Update on Decision Support Tools

0 downloads 0 Views 35MB Size Report
Actual Operator Decisions. Items that need to be continually monitored: Fault Correlation Reporting system must encompass: • how accurate is RTF for each OA/ ...
Update on Decision Support Tools Operations Management IPT update



Real Time Operations • CAP Tool • Fault Correlation



Next Steps for Steering Committee • Bus Fault Correlation • Action Tracker for CAP • Fault Correlation Reporting Analyzer • Integrated T&D Top Down Display

Machine Learning and CAP tool results for June 9-11, 2008 heat wave

Summary: 10 of 33 OA’s were RED (10% worst) in CAP 2 of 33 OA’s were GREEN (10% best) in CAP Random would have given 3 red and 3 green 6 SwitchChecks identified 3M transformer cooled 3Q transformer replaced on OOE2 6B PVL error and load pocket problem identified

Cumulative OA’s

ROC Heat Wave 1 Summer 2008 Machine Learning Citywide Excellent Prediction Performance from MartaRank

Feeder Percentile 08067-11

ML Attributes contributing to the excellent performance

XLP Joints PQ Infant Mortality Length Bad Xfmrs PQ Load Pocket Problems Load

Heat Wave 1 08067-11 ML M Network Average FS Yorkville #1

3M

FS

1M 2M Cleanup Moves

1M 2M 3M

Heat Wave 1 08067-11 ML M Network Average FS Yorkville Trouble Feeders

FS

3M50 3M60 3M63

3M52

3M43

3M PQ Chart Before and After Heat Wave 1 080609-11 Overvoltage Undervoltage

CAPT tool PILC replacement costs and MTBF improvement

PILC Replacement alone in Yorkville not enough. 3M needs Load Pocket work as well.

Since Heat Wave 1 08067-11 ML M Network Average FS Yorkville #1

3M 1M

FS

2M Cleanup Moves

1M 2M 3M

Post Heat Wave 1 08067-11 ML M Network Average FS more 2M Harlem Cleanup Moves

2M33

FS

2M19

Cumulative OA’s

Heat Wave 1 08067-11 ML X MartaRank

1X24 3X66

Feeder Percentile

Heat Wave 1 08067-11 ML X

1 in 4 OA’s were 10% worst FS

Cumulative OA’s

Heat Wave 1 08067-11 ML M MartaRank

11M48 3M52 3M43 3M43 3M60 3M63

Feeder Percentile

SwitchChecks Issued

CAP Yorkville 3rd

V5258 (3M49) no switch check

V2421 (3M50) – switch check V1567 (3M49) – switch check

SwitchChecks Issued

CAP Yorkville 4th

V5868 (3M42) switch check

V2421 (3M50) switch check

V5258 (3M49) no switch check

TLV

V5258 (3M49) no switch check

CAP Yorkville 3rd

V2421 (3M50) crew checked volt, load & temp. Unit left with NWP closed, temp 80 deg

V1367 (3M50) closed switch

V5868 (3M42) cooled

Action Tracker Transformer Load Variance >20% over expected load Nearby Open Switches or cool unit check what

description 1 Blown_Fuse_A 2 Blown_Fuse_B 3 Blown_Fuse_C 4 Open_NWP 5 Bad_RMS_A 6 Bad_RMS_B 7 Bad_RMS_C 8 HighWater 9 Unit Not Rptg 10 Oil Smpl Rqd

Field actions

description

1 Morning Switch Check 2 Critical Customers 3 Overloaded Transformer 4 Cooling/LVT 5 Outage Scheduling System 6 When Ready/OOE 7 Backfeed 8 Switch Repair 9 Other 10 Susceptibility 11 HighWater 12 Transformer Over-Temp

Closed Switch Cooled Unit Relieved Load Pocket

Cumulative OA’s

Heat Wave 1 08067-11 ML BQ MartaRank

6B53 6B54 3Q84

Feeder Percentile

Heat Wave 1 08067-11 ML BQ

ODDS: 1 in 5 OA’s were 10% worst FS

Since Heat Wave 1 08067-12-26 ML BQ 2 in 5 OA’s were 10% worst FS

Cumulative OA’s

ROC Heat Wave 1 Summer 2008 BQ ODDS

Feeder Percentile

08067-11

Cumulative OA’s

Since Heat Wave 1 08067-12-26 ML BQ MartaRank

Feeder Percentile

Feeder Percentile

CAP

V9922 switch check. Pressure below min. T=130 deg

Rego Park 6-09 6-10

3Q83 OOE2 Replaced Xfmr in v9922 on 6-11 6-11

6-12

CAP

PVL problem

Load problem

CAP LIC 080527

78 0 0

TLV VS326 LIC 080527 Open switch OA

% load kva

WR

Transformer Overloaded and Switchcheck issued to close Nearby Open Switch

Switch Closed on V558

Hipot Performance

Reg Hipot Performance 2008

Mod Hipots have been the problem

Bus Fault Correlation

Fault Correlation Reporting System Items that need to be continually monitored: • • • • •

Automated inputs from FMS, Rapid Restore, PI, PQNode Events and Microprocessor Relays Manual Inputs including from Rapid Restore and STAR System Protective Devices Recommended Decisions Actual Operator Decisions

Fault Correlation Reporting system must encompass: • how accurate is RTF for each OA/CIOA • how quickly are operational information and recommended decisions posted • how quickly do relay targets get reported • how many times are recommended decisions followed • how many times are OA/CIOA feeders successfully restored

Fault Correlation Reporting System RTF Structure/Visible Fault Structure Match – Y/N RTF Structure/Damage Structure Match – Y/N Variance- OA/CIOA Time to RTF/Inrush/Subcycle Posting Time RTF Fault Impedance/Damage Fault Impedance – Percentage (RTF Fault Impedance-Visible Fault Impedance)/Visible Fault Impedance (RTF Fault Impedance-Damage Fault Impedance/Damage Fault Impedance Visible Fault Impedance/Total Feeder Impedance (RTF Fault Impedance-Visible Fault Impedance)/Total Feeder Impedance Visible Fault Impedance/Total Feeder Impedance Target Phase/RTF Phases Match – Y/N Inrush-Feeder MVA>32 – Y/N Percentage RTF Structure/Visible Fault Structure Match Percentage RTF Structure/Damage Structure Match Average Variance -OA/CIOA Time to RTF/Inrush/Subcycle Posting Time Average RTF Impedance/Visible Fault Impedance-Percentage Average RTF Fault Impedance/Damage Fault Impedance-Percentage Average - (RTF Fault Impedance-Visible Fault Impedance)/Visible Fault Impedance Average -(RTF Fault Impedance-Damage Fault Impedance)/Damage Fault Impedance Average - Visible Fault Impedance/Total Feeder Impedance Average - Target Phase matches RTF Phase

This chart demonstrates that for the same 75 events where there was no more than an 18 second variation between the RTF timestamp and the PI breaker status change timestamp, the variation between the FMS/Rapid restore timestamp and the RTF timestamp varied from over 43 seconds to over 7 minutes. This is the type of detailed analysis which is needed to continuously improve the Fault correlation application.

Fault Impedances versus Total Feeder Impedances determine trends associated with location of feeder faults relative to substation for closing the feedback loop to produce more accurate RTF predictions Variance – CIOA/OA Time-Target Posting Time Variance – CIOA/OA Time – Damage Time Variance – CIOA/OA Time – Visible Fault Time Variance – CIOA/OA Time – Recommended Action Posting Time These metrics are designed to track the performance of the manual inputs which lead up the recommended decisions and ultimately full fault correlation and are designed to focus the operators on areas of improvement in utilizing the system. Recommended Action Matches FMS/Rapid Restore Action – Y/N Recommended Action Time vs. Actual Action Time These metrics are designed to track the design of the fault correlation decision matrix versus actual operator action and should point out areas for improvements in the matrix and ultimately the areas for improvements with individual operators.

Top Down Display An Integrated High Level Overview of the Transmission and Distribution Systems

Display Attributes • • • • • • • •

Integrates Transmission and Distribution High Level Display Event Driven Zoom In and Drill Down Capability 3D Rendering Animation Color Coding Icon Representation

High Level Trouble Indicator

Top Down Display

Heat Mapping

Integrates Transmission and Distribution • • • • • •

Present anomalies for subsystems: Transmission stations Area Substations Unit Stations Auto Loops Networks ATS Systems

Event Driven

• • • • •

Would provide high level indications: Load Shedding Voltage Reduction Low Voltage Transmission feeder and transformer overloads Area Station Feeder and transformer overloads

Event Driven • Transmission feeder, breaker and transformer trip outs • Area station trip outs of transformers, buses and feeders • Network contingencies • Auto-loop interruptions • Network Transformer Overloads

Zoom In/Drill Down Capability • Poke on event indication and bring up subsystem that provides operational detail. • All anomalies registering within a subsystem not represented preventing Display from being overloaded. • Only the more critical listed events would be included

3D Rendering • Present a geographical 3D representation of the Con Edison service Territory (i.e. Google) with Zoom in. Also provide switching to a 2D view for more clarity. • Drill down to sub-systems would be available from both the 3D and 2D modalities

Animation/Color Coding • Animation is essential to present events in easy to recognize formats. • Animation together with color coding conveys both specific area experiencing an anomaly and severity.

Icon Representation • Develop specific Icons for represented systems • Icons would be animated and color coded • Icons would be the poke points for sub-system drill down