Vis enkel innførsel

dc.contributor.advisorNygård, Heidi Samuelsen
dc.contributor.advisorTomic, Oliver
dc.contributor.advisorLiland, Kristian Hovde
dc.contributor.authorSheikh-Mohamed, Yasmin Bashir
dc.coverage.spatialNorwayen_US
dc.date.accessioned2021-02-08T14:45:28Z
dc.date.available2021-02-08T14:45:28Z
dc.date.issued2020
dc.identifier.urihttps://hdl.handle.net/11250/2726672
dc.description.abstractAs a part of the fossil fuel phase-out, Norway is investing in electrification and therefore expansion of the transmission grid. This requires efficient and accurate methods to assess long-term reliability and socioeconomic benefits of alternative expansions. The Norwegian Transmission System Operator, Statnett SF, has developed a probabilistic simulation tool, MONSTER, for reliability analysis and transmission system planning. This thesis evaluates methods used for verification of probabilistic results and assesses their suitability for MONSTER predictions. Through a case study, the accuracy of the tool is assessed for different time spans to investigate the tools performance for short-term reliability, and therefore possibility for future application in other areas. A sensitivity analysis is also performed to assess the simulation tool’s sensitivity to inputs. After an assessment of the results from MONSTER and methods used for verification of probabilistic forecasts, Continuous Rank Probability Score (CRPS) was chosen as the main method to evaluate the accuracy of the results. Reliability diagrams and percentile diagram are used as complementary visual tools of assessment. The CRPS score from probabilistic results – in form of a probability density function – can be directly compared to the Mean Absolute Error (MAE) of point-predictions. Therefore, the point-predictions from the simulation tool are assessed using MAE. The case study is based on the Greater Oslo Region over 9 years. The point predictions for yearly intervals have higher accuracy than for 6-month and monthly intervals. This indicates that the tool performs better for predictions with longer time spans. The resulting CRPS score indicates better accuracy for monthly predictions compared to yearly and 6-month predictions. Examining the results closer with the visual assessment tools shows that the CRPS score does not capture deficiencies in the probability distribution and has therefore computed better results for monthly predictions than expected. Use of score methods that detect probability distribution deficiencies is suggested for future evaluations. It is further concluded that the end results of this study are most sensitive to the remedial measures-input. Therefore an expanded use of this feature could result in better predictions.en_US
dc.description.abstractSom en del av utfasingen av fossile brensler satser Norge på elektrifisering. Dette krever utbygging av overføringsnettet. Nøyaktige pålitelighetsanalyser er nødvendig for å sikre pålitelig og samfunnsøkonomisk utbygging. Statnett SF har utviklet et probabilistisk simuleringsverktøy, MONSTER, for å vurdere langsiktig leveringspålitelighet ved å predikere sannsynlighet for ikke-levert energi (ILE) og kostnadene av dette (KILE). Denne oppgaven evaluerer ILE-resultater fra MONSTER ved å først vurdere ulike metoder som brukes i dag for verifisering av probabilistiske modeller. I en casestudie blir nøyaktigheten til verktøyet vurdert for forskjellige tidsperioder for å undersøke mulighetene til å utnytte verktøyet for kortsiktige analyser. Det utføres også en sensitivitetsanalyse for å vurdere sensitiviteten til simuleringsverktøyet for endring i input-variabler.en_US
dc.language.isoengen_US
dc.publisherNorwegian University of Life Sciences, Åsen_US
dc.rightsAttribution-NonCommercial-NoDerivatives 4.0 Internasjonal*
dc.rights.urihttp://creativecommons.org/licenses/by-nc-nd/4.0/deed.no*
dc.titleVerification of probbilistiac predictions for reliability analysis in the Norwegian transmission systemen_US
dc.typeMaster thesisen_US
dc.description.localcodeM-MFen_US


Tilhørende fil(er)

Thumbnail

Denne innførselen finnes i følgende samling(er)

Vis enkel innførsel

Attribution-NonCommercial-NoDerivatives 4.0 Internasjonal
Med mindre annet er angitt, så er denne innførselen lisensiert som Attribution-NonCommercial-NoDerivatives 4.0 Internasjonal