Breakout Session 1.5 Innovation in Electricity Network Design LCNI - PowerPoint PPT Presentation

Breakout Session 1.5 Innovation in Electricity Network Design LCNI Conference Wednesday 6 December 2017 1

The ATLAS project (Architecture of Tools for Load Scenarios) Dr Rita Shaw Model Development Lead 2

Future demand is uncertain Load may rise... ... and it may fall 3

Objectives of our work Enabling good Credible demand decisions about and generation solutions to Support scenarios, reflecting capacity problems, uncertainty well-justified and informed strategic planning of dialogue network capacity Tailored to our with National Grid region, assets and and other data stakeholders 4

This presentation Overview of the ATLAS New approach to MW project (P) forecasting New approach to MVAr Next steps (Q) forecasting 5

Two NIA projects on load scenarios Demand Scenarios with ATLAS Electric Heat and Commercial (Architecture of Tools for Load Capacity Options Scenarios) Winter / summer peak load Half-hourly (hh) through year Heat pumps & air con Demand & generation The Real Options CBA model Seasonal peak and min P (MW) & Q (MVAr) Nov 2015 – December 2017 April 2015 - October 2016 6

ATLAS scope Full half-hourly view of true MW demand MW scenarios Prototype tools learning from the for GSP, BSP and Demand Scenarios NIA, Primary with more customer detail scenarios MVAr scenarios learning from REACT NIA, for whole DNO network 7

ATLAS – demand definitions Measured demand True demand Latent demand Loads DG units 8

ATLAS – true demand Effects of DG Measured Monitored on reducing Non- True demand demand DG exports customer monitored DG demand Monitored component of true demand Latent demand 9

Data processing - monitored component Data corrections Identification of data problems (half-hourly & daily analyses) See detailed methodology at www.enwl.co.uk/atlas 10

Aggregated MW demand across GSPs Peak true demand (23/11/2016) Min true demand (05/07/2016) ( ) 4500 4000 3500 3000 3000 P (MW) 2500 2500 2000 2000 P (MW) 1500 1500 Generation 1000 1000 Generation Measured Demand 500 500 Measured Demand 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 time (hr) time (hr) Latent demand varies over time 11

Substation-specific weather correction Correlate daily weekday demand over five years, with temperature and daylight hours Scale half-hourly demand to the historic temperature range of that month 12

MW forecast model per G&P substation Integrated Scenarios Working with Baseline of Model on FY17 scenarios presenting Element processed baseline used approach peak/average/ Energy, half hourly (hh) for 2017 for all GSPs, min diurnal extending their true demand + scenarios BSPs and profiles work with database of primary of demand and UKPN and NPG installed DG substations generation 13

MW forecast approach Underlying demand based on 35 customer archetypes matched to substations Efficiency, demographics, economic activity Energy Storage Demand Technologies Generation Technologies Technologies Domestic storage Electric vehicles Solar PV (with solar PV) Heat pumps I&C storage behind the Wind (domestic and I&C) meter Air conditioning Micro and larger CHP Frequency response (domestic and I&C) Flexible generation Other generation 14

What does ATLAS add? 1 2 3 4 Full views of Not just peaks - New weather- New long-term true demand and 48hh per day correction MW forecast latent demand, approach approach linked to measured demand 5 6 7 Add connections New time-series Combine MW and activity MVAr forecast MVAr to meet all approach with reporting and network modelling planning needs All prototype development in 2017 – transfer to BAU in 2018 15

2017 peak true demand scenarios Using the ATLAS prototype approach 6,500 Green Ambition Active Economy MVA Peak Demand Scenarios 6,000 Central Outlook Focus on Efficiency 5,500 Slow Economy 5,000 4,500 4,000 FY17 FY18 FY19 FY20 FY21 FY22 FY23 FY24 FY25 FY26 FY27 FY28 FY29 FY30 FY31 2017 2024 2031 Long-term scenario adjusted for known major demand projects 16

Use scenarios to make decisions Inputs Calculations Summary metrics Cost and risk Site demand scenarios Repeat analysis for distributions Strategy A Choose timescale etc. and Strategy B Define strategies with up to 3 interventions, including post- fault DSR 17

Why forecast reactive power? Source: NG SOF 2016 Declining minimum Q (MVAr) demand from distribution  High voltage problem on transmission network Develop ATLAS method to put scale on future Q exports to transmission 18

Simplified view of MVAr (Q) flows Empirical Rule: Q GSP = Q primaries + Q EHV - absorbed - Q EHV-gains = Q EHV-absorbed - Q EHV-gains I 2 X V 2 C’ℓω Q primaries 19

ATLAS Q Forecasting method Historical Q/P-ratio at primaries Primary true P demand Primary latent P demand Primary latent P demand Historical Q/P-ratio at primaries Primary true P demand (linear fitting of seasonal trends per GSP) (scenario results) (scenario results) (scenario results) (linear fitting of seasonal trends per GSP) (scenario results) Primary Primary Substations Substations Future measured Q demand at primary substations Future measured Q demand at primary substations EHV generation EHV demand of large customers (P and Q of existing DG (P and Q demand of existing load & scenario results for P) & scenario results for P) GSP & BSP substations EHV Network Component Future measured Q demand at GSPs and BSPs Empirical or modelled approach? 20

Q forecasting – empirical rule Q absorption → 100 EHV absorbed reduced for more EHV gains Primaries lightly loaded EHV, but 75 not for reverse flows 50 Q gains → increased when more cables or Q(MVAr) 25 higher voltage targets are used 0 Q at primaries → -25 more capacitive primaries (declining -50 Q/P trends) 0 1000 2000 3000 4000 5000 6000 7000 8000 time(hr) 21

Q forecasting – network modelling Kearsley 132 GSP 500 simulation NG data Network Modelling CLAVA 400 Time-series P(MW) 300 analyses (i.e. daily simulation using 200 operational 100 aspects) Validation using historical network and 0 50 100 150 200 250 300 350 half-hourly monitoring data time(hr) REACT approach... 100 but with enhanced 50 inputs 0 P and Q profiles at Q(MVAr) primaries (and BSPs -50 for large customers) -100 -150 0 50 100 150 200 250 300 350 time(hr) 22

Central Outlook scenario, avg DG output , minimum Q demand = max Q exports sum of min Q at GSPs -500 Q exports in this scenario: Q(MVAr) +5% in 5 years -1000 min Q Q at min P +11% in 10 years +83% in 35 years -1500 5 10 15 20 25 30 35 year (starting from FY17) 2 But... in reality max Q 2 0 1 7 (pu exports) min Q (max Q exports) exports could be even higher 1.5 in different scenario and with different generation output Q/Q 1 5 10 15 20 25 30 35 year (starting from FY17) 23

Future application of the ATLAS methods By 2020: So next year we will: And in FY20 we will: NG as SO will use powers Use 2018 scenarios to Use 2019 scenarios to under RfG / DCC to set Q estimate max Q exports at estimate max Q exports at export limits at GSPs, via GSPs GSPs expanded NOA process Request NG’s expected Q Compare max Q exports Could add export limits at GSPs / in our scenarios to limits significant costs on DNOs compare to Q export per GSP in ED2 period scenarios Create high-level Scope interventions to intervention programme alter max Q in ED2 for ED2 WJBP 24

Final months of the project Scope approach Available for secondary Transition G&P capacity for networks, build approach to generation on improved BAU, but keep Thermal and baseline data in under review fault level new NMS 25

For more information www.enwl.co.uk/innovation e innovation@enwl.co.uk 0800 195 4141 @ElecNW_News linkedin.com/company/electricity-north-west facebook.com/ElectricityNorthWest youtube.com/ElectricityNorthWest Please contact us if you have any questions or would like to arrange a one-to-one briefing about our innovation projects 26