ROBUST OPTIMIZATION AND DYNAMIC SIMULATION OF A SMALL SCALE - PowerPoint PPT Presentation

ROBUST OPTIMIZATION AND DYNAMIC SIMULATION OF A SMALL SCALE POWER-TO-AMMONIA PLANT Robert Weiss, Jouni Savolainen, Matti Tähtinen, Teemu Sihvonen, VTT Technical Research Centre of Finland Ltd., Finland Yasmina Bennani, Vincent Hans, Hans Vrijenhoef, Proton Ventures BV, The Netherlands 15/03/17 IRES Energy Storage Europe

Contents  Introduction  Nitrogen fertilizer use in Europe  Small scale Power-to-Ammonia concept  Results: Sizing and operation of a Power-to-Ammonia plant  Dynamic simulation: Virtual tests of plant dynamics  Robust optimization and market operations: Scheduling, profitability and CO 2 -emissions reduction in wind power intensive West Denmark. 11/15/16 2

Nitrogen fertiliser usage Installed wind capacity in Europe 2005 in Europe 2011 Source: Eurostat, 2016 Source: European Spatial Monitoring System, 2011 11/15/16 3

Nitrogen fertiliser usage Installed wind capacity in Europe 2005 in Europe 2011 Source: Eurostat, 2016 Source: European Spatial Monitoring System, 2011 11/15/16 4

Power to ammonia Small scale ammonia production makes it possible to Feedstock: produce green decentralized ammonia which can be • Electricity from renewable sources (Wind turbines, PV) further used as: • Air (via N 2 /O 2 separation) • Nitrogen carrier (fertilizer) • Hydrogen carrier • Energy storage • Fuel 11/15/16 5 Source: http://www.protonventures.com

Wind-driven Small-scale Ammonia mill Nominal Full load Capacity Wind NH3 power capacity Wind power: - Intermittency - Forecasting errors: Uncertainty, depends on forecast horizon 11/15/16 6

Wind-driven Small-scale Ammonia mill Market Power PEM Overload Capacity 60% /30min Nominal Full load Capacity Part load 30-100% PEM Wind Flexible NH3 Electrolyzer power H 2 Ammonia H 2 capacity Buffer process Wind power: - Intermittency - Forecasting errors: Uncertainty, depends on forecast horizon 11/15/16 7

High-fidelty APROS simulator model for virtual test of process dynamics and control concepts Apros is a versatile high fidelty first principles dynamic simulator, including  thermal hydraulics networks,  automation & control, and  specialized dynamic models for  Power-to-gas electrolysis and  Haber-Bosch synthesis  etc 11/15/16 8

Virtual test of power-to-ammonia dynamics: Primary Frequency Control operation in West Denmark (DK1) In this example test we used an optimal 5-days hourly operation plan for the power-to-ammonia plant: - Available wind power (Day-ahead Forecast) - Wholesales spot power purchase and sales (Day-ahead hourly) - Primary Frequency Control capacity sold (Day-ahead 4-hour blocks) - Running schedule for power-to-ammonia plant units (Day-ahead hourly) 11/15/16 9

Virtual test of power-to-ammonia dynamics: Primary Frequency Control operation in West Denmark (DK1) With the contract schedules and TSO system frequency data (0.1-1s resolution), we calculate the actual primary frequency control response of the PEM electrolyser units. This allows us to calculate the fast dynamic responses of all subsequent process units, and the plant power consumption. Power consumption including executed frequency control 11/15/16 10

Virtual test of power-to-ammonia dynamics: Primary Frequency Control operation in West Denmark (DK1) The operation mode performed well in the virtual tests. - Despite fast fluctuating power input, we achieved stable and good operating conditions (p,T) for NH 3 synthesis reactor - Some process valve controls had to be updated, especially for very large and fast transients. - Gas storages operation schedules: Daily intial and end-target storage levels important, some revision on sizes/max pressures Similar test results for following wind power production. Power consumption including executed frequency control H 2 balance and NH 3 synthesis reactor storage pressure 11/15/16 11

Small scale ammonia plant Operation optimization – example cases  Case 1: Plant inside the fence of a Wind Park  Subcase G50%: Restricted grid connection to 50% of Wind Park max capacity  Intra-day trade balances WP forecast error except for the final hour-ahead WP forecast  Robust optimization reserves optimal capacity to mitigate worst-case errors in final hour-ahead WP forecast  Assumed automatic load following within execution hour  Case 2: Plant in DSO grid, hourly Net Settlement with Wind Park  Subcase PFC : Plant is also able to sell Primary Frequency Control on day-ahead basis.  Load following not allowed.  Robust optimization reserves optimal day-ahead capacity to mitigate worst-case distortions for prolonged PFC response.  Case 3: P2NH3 plant buying only market power and Certificates of Origin  Subcase PFC : like in case 2. 11/15/16 12

Example market area – West Denmark DK1 in 2016 For our example, we used • wind power production data, • power spot price data and • primary regulation price data of West Denmark, DK1 price and control area (Energinet.dk 2016; NordPool 2016). DK1 spot price varied between -54 to 105 EUR/MWh and had an average of 26.7 EUR/MWh. Grid fees in the cases ranged between 7.6 to 48.6 EUR/MWh. This area is well known for a high intensity of Wind Power, which during peak production times was up to 200% of the power consumption in the area during 2016. In 2015, 42% of the power used was wind power. 11/15/16 13

Case1: Plant inside the fence of a Wind Park Restricted grid connection to 50% of Wind Park max capacity 12.5 MW Wind park Hourly power production as in DK1 = same timing as DK1 market wind power Grid connection only 50% = 6.25 MW Grid connection capacity Without storage, 12.5% of the produced wind energy would be stranded 11/15/16 14

Case1: Plant inside the fence of a Wind Park Restricted grid connection to 50% of Wind Park max capacity 12.5 MW Wind park Hourly power production as in DK1 = same timing as DK1 market wind power Grid connection only 50% = 6.25 MW Grid connection capacity Without storage, 12.5% of the produced wind energy would be stranded -> This was avoided! 4000 tNH3/year ammonia plant 5 MWe electrolysis (nominal) - 4 parallel PEM electrolysers - 60% overload capacity 5 hours intermediate H2 buffer Haber-Bosch flexibility 30-100% 11/15/16 15

Wind power utilization and CO2 emission reduction High Wind power utilization was reached for case 1 ”Inside Fence of Wind park” and case 2 ”Net Settlement ”. For case 3 ”pure market purchase”, only slighly higher WP than yearly average of grid was reached. PFC operation increased slightly market purchases, decreasing WP share. Case 3 increased the CO 2 -emissions of NH 3 production , since the non-WP share of grid market power purchases contain a high share of Fossil based power, especially coal fired power. Wind power utilization includes here both • direct ”own” WP, and • indirect grid WP (hourly share of market purchases, via settlement calculation from DK1 official timeseries) 11/15/16 16

Production Costs Strongly dependent on operation case Primary Frequency control:  Good income potential  Assymmetric Bidding in DK1 area is beneficial for electrolysers! Compare to global market price range of 250-700 EUR /tNH 3 for conventional fossil NH 3 during 2010-2016 11/15/16 17

Production Costs Strongly dependent on operation case Primary Frequency control:  Good income potential  Assymmetric Bidding in DK1 area is beneficial for electrolysers! Some additional income potential from O 2 sales (estimate 50-70 EUR/tNH 3 ) CAPEX share still very large 11/15/16 18

Ammonia total production cost: Impact of CAPEX and Wind Power cost 11/15/16 19

Summary and conclusions  Small Scale Power to Ammonia example was shown for Danish wind power and market (DK1) conditions.  Operational planning needed to find economical case and CO2-reduction potential.  ”Net settlement with wind power” or ”Inside fence of Wind park” operation seems to be suitable in Denmark, while a pure market-based operation has clear drawbacks.  Process dynamics provides needed flexibility to meet intermittency of wind power. PEM Electrolyzer’s capacity to temporarily overload is beneficial.  Participation to electrical grid ancillary services (Primary Frequency control) is essential. DK1 assymetric bidding is beneficial for PtX.  Robust optimization was needed in the operational planning to include the effects of uncertainty for Primary Frequency control and mitigation of Wind Power forecast errors.  Competitive or at least acceptable NH 3 -prices seem to be reachable within near future. 11/15/16 20