management in smart grids
play

Management in Smart Grids Hoang Hai Nguyen 1 Rui Tan 1 David K. Y. - PowerPoint PPT Presentation

Safety-Assured Collaborative Load Management in Smart Grids Hoang Hai Nguyen 1 Rui Tan 1 David K. Y. Yau 2,1 1 Advanced Digital Sciences Center, Illinois at Singapore 2 Singapore University of Technology and Design Overloaded Grid is Unsafe


  1. Safety-Assured Collaborative Load Management in Smart Grids Hoang Hai Nguyen 1 Rui Tan 1 David K. Y. Yau 2,1 1 Advanced Digital Sciences Center, Illinois at Singapore 2 Singapore University of Technology and Design

  2. Overloaded Grid is Unsafe

  3. Overloaded Grid is Unsafe • Loss of generation – Unexpected failures

  4. Overloaded Grid is Unsafe • Loss of generation – Unexpected failures normal Time

  5. Overloaded Grid is Unsafe • Loss of generation – Unexpected failures cascading normal failure Time overloaded grid

  6. Overloaded Grid is Unsafe • Loss of generation – Unexpected failures • Transmission line short circuit – Hits by overgrown trees (2003 Northeast Blackout)

  7. Overloaded Grid is Unsafe • Loss of generation – Unexpected failures • Transmission line short circuit – Hits by overgrown trees (2003 Northeast Blackout)

  8. Overloaded Grid is Unsafe • Loss of generation – Unexpected failures • Transmission line short circuit – Hits by overgrown trees (2003 Northeast Blackout) cascading trip

  9. Existing Solution: Load Shedding • Disconnect some loads – When demand surges or failure detected – Resilient to (remaining) credible contingencies • Unfair, uncomfortable

  10. New Opportunity: Load Curtailment Large commercial and industrial curtailment Residential air conditioner moderated by programs [CenterPoint Energy] real-time electricity price [ComEd Illinois] • Collaborative load curtailment – Fair, less painful – Untrustworthy (human factors, huge # of edge devices) • Handle overload using curtailment with safety assurance?

  11. Approach Overview Safety Assessment V A How far from unsafe? A far No action

  12. Approach Overview • Close to unsafe – Load curtailment Safety Assessment V A How far from unsafe? A ≤ 3 MW ≤ 20 KW ≤ 5 KW far close ≤ 6 KW ≤ 2 MW ≤ 3 KW No action Load ≤ 1 MW ≤ 6 KW curtailment

  13. Approach Overview • Close to unsafe – Load curtailment • Already unsafe – Load shedding Safety Assessment V A How far from unsafe? A ≤ 3 MW unsafe ≤ 20 KW far close ≤ 3 KW No action Load Load ≤ 1 MW ≤ 6 KW shedding curtailment

  14. Challenges • Existing grid safety assessment tools – Time-domain simulators [PowerWorld] Slow! – Learning-based classifiers [Sun 2007, Amjady 2007] “Safe” or “unsafe” for triggering shedding

  15. Challenges • Existing grid safety assessment tools – Time-domain simulators [PowerWorld] Slow! – Learning-based classifiers [Sun 2007, Amjady 2007] “Safe” or “unsafe” for triggering shedding • Curtailment needs time to take effect – Too late to trigger curtailment if already unsafe – Predictive assessment needed

  16. Challenges • Existing grid safety assessment tools – Time-domain simulators [PowerWorld] Slow! – Learning-based classifiers [Sun 2007, Amjady 2007] “Safe” or “unsafe” for triggering shedding • Curtailment needs time to take effect – Too late to trigger curtailment if already unsafe – Predictive assessment needed • Safety: non-linear – Curtailment scheduling repeatedly invokes assessment – Rapid assessment needed

  17. Outline • Motivation, Approach Overview • Rapid and Predictive Grid Safety Assessment • Predictive Curtailment Scheduling • Simulations

  18. Background of Safety Assessment • Grid is safe if safety condition is met when contingency happens – Safety condition Example : All generators’ speed within (55 Hz, 62 Hz) – Contingency Example 1: Most overloaded line trips Example 2: Any single line trips • Safety depends on grid state – Load (dominating)

  19. Background of Safety Assessment • Grid is safe if safety condition is met when contingency happens – Safety condition Example : All generators’ speed within (55 Hz, 62 Hz) – Contingency Example 1: Most overloaded line trips Example 2: Any single line trips Basic requirement: Tolerate loss of any single line • Safety depends on grid state – Load (dominating)

  20. An Example G Load bus 8 transformer G Load bus 6 Load bus 5 G IEEE 9-bus system • Safety assessment – Contingency: short circuit on a line

  21. An Example G Load bus 8 transformer G Load bus 6 Load bus 5 Bus6 demand (MW) G Time-domain simulation result IEEE 9-bus system (Bus5 demand fixed) • Safety assessment – Contingency: short circuit on a line

  22. An Example G Load bus 8 unsafe transformer G safe Load bus 6 Load bus 5 Bus6 demand (MW) G Time-domain simulation result IEEE 9-bus system (Bus5 demand fixed) • Safety assessment – Contingency: short circuit on a line – Safety condition: speed dev < 3 Hz

  23. An Example G Load bus 8 unsafe transformer now G safe Load bus 6 Load bus 5 Bus6 demand (MW) G Time-domain simulation result IEEE 9-bus system (Bus5 demand fixed) • Safety assessment – Contingency: short circuit on a line – Safety condition: speed dev < 3 Hz • A grid becomes unsafe if demands increase – How much time from now?

  24. Time to Being Unsafe (TTBU) • TTBU is minimum time t grid with demand D + Δ ( t ) is unsafe max demand vector of buses’ increment over demands time period t

  25. Time to Being Unsafe (TTBU) • TTBU is minimum time t grid with demand D + Δ ( t ) is unsafe max demand vector of buses’ increment over demands time period t Δ( t ) for 3 load buses learned from New York ISO load data June-July, 2012 t (minute)

  26. Time to Being Unsafe (TTBU) • TTBU is minimum time t grid with demand D + Δ ( t ) is unsafe max demand vector of buses’ increment over demands time period t Δ( t ) for 3 load buses learned from New York ISO load data June-July, 2012 t (minute) • Predictive but compute-intensive safety metric – Run PowerWorld for each t 15 secs for 37-bus system on 4core @ 2.8GHz

  27. ELM-Based Assessment • Extreme Learning Machine [Huang 2006] – Neural network with one hidden layer • Training data set {<demand vector, TTBU>} – Demand history – TTBU from offline time-domain simulations

  28. ELM-Based Assessment • Extreme Learning Machine [Huang 2006] – Neural network with one hidden layer • Training data set {<demand vector, TTBU>} – Demand history – TTBU from offline time-domain simulations true value ELM Time (hour) avg err = 0.9% 10 5 x speed-up 37-bus system

  29. Outline • Motivation, Approach Overview • Rapid and Predictive Grid Safety Assessment • Predictive Curtailment Scheduling • Simulations

  30. Load Curtailment Scheme Demand at a bus Time TTBU safeguard threshold Time

  31. Load Curtailment Scheme Demand at a bus Time TTBU safeguard threshold Time

  32. Load Curtailment Scheme Load curtailment phase Demand at a bus Time Load curtailment phase TTBU safeguard threshold Time

  33. Load Curtailment Scheme Load curtailment phase desired demand Demand at a bus Time Load curtailment phase TTBU safeguard threshold Time

  34. Load Curtailment Scheme Load curtailment phase desired demand Demand } curtailment at a bus demand ceiling Time Load curtailment phase TTBU safeguard threshold Time

  35. Load Curtailment Scheme Load curtailment phase Demand at a bus demand ceiling Time Load curtailment phase TTBU safeguard threshold Time

  36. Load Curtailment Scheme Load curtailment phase Demand at a bus Time Load curtailment phase TTBU safeguard threshold Time

  37. Load Curtailment Scheme Load curtailment phase Demand at a bus Time Load curtailment phase TTBU safeguard threshold Time Unsafe!

  38. Load Curtailment Scheme Load curtailment phase Load shedding phase Demand at a bus Time Load curtailment phase Load shedding phase TTBU safeguard threshold Time Unsafe!

  39. Demand Prediction Model • Strong temporal correlation – One-step prediction ˆ   d f ( d , d , , d )    1 0 1 R 1

  40. Demand Prediction Model • Strong temporal correlation – One-step prediction ˆ   d f ( d , d , , d )    1 0 1 R 1 – Recursive prediction at horizon h ˆ ˆ ˆ    d f ( d , , d , d , , d )    h h 1 1 0 R h

  41. Demand Prediction Model • Strong temporal correlation – One-step prediction ˆ   d f ( d , d , , d )    1 0 1 R 1 – Recursive prediction at horizon h ˆ ˆ ˆ    d f ( d , , d , d , , d )    h h 1 1 0 R h New York ISO data f (·) = autoregressive model Cycle = 10 min R = 12 Prediction horizon h

  42. Demand Prediction Model • Strong temporal correlation – One-step prediction ˆ   d f ( d , d , , d )    1 0 1 R 1 – Recursive prediction at horizon h ˆ ˆ ˆ    d f ( d , , d , d , , d )    h h 1 1 0 R h New York ISO data f (·) = autoregressive model Cycle = 10 min R = 12 avg err = 1.3% at 1 hour horizon Prediction horizon h

  43. Curtailment Scheduling • Find curtailments { x 1 , x 2 , …, x H }   H  | TTBU safeguard | min. h 1 h

  44. Curtailment Scheduling • Find curtailments { x 1 , x 2 , …, x H } Predicted TTBU at horizon h   H  | TTBU safeguard | min. h 1 h

Recommend


More recommend