Sediment transport dynamics in dam-break modelling

Msadala, Vincent P. C. (2016-03)

Thesis (DEng)--Stellenbosch University, 2016.

Thesis

ENGLISH ABSTRACT: The prediction of breach characteristics during earth embankment dam-break modelling due to overtopping is dependent on the accurate modelling of the intrinsic breaching processes such as flow hydraulics, bed level changes, soil mechanics and sediment transport dynamics. Earth dams are characterised by steep embankment slopes (up to 1:2 (V:H), 50%). Commonly used sediment transport equations in dam-break modelling were based on slopes of up to 20%. The application of sediment transport equations that were derived from data on mild or moderately steep slopes is one of the sources of uncertainty in dam-break modelling. The general objective of this study was to develop new empirical sediment transport equations for steep slopes that can be applied in homogeneous earth embankment dam-break modelling and to assess the uncertainty that is associated with the input of sediment transport equations. Physical experimental studies were conducted at the Hydraulic Laboratory of the University of Stellenbosch, South Africa. A 4.7 m long, 0.25 m wide and 0.3 m deep flume was used. Three preselected bed slope configurations of 25%, 33% and 40% were implemented. A total of 87 tests were carried out with median (d50) sediment sizes of 0.2 mm, 1.0 mm and 2.4 mm. New empirical sediment transport equations for steep slopes were developed based on the data from the experimental study. One of the two proposed sediment transport equations incorporated six parameters in its formulation, namely the particle sizes of the sample for which 30 % and 90% were finer (d30 and d90); shear velocity, average flow velocity, friction slope, dimensionless shear stress and critical dimensionless shear stress. The validity or statistical significance of the proposed sediment transport equations (Equations 5.2-5 and 5.2-6) was determined using statistical tests (F-Test and r2). The predictive capability of the proposed sediment transport equations (Equations 5.2-5 and 5.2-6) was confirmed by the degree of correlation between measured and predicted sediment transport rates. There was a deviation of less than 22% between the measured and predicted sediment transport rates. A comparative analysis was done between measured and predicted sediment transport rates using selected sediment transport equations from literature. The analytical comparison showed that the sediment transport predictions by one of the newly calibrated sediment transport equations (Equation 5.2-6) were within the same order of magnitude as those of the Meyer-Peter Müller (1948) and Camenen & Larson (2005) sediment transport equations, even though the Meyer-Peter Müller (1948) equation overestimated sediment transport rates at higher unit discharges. The Smart and Jäeggi (1983) sediment transport equation predicted higher sediment transport rates than the measured sediment transport rates and those predicted by Equation 5.2-6. The MIKE 21C two dimensional hydrodynamic model of the DHI Group was successfully applied to evaluate and compare the performance of the newly calibrated sediment transport equations in dam-break modelling using five dam-break case studies. The numerical simulation results using the newly calibrated sediment transport equations (Equations 5.2-5 and 5.2-6) were compared with the output results that were simulated by two selected sediment transport equations from literature, namely those of Camenen & Larson (2005) and Smart & Jäeggi (1983). Specifically, Case studies 1 and 2 were applied to analyse the effect of sediment transport equations in the simulation of temporal bed level changes and the simulated breach shape respectively. Case studies 3 and 4 investigated the numerical modelling of dam-break outflow hydrographs on very steep slopes. Case Study 5 was used to analyse the effect of sediment transport equations on the simulated peak discharge and outflow volume and showed that the newly calibrated equations are applicable at prototype scale. The newly calibrated sediment transport equations performed better in Case studies 1 and 2 where the embankment slopes were within the equations’ calibration range (25% - 40% slopes). Consistent and realistic predictions of numerical model output parameters were not possible when the sediment transport equations were applied to Case studies 3 and 4 where the embankment slopes were outside the calibration range of all four sediment transport equations (Equations 5.2-5 and 5.2-6, Camenen & Larson (2005) and Smart and Jäeggi (1983)). The original contribution to knowledge of this dissertation is the development of sediment transport equations that are based on steep embankment slopes and are suitable for application in dam-break modelling. For dam-break studies, sensitivity analysis of model input parameters such as resistance and alternative sediment transport equations is recommended in order to evaluate a range of scenarios. The sensitivity analysis that was accomplished pertaining to Manning’s resistance coefficient (n) showed that a lower Manning’s resistance coefficient (n) in the numerical model resulted in a higher peak outflow discharge.

AFRIKAANSE OPSOMMING: Die voorspelling van breekkenmerke in die breekmodellering van gronddamme as gevolg van vloedwater wat oor die nie oorloopkruin van die dam spoel, berus op die akkurate modellering van die intrinsieke breekprosesse, soos vloeihidroulika, veranderinge in die bodemvlak, grondmeganika en sedimentvervoerdinamika. Gronddamme word gekenmerk deur steil wal hellings (tot 1:2 (V:H), 50%). Tog is die vergelykings wat meestal vir die berekening van sedimentvervoer gebruik word, gegrond op hellings van tot slegs 20%. Die toepassing van sedimentvervoervergelykings wat afgelei is van data vir matige hellings is dus een van die bronne van onsekerheid in dambreekmodellering. Daarom was die algemene oogmerk van hierdie studie om nuwe empiriese sedimentvervoervergelykings vir steil hellings te ontwikkel wat in homogene gronddambreekmodellering toegepas kan word, en om die onsekerheid te beoordeel wat met die inset van sedimentvervoervergelykings verband hou. Fisiese modelstudies is in die Hidrouliese Laboratorium van die Universiteit Stellenbosch onderneem. ’n Kanaal met ’n lengte van 4.7 m, ’n breedte van 0.25 m en ’n diepte van 0.3 m is gebruik. Drie voorafgekose damhellings van 25%, 33% en 40% is toegepas. Altesaam 87 toetse is met gemiddelde (d50) sedimentgroottes van 0.2 mm, 1.0 mm en 2.4 mm uitgevoer. Nuwe empiriese sedimentvervoervergelykings vir steil hellings is op grond van die data uit die proefstudies ontwikkel. Die formulering van een van die twee voorgestelde sedimentvervoervergelykings het ses parameters ingesluit, naamlik die deeltjiegrootte van die monster, waarvan 30% en 90% fyner was (d30 en d90); sleursnelheid, gemiddelde vloeisnelheid, wrywingshelling, dimensielose skuifspanning en kritieke dimensielose skuifspanning. Die geldigheid of statistiese beduidendheid van die voorgestelde sedimentvervoervergelykings (vergelykings 5.2-5 en 5.2-6) is met behulp van statistiese toetse (F-toets en r2) bepaal. Die voorspellingsvermoë van die vergelykings is bevestig deur die mate van korrelasie tussen die gemete en voorspelde sedimentvervoertempo’s, wat minder as 22% van mekaar verskil het. ’n Vergelykende ontleding van die gemete en voorspelde sedimentvervoertempo’s is met behulp van gekose sedimentvervoervergelykings uit die literatuur uitgevoer. Dít het getoon dat die sedimentvervoervoorspellings van die pas gekalibreerde vergelykings (vergelyking 5.2-6) in dieselfde grootteorde as dié van die vergelykings van Meyer-Peter Müller (1948) en Camenen & Larson (2005) val, hoewel Meyer-Peter Müller se vergelyking sedimentvervoertempo’s met hoër eenheidsdeurstromings oorskat het. Smart & Jäeggi se sedimentvervoervergelyking (1983) het hoër Stellenbosch University https://scholar.sun.ac.za v vervoertempo’s vergeleke met die gemete waardes sowel as die voorspelde tempo’s van vergelyking 5.2-6 voorspel. Hierna is die DHI Groep se tweedimensionele hidrodinamiese model MIKE 21C suksesvol toegepas om die akkuraatheid van die pas gekalibreerde sedimentvervoervergelykings in dambreekmodellering in vyf dambreekgevallestudies te beoordeel en te vergelyk. Die numeriese simulasieresultate van die pas gekalibreerde sedimentvervoervergelykings (vergelykings 5.2-5 en 5.2-6) is met die gesimuleerde uitsetresultate van twee gekose vergelykings uit die literatuur, naamlik dié van Camenen & Larson (2005) en Smart & Jäeggi (1983), vergelyk. Meer bepaald is gevallestudie 1 en 2 toegepas om die uitwerking van die sedimentvervoervergelykings op gesimuleerde veranderinge in die bodemvlak oor tyd en die gesimuleerde breekvorm onderskeidelik te ontleed. Gevallestudies 3 en 4 het die numeriese modellering van uitvloeihidrograwe by dambreke met baie steil hellings ondersoek. Gevallestudie 5 is gebruik om die uitwerking van die sedimentvervoervergelykings op die gesimuleerde piek uitvloei deurstroming en uitvloeivolume te ontleed, en het getoon dat die pas gekalibreerde vergelykings op prototipeskaal toegepas kan word. Die prestasie van die pas gekalibreerde vergelykings was beter in gevallestudies 1 en 2, waar die damhellings binne die kalibreerbestek van die vergelykings geval het (25% - 40% helling). Konsekwente en realistiese voorspellings van numeriese modeluitsetparameters was onmoontlik toe die vergelykings op gevallestudie 3 en 4 toegepas is, omdat die damhellings buite die kalibreerbestek van ál vier vergelykings (vergelyking 5.2-5 en 5.2-6, Camenen & Larson (2005) en Smart & Jäeggi (1983)) geval het. Hierdie verhandeling lewer ’n oorspronklike bydrae deur die ontwikkeling van sedimentvervoervergelykings vir steil damhellings wat in dambreekmodellering toegepas kan word. Sensitiwiteitsontleding van modelinsetparameters soos weerstand en alternatiewe sedimentvervoervergelykings word vir dambreekstudies aanbeveel sodat ’n verskeidenheid scenario’s beoordeel kan word. Die sensitwiteitsontleding wat met betrekking tot Manning se ruheid (n) uitgevoer is, het getoon dat ’n laer (n) in die numeriese model tot ’n hoër piekuitvloei gelei het.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/98395
This item appears in the following collections: