Changing Electricity System Fabian Hinz 15th IAEE European - PowerPoint PPT Presentation

Faculty of Business and Economics , Chair of Energy Economics, Prof. Dr. Möst The Influence of Voltage Stability on Congestion Management Cost in a www.ee2.biz Changing Electricity System Fabian Hinz 15th IAEE European Conference Vienna, September 2017

1 Motivation 2 Model Development 3 Staus Quo 2014 4 Future scenario 2025 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 2

Congestion management Challenge Current / Real power causes high cost Development of congestion mgmt. cost, causes 750 Congestion mgmt. cost Curtailment [mio. EUR] Redispatch & Countertrading 269 198 164 159 58 45 30 32 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Direct flow from North to South Load vs. generation  Power flows from North to South Loop flows cause loop flows via Eastern Europe  Phase shifting transformers being Load distribution Wind distribution installed Loop flow via  Load concentrated in the South and West PL, CZ and AT  Wind concentrated in the North Source: BNetzA Monitoring Reports 2007 - 2016 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 3

Availability of reactive power Challenge Voltage / Reactive power in the transmission grid declines Reactive power supply: conventional and future scenario Conventional supply Future supply Transmission grid 110 kV grid Medium / low Reactive power Reactive power consumption consumption voltage grid 8% 13% Reactive power supply  Conventional supply through 28% 42% 62% large power plants Electricity  Availability in the feed-in 59% 51% transmission grid decreases 38%  Supply can be replaced by RES in the distribution grid 2014 2025 2035 Source: Kraftwerksliste BNetA 2015, Netzentwicklungsplan 2015 Controllable reactive power Offshore TSO DSO 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 4

Redispatch measures conducted in order to solve current and voltage problems Current- and voltage-induced redispatch Real power Reactive power Current-induced redispatch Voltage-induced redispatch Voltage too low Voltage okay Q Current too high expan- expan- cheap cheap sive sive Situation Power plant 1 Power plant 2 Power plant 1 Power plant 2 not dispatched fully dispatched not dispatched Reactive power Voltage okay Voltage okay Q Q Current okay expan- expan- cheap cheap sive sive Redispatch Power plant 1 Power plant 2 Power plant 1 Power plant 2 ramped up ramped down Reactive power Reactive power More expansive power plant ramped up in More expansive power plant ramped up in order to alleviate transmission line order to provide reactive power  Redispatch cost 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 5

1 Motivation 2 Model Development 3 Staus Quo 2014 4 Future scenario 2025 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 7

Redispatch cost calculated in a 3-step approach Model approach  Electricity market model (copper plate) for Reactive power behavior of Germany and neighboring countries to generate 380 KV line power plant dispatch 600 Step 1  NTC-based trade between market zones 500 Market model Reactive power [Mvar]  Only real power (P) dispatch 400 300  Estimation of current-induced redispatch based on 200 a transmission & 110 kV distribution grid model Step 2 100  Usage of ELMOD to calculate load flows, overloads Real power: 0 and least-cost redispatch current-induced  Penalty cost for international redispatch -100 redispatch 0 500 1000 1500 Line load [MVA]  Estimation of reactive power dispatch and voltage- Q_cap Q_ind induced redispatch Q_tot Step 3  Usage of ELMOD LinAC , a linearized AC model to Iterative calculation of Reactive power: account for voltage stability and reactive power quadratic inductive reactive voltage-induced flows power behavior redispatch  Iterative approach to account for quadratic reactive power behavior of electricity lines 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 8

Redispatch models use linearized real and reactive power flow calculations Simplified model formulation of redispatch models Market Current-induced Redispatch Voltage-induced Redispatch model 𝒏𝒃𝒔𝒉 ∙ 𝑯𝒇𝒐 𝒐 𝑸 − 𝒉𝒇𝒐 𝒐 Target 𝑸,𝒏𝒃𝒔𝒍𝒇𝒖 𝐍𝐣𝐨 ෍ 𝒅𝒑𝒕𝒖 𝒐 function 𝒐∈𝑶 Thermal limit: 𝑴𝒋𝒐𝒇𝑫𝒗𝒔𝒔𝒇𝒐𝒖 𝒎 ≤ 𝑼𝒊𝒇𝒔𝒏𝒃𝒎𝒎𝒋𝒏𝒋𝒖 𝒎 Voltage TS: 𝟏, 𝟘𝟖 𝒒. 𝒗. ≤ 𝑽 𝒐 ≤ 𝟐, 𝟏𝟒 𝒒. 𝒗. Voltage DS: 𝟏, 𝟘𝟓 𝒒. 𝒗. ≤ 𝑽 𝒐 ≤ 𝟐, 𝟏𝟕 𝒒. 𝒗 Restrictions Gen P 𝑸 , 𝑯𝒇𝒐 𝒐 ∈ 𝑹 𝑯𝒇𝒐 𝒐 Gen Q 𝑸 − 𝑬𝒇𝒏 𝒐 𝑸 = σ 𝒏∈𝑶 𝒉 𝒐,𝒏 𝑽 𝒐 − 𝑽 𝒏 − 𝒄 𝒐,𝒏 (𝜾 𝒐 −𝜾 𝒏 ) Real power: 𝑯𝒇𝒐 𝒐 Grid Iterative calculation 𝑹 − 𝑬𝒇𝒏 𝒐 𝑹 − 𝑴𝒑𝒕𝒕 𝒐 𝑹 = balance Reactive power: 𝑯𝒇𝒐 𝒐 σ 𝒏∈𝑶 −𝒄 𝒐,𝒏 𝑽 𝒐 − 𝑽 𝒏 − 𝒉 𝒐,𝒏 (𝜾 𝒐 −𝜾 𝒏 ) 1) U n / U m ... Voltage magnitude at node n / m Θ n / Θ m ... Voltage angle at node n / m g n,m / b n,m ... Conductance / susceptance between node n and m 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 9

Current and voltage are represented reasonably well by the redispatch model Model quality of ELMOD AC and ELMOD LinAC Voltage [p.u.] Current [A] aMAPE 1) aMAPE 1) LinAC MAE RSME LinAC MAE RSME U [kV] 2) I [A] 22.9 39.6 0.69% 2.0 2.5 0.53% Good fit for current Reasonable fit for voltage Comparison between redispatch model (ELMOD LinAC) and AC load flow model (ELMOD AC), Germany, 16 grid situations 1) Adjusted Mean Absolute Percentage Error: adjusted in relation to nominal voltage / thermal limit 2) On 380 kV level 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 10

110 kV grid set developed based on OSM data and other public sources Data set for grid model OSM data 380 kV Power plants / RES 220 kV  Substations 110 kV Attribution to nodes 380 / 220 / 110 kV  Plants: based on  Electricity lines addresses and 380 / 220 / 110 kV coordinates  RES: based on OSM data / RES database  Nodes with generation and demand  Auxiliary nodes Load  Attribution based on  Lines start / end, technical GDP and population of surrounding area parameters updated with TSO static grid models  Transformers Nodes: ~5700 380 / 110 kV Lines: ~6500 220 / 110 kV Substations: ~370 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 11

1 Motivation 2 Model Development 3 Staus Quo 2014 4 Future scenario 2025 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 12

Good fit between congestions in model and reality Congested grid elements: Model results vs. reality Model results 2014 Monitoring report 2014 Frequency of congested grid elements  Good fit between for border areas to Poland, Czech Republic and Denmark  Fit for Remptendorf- Redwitz line  Congestions in the North West and Center not reliably recognized  Distribution grid congestions in the North fit local curtailment compensation Source: BNetzA Monitoring Report 2015 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 13

Taking into account voltage stability, redispatch patterns change High load and high wind feed-in situation: current- and voltage-induced redispatch Current-induced Current- and voltage induced  Ramp-down of power plants in the North  Additional redispatch in the South to cover Results  Curtailment mainly in Schleswig-Holstein reactive power requirements  Ramp-up in the South and Austria  Additional ramp-downs in the North 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 14

Reactive power from the 110 kV grid decreases voltage-induced redispatch cost Redispatch costs 2014 in Germany -13.4 (-8%) 174.7 180 Redispatch cost Germany 2014 161.3  Comparison of voltage- / 17.1 160 13.3 current- induced redispatch  Cost reduction potential 140 44.1 from 110 kV grid reactive 34.6 Cost p.a. [mio. EUR] 120 113.4 power sources 100  Redispatch and curtailment 80 78.1 78.1 78.1 cost is mainly current- 60 induced  8% reduction possible 40 through reactive power from the distribution grid 20 35.4 35.4 35.3 0 Current-induced Current- / with 110 kV sources Voltage-induced Curtailment U Redispatch U Curtailment I Redispatch I 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 15

1 Motivation 2 Model Development 3 Staus Quo 2014 4 Future scenario 2025 06.09.2017 TU Dresden, Chair of Energy Economics, Fabian Hinz 16