Impacts of scenario definitions on CO2 mitigation cost in energy system models
1 like475 views
The document discusses how different definitions of CO2 constraints in energy system models can impact modeling results. It analyzes two case studies using the TIMES and E2M2 models to compare scenarios with different CO2 constraints: a CO2 cap, CO2 budget, and additional constraints like energy autarky goals or coal phaseouts. The key findings are: 1) A CO2 budget constraint, which defines a total available CO2 budget over the entire modeling period, results in lower total system costs and average mitigation costs compared to an annual CO2 cap in both models. 2) Adding a second constraint, like an energy autarky goal or coal phaseout, increases total system costs but can further
![Click to edit Master subtitle style
Impacts of scenario
definitions on CO2
mitigation cost
in energy system models
Lukasz Brodecki,
Annika Gillich
Source: [1]](https://image.slidesharecdn.com/etsapbrodecki-181121072203/85/Impacts-of-scenario-definitions-on-CO2-mitigation-cost-in-energy-system-models-1-320.jpg)
![BUT: different ways of modelling CO2-targets in ESM can lead to different results!
How should those targets be modelled and which scenarios should be selected in order to derive sound
policy recommendations?
17-Nov-18IER University of Stuttgart 3
Ambitious greenhouse gas reduction goals defined at COP21 in Paris
Introduction
Source: [2]
Types of GHG targets differ across
countries, but a high share relies
on a maximum level of GHG
emissions in a target year!
Energy System Models (ESM):
used for planning on how to
achieve those targets and
assessment of progress](https://image.slidesharecdn.com/etsapbrodecki-181121072203/85/Impacts-of-scenario-definitions-on-CO2-mitigation-cost-in-energy-system-models-3-320.jpg)
![17-Nov-18IER University of Stuttgart 7
Model description TIMES Local
Source: [3-4]](https://image.slidesharecdn.com/etsapbrodecki-181121072203/85/Impacts-of-scenario-definitions-on-CO2-mitigation-cost-in-energy-system-models-7-320.jpg)
![17-Nov-18IER University of Stuttgart 9
System cost and average mitigation cost behave differently under CO2-constraints
Results TIMES Local
System cost:
• Definition of additional constraints increases overall system cost
• Slightly lower system cost in BUDGET compared to CAP due to
higher flexibility in selection of mitigation options
Average mitigation cost:
• BUDGET represents time-integral optimum for CAP reduction level
and therefore achieves lower system cost AND lower AMC!
• CAP+AUT leads to higher system cost but also to higher emission
reduction compared to CAP
• CAP+AUT results in lower AMC compared to CAP, although solution
space is smaller!
𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑚𝑖𝑡𝑖𝑔𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡 (𝐴𝑀𝐶) =
𝐶𝑂2
𝐵𝐴𝑆𝐸𝑇
𝑡=1 − 𝐶𝑂2
𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜𝑇
𝑡=1
𝑆𝑦𝑠𝑡𝑒𝑚𝑐𝑜𝑠𝑡 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 − 𝑆𝑦𝑠𝑡𝑒𝑚𝑐𝑜𝑠𝑡 𝐵𝐴𝑆𝐸
BICO (BIased COst) effect 0
50
100
150
200
250
0
200
400
600
800
1000
BASE CAP CAP+AUT BUDGET
Averagemitigationcost[€/tCO2]
Reducedemissions
comparedtoBASE[kt]
Reduced emissions compared to BASE
Average CO2 mitigation cost
232 € 226 € 219 €
0
20
40
60
80
3800
3900
4000
4100
4200
BASE CAP CAP+AUT BUDGET
Totaldiscounted
systemcosts[M€2010]
+3.9% +4.5% +3.5%
≙ higher absolut cost, higher emission
reduction, BUT lower average mitigation
cost!](https://image.slidesharecdn.com/etsapbrodecki-181121072203/85/Impacts-of-scenario-definitions-on-CO2-mitigation-cost-in-energy-system-models-9-320.jpg)
![17-Nov-18IER University of Stuttgart 10
How do emission pathways develop over time?
Results TIMES Local
Emission reduction through 2nd constraint approximates
BUDGET emission reduction path in the medium term
0
1.000
2.000
3.000
4.000
5.000
2010 2015 2020 2025 2030 2035 2040 2045 2050
CumulatedCO2Emissions[kt]
BASE CAP CAP+AUT BUDGET
2025 2026 2027 2028 2029 2030
BASE CAP
CAP+AUT BUDGET
• Profitable abatement
measures are already
drawn in BASE case
(degressive curve
character)
• 2nd constraint CAP+AUT
pushes emissions in 2050
below level of CAP and
BUDGET](https://image.slidesharecdn.com/etsapbrodecki-181121072203/85/Impacts-of-scenario-definitions-on-CO2-mitigation-cost-in-energy-system-models-10-320.jpg)
![17-Nov-18IER University of Stuttgart 12
Model description E2M2
Power plants
Dispatch
Energy generation
Results
System cost
Electricity prices
Market value
Input Model
Linear programming
Objective function
Restrictions
European Electricity Market Model – E2M2
Fundamental linear (mixed-integer) electricity market model for Europe
Investment decisions for plants, storages, transmission capacity and other flexibility options and simultanous optimization of dipatch
Provision of balancing energy and reserve capacity
Myopic optimization on yearly basis with hourly time resolution
Electricity prices for markets with perfect competition
Generation
Production from RES
Existing power plants
Techn. + econ. parameter
Investment
Power plants (therm. + RES)
Flexibility options
Restrictions
Satisfy demand
Upper and lower bounds
RES: Renewable Energy Sources Source: [5-6]](https://image.slidesharecdn.com/etsapbrodecki-181121072203/85/Impacts-of-scenario-definitions-on-CO2-mitigation-cost-in-energy-system-models-12-320.jpg)
![17-Nov-18IER University of Stuttgart 13
Scenario description E2M2
General scenario framework:
• 5-year-steps until 2050, 2-hourly time
resolution
• Germany as one region
• Constant domestic electricity demand,
development from exporting country in
2020 to importing country in 2050
• Must-run for CHP-plants considered
• Endogen investment in thermal and
renewable power plants
• Base year for weather and demand data:
2006
• Perfect foresight over full period 2020-2050
• No restrictions on CO2 (no upper bound, CO2-price = 0)BASE
• Yearly upper bound (UB) on CO2 according to 2030 energy
sector targets (Klimaschutzplan 2050 [7])
• Projection of targets until 2050 ( 95,5% reduction vs. 1990)
• Linear interpolation for years between target years
CAP
• Sum of yearly upper bounds from scenario CAP as one single
UB over entire modelling period
• Additional UB only for 2050 in order to reach 95,5%
reduction level (as in CAP and CAP+CPO)
BUDGET
CAP+CPO
• CPO = Coal-phase out
• UB on CO2 acc. to scenario CAP
• Additional early phase-out of lignite and hard coal power
plants in Germany until 2045](https://image.slidesharecdn.com/etsapbrodecki-181121072203/85/Impacts-of-scenario-definitions-on-CO2-mitigation-cost-in-energy-system-models-13-320.jpg)
![[1] Agora Energiewende (2017): Die Energiewende im Stromsektor: „Stand der Dinge 2016. Rückblick auf die wesentlichen Entwicklungen sowie
Ausblick auf 2017.“
[2] CAIT Climate Data Explorer, CAIT Paris Contributions Map, (2016). https://www.climatewatchdata.org/ndcs-content, accessed 02.09.2018.
[3] R. Loulou, G. Goldstein, A. Kanudia, A. Lettila, U. Remme, Documentation for the TIMES Model - Part I, (2016) 1–78.
[4] L. Brodecki, M. Blesl, Modellgestützte Bewertung von Flexibilitätsoptionen und Versorgungsstrukturen eines Bilanzraums mit hohen
Eigenversorgungsgraden mit Energie, in: EnInnov, Graz, 2018: pp. 1–15.
[5] N. Sun, Modellgestützte Untersuchung des Elektrizitätsmarktes, University of Stuttgart, 2012.
[6] S. Bothor, M. Steurer, T. Eberl, H. Brand, A. Voß, Bedarf und Bedeutung von integrations- und Flexibilisierungsoptionen in
Elektrizitätssystemen mit steigendem Anteil erneuerbarer Energien, in: 9. Int. Energiewirtschaftstagung an Der TU Wien, IEWT 2015, 2015.
[7] „Klimaschutzplan 2050 – Klimaschutzpolitische Grundsätze und Ziele der Bundesregierung“, Bundesministerium für Umwelt, Bau und
Reaktorsicherheit (BMUB), (2016) 1–96. doi:10.1016/j.aqpro.2013.07.003.
17-Nov-18IER University of Stuttgart 24
References](https://image.slidesharecdn.com/etsapbrodecki-181121072203/85/Impacts-of-scenario-definitions-on-CO2-mitigation-cost-in-energy-system-models-24-320.jpg)
Impacts of scenario definitions on CO2 mitigation cost in energy system models
- 1. Click to edit Master subtitle style Impacts of scenario definitions on CO2 mitigation cost in energy system models Lukasz Brodecki, Annika Gillich Source: [1]
- 2. • Introduction • TIMES: Model, Scenarios, Results • E2M2: Model, Scenarios, Results • Discussion and Summary • References 17-Nov-18IER University of Stuttgart 2 Agenda
- 3. BUT: different ways of modelling CO2-targets in ESM can lead to different results! How should those targets be modelled and which scenarios should be selected in order to derive sound policy recommendations? 17-Nov-18IER University of Stuttgart 3 Ambitious greenhouse gas reduction goals defined at COP21 in Paris Introduction Source: [2] Types of GHG targets differ across countries, but a high share relies on a maximum level of GHG emissions in a target year! Energy System Models (ESM): used for planning on how to achieve those targets and assessment of progress
- 4. 17-Nov-18IER University of Stuttgart 4 CO2-targets in Energy System Models: few model runs use budget Literature Review Total number of publications considered: 117 • Majority of publications consider a minimum share of renewables, • One third considers a CO2-price or cap, only 2% use a CO2 budget • Model foresight is often not mentioned explicitly, but relevant for interpretation of results
- 5. 1) How does the selection of CO2-constraint impact model results? 2) Which CO2-constraint should be used to assess mitigation pathways with energy system models? 17-Nov-18IER University of Stuttgart 5 Various CO2-constraints will be analysed in two case studies Modelling Approach Research questions Methodology E2M2 TIMES-Local BASE CAP BUDGETCAP-CPO CAP-AUT Comparison of emission reduction and mitigation cost Comparison of emission reduction and mitigation cost Result comparison and effect analysis
- 6. • Introduction • TIMES: Model, Scenarios, Results • E2M2: Model, Scenarios, Results • Discussion and Summary • References 17-Nov-18IER University of Stuttgart 6 Agenda
- 7. 17-Nov-18IER University of Stuttgart 7 Model description TIMES Local Source: [3-4]
- 8. 17-Nov-18IER University of Stuttgart 8 Scenario description TIMES Local General scenario framework: • Linear optimizaton, bottom-up model • Medium-sized municipality in Germany as one region • Focus on supply and demand processes relevant for a city/district model, all sectors • Starting point 2010, 5-year-steps until 2050 with perfect foresight • Hourly time resolution with 5 representative seasons (original seasons plus fall peak) adding up to 840 timeslices, • Endogen investment and dispatch in eletrical, thermal sevices and mobility technologies • No restrictions on CO2 (no upper bound, CO2-price = 0) • Extrapolation of local development based on statistical data BASE • Limit of total CO2 emissions according to 2050 state targets • Projection of targets until 2050 as yearly upper bound (UB) -90% vs. 1990 with linear interpolation for timesteps between target years CAP • Sum of yearly upper bounds from scenario CAP as one single UB over entire modelling period • Additional UB only for 2050 in order to reach same CO2 reduction level (as in CAP and AUT) BUDGET • UB on CO2 according to scenario CAP • Additional long term „energy-autarky (AUT) goal on local level“ until 2050 – level of self-sufficiency in 2050 75% • Linear interpolation for timesteps between years for AUT CAP+AUT
- 9. 17-Nov-18IER University of Stuttgart 9 System cost and average mitigation cost behave differently under CO2-constraints Results TIMES Local System cost: • Definition of additional constraints increases overall system cost • Slightly lower system cost in BUDGET compared to CAP due to higher flexibility in selection of mitigation options Average mitigation cost: • BUDGET represents time-integral optimum for CAP reduction level and therefore achieves lower system cost AND lower AMC! • CAP+AUT leads to higher system cost but also to higher emission reduction compared to CAP • CAP+AUT results in lower AMC compared to CAP, although solution space is smaller! 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑚𝑖𝑡𝑖𝑔𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡 (𝐴𝑀𝐶) = 𝐶𝑂2 𝐵𝐴𝑆𝐸𝑇 𝑡=1 − 𝐶𝑂2 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜𝑇 𝑡=1 𝑆𝑦𝑠𝑡𝑒𝑚𝑐𝑜𝑠𝑡 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 − 𝑆𝑦𝑠𝑡𝑒𝑚𝑐𝑜𝑠𝑡 𝐵𝐴𝑆𝐸 BICO (BIased COst) effect 0 50 100 150 200 250 0 200 400 600 800 1000 BASE CAP CAP+AUT BUDGET Averagemitigationcost[€/tCO2] Reducedemissions comparedtoBASE[kt] Reduced emissions compared to BASE Average CO2 mitigation cost 232 € 226 € 219 € 0 20 40 60 80 3800 3900 4000 4100 4200 BASE CAP CAP+AUT BUDGET Totaldiscounted systemcosts[M€2010] +3.9% +4.5% +3.5% ≙ higher absolut cost, higher emission reduction, BUT lower average mitigation cost!
- 10. 17-Nov-18IER University of Stuttgart 10 How do emission pathways develop over time? Results TIMES Local Emission reduction through 2nd constraint approximates BUDGET emission reduction path in the medium term 0 1.000 2.000 3.000 4.000 5.000 2010 2015 2020 2025 2030 2035 2040 2045 2050 CumulatedCO2Emissions[kt] BASE CAP CAP+AUT BUDGET 2025 2026 2027 2028 2029 2030 BASE CAP CAP+AUT BUDGET • Profitable abatement measures are already drawn in BASE case (degressive curve character) • 2nd constraint CAP+AUT pushes emissions in 2050 below level of CAP and BUDGET
- 11. • Introduction • TIMES: Model, Scenarios, Results • E2M2: Model, Scenarios, Results • Discussion and Summary • References 17-Nov-18IER University of Stuttgart 11 Agenda
- 12. 17-Nov-18IER University of Stuttgart 12 Model description E2M2 Power plants Dispatch Energy generation Results System cost Electricity prices Market value Input Model Linear programming Objective function Restrictions European Electricity Market Model – E2M2 Fundamental linear (mixed-integer) electricity market model for Europe Investment decisions for plants, storages, transmission capacity and other flexibility options and simultanous optimization of dipatch Provision of balancing energy and reserve capacity Myopic optimization on yearly basis with hourly time resolution Electricity prices for markets with perfect competition Generation Production from RES Existing power plants Techn. + econ. parameter Investment Power plants (therm. + RES) Flexibility options Restrictions Satisfy demand Upper and lower bounds RES: Renewable Energy Sources Source: [5-6]
- 13. 17-Nov-18IER University of Stuttgart 13 Scenario description E2M2 General scenario framework: • 5-year-steps until 2050, 2-hourly time resolution • Germany as one region • Constant domestic electricity demand, development from exporting country in 2020 to importing country in 2050 • Must-run for CHP-plants considered • Endogen investment in thermal and renewable power plants • Base year for weather and demand data: 2006 • Perfect foresight over full period 2020-2050 • No restrictions on CO2 (no upper bound, CO2-price = 0)BASE • Yearly upper bound (UB) on CO2 according to 2030 energy sector targets (Klimaschutzplan 2050 [7]) • Projection of targets until 2050 ( 95,5% reduction vs. 1990) • Linear interpolation for years between target years CAP • Sum of yearly upper bounds from scenario CAP as one single UB over entire modelling period • Additional UB only for 2050 in order to reach 95,5% reduction level (as in CAP and CAP+CPO) BUDGET CAP+CPO • CPO = Coal-phase out • UB on CO2 acc. to scenario CAP • Additional early phase-out of lignite and hard coal power plants in Germany until 2045
- 14. 17-Nov-18IER University of Stuttgart 14 BICO effect occurs also in power sector scenarios Results E2M2 System cost: • Coal phase-out as additional constraint results in higher system cost than CAP due to limited solution space • BUDGET shows lower system cost than CAP due to timely flexibility of reduction Average mitigation cost: • CAP+CPO: induces higher emission reduction but slightly lower average mitigation cost compared to BASE scenario! BICO effect appears again
- 15. 17-Nov-18IER University of Stuttgart 15 How do cost and emission pathways develop over time? Results E2M2 BICO effect: 2nd constraint pushes emission reduction more towards BUDGET scenario (higher emission reductions 2020 and 2025) an therefore towards a more cost-optimal solution! BUDGET and CAP+CPO show higher emission reduction in early years
- 16. • Introduction • TIMES: Model, Scenarios, Results • E2M2: Model, Scenarios, Results • Discussion and Summary • References 17-Nov-18IER University of Stuttgart 16 Agenda
- 17. 17-Nov-18IER University of Stuttgart 17 Generic mitigation cost curve explains BICO effect Effect Analysis Simplifications compared to model runs: • cost assumed constant over time • interest rate=0% • decommissioning of plants is possible anytime at no cost (lifetime of new plants = 1 year) 2020 2020 2025 reduced t CO2 € per reduced t CO2 2020 2025 2030 mitigation in BASE 2025 2030 2025 fuel switch low emission investment replaces high emission investment 2030 2030 2020 low emission investment replaces existing plant 2025 a b 2025 2020
- 18. 2020 2020 € per reduced t CO2 2020 2020 2030 2030 2025 2025 2025 2025 d e 2020 2025 2025 20 2030 2020 2020 € per reduced t CO2 2020 2020 reduced t CO2 2030 2030 2025 2025 2025 2025 d e f 2020 2025 2025 2030 2030 2020 2020 € per reduced t CO2 2020 2020 reduced t CO2 2030 2030 2025 2025 2025 2025 d e f 2020 2025 2025 2030 17-Nov-18IER University of Stuttgart 18 Generic mitigation cost curve explains BICO effect Effect Analysis BUDGET scenario sees all mitigation options and has full flexibility of choice: mitigation in BUDGET CAP scenario sees all mitigation options, but can only choose options that are effective to fulfill the yearly restrictions! d: emission reduction in CAP 2020 e: emission reduction in CAP 2025 f: emission reduction in CAP 2030 2020 2020 2025 reduced t CO2 € per reduced t CO2 2020 2025 2030 mitigation in BASE 2025 2030 2025 fuel switch low emission investment replaces high emission investment 2030 2030 2020 low emission investment replaces existing plant 2025 a b 2025 2020 c
- 19. 17-Nov-18IER University of Stuttgart 19 2nd constraint decreases average mitigation cost Effect Analysis Cause 1: Early use of low cost mitigation options avg. mitigation cost 2030 d* 2030 2020 € per reduced t CO2 2020 2020 reduced t CO2 2030 2030 2025 2025 2025 2025 d e f 2020 2025 2025 2030 avg. mitigation cost 2025 avg. mitigation cost for additional reduction through coal phase-out 2020 2020
- 20. 17-Nov-18IER University of Stuttgart 20 2nd constraint decreases average mitigation cost Effect Analysis Cause 2: Innovation of low emission technologies avg. mitigation cost 2030 e* 2030 2020 € per reduced t CO2 2020 2020 reduced t CO22030 2025 2025 d e f 2020 2025 2025 2030 avg. mitigation cost 2025 avg. mitigation cost 2020 2020 20302025
- 21. 1) … the definition of model constraints plays a crucial role in energy system analysis and the evaluation of CO2 mitigation pathways, as costs differ significantly and distortion of AMC can appear! 2) … no general answer to when the BICO effect appears can be given, but it has been shown in two different ESMs for two different research subjects. 3) … above explained two causes are catalyst for the effect, but whether it occurs, depends on model type, time horizon and parameterization. 17-Nov-18IER University of Stuttgart 21 Our research has shown that… Conclusion Avoidance of BICO effect: compare CO2-cap and -price model runs with a BUDGET scenario! Considering the following limtations…
- 22. 17-Nov-18IER University of Stuttgart 22 Careful when using a BUDGET run as comparison Discussion and OutlookQualitativeQuanti- tative Upper bound of emissions in BUDGET shall equal resulting sum of emissions in CAP scenario. Additional upper bound in final year shall be set and be equal to the one in CAP to achieve same reduction level. Compare resulting technology portfolio at the end of the modelling period (and therefore remaining reduction potential of energysystem after final year). Consider salvage cost or use annuities in ESM with short/limited time horizon. Further analyses should examine… • Robustness of results regarding temporal resolution, • Sensitivity of the models for technology parameterization, • Impact of discount rate (highly relevant for results), • Use of non-perfect-foresight models, e.g. myopic optimization, may increase the BICO effect.
- 23. • Introduction • TIMES : Model, Scenarios, Results • E2M2: Model, Scenarios, Results • Discussion and Summary • References 17-Nov-18IER University of Stuttgart 23 Agenda
- 24. [1] Agora Energiewende (2017): Die Energiewende im Stromsektor: „Stand der Dinge 2016. Rückblick auf die wesentlichen Entwicklungen sowie Ausblick auf 2017.“ [2] CAIT Climate Data Explorer, CAIT Paris Contributions Map, (2016). https://www.climatewatchdata.org/ndcs-content, accessed 02.09.2018. [3] R. Loulou, G. Goldstein, A. Kanudia, A. Lettila, U. Remme, Documentation for the TIMES Model - Part I, (2016) 1–78. [4] L. Brodecki, M. Blesl, Modellgestützte Bewertung von Flexibilitätsoptionen und Versorgungsstrukturen eines Bilanzraums mit hohen Eigenversorgungsgraden mit Energie, in: EnInnov, Graz, 2018: pp. 1–15. [5] N. Sun, Modellgestützte Untersuchung des Elektrizitätsmarktes, University of Stuttgart, 2012. [6] S. Bothor, M. Steurer, T. Eberl, H. Brand, A. Voß, Bedarf und Bedeutung von integrations- und Flexibilisierungsoptionen in Elektrizitätssystemen mit steigendem Anteil erneuerbarer Energien, in: 9. Int. Energiewirtschaftstagung an Der TU Wien, IEWT 2015, 2015. [7] „Klimaschutzplan 2050 – Klimaschutzpolitische Grundsätze und Ziele der Bundesregierung“, Bundesministerium für Umwelt, Bau und Reaktorsicherheit (BMUB), (2016) 1–96. doi:10.1016/j.aqpro.2013.07.003. 17-Nov-18IER University of Stuttgart 24 References
- 25. e-mail phone +49 (0) 711 685- fax +49 (0) 711 685- Universität Stuttgart Thank you! IER Institute for Energy Economics and Rational Energy Use Lukasz Brodecki, Annika Gillich 878 49 878 73 Institut für Energiewirtschaft und Rationelle Energieanwendungen (IER) annika.gillich@ier.uni-stuttgart.de, lukasz.brodecki@ier.uni-stuttgart.de Heßbrühlstraße 49a, 70565 Stuttgart