Vis enkel innførsel

dc.contributor.authorNezhadali, Mohammad
dc.date.accessioned2023-03-21T07:56:47Z
dc.date.available2023-03-21T07:56:47Z
dc.date.issued2023-04-14
dc.date.submitted2023-03-06T11:04:12.594Z
dc.identifiercontainer/41/94/23/c9/419423c9-3eae-4d18-b174-d313ab71b647
dc.identifier.isbn9788230859636
dc.identifier.isbn9788230855935
dc.identifier.urihttps://hdl.handle.net/11250/3059410
dc.description.abstractI ensemble-basert data-assimilering (DA) er størrelsen på ensemblet vanligvis begrenset til hundre medlemmer. Rett frem bruk av ensemble-basert DA kan resultere i betydelig Monte Carlo-feil, som ofte viser seg som alvorlig undervurdering av parameterusikkerheter. Assimilering av store mengder samtidige data forsterker de negative effektene av Monte Carlo-feilen. Avstandsbasert lokalisering er det konvensjonelle middelet for å begrense dette problemet. Denne metoden har imidlertid sine egne ulemper. Den vil, f.eks., fjerne sanne korrelasjoner over lange distanser og det er svært vanskelig å benytte på data som ikke har en unik fysisk plassering. Bruk av modeller med lavere kvalitet reduserer beregningskostnadene per ensemble-medlem og gir derfor muligheten til å redusere Monte Carlo-feilen ved å øke ensemble-størrelsen. Men, modeller med lavere kvalitet øker også modelleringsfeilen. Data-assimilering på flere nivåer (MLDA) bruker et utvalg av modeller som danner hierarkier av både beregningskostnad og beregningsnøyaktighet, og prøver åå oppnå en bedre balanse mellom Monte Carlo-feil og modelleringsfeil. I dette PhD-prosjektet ble flere MLDA-algoritmer utviklet og deres kvalitet for assimilering av inverterte seismiske data ble vurdert på forenklede reservoarproblemer. Bruk av modeller på flere nivå innebærer introduksjon av noen numeriske feil (multilevel modeling error, MLME), i tillegg til de allerede eksisterende numeriske feilene. Flere beregningsmessig rimelige metoder ble utviklet for delvis å kompansere for MLME i gjennomføring av data-assimilering på flere nivåer. Metodene ble også undersøkt under historie tilpassing på forenklede reservoar problemer. Til slutt ble en av de nye MLDA-algoritmene valgt og ytelsen ble vurdert på et historie tilpassings problem med en realistisk reservoar modell.en_US
dc.description.abstractIn ensemble-based data assimilation (DA), the ensemble size is usually limited to around one hundred. Straightforward application of ensemble-based DA can therefore result in significant Monte Carlo errors, often manifesting themselves as severe underestimation of parameter uncertainties. Assimilation of large amounts of simultaneous data enhances the negative effects of Monte Carlo errors. Distance-based localization is the conventional remedy for this problem. However, it has its own drawbacks, e.g. not allowing for true long-range correlations and difficulty in assimilation of data which do not have a specific physical location. Use of lower-fidelity models reduces the computational cost per ensemble member and therefore renders the possibility to reduce Monte Carlo errors by increasing the ensemble size, but it also adds to the modeling error. Multilevel data assimilation (MLDA) uses a selection of models forming hierarchies of both computational cost and computational accuracy, and tries to obtain a better balance between Monte Carlo errors and modeling errors. In this PhD project, several MLDA algorithms were developed and their quality for assimilation of inverted seismic data was assessed in simplistic reservoir problems. Utilization of multilevel models entails introduction of some numerical errors (multilevel modeling error, MLME) to the problem in addition to the already existing numerical errors. Several computationally inexpensive methods were devised for partially accounting for MLME in the context of multilevel data assimilation. They were also investigated in simplistic reservoir history-matching problems. Finally, one of the novel MLDA algorithms was chosen and its performance was assessed in a realistic reservoir history-matching problem.en_US
dc.language.isoengen_US
dc.publisherThe University of Bergenen_US
dc.relation.haspartPaper A: Mohammad Nezhadali, Tuhin Bhakta, Kristian Fossum, and Trond Mannseth. A novel approach to multilevel data assimilation. In ECMOR XVII, vol. 2020, no. 1, pp. 1-13. European Association of Geoscientists & Engineers, 2020. Full text not available in BORA due to publisher restrictions. The article is available at: <a href="https://doi.org/10.3997/2214-4609.202035091" target="blank">https://doi.org/10.3997/2214-4609.202035091</a>en_US
dc.relation.haspartB Mohammad Nezhadali, Tuhin Bhakta, Kristian Fossum, and Trond Mannseth. Multilevel assimilation of inverted seismic data with correction for multilevel modeling error. Frontiers in Applied Mathematics and Statistics 7 (2021): 673077. The article is available in the thesis. The article is also available at: <a href="https://doi.org/10.3389/fams.2021.673077" target="blank">https://doi.org/10.3389/fams.2021.673077</a>en_US
dc.relation.haspartC Mohammad Nezhadali, Tuhin Bhakta, Kristian Fossum, and Trond Mannseth. Iterative multilevel assimilation of inverted seismic data. Computational Geosciences 26, no. 2 (2022): 241-262. The article is available in the thesis. The article is also available at: <a href="https://doi.org/10.1007/s10596-021-10125-3" target="blank">https://doi.org/10.1007/s10596-021-10125-3</a>en_US
dc.relation.haspartD Mohammad Nezhadali, Tuhin Bhakta, Kristian Fossum, and Trond Mannseth. Sequential multilevel assimilation of inverted seismic data. Computational Geosciences 27, (2023): 265-287. The article is available at: <a href="https://hdl.handle.net/11250/3073272" target="blank">https://hdl.handle.net/11250/3073272</a>en_US
dc.relation.haspartE Mohammad Nezhadali, Tuhin Bhakta, Kristian Fossum, and Trond Mannseth. Towards Application of Multilevel Data Assimilation in Realistic Reservoir History- Matching Problems. In ECMOR 2022, vol. 2022, no. 1, pp. 1-12. European Association of Geoscientists & Engineers, 2022. Full text not available in BORA due to publisher restrictions. The article is available at: <a href="https://doi.org/10.3997/2214-4609.202244029" target="blank">https://doi.org/10.3997/2214-4609.202244029</a>en_US
dc.rightsIn copyright
dc.rights.urihttp://rightsstatements.org/page/InC/1.0/
dc.titleMultilevel assimilation of inverted seismic dataen_US
dc.typeDoctoral thesisen_US
dc.date.updated2023-03-06T11:04:12.594Z
dc.rights.holderCopyright the Author. All rights reserveden_US
dc.contributor.orcid0000-0002-9151-8547
dc.description.degreeDoktorgradsavhandling
fs.unitcode12-11-0


Tilhørende fil(er)

Thumbnail

Denne innførselen finnes i følgende samling(er)

Vis enkel innførsel