Development of a calibration method for coupled CFD-DEM

Simple item page
dc.contributor.advisorCoetzee, Corneen_ZA
dc.contributor.advisorMeyer, Chrisen_ZA
dc.contributor.authorWasserfall, Jacob Gabrielen_ZA
dc.contributor.otherStellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering.en_ZA
dc.date.accessioned2023-03-02T12:49:48Z
dc.date.accessioned2023-05-18T07:08:36Z
dc.date.available2023-03-02T12:49:48Z
dc.date.available2023-05-18T07:08:36Z
dc.date.issued2023-03
dc.descriptionThesis (MEng)--Stellenbosch University, 2023.en_ZA
dc.description.abstractENGLISH SUMMARY: A fully coupled Computational Fluid Dynamics (CFD) and Discrete Element Method (DEM) model was developed for a granular material submerged in water. The granular material was divided into three distinct particle size distributions (PSD). The particle size and shape were measured and used to create multi-sphere (composite) particles in the model. The bulk density was measured and used to calibrate the particle density, which was also measured directly. The remaining input parameters were calibrated using a large-scale Draw Down Test (DDT) rig under dry, wet and submerged saturation conditions. Due to geometric constraints and the relatively large size of the particles (4 mm to 32 mm), momentum source smoothing had to be used to allow for the large particle-to-cell size ratios. This technique was implemented using the software Simcenter StarCCM+. The effects of the smoothing were investigated and it was found that while the use of smoothing improved model stability, the use of large smoothing lengths resulted in instability. Additionally, a drag coefficient modifier was implemented to improve fluid-particle interactions which were reduced due to the momentum source smoothing method. First, the granular material was calibrated for the dry and wet conditions and the input values for the particle-particle coefficient of friction and the coefficient of restitution determined. For the larger PSD, it was found that the change in saturation condition did not affect the results of the DDT which meant the same calibration parameters were found for both. For the smaller PSD, some change was seen with the increase in saturation but similar input parameters were found for the two conditions. The smallest PSD was influenced the most by the change in saturation which resulted in two distinct sets of input parameters being found for the two conditions. Secondly, the model was calibrated for submerged conditions using the Haider- Levenspiel drag model. The input parameter values were initially set equal to those calibrated for the dry/wet conditions. Under submerged conditions, results showed that the particle-particle coefficient of friction and the drag modifier had a significant influence on the results. Therefore these two parameters were calibrated. It was found that the drag modifier had to be calibrated, while the particle-particle coefficient of friction, calibrated under dry conditions, could be used for the submerged conditions. Thus, the particle-particle coefficient of friction calibrated under dry conditions, could be used for the wet and submerged conditions for the larger PSDs. Finally, a vertical suction pipe validation experiment was conducted. The suction pipe had a constant diameter, but the fluid velocity and the distance the pipe opening was held from the granular bed were varied. The amount of mass (particles) removed as well as the size of the cavity that formed in the material bed was measured and compared to model predictions. The results showed that using the parameter values calibrated in the DDT, too much material was removed (error of 30%). Removing the drag modifier (setting it equal to unity) significantly improved the results (error of 6%). It is concluded that due to the difference in flow mechanism (particle-induced in the DDT versus fluid-induced in the suction pipe), the DDT is not a suitable experiment to calibrate the input parameter values for a suction pipe. It is hypothesized that the flow mechanism and dynamics of the granular material and the fluid in the calibration experiment should be similar to that of the final application being investigated.en_ZA
dc.description.abstractAFRIKAANSE OPSOMMING: ’n Volledig gekoppelde Berekenings Vloei Dinamika (BVD) en Diskreet Element Metode (DEM) model is ontwikkel vir ’n onderwater korrelmateriaal. The materiaal was verdeel in drie Partikel Grootte Verspeidings (PGV). Die partikel groote en vorms is gemeet en gebruik om multi-sfeer partikel samestellings te vorm. Die grootmaat digtheid is gemeet en gebruik om die partikel digtheid te kalibreer wat ook direk gemeet is. The res van die invoer parameterwaardes is gekalibreer met die gebruik van ’n grootskaalse Aftrek Toestel (AFT) onder droe, nat en onderwater versadigingstoestande. As gevolg van geometriese beperkings en die groot partikels wat gebruik is (4 mm tot 32 mm), moes ’n momentum bron vergladding toegepas word om die groot partikel tot sel verhoudings toe te laat. Die effekte van die vergladding is ondersoek en die daar is gevind dat alhoewel dit die stabiliteit van die model verbeter het, het die gebruik van groot vergladding lengtes ook onstabiliteit veroorsaak. ’n Sleurweerstand vermenigvuldiger is toegepas om die vloeistof- partikel-interaksie te verbeter wat verminder is deur die gebruik van die momentum bron vergladding. Eerstens is die korrelmateriaal gekalibreer vir nat en droe toestande en die invoer waardes vir die partikel-partikel wrywing koeffisient en koeffisient van restitusie bepaal. Die groter PGV het geen verandering getoon met die verandering in versadiging toestand nie en daar was dus eenderse kalibrasie parameters gevind vir beide toestande. Die kleiner PGV het wel ’n mate van verandering getoon, maar baie eenderse invoer waardes was gevind aangesien die verandering baie klein was. Die heel kleinste PGV het ’n duidelike verandering getoon soos wat die versadiging verander is en daar was dus twee baie verskillende invoer parameterwaardes gevind vir die twee toestande. Tweedens is die model gekalibreer in onderwater toestande en ’n Haider-Levenspiel sleurmodel gebruik. Die invoer parameterwaardes was aanvanklik gestel na die droog en nat gekalibreerde waardes. Die onderwater toestand resultate het gewys dat die partikel- partikel wrywing koeffisient en sleurweerstand vermenigvuldiger ’n beduidende invloed gehad het op die resultate. Hierdie invoer parameters is dus gekalibreer. Daar is gevind dat die sleurweerstand vermenigvuldiger kalibrasie benodig het, terwyl die partikel-partikel wrywing koeffisient, wat gekalibreer is in droe toestande, net so gebruik kon word in die onderwater toestande. Die partikel-partikel wrywing koeffisient, wat gekalibreer was in droe toestande kon dus toegepas word in beide nat en onderwater toestande vir die groter PGVs. Laastens is ’n vertikale suigpyp-valideringseksperiment uitgevoer. Die suigpyp het ’n konstante diameter gehad maar die afstand tussen die pyp inlaat en die materiaal bed asook die snelheid in die pyp is gevarieer. Die hoeveelheid massa (partikels) wat opgesuig is en die grootte van die gat wat gevorm is gemeet en vergelyk met die model voorspellings. Die resultate het gewys dat die gebruik van die parameterwaardes gekalibeer met die AFT te veel materiaal verwyder het (fout van 30%). Deur die sleurweerstand vermenigvuldiger te verwyder (gelyk te maak aan 1) het die resultate aansienlik verbeter (fout van 5%). Ten slotte word die gevolgtrekking gemaak dat as gevolg van die verskil in vloei meganismes (partikel gedrewe in die AFT en vloeistof gedrewe in die suigpyp), is die AFT nie geskik om die invoer parameterwaardes vir die suigpyp te kalibreer nie. Die hipotese word voorgestel dat die vloei meganisme en dinamika van die korrelmateriaal en die vloeistof in die kalibrasie eksperiment vergelykbaar moet wees met die finale toepassing wat ondersoek word.en_ZA
dc.description.versionMasters
dc.format.extentxvi, 74 pages : illustrations
dc.identifier.urihttp://hdl.handle.net/10019.1/127183
dc.language.isoen_ZAen_ZA
dc.publisherStellenbosch : Stellenbosch University
dc.rights.holderStellenbosch University
dc.subject.lcshComputational fluid dynamicsen_ZA
dc.subject.lcshDiscrete element methoden_ZA
dc.subject.lcshFluid dynamicsen_ZA
dc.subject.nameUCTD
dc.titleDevelopment of a calibration method for coupled CFD-DEMen_ZA
dc.typeThesisen_ZA
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
wasserfall_development_2023.pdf
Size:
6.45 MB
Format:
Adobe Portable Document Format
Description:
Download
Collections
Masters Degrees (Mechanical and Mechatronic Engineering)