Download presentation
Presentation is loading. Please wait.
Published byGladys Melton Modified over 9 years ago
1
Creating LV network templates Gavin Shaddick 5 th November 2012
2
Outline Why do we need LV templates Monitoring and data Statistical analysis Overview Clustering methodology Classification Results from summer 2012 Future developments
3
Acknowledgements Electrical and electronic engineering – Furong Li – Chenghong Gu – Ran Li Mathematical Sciences – Haojie Yan WPD – Mark Dale
4
Why do we need LV Network Templates? New Challenges Lack of real data/ understanding of the LV Network Assets on the network are expensive Growing consumer market more connections Consumer Prosumer adopting more green technology e.g. PV, SHP LV Network Templates
5
Monitoring and Data
6
Data Substation SMOS Fixed data Generation Wales Arbed PV
7
Sub-station Enclosure EDMI Customer Monitoring Ground Mounted Pole Mounted Street Furniture: Haldo Pillar How are we monitoring? 800 Sub-station Monitors 3500 Voltage Monitors 120 PV Monitors
8
138.38.106.138 Substation data – May – June: Emailed zip files (~380 sites) – July 4 th – present sftp transfer (~800 sites) SMOS data – July 16 th - present sftp transfer (~3500)
9
Substation data availability over period 16 th April – 17 th September Blue lines represent times when a substation supplying data Yellow lines represent dates without data.
10
SMOS data Voltage data over period Monday 23 rd July – Sunday 30 th July, 2012 TOTAL_CUSTOMERS: 197 PROFILE_1_CUSTOMER_COUNT : 136 PROFILE_2_CUSTOMER_COUNT : 49 PROFILE_3_CUSTOMER_COUNT: 9 PROFILE_4_CUSTOMER_COUNT: 2 PROFILE_5_CUSTOMER_COUNT: 0 PROFILE_6_CUSTOMER_COUNT: 0 PROFILE_7_CUSTOMER_COUNT: 0 PROFILE_8_CUSTOMER_COUNT: 1
11
PV data Left: Pattern of generated energy (units on y-axis: kWh) by 10 minutes interval over 24 hours period recorded at PV monitor on 4 selected dates: (a) 23 rd July; (b) 28 th July; (c) 7 th Aug. (d) 14 th Aug. Right : Records of sunshine hours (on y-axis, units: number of hours) at St. Athan (Wales) weather monitoring station (http://www.weatheronline.co.uk/) covering period from 6 th July to 31 st Aug. 2012.http://www.weatheronline.co.uk/
12
Fixed data Primary station number Primary station name HV feeder index Substation number Substation name Transformer Transformer type Rating LV feeder Grid reference Total number of customers Number of customers in profiles 1-8 …
13
Generation Wales data Substation number Site name Feeder number Generator type Number of phases Installed size kW Number of units Installed size*Number of units
14
ARBED Delivery partner Town Local authority post code LSOA Standard property type Standard construction type Age band Wall construction Wall insulation Built form Detachment position Mains Gas Available Primary fuel type Standardised fuel type Arbed package (values: combination of ‘SWI’, ‘PV’, ‘SHW’, ‘ASHP’, ‘FS’, ‘CESP’, ‘CERT’*, where SWI = Solid Wall Insulation, PV = Photo-Voltaic, SHW - Solar Hot Water, ASHP = Air Source Heat Pump, FS = Fuel Switching, CESP = Community Energy Savings Programme, CERT = Carbon Emission Reduction Target) Arbed SWI Delivery partner Measure Completion Date or Proposed Date
15
Statistical analysis
16
Statistical analysis: Overview Develop network templates Use statistical clustering techniques group sub-stations based on load profiles Overlay fixed data onto resulting clusters Create set of classification rules Using three months of data we have already started to identify clusters which form basis of templates Example of data received over 24 hour period (Wed. 25 th July)
17
Statistical methodology: clustering Based on (dis)similarities in the data Data structured according to - Time (within days, 10 min intervals) - Date (days, months, season) - Sub-stations Real Power Delivered (RPD) Allocates ‘units’ to groups (clusters)
18
Dendrogram Black vertical lines indicate how sub-station clusters join together Highest level split is commercial (left) and residential (right) Further splits are based on magnitude (RPD) and temporal patterns
19
Determining number of clusters Decreasing curve of within groups sum of squares (y-axis) against increasing number of possible clusters (x-axis) by using K-means.
20
Classification 1.Domestic Unrestricted (single rate) 2.Domestic Economy 7 (two rate) 3.Non-Domestic Unrestricted (single rate) 4.Non-Domestic Non-Maximum Demand Economy 7 type (two rate) 5.Non-Domestic Max. Demand Customers with Load Factor 0-20% 6.Non-Domestic Max. Demand Customers with Load Factor 20-30% 7.Non-Domestic Max. Demand Customers with Load Factor 30-40% 8.Non-Domestic Max. Demand Customers with Load Factor >40% Sub-station clusters contain a mix of customer profile classes
21
Clusters based on overall levels of RPD (kW) and different temporal patterns Sub-station profiles over time (within day)
22
Sub-station profiles over time (weekly) Clusters exhibit different patterns of RPD (kW)
23
Future developments Primary data is being continuously received Update clustering Refine classification rules Create a set of LV templates Identify stresses on LV networks due to low carbon technologies Liaise with DNOs to discuss LV Network Templates and application in other areas
24
Questions?
Similar presentations
© 2024 SlidePlayer.com Inc.
All rights reserved.