Turitsyn_MS78-3_5-24-11_3-50pm

Embed Size (px)

Citation preview

  • 7/27/2019 Turitsyn_MS78-3_5-24-11_3-50pm

    1/18

    Cascading dynamics of power grid networks

    R. Pfitzner1,2

    , K. Turitsyn3,2

    , M. Chertkov2

    1 ETH Zurich, 2 LANL, 3 MIT

    May 24, 2011

    Konstantin Turitsyn Cascading failure dynamics

    http://find/
  • 7/27/2019 Turitsyn_MS78-3_5-24-11_3-50pm

    2/18

    Outline

    Motivation: basic facts about cascading failures

    Dynamical model

    Results: effect of load fluctuations (arXiv:1012.0815)

    Results: mitigation techniques (arXiv:1104.4558)

    Path forward

    Konstantin Turitsyn Cascading failure dynamics

    http://find/http://goback/
  • 7/27/2019 Turitsyn_MS78-3_5-24-11_3-50pm

    3/18

    Cascading failures

    Cascades in power grids:About 20 large blackouts (>300MW ) happen every year.

    Annual damage to US economy around $100 billions.

    Power-law probability distribution, i.e. enhanced probability of

    large cascades.Will get only worse: more demand, intermittent generation.

    Unlike epidemic cascades

    Non-local response (failure of one component affect all the

    network).Hierarchical (i.e. not scale-free) networks.

    Konstantin Turitsyn Cascading failure dynamics

    http://find/
  • 7/27/2019 Turitsyn_MS78-3_5-24-11_3-50pm

    4/18

    Model

    Potential flow on the graphs:pij- flow over the edge (ij):

    xijpij=i j

    gi - generator nodes (sources)di - load nodes (sinks)

    Kirchchoff Law:

    ji

    pij= gi, i Gg

    di, i Gd0, i G0\ (Gg Gd)

    G 1

    G 2

    3

    4

    5

    6

    7

    8

    10

    11

    12

    G 13 14

    15

    16

    17

    18

    19

    20

    G 22

    G 23

    24

    25

    26

    G 27

    28

    29

    30

    9 21

    Konstantin Turitsyn Cascading failure dynamics

    http://find/
  • 7/27/2019 Turitsyn_MS78-3_5-24-11_3-50pm

    5/18

    Tripping events

    Each line and generator have limited capacities:

    |pij|

  • 7/27/2019 Turitsyn_MS78-3_5-24-11_3-50pm

    6/18

    Algorithm: load fluctuations

    Loads are assumed to be random and increase in linear fashion:

    di =d0i + dit.

    with d0i chosen from optimal power flow solution anddibeing an

    i.i.d random variable with variance d0i .

    At each step the moment of next overloading event is calculated:

    t = min(1, tgen, tline)

    tgen= min

    iG

    ggmaxi g

    0i

    gi

    tline= min(i,j)

    pmaxij p

    0ij

    pij

    Konstantin Turitsyn Cascading failure dynamics

    http://find/http://goback/
  • 7/27/2019 Turitsyn_MS78-3_5-24-11_3-50pm

    7/18

    Optimal Power Flow: d0, g0

    choose d

    Identify islands

    t = min(1, tl, tp)

    dt =d0 +td

    Droop Control Identify islands

    Power Flow cut one overloaded line

    Line Check

    Finish

    no violation and t = 1

    violation

    no

    violation

    and

    t