Slideshare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our User Agreement and Privacy Policy.

Slideshare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our Privacy Policy and User Agreement for details.

Successfully reported this slideshow.

Like this presentation? Why not share!

- AI and Machine Learning Demystified... by Carol Smith 2922195 views
- The AI Rush by Jean-Baptiste Dumont 760625 views
- 10 facts about jobs in the future by Pew Research Cent... 434165 views
- 2017 holiday survey: An annual anal... by Deloitte United S... 713568 views
- Harry Surden - Artificial Intellige... by Harry Surden 389685 views
- Inside Google's Numbers in 2017 by Rand Fishkin 945701 views

168 views

Published on

Short-term uncertainty in long-term energy models

Published in:
Data & Analytics

No Downloads

Total views

168

On SlideShare

0

From Embeds

0

Number of Embeds

0

Shares

0

Downloads

27

Comments

0

Likes

1

No embeds

No notes for slide

- 1. v v Short-term uncertainty in long-term energy models 71TH SEMI-ANNUAL ETSAP MEETING Maryland, USA 10.07.2017 Pernille Seljom (pernille.seljom@ife.no) & Asgeir Tomasgard (asgeir.tomasgard@ntnu.no)
- 2. v v PhD thesis: Stochastic modelling of short-term uncertainty in long-term energy models - Applied to TIMES models of Scandinavia • Seljom, P., Tomasgard, A., 2015. Short-term uncertainty in long-term energy system models — A case study of wind power in Denmark. Energy Economics 49, 157-167. • Seljom, P., Tomasgard, A., 2017. The impact of policy actions and future energy prices on the cost-optimal development of the energy system in Norway and Sweden. Energy Policy 106, 85-102. • Seljom, P., Lindberg, K.B., Tomasgard, A., Doorman, G., Sartori, I., 2017. The impact of Zero Energy Buildings on the Scandinavian energy system. Energy 118, 284-296. • Seljom, P., Tomasgard, A. Sample Average Approximation and stability tests applied to energy system design. Submitted to an international peer-reviewed journal. 10.07.2017 Background 2
- 3. v v • The thesis uses Stochastic Programming to consider short-term uncertainty in TIMES models • Methodology is applicable to long-term energy models • Endogenous investments • Energy system and electricity models • First time this methodology is used in • energy system models • Scandinavian models • Denmark, Norway & Sweden 10.07.2017 Background 3
- 4. v v • The future climate is uncertain • Temperature, solar irradiation, wind speed & precipitation • Short-term & long-term uncertainty • Short-term uncertainty = periodically recurring • Short-term climate uncertainty → • Uncertain renewable electricity generation & building heat demand • A higher share of renewables in the electricity generation mix → • More short-term uncertainty in long-term energy models • To value flexibility in energy models it is important to consider of short-term uncertainty 10.07.2017 Motivation 4
- 5. v v Motivation Hourly wind power availability in Denmark West Availability = hourly generation/ capacity 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 20 40 60 80 100 120 140 160 Hourlywindavailability, week1DK-W 2008 2009 2010 2011 10.07.2017 5
- 6. v v Motivation Hourly heat demand for non-residential buildings in the spot price area of Oslo in winter 10.07.2017 0 100 200 300 400 0 2 4 6 8 10 12 14 16 18 20 22 24 WinterheatdemandNO12050,GWh Hour 6
- 7. v v • Methods to consider short-term uncertainty in long-term energy models 1. Ignore the uncertainty and use the expected value 2. Add constraints to ensure investments in flexible technologies 3. Link with other models 4. Run the model with different outcomes of the uncertainty 5. Use Stochastic Programming • A mathematical framework to consider uncertainty and to value flexibility in optimisation models 10.07.2017 Methodology 7
- 8. v v • Two-stage stochastic model • S scenarios = possible outcomes of the short-term uncertainty • Uncertain parameters ds with a probability of ps to occur • First stage decisions, x, investments in new capacity • Second stage decisions, ys, operation of the energy sector 10.07.2017 Methodology min . ( ) 0 (x, ) 0 s s s s S s v c x p d y s t f x g y s S 8
- 9. v v • The investments • consider the outcomes of short-term uncertainty • are feasible for all scenarios • minimise the expected cost 10.07.2017 Methodology 2010 2050 2010 2050 Stage 1 Stage 2 Investment decisions Operational decisions 9
- 10. v v • The computational effort increases with number of scenarios • In most cases, the true distribution of the short-term uncertainty cannot be used • Using a subset of the true distribution gives an estimate of the optimal value and solution • Poor scenarios can give inadequate model results and misleading model insights • It is important to evaluate the quality of the stochastic model solution! 10.07.2017 Methodology 10
- 11. v v • A solution evaluation requires • scenarios • an optimised model • optimal value and solution • Satisfactory solution criteria → Model analysis • Unsatisfactory solution criteria → new scenarios & model run • New random scenarios • Different scenario generation method • Higher number of scenarios 10.07.2017 Methodology Data Scenario generation Optimisation Solution evaluation Value, v Solution, x Scenarios, N Model analysis No Yes Satisfactory solution criteria 11
- 12. v v • Scenario generation methods • Random sampling • Statistical methods: Moment matching & Distance measures • It is more important with a good quality solution than to use scenarios that accurately replicate the uncertainty • Solution evaluation methods • General: In-sample and out-of-sample stability tests • Linked to scenario generation method • Sample Average Approximation (SAA) - Random Sampling • Optimality gap of clustered scenarios - Distance measures 10.07.2017 Methodology 12
- 13. v v Scenario generation • Random sampling • Moment matching Solution evaluation • Stability tests • SAA • Used short-term uncertainty • Wind power production • Hydropower production • PV production • Building heat demand • Electricity trade prices 10.07.2017 13 Methodology Random sampling Stability tests Scenario generation Solution evaluation Moment matching Distance measures SAA Optimality gap Grey - thesis
- 14. v v • Examples from the thesis publications, Paper I - Paper IV • The papers use different model assumptions and various uncertain parameters • Focus on the difference between stochastic and deterministic model results & solution quality • Deterministic model • One scenario • Assume all input parameters are certain • Traditional energy system approach 10.07.2017 Results 14
- 15. v v • Electricity generation capacity in Denmark - 2050 • Uncertain wind power production → 43 % lower wind power capacity with a stochastic approach 10.07.2017 Result – Paper I 15
- 16. v v • Difference in heat capacity (deterministic – stochastic) in buildings in Norway and Sweden - 2030 • Uncertain hydro production, wind production & electricity trade prices → NNU: 24 % higher low-cost electricity capacity with a stochastic approach 10.07.2017 Result – Paper II -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 REF NNU HYD LTD Heatcapacity,GW Gas Heat pump Electricity Biomass 16 Policy assumptions: REF, NNU, HYD, LTD
- 17. v v • Difference in Scandinavian heat capacity in buildings • Uncertain hydro production, wind production, electricity trade prices, PV production & heat demand → REF 2050: 71 % higher low-cost electricity capacity with a stochastic approach 10.07.2017 Result – Paper III -10 -8 -6 -4 -2 0 2 REF PBU SUN ZEB ZEB* REF PBU SUN ZEB ZEB* 2030 2050 Deterministic-Stochasticheat capacity,GW Bio Gas Heat Pump Electricity 17 Model instances: REF, PBU, SUN, ZEB, ZEB*
- 18. v v • Difference in Scandinavian electricity generation capacity • Uncertain hydro production, wind production, electricity trade prices, PV production & heat demand → ZEB* 2030: 24 % lower wind power capacity with a stochastic approach 10.07.2017 Result – Paper III -1 0 1 2 3 4 REF PBU SUN ZEB ZEB* REF PBU SUN ZEB ZEB* 2030 2050 Deterministic-Stochastic electricitycapacity,GW Wind Non-flexible hydro CHP 18 Model instances: REF, PBU, SUN, ZEB, ZEB*
- 19. v v • Wind capacity in Denmark – 3 random samples scenarios • Uncertain wind power production → Few scenarios can give poor solutions! 10.07.2017 Result – Paper IV 0 2,000 4,000 6,000 8,000 10,000 12,000 2020 2025 2030 2035 2040 2045 2050 Windcapacity,MW Det M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 19 M = candidate samples
- 20. v v • Wind capacity in Denmark – 60 random samples scenarios • Uncertain wind power production • → SAA identify M5 to be the best solution 10.07.2017 Result – Paper IV 20 M = candidate samples 0 2,000 4,000 6,000 8,000 10,000 12,000 2020 2025 2030 2035 2040 2045 2050 Windcapacity,MW Det M1 M2 M3 M4 M5 M6 M7 M8 M9 M10
- 21. v v • Stochastic Programming is a suitable tool to explicitly consider short- term uncertainty in long-term energy models • It is important to use a scenario representation that gives a good quality of the stochastic solution • For our analyses, a stochastic approach lowers investments in intermittent electricity generation & increases investments in low-cost electricity heating compared to a deterministic approach • We recommend using a stochastic representation of short-term uncertainty in long-term energy models for more solid model insights 10.07.2017 Conclusions 21
- 22. v v Thank you for the attention pernille.seljom@ife.no 10.07.2017 22

No public clipboards found for this slide

Be the first to comment