, 2020;
https://www.theccc.org.uk/wp-content/uploads/2020/12/The-Sixth-Carbon-Budget-The-UKs-path-to-
Net-Zero.pdf

Customer Behaviour
assumptions
Generation and storage
technologies

Solar Generation
Table 4: Table of solar generation technology types used in the DFES 2022 analysis
Technology Subtechnology Units used in DFES volume projections
Solar Commercial rooftop (10 kW – 1 MW of installed capacity

Generation MW)
Domestic rooftop (<10 kW)
Ground mounted (>1 MW)
Methodology
Each solar generation customer is geographically allocated to an Electricity Supply Area where it would
be most likely to connect to the distribution network. Solar generation volumes are provided as the
installed capacity (MW) of generation connected.
Real power output data from all solar generation sites across the NG distribution licence areas was
collected and aggregated by each half hour for the three years prior to this analysis. Table 5 shows the
solar generation data sample:
Table 5: Sample size of solar generation site used to create profiles
Licence Area Number of sites in sample Total installed capacity of sites in sample
East Midlands 127 868 MVA

South Wales 57 322 MVA
South West 176 823MVA
West 60 344 MVA

Midlands
Half hourly generation profiles were created for each of the five representative days used for network
analysis. This methodology was changed in DFES 2022 to normalise the generation output observed
at each solar site by installed capacity to give a per unit value, before finding the maximum and minimum
half hourly profile for each season by aggregating the sites to licence area. This has the benefit of
ensuring equal weighting of each site’s normalised generation output and prevents results being
skewed by larger sites.
To account for varying levels of diversity of solar generation output across the network, the analysis
was completed with different subcategory groups that correlate to the subtechnologies used in the
analysis. This allows for the capturing of the unique behaviour of each of these categories, and also
captures the diversity associated with different scales of sites being connected at different voltages.
Further development of generation analysis tools allowed for better understanding of behaviour at
different installed capacities level. For sites under 1MW, self-consumption behaviour was observed as
shown in Figure 5. This lower perceived metered generation is due to site self-consumption, which not
only means that less generation is exported to the network, but that the site has lower demand. Due to
this, only sites above 1MW were included in the analysis to create the Solar Generation profile as they
have lower or almost no demand.

Figure 5: Comparison of behind-the-meter consumption of commercial rooftop solar generation against
the total generation for the installed solar arrays.
As each licence area was analysed separately, there are four sets of solar generation profiles used for
network analysis.
Representative Day Profiles

South West
Figure 6: Representative solar generation profiles for customers in the South West licence area

South Wales
Figure 7: Representative solar generation profiles for customers in the South Wales licence area
East Midlands
Figure 8: Representative solar generation profiles for customers in the East Midlands licence area

West Midlands
Figure 9: Representative solar generation profiles for customers in the West Midlands licence area
How will these profiles change over time

It is assumed that the load profiles for solar generation sites will not change in the future.
These profiles are normalised around the installed capacity, rather than the contracted export capacity.
For instances where a customer installs much more generating plant than the contracted capacity, an
export limitation scheme is implemented in the network analysis stage to limit the export of any over-
installed generation. This is also the case for any generation sites with an installed Active Network
Management (ANM) scheme, where the logic for the load management scheme is incorporated into the
network analysis.
Known Limitations
As some of the solar generation sites across the NG distribution network are due to reach the end of
the operational life in the next 30 years, there may be opportunities for customers to replant with more
efficient equipment, as well as utilising behind the meter storage and customer-led load management
schemes. These potential changes have not been accounted for in this analysis. Behind-the-meter
consumption of the generation was not part of this analysis, but a great consideration was given which
resulted in the removal of all sites under 1MW from the analysis.
Future Developments
Further Analysis will be undertaken to better understand self-consumption, which will enable the
production of Photovoltaic sub technology specific profiles. More spatially granular profiles could be
something to investigate in the future.

Onshore Wind Generation
Table 6: Table of onshore wind generation technology types used in the DFES 2022 analysis
Wind Onshore Wind <1MW MW of installed capacity

Onshore Wind >=1MW
Methodology
Each onshore wind generation customer is geographically allocated to an Electricity Supply Area where
it would be most likely to connect to the distribution network. Onshore wind generation volumes are
provided as the installed capacity (MW) of generation connected.
Real power output data from all onshore wind generation sites across the NG distribution licence areas
was collected and aggregated by each half hour for the three years prior to this analysis. Only onshore
wind sites with an installed capacity greater than or equal to 100 kW were considered. Table 7 shows
the onshore wind generator data sample:
Table 7: Sample size of onshore wind generation site used to create profiles
Subtechnology Licence Area Number of sites in Total installed

sample capacity of sites in
sample
Onshore Wind East Midlands 80 25 MVA

<1MW
South West 150 34 MVA
West Midlands 28 73 MVA
Onshore Wind East Midlands 23 199 MVA

>=1MW
South West 25 208 MVA
West Midlands 4 39 MVA
analysis. This methodology was changed in DFES 2022 to normalise the generation output observed
at each wind site by installed capacity to give a per unit value, before finding the maximum and minimum
half hourly profile for each season by aggregating the sites to licence area.
To account for varying levels of diversity of onshore wind generation output across the network, the
analysis was completed with different subcategory groups of licence area totals, 132 kV connected, 33
kV connected and Primary substation connected sites. The licence area aggregated profiles represent
the expected per unit output of onshore wind generation when viewed as part of the total generation
installed across a licence area. At lower levels of diversity the profiles represent the worst case
observed profile of all sites considered in the sample group.
The grouped profiles represent the maximum output of each group combined, so lead to a more
pessimistic profile showing greater generation output due to local diversity being less than that of the
whole licence area. Given the purpose of the analysis for the studies was focussed on the EHV
networks, a worst case Primary substation diversified profile was chosen for this analysis.
As each licence area was analysed separately, there are four sets of onshore wind generation profiles
used for network analysis.

South West
Figure 10: Representative onshore wind profiles for customers in the South West licence area
South Wales
Figure 11: Representative onshore wind profiles for customers in the South Wales licence area

East Midlands
Figure 12: Representative onshore wind profiles for customers in the East Midlands licence area
West Midlands
Figure 13: Representative onshore wind profiles for customers in the West Midlands licence area

It is assumed that the load profiles for onshore wind generation sites will not change in the future.

Management scheme, where the logic for the load management scheme is implemented in the network
analysis.
Known Limitations
The DFES volume projections assume that existing onshore wind sites will replant with a larger installed
capacity when these sites reach the end of their design life. In addition, there may be future
opportunities for onshore wind sites to co-locate with demand sources, such as hydrogen electrolysis
or energy storage. Any potential changes in onshore wind turbine efficiency and change of import
requirements have not been accounted for in these profiles.
Future Developments
Further analysis will be undertaken to develop more spatially granular profiles to account for regional
variations in onshore wind generation output within a licence area.

Offshore wind Generation
Table 8: Table of offshore wind generation technology types used in the DFES 2022 analysis
Wind Offshore Wind MW of installed capacity
Methodology
Each offshore wind generation customer is geographically allocated to an Electricity Supply Area where
it would be most likely to connect to the distribution network. Offshore wind generation volumes are
Real power output data from all offshore wind generation sites across the NG distribution licence areas
was collected and aggregated by each half hour for the three years prior to this analysis. Only two
offshore wind generation customers are connected to the NG distribution network, both situated off the
coast of the East Midlands licence area.
analysis. This was achieved by considering the maximum and minimum generation output observed
during each half hour across the whole of each season in the aggregated generation data and
normalising this by installed capacity of the sample to give a per unit value.
Both offshore wind sites currently connected to the NG distribution network are connected at 132 kV,
so no study of different diversity levels is applicable. However, as any future offshore wind sites
connected to the distribution network would be most likely to connect at the 132 kV voltage level, the
generated profiles are deemed at a suitable level to be used for network analysis. This profile is applied
to all NG distribution licence areas where offshore wind is projected to connect.

Figure 14: Representative offshore wind profiles

It is assumed that the load profiles for offshore wind generation sites will not change in the future.
analysis.
Known Limitations
The sample size used to generate offshore wind profiles is very small and may not be representative of
the behaviour of a larger number of customers connected across the NG distribution network.
Future Developments
Given the small sample size of existing customers and the relatively low projections of future offshore
wind connections, no future developments have been identified for offshore wind customers.

Non-weather Dependent Generation
Table 9: Table of non-weather dependent generation technology types used in the DFES 2022 analysis
Technology Subtechnology Units used in DFES volume

projections
Biomass & Energy Crops (including - MW of installed capacity

CHP)
CCGTs (non CHP) -
Geothermal -
Hydro -
Hydrogen-fuelled generation -
Marine Tidal stream
Wave energy
Micro CHP Domestic

(G98/G83)
Non-renewable CHP <1MW
>=1MW
Non-renewable Engines (non CHP) Diesel
Gas
OCGTs (non CHP) -
Other generation -
Renewable Engines (Landfill Gas, -
Sewage Gas, Biogas)
Waste Incineration (including CHP) -
Retained Connection -
Methodology
All non-weather dependent generation customers are geographically allocated to an Electricity Supply
Area where they would be most likely to connect to the distribution network. Generation volumes are
In the case of infrequently despatched, non-intermittent generation, measured flows may not reflect the
potential network impact. Instead, a flat (continuous output) profile was assumed for each
representative day, representing the realistic behaviour that would have the worst impact upon the
network. These were assumed as follows:
 Summer Peak Generation day: continuous export at agreed supply capacity; and
 Peak Demand days (all seasons): zero export.
Generation output may in reality be limited by load management schemes, such as Export Limited
connections or Active Network Management schemes. In addition to this, some generation customers
may hold flexibility contracts with NG that mandate a certain profile at certain times of day or year. The
behaviour of these customers is included, as part of the network analysis, but the output of non-weather
dependent generation must be assumed full before network management systems can take effect in
order to accurately assess network capability.

In DFES 2022, an additional category of generation has been used to represent retained connection.
Across the four DFES scenarios, some technologies will see a level of decommissioning between the
baseline year and 2050. This largely consists of technologies that are incompatible with net zero carbon
emissions, such as unabated fossil fuel power generation.
Upon ceasing of conventional operation, the connection agreement held by the operator and the
associated contracted export capacity secured with NG is not automatically relinquished and some sites
will likely retain this connection capacity. The motivation behind retaining this capacity will be to connect
an alternative generation or storage technology that is more compatible with net zero emission targets.
Figure 15: Non-weather dependent generation representative day profiles

It is assumed that the load profiles for non-weather dependent generation sites will not change in the
future.
analysis.

Known Limitations
These profiles are considered to be pessimistic as rarely all generation installed across an area of the
network is exporting at the full installed capacity at any single time. When assessing each connected
customer in isolation this approach is justifiable, but when assessing a group of generators connected
at a Primary substation the coincident behaviour of all generators need to be used to assess a credible
edge-case profile.
Future Developments
NG is continuing to develop the tools to assess the coincident behaviour of generators connected at a
Primary substation level to determine a suitable profile per generator that balances the safe design and
operation of the distribution network and the design of an economic and efficient network.

Battery storage
Table 10: Table of battery storage technology types used in the DFES 2022 analysis
Storage Co-location MW of installed capacity

Domestic Batteries (G98)
Grid services
High Energy User

Other
Methodology
All battery storage customers are geographically allocated to an Electricity Supply Area where they
would be most likely to connect to the distribution network. Battery storage volumes are provided as
the installed capacity (MW) of storage connected.
NG previously worked with Regen to develop an approach to model the growth and operation of
storage. As part of this modelling work, a consultation paper was developed and issued, aiming to
validate some of the key assumptions used to model energy storage. The results from the consultation
paper are published on the NG website.
The consultation paper proposed different energy storage business models and asked for feedback on
the behaviour of energy storage in each of these business models. One noteworthy response to the
consultation was that customers expressed a desire to be able to ‘stack’ different business models and
revenue streams. Respondents also identified a preference not to commit to a specific operating mode,
as the evolving nature of procurement of balancing services by the Great Britain System Operator
(GBSO) in the future may change some of the proposed operating modes.
The consultation responses demonstrated that energy storage customers prefer flexibility to operate
energy storage without a specific operating profile. As a result, the profile assumptions used in this
study are:
 Summer Peak Generation day: continuous export at agreed supply capacity; and
 Peak Demand days (all seasons): continuous demand at agreed import capacity; and zero
export.
This unconstrained mode of operation is onerous for networks. In some cases, it may trigger major
reinforcements that would prove unnecessary with relatively minor changes in the behaviour of energy
storage connections. Where battery storage customers hold flexibility contracts with NG that mandate
a certain profile at certain times of day or year, the behaviour of these customers is included, as part of
the network analysis. However; in the absence of load management schemes to limit battery storage
usage, the output of battery storage must be assumed to be worst-case in order to assess network
capability.

Figure 16: Battery storage representative day profiles
In DFES 2022 we have included flexible behaviour of domestic batteries, using the same customer
behaviour assumptions regarding uptake and response to domestic flexibility as those observed in the
domestic demand.
Figure 17: Domestic battery storage flexed profile

It is assumed that the load profiles for battery storage sites will not change in the future. These profiles
are normalised around the installed capacity, rather than the contracted export capacity.
Known limitations
The projections and profiles only consider the impacts of battery storage. It is noted that there are other
forms of energy storage, which may connect to the distribution network in the next 30 years. This
feedback was given as part of the DFES 2022 stakeholder engagement exercise and will be
incorporated into future DFES studies.
These profiles are considered to be pessimistic as rarely all storage capacity installed across an area
of the network will be operating in a way to increase network loadings for each of the representative
days studied. When assessing each connected customer in isolation this approach is suitable, but when
assessing a group of generators connected at a Primary substation the behaviour of storage coincident
to local network loadings need to be used to assess a credible edge-case profile.
Future Developments
The energy storage profiles will be reviewed in future studies, with the expansion of the suite of
representative days to further assess the energy curtailment impact of measures such as ANM and
DSR. NG and National Grid ESO have collaborated on the Regional Development Plan 4 (RDP) 8 to
investigate how energy storage or other customers can provide flexibility to the system. This RDP looks
to extend the flexibility arrangements given to generation so that they apply to storage demand. This
will allow storage projects to become part of the solution to network capacity issues, rather than capacity
planning standards being a potential barrier to them.
8 Regional Development Programmes, West Midlands RDP, National Grid, 2020;

https://www.nationalgrid.co.uk/downloads-view/106888

Customer Behaviour
assumptions
Demand Technologies

Underlying Demand
For the purposes of this document, underlying demand refers to the aggregate behaviour to the existing
NG customer base currently connected to the distribution network, and how electricity demands from
existing customers will change over time.
Methodology
The underlying demand profiles are not included as part of the DFES Part 1: Volumes project, as this
is data that NG has access to as part of internal network design processes. As the purpose of the
analysis is to study the network impact of the DFES projections on the 33 kV, 66 kV and 132 kV
networks, the demand at each Primary substation needed to be modelled individually.
NG undertake an engineering load survey on an annual cycle to update the Primary demand sets for
network design purposes. This focuses on an annual peak demand figure and accounts for abnormal
network running arrangements and the export of any downstream connected generation customers.
However to be used in the half-hourly analysis, further analysis was required to determine
representative half hourly profiles for different Primary substations.
Due to the absence of directional MW/MVAr monitoring at all Primary substations, it was not possible
to use data directly for each Primary across the NG distribution network. Instead, a sample of 200
Primary substations with directional MW/MVAr monitoring was used to determine a set of representative
profiles that could be retrospectively applied to other Primary substations with similar metrics.
The Primary underlying demand profiles are created as a profile normalised around the peak demand
observed as part of the engineering load survey. For each half hour and representative day, the peak
demand multiplied by the profile value gives an expected MW demand at each Primary substation.
Clustering Methodology
A bespoke machine learning Python-based program was written to cluster the Primary substations in
the sample into groups with similar profile characteristics for the representative days used for analysis.
This used metering data for a yearly period for all sample sites. More information about the cluster
methodology is contained in Appendix B: Primary substation clustering of this report.
The output of the clustering was a list of Primary substations with ‘similar’ behaviour (in terms of the
time of peak and the profile shape) for all five representative days.
Grouping Methodology
The most important part for categorising primaries is using its time of peak. As part of the engineering
load survey two distinct groups of Primary substations were found, one with a morning peak and another
with an evening peak. The Primary substations with a morning/midday peak were identified and these
substations grouped to create a representative morning/midday peaking profile. The clustering
methodology outlined above was used for the evening peaking Primaries in the sample to identify two
different demand profiles for evening peaking Primary substations.
After profiles were clustered there was a need to find available data that could correlate with each group
to define the profile. This involved collating a list of publicly available data for the geographic areas
supplied by Primary substations, including the customer density and average energy consumption for
domestic and non-domestic customers.

Creating the profiles
Combining both grouping methods Primary substations are grouped into three demand types:
 Morning peaking (usually industrial type profiles), with a period of sustained high levels of
demand throughout a daily period for each representative day,
 Evening peaking with a relatively high output in early morning and more notable higher peak in
the intermediate warm and summer seasons. This sample of sites correlated with an expected
rural demand type, which was supported by an average domestic customer density of fewer
than 150 domestic customers per km².
 Evening peaking with no noticeable drop in demand between the hours of 09:00 and 17:00,
with a lower peak in the intermediate warm and summer seasons. This sample of sites
correlated with an expected urban demand type, which was supported by an average domestic
customer density of greater than 150 domestic customers per km².
Using the methodologies described in Figure 18 the sample of 200 Primary substations was grouped
into the three demand types, with profiles being generated for each representative day. For the peak
demand representative days, the average profile of all sites within the group was taken for each season,
this was renormalized to ensure there was a 1 per unit peak to assign each Primary substation to its
peak annual demand. For the peak generation representative day the mean of the minimum profile for
the sample was used.
Application methodology
After profiles were created for the sample of 200 Primary substations, the remaining Primary
substations needed to be categorised into one of the three demand types. Figure 18 shows how the
two methods were applied to categorise the remaining Primary substations across the NG distribution
network not originally identified in the sample for clustering.
Demand type:
Time of peak before 14:00
Morning
Primary substation on the

NG distribution network
Domestic customer density Demand type:
<= 150 customers per km² Rural
Time of peak after 14:00
Domestic customer density Demand type:

> 150 customers per km² Urban
Figure 18 - Primary substation category by underlying demand

The map in Figure 19 demonstrates a geographic view of how the Primary substations were categorised
into the three demand types.
Figure 19 – Primary Categorisation Map
Single customers
There are a number of customers connected to the NG distribution network with connections at 132 kV,
66 kV, and 33 kV or by a dedicated Primary transformer at a Primary substation. Such customers do
not have a regular daily or seasonal demand profile. As a result, the assumed profile for these
customers is:
 Peak Demand days: continuous demand at peak annual demand observed in the engineering
load survey; and
 Summer Peak Generation day: zero demand.

Profiles for both active power and reactive power were generated from this analysis. For detailed
network design, a power factor per site may be used instead of the reactive profiles generated using
this process where detailed network monitoring can be used.

Morning
Figure 20: Representative morning peaking underlying demand active power profiles
Figure 21: Representative morning peaking underlying demand reactive power profiles

Rural
Figure 22: Representative rural underlying demand active power profiles
Figure 23: Representative rural underlying demand reactive power profiles

Urban
Figure 24: Representative urban underlying demand active power profiles
Figure 25: Representative urban underlying demand reactive power profiles

Single Customer
Figure 26: Representative single customer underlying demand profiles

Investigations of the following three variables as published in the RIIO-ED1 Network Performance
Summary reports9 demonstrate that aggregate customer demand usage has changed over the previous
10 years. Figure 27 plots the total customers connected to the NG distribution network over a ten-year
period, alongside the total NG distribution electrical energy consumption and sum of licence area peak
demand.
9Ofgem, RIIO-ED1 Electricity Distribution Network Performance Summary 2019-20 [Data set]
https://www.ofgem.gov.uk/system/files/docs/2021/03/riio-ed1_annual_report_2019-
20_supplementary_data_file_0.xlsm

Figure 27: Total change from 2011 baseline of customer numbers, annual energy consumption and peak
demand for NG distribution licence areas and Great Britain
These figures demonstrate that although greater numbers of customers have connected to the
distribution network, the total observed peak demand per licence area and the total energy consumption
has reduced over the same period. This aligns with increasing customer uptake of more energy efficient
devices and a more energy conscious consumer.
However, extrapolating from these datasets to inform future peak power demand reduction due to
energy efficiency does not provide a complete picture. Further analysis to estimate the additional
demand contribution from newly connected customers over a three year period was used to determine
a more accurate underlying demand reduction without accounting for any new connections. These
assumptions have been used to infer a licence area specific peak demand reduction for each year and
scenario using measureable data observed from historic NG distribution licence area peak demand
figures.

Figure 28: Expected underlying demand change for each licence area as a result of energy efficiency
assumptions.
Energy Assumptions
The energy consumption figures reported on for DFES 2022 focus on the licence area total energy
consumption, and are not provided at a more granular level. As demonstrated in Figure 27, the energy
consumption across NG’s distribution network has decreased in the previous ten years. A similar
process to the underlying demand decreases was undertaken to extrapolate these trends into the
future, also accounting for the fact that new connections made in the previous 10 years. These
assumptions have been used to infer a licence area specific energy reduction for each year and
scenario using measureable data observed from historic NG specific energy consumption figures. The
2020 data was excluded from this analysis due to the impact of Covid-19 on customer behaviour.

Figure 29: Expected annual energy consumption change for each licence area as a result of energy
efficiency assumptions.
Known Limitations
The underlying demand profiles currently all assume a profile peak which occurs in the winter season.
Analysis from the engineering load survey demonstrates that not all Primary substations across NG
peak in the winter season.
The expected underlying demand reduction due to energy efficiency analysis was completed at a
licence area level, and may not account for the more granular changing demand trends on a per Primary
substation basis.
Future Developments
NG has committed to improving the network monitoring installed across Primary substations during
ED210. Once the data for these substations is available, this can be used directly in the analysis to
provide more accurate data without the need for assigning a representative profile.
Where Primary substation monitoring is not suitable to use the data directly in network analysis, the
above process has demonstrated to provide useful clusters of sites with similar behaviour based on
observed data. This can further be improved by subcategorising the sample by the season that the
peak demand occurred, also further subcategorising the time of peak to capture early morning and
midday peaking substations as different categories.
10 National Grid, RIIO-ED2 Business Plan 2023-2028, 2021;

https://yourpowerfuture.nationalgrid.co.uk/downloads/40290

Domestic
Table 11: Table of domestic technology types used in the DFES analysis
Domestic - Number of new dwellings
Methodology
The DFES volumes project includes an analysis of all local authority development plans to identify new
domestic developments. Each development is geographically allocated to an Electricity Supply Area
where it would be most likely to connect to the distribution network. This only accounts for the electrical
household demand due to lighting, cooking and entertainment. Any additional domestic demand due to
the installation of Heat Pumps, Direct Electric Heating, Air Conditioning or Electric Vehicle chargers are
covered in other sections of this report.
Domestic developments are provided with units of the number of domestic dwellings. Each new
domestic customer is assigned an electrical demand profile for each of the five representative days
considered in the analysis.
The customer behaviour assumptions are based on the Elexon Profile Class 1 profiles 11 used in the
electricity settlement purposes, and are consistent with the NG Policy Document ST:SD5A (Design of
Low Voltage Domestic Connections) 12. This process is developed for the purposes of Low Voltage
network design, and uses a statistical methodology consistent with that published in the ACE49
methodology13. For the application of these domestic profiles for use in strategic analysis of the EHV
networks, a diversity level of 57 customers was chosen. This represents the profile to be a credible
usage profile for a single domestic customer, aggregated as part of a wider group of 57 domestic
customers. The diversity level was chosen as the average number of additional domestic dwellings per
Primary substation per year considered in the DFES volumes projections.
A limitation of the ACE49 methodology is it does not produce a half-hourly profile for all representative
days assessed. To create a profile for all demand representative days, the ratio of the urban underlying
demand profile referenced to winter peak was used as an approximation for seasonal scaling.
In addition to the increased level of diversity to make the profiles suitable for network analysis, an
Estimated Annual Consumption (EAC) was used that is consistent with the Total Domestic
Consumption Values (TDCVs) recommended by Ofgem in January 2020 14. This was deemed a suitable
figure for new-build domestic properties. The application of revised TDCV values into existing NG
system design policy is under review.
To account for customers who alter their electricity demand in response to flexibility services, a flexed
profile is also used. This assumes a domestic demand of zero for the time of day where a GB electricity
system peak is assumed to occur in the winter and intermediate cool seasons. The total energy
consumption of the flexed profile is the same as used in the unabated profile, however it allows a total
domestic energy reduction at time of system peak to be consistent with the residential white good DSR
at peak figures published in the Future Energy Scenarios (FES) data workbook15.
11 Elexon, Load Profiles and their use in Electricity Settlement, 2018;

https://www.elexon.co.uk/documents/training-guidance/bsc-guidance-notes/load-profiles/
12 National Grid, Standard Technique: SD5A/5, Design of Low Voltage Domestic Connections, 2020;
13 Energy Networks Association, ACE Report no.49, p5
14 Ofgem, Open letter: Decision on revised Typical Domestic Consumption Values for gas and
electricity and Economy 7 consumption split, 2020;
https://www.ofgem.gov.uk/system/files/docs/2020/01/tdcvs_2020_decision_letter_0.pdf
15 National Grid ESO, Future Energy Scenarios 2022 Data Workbook v3 [Data set]
https://www.nationalgrideso.com/future-energy/future-energy-scenarios/fes-2020-documents
(Accessed 01/11/2022)

Domestic (unabated)
Figure 30: Representative unabated domestic profiles
Domestic (flexed)
Figure 31: Representative flexed domestic profiles

As noted above, the flexed domestic profile is used to account for customers engaged to flex their
domestic demand at times of system peak. For the demand reduction due to energy efficiency
measures, the same profile scaling figures as applied to the underlying demand shown in Figure 28 are
used for new domestic customers. This assumption is made on the basis that a large proportion of the
customers across NG are domestic and energy efficiency measures (such as replacing appliances) will
also apply for new customers. The lower value of Total Domestic Consumption Values applied also
accounts for a more energy efficient home as is currently being connected to the NG distribution
network.
Energy Assumptions
The energy assumptions for new domestic customers projected to connect to the distribution network
are consistent with the Total Domestic Consumption Values recommended values of 2900 kWh per
year, scaled to the new baseline of 2021. This is projected to decrease further, in line with the expected
annual energy consumption reduction assumptions used in Figure 29.
Known Limitations
The domestic profile used for Profile Class 1 does not fully reflect the varying energy requirements of
new domestic customers. Further subcategorising the domestic customer type by the size of house and
number of electrical appliances could improve the granularity and accuracy of domestic customer
behaviour.
The assumptions for the demand contribution and energy consumption reductions for domestic
customers align with the licence area observed demand and energy reductions. Further
subcategorisation by customer type of the existing underlying demand could improve the assumptions
for expected customer behaviour change for domestic customers.
Future Developments
With an increased amount of domestic customers switching to smart meters and half-hourly metering,
this data could be used to infer how customer behaviour changes with reference to price signals which
a DNO does not directly impact. NG plan to investigate the suitability of aggregated smart meter data
to inform assumptions on the behaviour of existing and future domestic customers.

Non Domestic
Table 12: Table of distinct non-domestic technology types used in the DFES analysis, with description of
the planning use class definition for each category
Technology Subtechnology Description of customer types Units used in DFES

volume projections
Non A1/A2 Shops, financial and professional Floorspace (metres

domestic services squared) of new I&C
developments
A3/A4/A5 Restaurants, cafes, drinking
establishments and hot food
takeaways.
B1 Business (including offices)

B2 General industrial processes
B8 Storage and distribution
C1 Hotels
C2 Residential institutions (including

secure residential institutions)
D1 Non-residential institutions
D2 Assembly and leisure
Sui Generis Other customer types not covered

by the categories above
N.B. category C3/C4 (Dwelling and houses of multiple occupation) not included as the projections as covered by
domestic projections.
Methodology
The DFES volumes project includes an analysis of all local authority development plans to identify new
industrial and commercial developments. Each development is geographically allocated to an Electricity
Supply Area where it would be most likely to connect to the distribution network.
Non-domestic developments are provided with units as m² of floor space in the development. There are
ten difference categories used in the DFES volume analysis, which encompass different industrial and
commercial customer types. These are consistent with the planning use classes, before they were
updated in September 202016. The units and categories of industrial and commercial customer were
chosen as they are the most consistently used in local authority development plans and the Energy
Performance Certificate (EPC) database.
For large non-domestic customers with an accepted connection offer to the NG 33 kV, 66 kV or 132 kV
networks, these customers are modelled as a distinct category and profiled based on the requested
import capacity as per the connection offer.
Creating a sample of connected customers
To create representative profiles for each non-domestic customer type suitable for strategic network
analysis, a demand profile per m² of development floorspace is required. This requires two datasets
with the necessary information to be joined together as shown in Table 13.
16 Planning Portal, Change of Use – Use Classes, 2020;

https://www.planningportal.co.uk/info/200130/common_projects/9/change_of_use

Table 13: Summary of input data required for non-domestic customer behaviour analysis
Data source Information required
NG asset management systems  Meter Point Administration Number (MPAN)

 Address
Energy Performance Certificate  Address
database17  Property use class (matching the list in Table 12)
 EPC rating
 Floor space (m²)
As the address is the only common field to join the two datasets, a bespoke Python string-matching
program was developed match the different address fields for each customer. This uses an adaption of
an algorithm published by Ratcliff and Obershelp (commonly known as the Getsalt Pattern Matching
algorithm)18. This process generates a large sample of half-hourly metered customers with the required
fields as in Table 13.
For the purposes of DFES analysis, it is assumed that new industrial and commercial customers
connected to the network will have an EPC rating of C and above. This assumption is intended to
remove the impacts of customers with low EPC ratings influencing the representative profiles for future
non-domestic connections, and broadly aligns with proposals made in the Future Building Standard
consultation19.
Creating Representative Demand Profiles
For each customer in the sample above, half-hourly metering data is collected for the calendar year of
2019. Analysis of Valuation Office Agency (VOA) data on the number of non-domestic properties in
each Local Authority was cross-referenced against the NG geographic polygon datasets for Primary
substations. This process creates a representative sample size of the number of customers of each
non-domestic customer type connected to each Primary substation. The table below shows the total
sample size available of customers used to generate non-domestic profiles, alongside the
representative sample size for the average number of customers per non-domestic customer type
connected to a Primary substation.
17Ministry of Housing, Communities and Local Government, Energy Performance of Buildings Data
[Data set], https://epc.opendatacommunities.org/ (Accessed 13th September 2019)
18 Python Software Foundation, difflib – Helpers for computing deltas, 2020;
https://docs.python.org/2/library/difflib.html (Accessed 19th August 2020)
19 Ministry of Housing, Communities and Local Government, Future Buildings Standard: Consultation
on changes to Part L (conservation of fuel and power) and Part F (ventilation) of the Building
Regulations for non-domestic buildings and dwellings; and overheating in new residential buildings,
2021;
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/95
6037/Future_Buildings_Standard_consultation_document.pdf

Table 14: Total sample size and representative sample size per Primary substation of customers used to
derive non-domestic profiles
Property Use Class Number of customers in sample Representative sample size used
A1/A2 276 150
A3/A4/A5 68 65
B1 129 100
B2 64 60
B8 79 75
C1 60 30
C2 60 25
D1 69 35
D2 38 20
Sui Generis 9 5
For each customer type, the representative sample size was randomly selected from the input dataset.
The total demand for all half-hourly periods across a year was extracted from metering data and
normalised over the total floor space of the sample. For each of the seasons identified in the
Representative Days section of this report, a daily demand profile was generated for each
representative day using the following logic:
 Peak demand representative day: for each season studied, the maximum normalised value
for the sample was extracted for each half hour; and
 Peak generation representative day: for each season studied, the minimum normalised value
for the sample was extracted for each half hour; and zero export.
To ensure the worst-case network conditions were captured, the above process was repeated 1,000
times with different randomly selected input data for each customer type. The maxima and minima of
all repeated samples taken was used to derive the demand profiles.
Profile Benchmarking
The representative profiles were benchmarked against a range of data sources to validate their
suitability for use in network analysis, including similar data shared by other DNOs.
 Non-domestic profiles used by NG in previous strategic studies: NG has used non-
domestic normalised around the industrial and commercial floorspace in previous strategic
studies20. The profiles generated using the above process are similar when compared for
network impact totals for the same input volume data. A key improvement is the ability to split
the ‘Factory and warehouse’ customer type into distinct ‘General Industrial/B2’ and ‘Storage
and Distribution/B8’ categories, as it is observed they follow different electrical profiles.
 BSRIA Rules of Thumb21: This document is for building services engineers to specify the
electrical requirements for non-domestic buildings. The ‘rules of thumb’ for the after diversity
maximum demand figures (of kW/m²) compared favourably to the newly generated profiles.
Any differences can be explained by the aggregate behaviour of a group of customers
connected at a Primary substation level, rather than the individual customer demand.
20 National Grid, Shaping Subtransmission East Midlands, 2020;

https://www.nationalgrid.co.uk/downloads-view/134116
21 Hawkins, G, (2011). Rules of Thumb – Guidelines for building services (5th edition). BSRIA

A1/A2 (shops and retail)
Figure 32: Representative A1/A2 non-domestic profiles
A3/A4/A5 (eating and drinking establishments)
Figure 33: Representative A3/A4/A5 non-domestic profiles

B1 (office)
Figure 34: Representative B1 non-domestic profiles
B2 (general industrial)

B8 (storage and distribution)
C1 (hotel)
Figure 37: Representative C1 non-domestic profiles

C2 (residential institutions)
Figure 38: Representative C2 non-domestic profiles
D1 (non-residential institutions)
Figure 39: Representative D1 non-domestic profiles

D2 (assembly and leisure)
Figure 40: Representative D2 non-domestic profiles
Sui Generis
Figure 41: Representative Sui Generis non-domestic profiles

No flexed profile for non-domestic customers is used as part of the customer behaviour analysis. Whilst
the market for industrial and commercial DSR is still relatively immature, suitable data to inform
expected uptakes of load management for the purposes of 33 kV, 66 kV and 132 kV network analysis
is not currently available for the representative days studied.
For the demand reduction due to energy efficiency measures, the same profile scaling figures as applied
to the underlying demand shown in Figure 28 are used for new non-domestic customers. This
assumption is made on the basis that a large proportion of the energy consumption across NG is non-
domestic and energy efficiency measures are expected to continue.
Energy Assumptions
As part of the non-domestic profiling exercise, the years’ worth of metering was collated for the sample
and normalised over the total floorspace. An annual scaling factor for the floorspace to annual energy
consumption in kWh was generated. To obtain total estimated annual energy consumption from new
non-domestic customers, the values in Table 15 can be used. The energy consumption for new non-
domestic customers is expected to decrease further in line with the expected annual energy
consumption reduction assumptions used in Table 15: Average scaling factor from floorspace (m²) to
total annual energy consumption (kWh) for each non-domestic customer type
Property Use Scaling factor from development floorspace to total annual energy
Class consumption (m² to kWh/year)
A1/A2 360.6
A3/A4/A5 742.2
B1 186.9
B2 132.6
B8 166.8
C1 147.8
C2 211.1
D1 117.2
D2 144.3
Sui Generis 312.7
Known Limitations
The non-domestic profiling exercise is reliant on the accuracy of the EPC database of non-domestic
properties. There are some instances whereby the floorspace did not correlate to the floorspace of a
customer connected to the NG distribution network.
The electrification of industrial processes for existing customers is not captured as part of this analysis.
As the profiles are based on existing customer metering data, this covers the existing penetration of
electric heating and cooling for non-domestic customers.

Future Developments
As a large proportion of existing non-domestic customers switch to half-hourly settlement22, half hourly
data could be used to infer how customer behaviour changes with reference to price signals which a
DNO does not directly impact. NG plan to investigate the suitability of aggregated smart meter data to
inform assumptions on the behaviour of existing and future non-domestic customers.
22Ofgem, Modification Proposal: Balancing and Settlement Code (BSC) P272: Mandatory Half-Hourly
Settlement for Profile Classes 5-8, 2014; https://www.elexon.co.uk/wp-
content/uploads/2012/02/P272D.pdf

Electric Vehicles
Table 16: Electric vehicle and charger types used in the DFES analysis
Electric Hybrid bus and coach Number of electric vehicles

vehicles
Hybrid car (autonomous)
Hybrid car (non-autonomous)
Hybrid HGV
Hybrid LGV
Hybrid motorcycle
Pure electric bus and coach

Pure electric car (autonomous)
Pure electric car (non-autonomous)
Pure electric HGV

Pure electric LGV
Pure electric motorcycle

EV Charge Car parks Number of EV charge points
Point
Destination
Domestic off-street
Domestic on-street
En-route / local charging stations
En-route national network
Fleet/Depot
Workplace
Methodology
The DFES has been developed to include vehicle and charger projections for all NG distribution licence
areas. There are 12 vehicle and 8 charger types projected for all years and scenarios. Each vehicle
and charger is geographically allocated to an Electricity Supply Area where it would be most likely to
be registered or connect to the distribution network; noting that EVs will charge at multiple locations. All
vehicle and charger types projected are given in Table 16. Out of the 12 vehicle types assessed, hybrid
bus and coach, hybrid HGV, hybrid motorcycle and hybrid car (non-autonomous) were identified as not
being credible vehicle types and the volumes for all years and scenarios remain at zero. They are
included in the DFES projections volumes for completeness. All charger types are considered viable
and the volumes connected increase under all scenarios.
The distribution of EVs in the near term is based on affluence, rurality, existing vehicle baselines and
the distribution of on and off street parking. However, in the late 2020s under all Net Zero scenarios
uptake is assumed ubiquitous. This means that almost all consumers are assumed to have the same
likelihood of adopting an EV.

Early EV forecasting typically sub-categorised vehicle types into Battery Electric Vehicles (BEV) and
Plug-in Hybrids (PHEV). The majority of BEV and PHEV early adopters had access to off-street
charging; meaning the majority of early charging data available was from residential off-street chargers.
EV adoption has increased significantly in recent years and the type of vehicles and chargers available
are more diverse. By including eight viable vehicle types in the projections, differences in annual
mileage and energy are used to build a more representative energy model. The Department for
Transport (DfT) and FES datasets were used to determine average mileage and energy figures for each
vehicle subtechnology. More information on the energy assumptions for each vehicle type can be found
in the Energy Assumptions section.
To assess the impact EVs will have on the NG distribution network, profiles for all five representative
days are required. These need to capture EV demand at existing peak demand periods and possible
new peaks caused by the largescale adoption of EVs. There are two methods for profiling the impact
of EVs:
1. EV Profile – Assign a profile to each EV; or
2. Charger Profile – Assign a profile to each charger
There are a number of limitations of directly profiling vehicles:
 EV charging location – EVs will use multiple charger types and locations to fulfil their energy
requirement. Profiling an EV based on its registered location does not account for this. EVs
registered at addresses without access to off-street charging will use a combination of work,
slow/fast and rapid chargers to fulfil their energy requirements. These charger types may be
located in a separate ESA and will have a notably different charging profile and utilisation.
 Vehicle stock – The total vehicle stock needs to be accurately represented when profiling EVs.
This is more important as the volumes of electric LGVs and HGVs increase, as they will have
a significantly different charging behaviour and energy requirement.
 Multiple EVs to a charger - Modelling 1 EV to a charger, does not assess the utilisation and
diversity impact of multiple EVs on a house; particularly at peak demand periods.
 Charger uptake – The volumes and proportion of charger types available are projected to
change under each scenario. Different charger categories have notably different charging
behaviours, which is particularly important when profiling for network edge-cases.
 EV to charger ratio –By profiling each EV on a static profile it assumes the EV to charger ratio
remains constant.
For the reasons described above, only assessing vehicles or chargers independently does not give
sufficient information to determine the network impact of the EV transition. The DFES projections of
vehicles and chargers allows the creation of a more representative EV model that accounts for scenario
and year dependent energy requirement and the chargers that are available to deliver this energy.
This methodology has a number of benefits over directly profiling EVs:
 It accounts for the total vehicle stock yearly energy requirement
 It is profiling the chargers that are actually connected to the distribution network, based on
location and type
 Factors in the vehicle to charger ratio and the utilisation of chargers
A number of projects, datasets and reports were assessed to determine the most appropriate profiles.
The main projects used to derive and validate these profiles are:
 Element Energy’s EV Charging Behaviour Study (29th March 2019) 23 - This project
developed a set of annual charging demand profiles, covering all 8,760 hours within a year,
based upon a dataset of over 8 million real-world charging events collected from major charge
point operators.
 Electric Nation24 - When launched, Electric Nation was the world’s largest home smart
charging trial with nearly 700 Electric Vehicle (EV) owners taking part in the project. The large-
scale smart charging trial provided invaluable data on how EV owners charge their vehicle at
home and included a trial looking at managed charging.

Charger Categorisation
The Element Energy model is one of the most comprehensive assessment of charging behaviour (~8
million transactions) within the UK. The charging behaviour demand profiles produced as part of this
project were used as a starting point to produce charger specific profiles for the DFES EV Charge Point
subtechnologies. The Element Energy work grouped all chargers into one of four categories:
 Residential - Charge points located at or near EV drivers’ homes
 Work - Charge points installed in workplaces, for use by employees who commute to work
using an EV
 Slow/Fast Public - Publicly accessible charge points, excluding those classified as Work or
Residential.
 Rapid Public - Publicly accessible charge points with a charging capacity ≥43 kW
The mapping from the NG charger subtechnologies to the Element Energy charger types are given in
Table 17.
Table 17: Mapping of NG EV Charge Point subtechnologies to corresponding Element Energy
categorisations
Charger Grouping NG Charger Subtechnology
Residential Domestic off-street

Domestic on-street
Work Fleet/Depot
Workplace
Slow/Fast Public En-route / local charging stations
Car parks
Destination
Rapid Public* En-route national network
*”The rapid public charger data was not made publically available, as the relatively low sample size could be traced
back to individual customers. As stated in the Element Energy report “this charger type classification provides an
effective trade-off between distinctions in usage while ensuring each type has a large enough data volume”.
The NG DFES charger types are forecast at a more granular level to help inform interested stakeholders
and aid in the future development of EV customer behaviour modelling. The profiles described below
were created for the four overarching charging groups, and then applied to the more granular NG
charger types.
Normalised Profile Creation
The Element Energy dataset provides a whole yearly hourly profile for residential, work and slow/fast
for each NG distribution licence area. These profiles underwent detailed validation, including removal
of erroneous charging events, correcting for increasing EV and charger stock and ensuring anomalously
high and low demand were reflected in the final profiles. The final profiles used in this analysis are
normalised hourly profiles that give a kW per yearly kWh.
23Element Energy, Electric Vehicle Charging Behaviour Study, Final Report for National Grid ESO,
2019; https://www.smarternetworks.org/project/nia_ngso0021/documents
24 National Grid, Electric Nation, 2019; https://www.nationalgrid.co.uk/projects/electric-nation

The process below was used to create the representative day profiles per charger category:
1. Group the yearly profiles from each NG distribution licence area into the corresponding charger
category.
2. Discount any licence area profiles that did not meet the Element Energy diversified threshold
check. This is to exclude profiles that have an insufficient sample size to be statistically
significant.
3. Assign the relevant season from the
4. Representative Days to each day and hour within the year.
5. For each hour within a day, find the maximum across all licence areas within each season for
all 3 available charger categories. This give a credible worst-case peak demand profile for each
season.
6. For each half hour within a day, find the minimum across all licence areas for the summer
season for all 3 available charger categories. This give a credible worst-case summer minimum
demand profile.
The output of this process gave a profile for each representative day for the residential, work and
slow/fast charger categories. These profiles are kW per yearly kWh. This approach enables the energy
from a changing vehicle stock to be distributed across a year; accounting for increased utilisation and
numbers. These profiles are shown in Figure 42, Figure 44 and Figure 45.
The maximum for each licence area was taken and applied to all licence areas, rather than attempting
to apply licence area specific diversity. This is because the South Wales profiles did not meet the
diversified threshold for Work and Slow/Fast chargers. There is also a risk that any differences are due
to uptake rate and sample size limitations, rather than actual licence area specific variances. Note there
is not a significant difference between the fully diversified profiles.
Due to the lack of available data on rapid charger utilisation and the low number of EVs that are capable
of fully utilising the rapid charge, the profile for rapid chargers is modelled at a constant 350 kW for all
demand representative days. This does not represent how rapid chargers are likely to operate, but is a
worst-case Managed Charging Residential Profile.
To account for residential smart charging, the Electric Nation managed charging trial data was
analysed. The amount of energy time-shifted due to managed charging was calculated and this
assumption was applied to the residential charger category to create a residential flexed profile. This
final flexed profile is shown in Figure 43. It is important to note that the profile assumes no charging
between 5-6pm, to enable the FES assumption of energy time-shifted at peak to be applied.

Apportioning Energy
To determine the energy requirement for each representative day, the total vehicle stock energy is
apportioned to each charger type. The proportion of energy delivered via each charger type is taken
from the Element Energy model.
Table 18: Share of charging demand across charger categories
Charger Category Share of charging energy (%)
Residential 74.7
Work 14.7
Slow/Fast Public 5.8

Rapid Public 4.8
To obtain an hourly profile for each charge category for all representative days the equation below was
used to apportion the energy across each charger category.
𝑘𝑊
𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑠𝑒𝑑 𝑝𝑟𝑜𝑓𝑖𝑙𝑒 ( ) ∗ 𝑠ℎ𝑎𝑟𝑒 𝑜𝑓 𝑐ℎ𝑎𝑟𝑔𝑖𝑛𝑔 𝑑𝑒𝑚𝑎𝑛𝑑(%) ∗
𝑘𝑊ℎ
𝑇𝑜𝑡𝑎𝑙 𝐸𝑉 𝑒𝑛𝑒𝑟𝑔𝑦(𝑘𝑊ℎ)
𝐶ℎ𝑎𝑟𝑔𝑒𝑟 𝑝𝑟𝑜𝑓𝑖𝑙𝑒 (𝑘𝑊 𝑝𝑒𝑟 𝑐ℎ𝑎𝑟𝑔𝑒𝑟) =
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐ℎ𝑎𝑟𝑔𝑒𝑟𝑠 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑒𝑑
This final output is a year and scenario dependent profile that accounts EV energy requirements and
how this will be apportioned across the available chargers. These profiles are then applied to the NG
EV Charge Point subtechnologies using the mapping given in Table 17.

Residential EV Charge Point (unabated)
Figure 42: Residential unabated kW/yearly kWh representative day charging profile

Residential EV Charge Point (flexed)
Figure 43: Residential managed charging (flexed) kW/yearly kWh representative day charging profile
Work EV Charge Point (unabated)
Figure 44: Work unabated kW/yearly kWh representative day charging profiles

Slow/fast EV Charge Point (unabated)
Figure 45: Slow/Fast unabated kW/yearly kWh representative day charging profile

As described in detail in the
Methodology section, the EV charger profiles will change over time due to:
 Total EV energy based on vehicle stock
 EV to charger ratio
 Increase in managed residential charging
In addition to the above, the improvement in car (including autonomous), motorbike and LGV energy
efficiency (kWh/mile) is modelled. These figures are taken from the DfT Transport Analysis Guidance25
model and are applied to all scenarios equally. The energy efficiency improvements are shown in Figure
46. The DfT Transport Analysis Guidance model does not currently provide energy and efficiency
figures for HGV or bus and coaches. Due to this and the relative uncertainty surrounding the immature
HGV and the bus and coach category, no energy efficiency figures were applied; this assumption will
be periodically reviewed as more data becomes available.
A review of Electric Nation Powered Up26 was carried out for investigating the feasibility of incorporating
the V2G behaviours into the analysis. This review found that due to the sensitivity of customers to
pricing (which is supplied-led not as a result of a DNO flexibility contract) and the purpose of this
modelling, additional understanding of customer behaviour is required to accurately incorporate V2G
into the DFES analysis.
25 Department for Transport, Transport Analysis Guidance Data Workbook v1.13.1 [Data set]
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/89
8797/tag-data-book.xlsm (Accessed 15th December 2022)
26National Grid, Electric Nation – Powered Up, 2020; https://www.nationalgrid.co.uk/projects/electric-
nation-powered-up

Figure 46: Car and LGV efficiency improvement assumptions taken from DfT TAG model
Energy Assumptions
The baseline energy assumptions per vehicle category are given in Table 19.
Table 19: Electric vehicle average annual mileage, kWh/mile and yearly energy requirement in Baseline
year
Subtechnology Average Mileage kWh/mile kWh per Year

(Baseline) (Baseline)
Hybrid car (non-autonomous) 4799* 0.3059 1,468

Hybrid LGV 7410* 0.3724 2759
Pure electric bus and coach 15943 1.6030 25557
Pure electric car (autonomous) 8726 0.3059 2669
Pure electric car (non-autonomous) 8726 0.3059 2669
Pure electric HGV 34693 1.33925 46462
Pure electric LGV 13472 0.3724 5017

Pure electric motorcycle 2386 0.29058 693
*Mileage when running on electric only
The efficiency improvements described in the above section are also applied for years after baseline.
Table 19 and the DFES volumes data are used to determine the yearly energy requirement. This can
be calculated at an ESA level or aggregated to a licence area or NG view.

Known Limitations
Rapid chargers are currently not profiled due to the limited number of providers of rapid public charger.
Where rapid charger data is available, it includes charging of vehicles like the Nissan leaf that can only
charge at < 50kW. The risk of using existing rapid charger data to determine future utilisation is that it
does not represent new vehicles with larger batteries that are able to charge at a higher rate.
No modelling of Vehicle to Grid (V2G) has been assumed due to the uncertainty around behaviour at a
granular substation level, in particular if V2G will resolve network constraints that do not align with
system peak demand.
Customers charging directly from a 13A plug are not explicitly captured as a charger type. This is largely
mitigated, as the total energy requirement of all EVs is captured and apportioned across the available
chargers.
The proportion of energy apportioned to each charger category is modelled as constant, based on
existing vehicle stock. The total energy required for the vehicle stock is still delivered, but it does not
fully capture an increased proportion of total energy being supplied via a growing charger type segment.
This is the first year where autonomous vehicles have been included in the DFES volumes. The
operation of autonomous vehicles is still uncertain and potential annual mileage and charging
behaviours are not well understood. Autonomous vehicles were assigned the same annual mileage as
their non-autonomous counterparts. The growth of autonomous vehicles only significantly increases in
the late 2030s.
Profiles derived from the Element Energy model are only hourly, rather than half-hourly. Distribution
networks are traditionally designed at a half-hour granularity. Benchmarking the residential off-street
profiles against the Electric Nation profile shows that the hourly profile is largely representative of a full-
diversified profile.
Future Developments
As more data becomes available, improve the modelling assumptions on autonomous vehicles. This
includes understanding credible autonomous market models, including annual mileage and charging
behaviour.
Improve the constant EV energy apportionment assumption to charger category based on number of
available chargers and the chargers each vehicle type will utilise.
Evaluate ways to better capture EV customers who are charging via 13A plugs, who do not have a
dedicated EV charger.
Assess the outputs of the ongoing SP Energy Networks CHARGE 27 innovation project, which is
combining transport and electricity network data to highlight the best location for public EV charge points
at the lowest cost. It will also trial deployment of smart management of public chargers. The output of
this project should help inform public charger profiles and quantify the potential reduction due to smart
public charging.
Planned review of UK Power Networks OPTIMISE PRIME28 innovation project that is the world’s biggest
trial of commercial EVs. It is looking at how to minimise the impact the electrification of commercial
vehicles will have on the distribution network. The results of this project will help inform our commercial
EV charging behaviour.
As more charging data becomes available, producing EV charging profiles at a higher granularity (half-
hourly or better) will help identify any short duration peaks that will not be fully captured with a hourly
profile. Additional charging data will also enable explicit profiling of all NG EV Charge Point
subtechnologies, rather than grouping into high-level categories.
Better understand the behaviour of domestic chargers where users are utilising EV Charge Point
leasing applications.
27 SP Energy Networks, CHARGE, 2020; https://www.spenergynetworks.co.uk/pages/charge.aspx

Resistive Electric Heating
Table 20: Resistive electric heating technology types used in the DFES analysis
Resistive electric Direct electric heating Number of customers with resistive electric
heating heating
Night storage heating
Methodology
Resistive electric heating is a system using electricity to provide primary space heat and hot water to
domestic buildings that is not driven by a heat pump. Typically, this is night storage heating or direct
electric heating. This does not include heat networks.
The baseline number of resistive electric heating units is based on analysis of domestic heating
technology types from EPC data, census data and NG connected customers. The installation rate of
direct electric heating in new builds is also based on local EPC data. The most recent national data
shows that c.11% of new builds are heated by resistive electric heating, a proportion which has been
relatively stable over recent years. The NG DFES analysis of new build domestic properties is used to
project increases in the number of resistive electric heating installations. Resistive electric heating has
a higher running cost than a heat pump, they are assumed not to be the target of national policy to
decarbonise domestic heating.
The customer behaviour assumptions for resistive electric heating and night storage are based on the
Elexon profile classes29. They are consistent with the NG Policy Document ST:SD5A (Design of Low
Voltage Domestic Connections)30. This process is developed for the purposes of Low Voltage network
design, and uses a statistical methodology consistent with that published in the ACE49 methodology 31.
For the application of these resistive heating profiles for use in strategic analysis of the EHV networks,
a diversity level of 50 customers was chosen. This represents a credible usage profile for a single
resistive electric heating installation, aggregated as part of a wider group of 50 domestic customers with
night storage.
A limitation of the ACE49 methodology is it does not produce a half-hourly profile for all representative
days assessed. To create a profile for all demand representative days, the ratio of HP energy
requirement referenced to winter peak was used as an approximation for seasonal scaling.
Night storage
The Elexon profile class 2 represents domestic economy 7 customers. This is the profile class normally
allocated to domestic customers with night storage. Profile class 2 includes the energy requirement for
domestic demand, not just the night storage heating. To derive a night storage only profile, the delta
between Elexon profile class 1 and profile class 2 was calculated. The profiles for night storage in the
baseline year is shown in Figure 47.
28 UK Power Networks, Optimise Prime Project Overview, 2020; https://www.optimise-

prime.com/project-overview
29 Elexon, Load Profiles and their use in Electricity Settlement, p41
30 National Grid, Standard Technique: SD5A/5, Design of Low Voltage Domestic Connections, p41
31 Energy Networks Association, ACE Report no.49, p5

Direct Electric Heating
The direct electric heating profile was created using a similar methodology to the night storage profile.
The delta between profile class 1 and the unrestricted profile with electric heating was taken to create
a direct heating only profile. The energy per installation for each year and scenario is detailed in the
Energy Assumptions section.

Night storage profile
Figure 47: Night storage profile per installation in baseline year
Direct electric heating profile
Figure 48: Direct Electric heating profile per installation in baseline year

The yearly energy requirement per installation is projected to reduce under all scenarios; more
information can be found in the Energy Assumptions section. The ACE49 methodology used to derive
both profiles are scaled around the yearly energy requirement per resistive electric heating customer.
To account for how the reducing energy per installation will affect the MW profiles, a scaling factor
based on the data in
Figure 49 and Figure 50 is applied. The scaling factor is normalised around the baseline for each year
and scenario and can be linearly applied to the profiles in Figure 47 and Figure 48.
Energy Assumptions

The yearly energy for direct electric heating and night storage broken down by year and scenario are
taken from the FES data workbook32. The yearly energy is assumed to remain the same for direct
electric heating and night storage. The yearly energy requirement is projected to reduce under all
scenarios, due to improvements in existing and new housing stock thermal efficiency. The projected
efficiency improvements in heating are shown in
Figure 49.
Annual energy consumption percentage change from baseline per direct electric heating customer
Figure 49: Direct electric heating energy percentage change from baseline by scenario

Annual energy consumption percentage change from baseline per night storage heating customer
Figure 50: Night storage heating energy percentage change from baseline by scenario
Known Limitations
There are known limitations of these assumptions used for night storage and direct electric heating
where the profiles do not currently vary by the archetype of house that they would be found on. Due to
the varying thermal properties of different buildings, this could result in variations from the modelling
used.
Future Developments
In future there is scope to integrate these heating profiles into a heat model that also includes both heat
pumps and district heating, along with further categorisation into building types.

Heat Pumps
Table 21: Heat pump technology types used in the DFES analysis
Technology Subtechnology Units used in DFES volume

projections
Heat pumps Domestic - Hybrid Number of heat pumps

Domestic - Hybrid + thermal storage
Domestic - Non-hybrid ASHP
Domestic - Non-hybrid ASHP + thermal

storage
Domestic - Non-hybrid GSHP
Domestic - Non-hybrid GSHP + thermal

storage
District heating (non-bulk) Number of customers on district

heating scheme
District heating (bulk)
Methodology
Each heat pump connection is geographically allocated to an Electricity Supply Area where it would be
most likely to connect to the distribution network. Heat pump volumes are provided in number of heat
pumps. As described in the Domestic section, heat pumps are forecast independently to new domestic
properties. This allows the retrofitting of heat pumps in the existing housing stock to be more accurately
captured.
Ground Source Heat Pumps (GSHP) and Air Source Heat Pumps (ASHP) have been forecast
separately, an addition compared to last year’s analysis; however, it is expected that GSHPs will be
less prevalent due to GSHPs space requirements for the ground source loop and cost of installation. It
is worth noting that GSHPs do have a higher coefficient of performance, particularly at times of low
ambient temperatures.
Recent developments in hybrid heat pumps, which work with a backup technology (primarily gas), have
started to reduce some of the barriers and raise potential for much higher growth in the sector. As well
as starting to make it a cost-effective option for an on-gas grid customer, a hybrid system also requires
less disruptive change. The higher temperature heat can use existing radiators and the heat pump
operates at times when it is most efficient (e.g. low electricity prices or moderate heat requirements),
with back up sources taking over when it is not.
The majority of substations see a yearly peak demand during the winter or intermediate cool seasons,
at times of cold ambient temperatures. The majority of heat pump energy demand is also within the
winter and intermediate cool seasons, with a peak demand that is shown to correlate very closely with
existing peak demand.
With non-hybrid heat pumps, all energy is provided via the electricity network, compared with hybrid
systems that can switch between electricity and gas. At times of high network demand and low ambient
temperatures, non-hybrid systems coefficient of performance can drop significantly. An additional
electrical backup source is used where the non-hybrid heat pump is unable to maintain the required
temperature. A hybrid system is able to switch over to its alternate fuel source at times of high electricity
demand.
When determining the heat pump profiles, it was important to consider the coincident nature of existing
peak demand and heat pump peak demand. The use of average or typical profiles does not capture
the onerous network loading that will be seen for a 1-in-20 winter.

Non-hybrid Profiles
A review of available heat pump data was undertaken to determine the best source to derive edge-case
heat pump profiles. The projects, data sources and reports evaluated include:
 Electricity North West Limited Network Innovation Allowance (NIA) funded study: Managing the
Impact of Electrification of Heat.33
 Customer-Led Network Revolution Insight Report 34: Domestic Heat Pumps, dated January
2015. This included the associated TC3 datasets 35with heat pump mean and standard
deviation for a 48 half hourly period for all months.
 Energy Policy report on “Decarbonising domestic heating: What is the peak GB demand?” 36
 Applied Energy “The addition of heat pump electricity load profiles to GB electricity demand:
Evidence from a heat pump field trial” 37(Applied Energy 204 (2017) 332–342)
 National Grid FES 2021 Report and Data workbook 38
The non-hybrid heat pump profiles needed to capture the maximum electrical demand for all
representative days. This includes demand from any electrical backup that may operate at times of
extreme cold.
The Customer-Led Network Revolution TC3 dataset provide an average and standard deviation for a
full 48 half hour period for each month in the year. This data enabled production of a half-hourly profile
for each network capability season. It was determined that 3 standard deviations from the average was
a credible peak, representing a cold period, where a large portion of electric backup heating is required.
The unabated non-hybrid profiles are given in Figure 51.
A development in DFES 2022 is splitting the profiles of hybrid and non-hybrid heat pumps that are co-
located with thermal storage. Thermal storage is the capturing of energy at times of low network
demand, when electricity prices are lower. The thermal storage subtechnology profiles use the same
underlying profiles as the profiles without thermal storage, but with no electrical demand during the
morning or evening period. The non-hybrid thermal storage co-location profile is shown in Figure 52.
33 Electricity North West Lid, NIA ENWL001 – Demand Scenarios with Electric Heat and Commercial
Capacity Options, (2017) https://www.enwl.co.uk/globalassets/innovation/enwl001-demand-scenarios-
-atlas/enwl001-closedown-report/nia-enwl001-closedown-report-final.pdf
34Northern Powergrid, Customer Led Network Revolution Insight Report: Domestic Heat Pumps, 2015;
http://www.networkrevolution.co.uk/wp-content/uploads/2015/01/CLNR-L091-Insight-Report-
Domestic-Heat-Pumps.pdf
35 Northern Powergrid, Customer Led Network Revolution TC3 Dataset, [Data set] (2014)
http://www.networkrevolution.co.uk/wp-content/uploads/2015/01/TC3-Dataset_December-2014.xlsx
(Accessed July 2019)
36 Watson, S.D. et al, (2019), Decarbonising Domestic Heating: What is the peak GB demand? Energy
Policy 126, p533-544
37 Love, J et al, (2017), The addition of heat pump electricity load profiles to GB electricity demand:
Evidence from a heat pump field trial, Applied Energy 204, p332-342
38 National Grid ESO, FES Data Workbook 2022 v03, pED1

Hybrid Profiles
Hybrid heat pumps are an emerging technology and there is still a level of uncertainty as to how they
will operate. A hybrid system manages heat delivery from both a gas boiler, meaning the level of energy
delivered from electricity can vary significantly dependent on mode of operation and price signals.
A key source of information on hybrid heat pump operation is the Freedom Project39, which was a joint
innovation project between National Grid, Wales & West Utilities and Passive Systems. It used air
source heat pump and high-efficiency gas boiler hybrid system in 75 residential properties, the project
has demonstrated the significant benefits that an integrated whole energy systems approach to
deploying smart dual-fuel technologies can deliver.
The systems were operated in a range of different fuel price scenarios, including different ranges of gas
pricing and with both fixed and variable rate electricity tariffs. When optimising for consumer cost with
today’s energy prices, the systems strongly favour gas boiler usage due to the very low cost of gas
compared with electricity.
The FREEDOM Final Report states:
“The field trial demonstrated that hybrids can provide fully flexible loads with the
ability to: constrain peak whole-home demand below the existing Elexon Profile
Class 1 peak whilst still delivering 50% of the heat demand through the heat
pump; enforce a capacity cap across a population, including a cap of zero ASHP
demand; increase ASHP demand at times of plentiful low-cost renewable
electricity; for the first time ever live carbon forecasts were used so that the ASHP
could track grid carbon intensity and avoid times of high carbon peaking plant
generation”.
Based on the FREEDOM project findings, hybrid heat pumps are modelled as running on gas for the
Winter and Intermediate Cool Peak Demand representative days. Both of these representative are
focussed on extreme cold ambient temperatures, where the ASHP coefficient of performance will be
greatly reduced and it is presumed that the heat pump control system and price signals will incentive
gas sufficiently during these representative days.
Summer and intermediate warm representative days are profiled at 80% of the non-hybrid profile,
accounting for hybrid systems that are still operating on gas even during less onerous periods. The
profiles used for hybrid heat pumps can be found in Figure 54.
District heating profiles
New to DFES 2022, a profile has been produced to model district heating schemes. District heating has
been split into two categories to support this – bulk and non-bulk. We are treating the bulk district
heating schemes (those heat pump powered systems that can be found in a city centre or other high
density area) as a single customer with a unique operating profile, and non-bulk district heating
schemes (a few properties connected to a single large heat pump) as a profile per customer connected
to said heat pump.
Through using this methodology, the operating behaviour of bulk district heating schemes is better
captured, as this varies by network characteristic such as storage capacity and topography of the
network. This is particularly true where a heat network extends across multiple primary substations,
and the point of connection to the distribution network falls within only one of the primary electricity
supply areas.
39National Grid, FREEDOM: Flexible Residential Energy Efficiency Demand Optimisation and
Management, 2019; https://www.nationalgrid.co.uk/projects/freedom

Non-hybrid heat pump profile
Figure 51: Non-hybrid heat pump profiles in baseline year
Non-hybrid heat pump profile with thermal storage
Figure 52: Non-hybrid heat pump with thermal storage profiles in baseline year

Hybrid heat pump profile
Figure 53: Hybrid heat pump in baseline year
Hybrid heat pump profile with thermal storage
Figure 54: Hybrid heat pump with thermal storage profiles in baseline year

District heating (non-bulk) heat pump profiles
Figure 55: Heat pump profile of small-scale district heating scheme in baseline year

The uptake of non-hybrid and hybrid systems vary significantly with each scenario. This has a notable
impact when assessing the MW growth on the network, particularly for the Winter and Intermediate
Cool Peak Demand representative days. Hybrid systems are able to switch to an alternate fuel source
at times of high electricity demand; scenarios where a higher proportion of hybrid systems are forecast
will have a reduced MW growth at times of high demand due to heat pumps.
The energy requirement of non-hybrid systems is forecast to reduce as the thermal efficiency of housing
is improved and the coefficient of performance of heat pumps increases. The profiles are scaled linearly
with the overall energy requirements for each year and scenario, as described in the Energy
Assumptions section below.
A new development in this year’s DFES is splitting out of heat pump systems co-located with thermal
storage. The operating behaviour of these are represented by the profiles shown in Figure 52. The
profile creation is described in more detail in the Methodology section. The residential thermal storage
adoption rate (%) and the residential demand shaved at peak (%) are taken from the FES workbook.
Energy Assumptions
The energy requirements of non-hybrid (ASHP and GSHP) and hybrid heat pumps for each year and
scenario were derived from the FES workbook.
As described in the above section, these energy improvement figures were applied to the non-hybrid
profiles to represent the reduction in peak MW requirement as housing stock thermal efficiency
improves and coefficient of performance of heat pumps increases. This does not apply to the winter
and intermediate cool hybrid profiles, where they are already assumed to be operating on an alternate
fuel source.

Non-hybrid Air Source Heat Pump (and Non-hybrid Air Source Heat Pump co-located with thermal
storage) annual energy consumption
Figure 56: Yearly energy requirement per non-hybrid air source heat pump
Non-hybrid Ground Source Heat Pump (and Non-hybrid Ground Source Heat Pump co-located with
thermal storage) annual energy consumption
Figure 57: Yearly energy requirement per non-hybrid ground source heat pump

Hybrid Heat Pump annual energy consumption
Figure 58: Yearly energy requirement per hybrid heat pump
Known Limitations
Currently only domestic heat pumps are forecast as part of the DFES volumes process. The inclusion
of non-domestic heat pump volumes would enable a more representative profile to be applied to the
entire building stock across the NG distribution network.
The current profiles used are produced from the CLNR project, which is a static dataset. This does not
account for differences in property size, or thermal efficiency of the properties, which may result in
variations to the real life demand of these technologies.
Future Developments
NG will continue to monitor the progress of existing projects, trials and business as usual processes for
assigning customer behaviour for Heat Pumps suitable for network analysis of the EHV networks and
update the assumptions as necessary.
The outputs of the Defender NGED Innovation project will provide heat pump profiles that are better
aligned to archetypes, and would improve the ability to increase the granularity of the profiles while still
reflecting the area that the profile is representing. Through using smart meter data, this also ensures
that, should customer behaviours change, this can be identified and incorporated into the profiles. Some
work to ensure that the profiles are of appropriate diversity for the analysis being carried out will need
to be done prior to using these profiles.

Air Conditioning
Methodology
Each air-conditioning unit is geographically allocated to an Electricity Supply Area where it would be
most likely to connect to the distribution network. Air-conditioning volumes are provided in number of
air-conditioners.
The daily profile for all of the demand representative days was assumed to be zero. The reasoning for
this was that the peak demand representative days in winter, intermediate warm and summer all
coincide with a cold day where domestic air conditioning was assumed not to be in use.
The profile for the Summer Peak Generation representative day was also modelled as zero for network
assessments. This does not necessarily coincide with high ambient temperatures, particularly on
networks with high wind penetration. There is a risk that modelling air-conditioning demand for the
summer peak generation day will mask the worst-case condition.
Figure 59: Representative air-conditioning profiles

These profiles do not change for any year and scenario.
Energy Assumptions
As described in the Methodology section, the air-conditioning demand is assumed as zero for all existing
representative days. However, the overall energy is modelled as 500 kWh/year for each installation.
This figure is taken from the FES workbook and is assumed to not change by year and scenario.

Known Limitations
The current representative profiles do not capture the minimum coincident air-conditioning demand at
time of peak generation.
This air-conditioning technology currently only captures domestic installations. Non-domestic units are
more prevalent and the impact is largely captured in the existing demand behaviour described in the
non-domestic section.
As more domestic air-conditioning units connect, demand at times of high ambient temperature could
cause a new network edge-case. Similar to hot countries with high levels of air-conditioning, the peak
demand can actually occur at high ambient temperatures. All four demand representative days are
currently focussed on peaks due to cold ambient temperatures.
Future Developments
NG plan to undertake analysis on domestic and non-domestic air-conditioning operating behaviour.
Focussing on existing behaviour at time of network peak and potential for new edge-cases to occur as
uptake increases. Consideration of increased energy requirement as average temperature increases
could also be reviewed.

Hydrogen Electrolysis
Methodology
Each hydrogen electrolysis plant is geographically allocated to an Electricity Supply Area where it would
be most likely to connect to the distribution network. Hydrogen electrolysis volumes are provided in
installed capacity (MW).
The daily profile for all of the demand representative days was assumed to be one. The reasoning for
this was that the production of hydrogen from hydrogen electrolysis, although likely to be coincident
with peaking renewable generation, is not limited to times of high renewable generation. As our
modelling only took into account those hydrogen electrolysers that are directly connected to our
network, not located behind the meter at weather dependent generation sites, there is increased
likelihood of the demand not aligning with renewable generation peaks.
The profile for the Summer Peak Generation representative day was modelled as zero for network
assessments. This is to remove the risk of worst-case conditions for the summer peak generation days
being masked when hydrogen electrolysis is not guaranteed to be operating at times of high generation.
Figure 60: Representative hydrogen electrolysis profiles

These profiles do not change for any year and scenario.

Energy Assumptions
As described in the Methodology section, the hydrogen electrolysis demand is assumed as one for all
existing representative days. Previously, the load factor was modelled as increasing 40, in line with
hydrogen electrolysis using otherwise curtailed renewable energy; however, the overall energy in DFES
2022 has been modelled using 35% as the load factor. This is in line with stakeholder consultation with
industry and aligns with FES. Based on an extensive literature review, the load factor has the potential
to vary from 25% up to 95% depending upon the running style of the electrolysers, and the amount of
storage available to the plant40. The selected load factor is constant across the scenarios.
Known Limitations
There is currently limited knowledge about the running of hydrogen electrolysers as they are in the early
stages of development. As more hydrogen electrolysers are rolled out we will be able to apply real-
world use cases and data to our forecasting.
Future Developments
NG plan to undertake agile analysis on the operating behaviour of hydrogen electrolysers and continue
to develop the forecasts as the industry gains further insight into the operational workings of this new
technology. Focussing on existing behaviour at time of network peak and potential for new edge-cases
to occur as uptake increases.
40 BEIS, Hydrogen Production Costs 2021, (2021)

Summary of future developments
NG welcome any feedback on the profiles presented in this document. Any improvements in the
analysis used to create these profiles will be incorporated into future editions of the DFES process and
strategic investment planning activities within NG, with an annual review of each technology type
planned. Publication of customer behaviour assumptions allows for transparency in the strategic
investment planning activities, also to drive further discussion and data sharing between stakeholders
to improve the assumptions used. Regional variations in profiles across the UK are expected and any
differences in profile behaviour should be justified by the DNO. The list below covers some of the areas
of focus for improvements to the DFES customer behaviour analysis in 2023:
 Further considerations of the behaviour of domestic storage and response to flexibility signals.
 Further update and application of new national level policies into the customer behaviour
assumptions of new domestic and non-domestic connections.
 Developing a greater understanding of the coincidence between demand behaviour of different
technologies.
 Further extending the engagement with major energy users to better capture any anticipated
changes in consumption and generation over a 7 year horizon.
 Investigation into using more granular profiles and energy efficiency assumptions based on
existing customer breakdown within a smaller geographic area. This area of focus will utilise
smart meter data to inform better assumptions on how customers react to price signals and
how coincident price signals are to existing times of increased network loading.
 Development of a suite of profiles further accounting for differing levels of diversity across the
network to suit all of the required analysis for strategic investment planning purposes, in line
with the creation of the Secondary System Planning team within DSO.
 Introduce industrial & commercial heat pump profiles, to enable modelling of the heat demand
from this sector.

Appendix A: Technology comparison
to Open Networks building blocks
As part of the Open Networks projects led by the Energy Networks Association, distribution network
operators proactively work with National Grid ESO (NGESO) to drive further standardisation between
Future Energy Scenarios and Distribution Future Energy Scenarios processes. As part of the work
delivered in 2020, common ‘building blocks’ were agreed between member companies. This allows for
easy comparison between NGESO and DNOs of the forecast volumes. A list of the technology types
considered in the NG DFES analysis is included below, with the relevant building block number to which
it refers.

Table 22: DFES technology to building block lookup
FES technology DFES sub-technology Equivalent Building

block ID number
Air conditioning - -
Battery storage Domestic batteries (G98) Srg_BB002

Battery storage Grid services Srg_BB001
Battery storage Co-location Srg_BB001
Battery storage High energy user Srg_BB001
Biomass & Energy Crops - Gen_BB010

(including CHP)
CCGTS (non CHP) - Gen_BB009

Electric vehicles Pure electric motorcycle Lct_BB001
Electric vehicles Pure electric car (non-autonomous) Lct_BB001
Electric vehicles Hybrid car (non-autonomous) Lct_BB002
Electric vehicles Hybrid motorcycle Lct_BB002

Electric vehicles Pure electric bus and coach Lct_BB003
Electric vehicles Pure electric LGV Lct_BB003
Electric vehicles Pure electric HGV Lct_BB003

Electric vehicles Hybrid LGV Lct_BB004
Electric vehicles Hybrid bus and coach Lct_BB004
Electric vehicles Hybrid HGV Lct_BB004

Electric vehicles Pure electric car -
(autonomous)
Electric vehicles Hybrid car (autonomous) -
EV charge point Domestic -
EV charge point Workplace -
EV charge point En route -

EV charge point Destination -
Floating wind - Gen_BB014
Geothermal - Gen_BB019
Heat pumps Electric back-up Lct_BB005
Heat pumps Gas back-up Lct_BB006
Hydrogen electrolysis - Dem_BB009

Hydrogen-fuelled generation - Gen_BB023
Hydropower - Gen_BB018

Marine Tidal stream Gen_BB017
Marine Wave energy Gen_BB017
Non-renewable engines > 1 MW Gen_BB001
(CHP)
Non-renewable engines < 1 MW Gen_BB002
(CHP)
Non-renewable engines (G98/G83) Gen_BB003

(CHP)
Non-renewable engines (non Diesel Gen_BB005
CHP)
Non-renewable engines (non Gas Gen_BB006

CHP)
OCGTS (non CHP) - Gen_BB008
Offshore wind - Gen_BB014
Onshore wind Large scale (>1MW) Gen_BB015
Onshore wind Small scale (<1MW) Gen_BB016
Other generation - -
Renewable engines (landfill - Gen_BB004
gas, Sewage Gas, Biogas)
Retained connection - -
Solar PV Ground mounted (>1MW) Gen_BB012
Solar PV Commercial rooftop (10kw - 1MW) Gen_BB013
Solar PV Domestic rooftop (<10kw) Gen_BB013

Waste Incineration - Gen_BB011
(including CHP)

Appendix B: Primary substation
clustering
Methodology
In order for a computer to imitate the behaviour of a human manually grouping profiles, Machine
learning (ML) techniques were used. ML can be defined as a field of study which allows computers to
learn without being clearly programmed to do so
Specifically, unsupervised ML techniques can be used to produce profile groups for further human
analysis. The term “unsupervised” refers to the ML algorithm working with unlabelled data. In this
context of this project, a “label” would be a descriptor for a group of profiles with similar behaviours.
This section aims to give a high-level overview of the ML techniques used to produce the profile clusters;
the reader should consult other reference material for an in-depth explanation of how the ML techniques
work.
The starting point for devising the ML clustering methodology was to consider how a human would
group similar demand profiles. An intuitive way of doing this would be to plot the demand profiles, and
match those with similar profile shapes. For example, one group may consist of profiles with morning
and evening peaks, and another group may consist of profiles with a midday peak. When creating these
groups, a human would also consider the “context” of a profile peak. Peaks do not occur at the same
time, but fall within a range. For example, the evening peak could fall between 5 pm to 7 pm.
The ML clustering is divided in to two stages:
1. Calculating a metric to determine the similarity between profiles
2. Using the profile similarity metrics to divide the profiles in to clusters
For determining the similarity of profile shapes, dynamic time warping (DTW) was used. Given a
temporal sequence, DTW calculates an optimum matching between the two sequences. Figure 61
provides an example of where DTW has mapped the minimum and maximum points of two sinusoidal
signals.
Figure 61: Dynamic time warping for two signals
This is analogous to a human matching up profiles by examining the similarity of their shapes. To
prevent DTW from excessively warping the profiles in attempt to match the profile shapes (e.g.
attempting to match evening peaks with midday peaks), a constraint was set so that the algorithm could
only match points within a 90-minute range.
The DTW algorithm was applied to all pairs of profiles. In turn, this produced a matrix containing a
measure of similarity between the profiles. This matrix was used as the input for second stage of ML
clustering. To partition the profiles in to groups using the similarity matrix, the k-means algorithm was
used. This algorithm produces k clusters by minimising the within-cluster-sum-of-squares criterion. In
other words, the algorithm aims to produce k clusters where the variance between samples inside each
cluster is minimised.

The optimum value of k was chosen by sweeping across a range of k for each season used in the
representative days and recording the sum of squared distance of samples to the closest cluster centre.
Figure 62 is an example of a plot produced after sweeping through a range of k. The optimum value of
k is selected by picking the “elbow” of the plot (i.e. the point at which the curve bends from a high slope
to a low slope). In the case of Figure 62, the elbow value would be identified at k = 5.
Figure 62: Elbow method
Table 23: Example of substation cluster similarities
Substation Name Cluster Assignment
Winter Intermediate Cool Summer Intermediate Warm

A 1 4 0 1
B 2 3 0 2
C 1 4 0 1
D 0 3 1 2
Table 23 provides an example of this analysis. As substations “A” and “C” have identical cluster
assignments, they have the same profile behaviour over the course of the year. This can be used as
an input for mapping behaviour to categorisation metrics (e.g. housing density, proportion of
domestic/commercial customers etc.). The k-means algorithm was repeated for each seasonal rating,
using the optimum value of k identified. This clustering provides a starting point for further analysis.
Given that each substation has been assigned a cluster for each seasonal rating, substations with
identical cluster assignments can be viewed as having similar behaviour. Figure 63 is an example of
the clusters that are produced using ML clustering.

Figure 63: Example of ML clustering output

Customer Behaviour
assumptions
Glossary

Acronym Term Definition
– Access Window The period of spring, summer and autumn in which arranged
outages are normally taken.
ANM Active Network The ENA Active Network Management Good Practice Guide
Management summarises ANM as:
Using flexible network customers autonomously and in real-
time to increase the utilisation of network assets without
breaching operational limits, thereby reducing the need for
reinforcement, speeding up connections and reducing costs.
ASHP Air Source Heat Type of Heat Pump that absorbs heat from outside air for the
Pump purposes of space heating and hot water.
BEV Battery Electric Electric vehicle with a battery as the only means of propulsion
Vehicle
BSP Bulk Supply Point A substation comprising one or more Grid Transformers and
associated switchgear
– Demand The consumption of electrical energy.
DSR Demand Side Ofgem led tariffs and schemes which incentivise customers to
Response change their electricity usage habits
DfT Department for The governmental department responsible for the transport
Transport network in England and part of Scotland, Wales and Northern
Ireland which are not devolved.
DFES Distribution Future An annual process undertaken by Distribution Network
Energy Scenarios Operations to forecast future growth on the distribution network
DG Distributed Generation connected to a distribution network. Sometimes
Generation known as Embedded Generation.
DNO Distribution Network A company licenced by Ofgem to distribute electricity in the
Operator United Kingdom who has a defined Distribution Services Area.
DSOF Distribution System A document published by National Grid that assesses the
Operability technical issues facing Distribution Network Operators as they
Framework transition to Distribution System Operator (DSO).
DTW Dynamic Time An algorithm to measure similarity between two time series
Warping
EV Electric Vehicle General term for a vehicle which uses electric motors as its
method of propulsion.
ESA Electricity Supply Each ESA represents a block of demand and generation as
Area visible from the distribution network. For the 2022 DFES
studies, each ESA represents the geographic area supplied by
a Primary Substation (which contains NG-owned distribution
substations) providing supplies at a voltage below 33 kV, or a
customer directly supplied at 132, 66 or 33 kV or by a
dedicated Primary Substation.
ENA Energy Networks The Energy Networks Association is an industry association
Association funded by gas or distribution or transmission licence holders.
EPC Energy Performance Rating scheme to summarise the energy efficiency of buildings.
Certificate

ER Engineering A document published by the Energy Networks Association.
Recommendation
EAC Estimated Annual An estimated rate of consumption, nominally expressed in
Consumption kWh/year that is used in electricity settlement.
FCO First Circuit Outage P2/7 defines a First Circuit Outage as:
“a fault or an arranged Circuit outage…”
Also referred to as N-1 in some contexts.
FES Future Energy A set of scenarios developed by Nation Grid to represent
Scenarios credible future paths for the energy development of the United
Kingdom.
GB Great Britain A geographical, social and economic grouping of countries that
contains England, Scotland and Wales.
GBSO Great Britain System National Grid is the system operator for the National Electricity
Operator Transmission System (NETS) in Great Britain. Responsible for
coordinating power station output, system security and
managing system frequency.
GSP Grid Supply Point A substation comprising one or more Super Grid Transformers
and associated switchgear
GSHP Ground Source Heat Type of Heat Pump that absorbs heat from the ground for the
Pump purposes of space heating and hot water.
HP Heat Pump General term for a heating system that extracts heat from
surroundings which can then be used to produce hot water or
space heating.
HGV Heavy Goods A large goods vehicle with a gross mass greater than 3500 kg
Vehicle
LGV Light Goods Vehicle Commercial vehicle with a gross mass of less than or equal to
3500 kg
LTDS Long Term A document published by all DNO’s to assist current and future
Development users of the distribution network to identify and assess
Statement opportunities available to them for making new or addition use
of the network.
ML Machine Learning A field of study which allows computers to learn without being
clearly programmed to do so
MPAN Meter Point Unique reference number used in Great Britain to identify
Administration electricity supply points, such as individual domestic
Number residences
NGESO National Grid National Grid Electricity System Operator is the electricity
Electricity System system operator for Great Britain.
Operator
NDP Network Requirement as part of Electricity Distribution Licence
Development Plan Condition 25B for Distribution Network Operators to cover the
investments planned for the next 5 to 10 year period in relation
to the 11 kV network and above.
NIA Network Innovation Funding scheme for innovation projects introduced as part of
Allowance RIIO-ED1. For the RIIO-ED1 period, NG requested the
minimum 0.5% of total regulated income.

Ofgem Office for Gas and Ofgem is responsible for regulating the gas and electricity
Electricity Markets markets in the United Kingdom to ensure customers’ needs are
protected and promotes market competition.
– Open Networks The Open Networks Project is a major energy industry initiative
that will transform the way our energy networks work,
underpinning the delivery of the smart grid. This project brings
together 9 of UK and Ireland's electricity grid operators,
respected academics, NGOs, Government departments and
the energy regulator Ofgem.
Note: Open Networks was previously known as the ENA TSO-
DSO Project.
PV Photovoltaic Type of distributed generation which uses solar irradiance to
generate electricity.
PHEV Plug-in Hybrid Electric vehicle with a battery and a supplementary engine,
Electric Vehicle which is able to run on both modes of propulsion.
– Primary Distribution The sections of an electrical distribution network which provide
the interface between transmission and Primary or Secondary
Distribution.
In National Grid’s distribution network the 33kV circuits and
Primary Substations are considered to be Primary Distribution.
– Primary Substation A substation comprising one or more Primary transformers and
associated switchgear
– Primary Transformer A transformer that steps voltage down from 66 or 33kV to 11kV
or 6.6kV
RDP Regional A study which looks at the complex interaction between the
Development Plan distribution and transmission network, also between different
distribution networks.
TDCV Total Domestic Industry standard values for the annual gas and electricity
Consumption Values usage of a typical domestic customer
V2G Vehicle to Grid Where a plug-in EV can export to the power grid.
VOA Valuation Office UK government body responsible for the valuation of properties
Agency for the purpose of council tax and for non-domestic rates in
England and Wales
NG National Grid A Distribution Network Operator (DNO) that is licenced by
Electricity Ofgem to distributed electricity in the East Midlands, West
Distribution Midlands, South West, and South Wales regions of United
Kingdom.

Table of Units
Term Definition
kV Kilovolt, a unit of Voltage (x103)

LV This refers to voltages up to, but not including, 1kV
HV Voltages over 1kV up to, but not including, 22kV
EHV Voltages over 20kV up to, but not including, 132kV
kW Kilowatt, a unit of Power (x103)
MW Megawatt, a unit of Active Power (x106)
MVA Mega volt-ampere, a unit of Apparent Power (x106)
MVAr Mega volt-ampere (reactive), a unit of Reactive Power (x106)

kWh Kilowatt hour, a unit of energy (x103). Equivalent to a constant 1kW of Active Power
delivered for an hour

Disclaimer
Neither National Grid, nor any person acting on its behalf, makes any warranty, express or implied, with respect
to the use of any information, method or process disclosed in this document or that such use may not infringe
the rights of any third party or assumes any liabilities with respect to the use of, or for damage resulting in any
way from the use of, any information, apparatus, method or process disclosed in the document.
© National Grid 2022
Contains OS data © Crown copyright and database right 2022
No part of this publication may be reproduced, stored in a retrieval system or transmitted, in any form or by any
means electronic, mechanical, photocopying, recording or otherwise, without the written permission of the DSO
Manager, who can be contacted at the addresses given below.

DFES Customer Behaviour Report

Uploaded by

Copyright:

Available Formats

DFES Customer Behaviour Report

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DFES Customer Behaviour Report

Uploaded by

Copyright:

Available Formats

Distribution

Customer behaviour profiles

Produced: February 2022

National Grid | Distribution Future Energy Scenarios 2022

Forecasting Network Impact Assessment Optioneering

The Distribution Future Energy The Distribution Network Options

National Grid | Distribution Future Energy Scenarios 2022

Part 2: Customer Behaviour

Part 3: MW growth data and application of growth rates

National Grid | Distribution Future Energy Scenarios 2022

Assets Network topology and connectivity Customers connected to the distribution

Intended use of profile assumptions

National Grid | Distribution Future Energy Scenarios 2022

National Grid | Distribution Future Energy Scenarios 2022

National Grid | Distribution Future Energy Scenarios 2022

Substation Primary Time of Peak Date of Peak Primary @

Portishead Ashlands 11kV 6.17 19:30:00 09/01/2020 5.92

Weston in Gordano 11kV 4.55 18:00:00 12/01/2020 5.01

Total 51.43 45.12

National Grid | Distribution Future Energy Scenarios 2022

2 Engineering Recommendation P2 Issue 7, Security of Supply, Energy Networks Association, 2019;

National Grid | Distribution Future Energy Scenarios 2022

Table 3: Substation seasonal peak demand and rating comparison

Demand Edge-case Month Demand (MVA) Asset Rating (MVA) Utilisation

Winter Peak February 25.4 31 81.9%

Intermediate Warm Peak October 26.2 29 90.3%

National Grid | Distribution Future Energy Scenarios 2022

Changing Nature of Demand

4Energy Networks Association, Engineering Recommendation P2 Issue 7, p9

National Grid | Distribution Future Energy Scenarios 2022

Application of customer behaviour profiles

National Grid | Distribution Future Energy Scenarios 2022

Proportion of customer demand at each

Scenario and year dependent

Unabated Profile Flexed Profile

Based on existing customer data representing Representative profile to simulate customer

No scenario or year dependency No scenario or year dependency

Unabated Profile Scaling Flexed Profile Scaling

When all profile elements are combined

National Grid | Distribution Future Energy Scenarios 2022

National Grid | Distribution Future Energy Scenarios 2022

National Grid | Distribution Future Energy Scenarios 2022

Technology Subtechnology Units used in DFES volume projections

Solar Commercial rooftop (10 kW – 1 MW of installed capacity

Ground mounted (>1 MW)

East Midlands 127 868 MVA

South West 176 823MVA

West 60 344 MVA

National Grid | Distribution Future Energy Scenarios 2022

Representative Day Profiles

National Grid | Distribution Future Energy Scenarios 2022

National Grid | Distribution Future Energy Scenarios 2022

How will these profiles change over time

National Grid | Distribution Future Energy Scenarios 2022

Technology Subtechnology Units used in DFES volume projections

Wind Onshore Wind <1MW MW of installed capacity

Subtechnology Licence Area Number of sites in Total installed

Onshore Wind East Midlands 80 25 MVA

South West 150 34 MVA

West Midlands 28 73 MVA