A Modeling Framework for Future Energy Systems Gran Andersson, ETH - PowerPoint PPT Presentation

A Modeling Framework for Future Energy Systems Göran Andersson, ETH Zürich ETH Power Systems Laboratory

Content • Energy Hub - Multi energy-carrier systems • Power Node - Incorporation of fluctuating power sources - Incorporation of demand side participation - Incorporation of storage 2 ETH Power Systems Laboratory

The Energy Hub L = Loads (Output) M = Output side storage flows C = Coupling matrix P = Input power flows Q = Input storage flows 3 ETH Power Systems Laboratory

Hub Equations and Results 4 ETH Power Systems Laboratory

Applications (so far) • Long term energy planning of the city of Bern • Energy planning of several Swiss municipalities • Analysis of e-mobility • Energy/Exergy analysis of city of Zürich • … 5 ETH Power Systems Laboratory

Status Quo in Power Systems Modelling Traditional power system modeling is “fractional“:  Separate models are used for capturing information of  Transmission & distribution grid (topology, voltage & frequency dynamics, voltage & line limits)  Power generation (generator dynamics, ramp constraints, wind and PV in-feed predictions)  Load models (dynamics, load demand predictions)  Storage models (capacity, storage levels, dynamics)  Modelled interaction between individual power system units and grid does not necessarily capture all relevant aspects  No interaction with other energy carriers modeled (cf Energy Hub) 6 ETH Power Systems Laboratory

Status Quo in Power Systems Modelling  Example: optimal power dispatch simulations do consider units that inject or absorb power from the grid.  Which of these units are storages (energy-constrained)?  Which of these units provide fluctuating power in-feed?  What controllability (full / partial / none) does the operator have over fluctuating generation and demand processes? Power P in Grid System P out Unit 7 ETH Power Systems Laboratory

Status Quo in Power Systems Modelling  Example: optimal power dispatch simulations do consider units that inject or absorb power from the grid.  Which of these units are storages (energy-constrained)?  Which of these units provide fluctuating power in-feed?  What controllability (full / partial / none) does the operator have over fluctuating generation and demand processes? Power Energy ? P in Grid System provided / P out demanded Unit ? Storage 8 ETH Power Systems Laboratory

Motivation for Power Nodes Modeling Framework  Create unified framework for modeling power system units (incl. relevant operation constraints, power supply and demand processes and the controllability)  Diverse storage units (battery, pumped hydro, …)  Diverse generation units (fully dispatchable conventional generators, fluctuating in-feed of wind turbines and PV)  Diverse load units (conventional, interruptible, thermal, ...)  Operation constraints: ramp rates, storage capacity, current storage level (SOC)  Operation controllability over underlying process (=“flexibility“): fully controllable, curtailable / sheddable, non-controllable 9 ETH Power Systems Laboratory

The Power Nodes Framework  Modeling of three domains and their interactions 10 ETH Power Systems Laboratory

One Power Node − = η − η + ξ − −  1 C x u u w v i i load load gen i i i i i gen i i 11 ETH Power Systems Laboratory

One Power Node Storage capacity × state-of- Internal losses charge Power out- Shedding term Power in- feed from grid feed to grid − = η − η + ξ − −  1 C x u u w v < < i i load load gen i i i i i gen i i Provided / demanded power Efficiency factors 12 ETH Power Systems Laboratory

One Power Node (including constraints)  Power constraints defined by: min/max power, ramp rates, storage capacity Operation flexibility defined by: shedding term w i , storage term C i x i , ξ i  13 ETH Power Systems Laboratory

Power Node without storage (e.g. non-controllable load)  Power node equation degenerates to algebraic equality constraint (for classical load: u gen , i = 0 )  Power node’s power in-feed / out-feed is  Partially controllable, if shedding term adjustable ( w i (k) > 0 )  Non-controllable, if shedding term is zero ( w i (k) = 0 ) 14 ETH Power Systems Laboratory

Variety of Power Node modelling definitions Load Gener- ation Storage 16 ETH Power Systems Laboratory

Power Node Modelling Examples PV with local storage unit, no RES feed-in tariff u PV Panel gen, PV η = ξ -1 u gen, PV gen, PV gen, PV u Power Grid gen, Bat Battery storage u = η − η − − u = ξ  1 load, Grid C x u u u u Bat Bat load, Bat load, Bat gen, Bat gen, Bat gen, Grid load, Grid Grid load, Bat (modelled as a slack u power node) u Controllable Local Load gen, Grid gen, CL = η − − + ζ  ฀ C x u a x ( x ) CL CL load,CL load,CL CL CL,0 CL Non-Controllable Local Load η = − ξ u load,NCL load,NCL load,NCL u load, NCL 17 ETH Power Systems Laboratory

Power Node Modelling Examples PV with local storage unit, RES feed-in tariff u u PV Panel (subject to RES FIT) gen, PV load, Grid, RES η = ξ -1 u gen, PV gen, PV gen, PV Node 1 P 12 Node 2 Power Grid u Battery storage load, Bat − = η − η − u u  1 C x u u gen, Grid load, Grid Bat Bat load, Bat load, Bat gen, Bat gen, Bat u − = ξ u gen, Bat u load, Grid, RES Grid gen, Grid (additional variable u for FIT energy) Controllable Thermal Load u load, CL load, Grid = η − − + ζ ฀  C x u a x ( x ) CL CL load,CL load,CL CL CL,0 CL This enables the modelling of Non-Controllable Local Load u η = − ξ load, NCL differentiated feed-in tariffs u load,NCL load,NCL load,NCL incl. options for local PV energy usage (e.g. Germany). 18 ETH Power Systems Laboratory

Power Node Modelling Examples Joint Provision of Load Frequency Control Convenient representation: ∆ Control signal modelled as u Battery storage load, Bat a load to be served − ∆ = η ∆ − η ∆  1 ∆ C x u u u Bat Bat load, Bat load, Bat gen, Bat gen, Bat gen, Bat Secondary ∆ u Controllable Thermal Load Frequency load, LFC ∆ ∆ = η ∆ − ∆ − ∆ + ∆ ζ Controller u  ฀ C x u a ( x x ) CL CL load,CL load,CL CL CL,0 CL load, CL ∆ = ⋅ ˆ u P Y load, LFC LFC LFC Dispatchable Generator ∆ Y η ∆ u = ∆ ξ : control signal [-100%, +100%] -1 u LFC gen, CG gen, CG gen, CG gen, CG ˆ P : offered control band [MW] LFC Power Balance: ∆ + ∆ − ∆ − ∆ = ∆ u u u u u gen, Bat gen, CG load, Bat load, CL load, LFC 19 ETH Power Systems Laboratory

Power Node Modelling Examples Demand response (driven by dynamic electricity tariff) [ ] = − k N 1 ( ) ( ) ( ) ∑ ∗ = ⋅ + u min elec . tariff k u k v k load losses i i = k 0 ( ) = η + ξ −  C x u v x , s . t . i i load demand losses i load i i i i ≤ ≤ 0 x 1 i ≥ u 0 , load i 20 ETH Power Systems Laboratory

Power Node Modeling Example: Predictive power dispatch  Conventional generation unit [6]  Conventional (uncontrolled) load [1] + load predictions  Pumped-hydro storage units [4+5] and flexible loads (DSM) [7]  Wind/PV units (curtailable) [2-3] + Wind/PV power in-feed predictions 21 ETH Power Systems Laboratory

Power Node Modeling Example: Predictive power dispatch 22 ETH Power Systems Laboratory

 Optimal predictive power dispatch (Germany)  T pred. = 72h, T upd. = 4h, T sample =15min.  Simulation Period: May 2010 (30% Wind, 20% PV) – Calc < 4min. 23 ETH Power Systems Laboratory

 Optimal predictive power dispatch (Germany, high PV )  T pred. = 72h, T upd. = 4h, T sample =15min.  Simulation Period: May 2010 (30% Wind, 50% PV , no DSM ) 25 ETH Power Systems Laboratory

Power Nodes and Energy Hubs  Partial transformation between Power Nodes and Energy Hubs is possible  Converter: natural gas → electricity ( u load = 0, M β = 0 ) = η − η − + ξ = − η − + ξ gas  el 1 el gas 1 el gas C x u u u load load gen gen in gen gen in ( ) ( ) = η ξ − ⇔ = − el gas gas  u C x L c P Q β αβ α α gen gen in Q α = C gas ẋ [ ] [ ] + = − E α = C gas x L M C P Q E α P α = ξ gas L β Q α Gas Grid Converter El. Grid P α 27 ETH Power Systems Laboratory

Goals of Power Node Approach  Goal is to better evaluate performance of power system operation and to improve performance  Storage utilisation (What is its best use?)  Integrating fluctuating power in-feed  Integrating demand-side management (DSM)  Reduce forced ramping of conventional generators for load following and balancing of fluctuating power in-feed  Examples of performance criteria  power system operation cost  curtailment of RES in-feed  Power system CO 2 emissions 29 ETH Power Systems Laboratory

Contributions from Kai Heussen (DTU) Stephan Koch Andreas Ulbig Martin Geidl Gaudenz Koeppel Thilo Krause Florian Kienzle …….. 30 ETH Power Systems Laboratory