SWASH: A robust numerical model for shallow water coastal engineering applications

Van Eeden, Frans (2017-03)

Thesis (MEng)--Stellenbosch University, 2017.

Thesis

ENGLISH ABSTRACT: Integral to coastal engineering design is the understanding and analysis of the wave climate at an arbitrary site under consideration. In most cases, these wave climate and extreme conditions are simulated by means of numerical computer models. These models a ord the coastal engineer the tools to test a wide range of climatic and extreme conditions applicable to the design. The output from these models is used as input into the design formulas for the various structural elements. However, in order for these numerical models to be reliable, some form of calibration is needed. The surf zone is an area of coastal engineering that is generally di cult to simulate numerically, as this is a highly dynamic zone with numerous interactive processes. The nature of the surf zone makes collecting calibration data to use in numerical models fairly di cult to nearly impossible. Researchers and coastal engineers turn to laboratory measurements to quantify surf zone processes and to calibrate numerical models. A non-hydrostatic numerical model SWASH (SimulatingWAves till SHore) is analysed in terms of its numerical capability to simulate wave breaking, in particular, spilling and plunging waves. Laboratory data were made available for calibration purposes. The data describe regular spilling and plunging breakers on plane slopes of 1:20 and 1:35. Water levels and instantaneous horizontal and vertical velocities at various positions pre- and post-wave breaking were measured using particle image velocimetry and laser-Doppler velocimetry. The laboratory data are used to test the sensitivity of the SWASH model in terms of numerical parameters; i.e. the explicit conservation of momentum, bed friction and various discretisation and interpolation schemes that in uence the stability and accuracy of the numerical model. The sensitivities are compared to the laboratory measured data sets in terms of the incipient breaking point, phase averaged water levels and the phase averaged horizontal and vertical particle velocities. The results of the sensitivity analysis are subsequently used to test the robustness of the SWASH model by comparing three modes of the SWASH model: a calibrated wave model, an uncalibrated model (using default or recommended parameters) and the hydrostatic front approximation. The three modes are compared to the measured data and to each other in terms of breaker depth index, wave setup, phase averaged water levels and the phase averaged horizontal and vertical particle velocities and modelled turbulence. The root mean squared error is calculated between the model data and the measured data in order to gauge the accuracy of the three modes. It can be concluded that the SWASH model is capable of numerically representing surf zone processes that can assist the engineer in his designs. In addition, the model performs very well in most areas when an uncalibrated SWASH model is employed with accuracy limited to that of the preliminary design stage. The SWASH model offers a vivid array of numerical schemes and parameters that can be used for calibration purposes. However, the exibility of the model does not overshadow the robustness and simplicity with which the model can be set up. Regardless, the model does live up to its claim of robustness and accuracy without specifying external parameters to a preliminary design level. Beyond this design level, proper calibration of the model is necessary before the model results can be utilised in detailed design.

AFRIKAANSE OPSOMMING: 'n Deel van kusingenieursontwerp is die begrip en ontleding van die golfklimaat van toepassing op 'n bepaalde terrein onder oorweging. In die meeste gevalle word hierdie golfklimaat en uiterste toestande gesimuleer met behulp van numeriese rekenaarmodelle. Hierdie modelle laat die kusingenieur toe om 'n wye verskeidenheid van klimaatstoestande en uiterstes van toepassing op die ontwerp te toets. Die uitvoer van hierdie modelle word gebruik as insette in die ontwerpformules vir die verskillende strukturele elemente. Ten einde die resultate van hierdie numeriese modelle te vertrou, is 'n vorm van yking nodig. Die brandersone is 'n studiegebied van kusingenieurswese wat oor die algemeen moeilik is om numeries te simuleer omrede dit 'n baie dinamiese gebied met talle interaktiewe prosesse bevat. Die aard van die brandersone maak die meting en versameling van ykingsdata om te gebruik in numeriese modelle baie moeilik of selfs bykans onmoontlik. Navorsers en kusingenieurs moet hulle dus wend na laboratoriummetings om die brandersone-prosesse te kwanti seer en die numeriese modelle te kan yk. 'n Nie-hidrostatiese numeriese model SWASH (SimulatingWAves till SHore) word ontleed in terme van sy numeriese vermoë om golfbreking te simuleer, spesi ek met betrekking tot rollende- en plonsende branders. Laboratoriumdata vir yking is beskikbaar gemaak. Die data beskryf reëlmatige rollende- en plonsendebranders wat oor reëlmatige hellings van 1:20 en 1:35 beweeg. Watervlakke en oombliklike horisontale en vertikale snelhede by verskeie posisies voor- en na golfbreking is deur Particle Image Velocimetry en Laser Doppler Velocimetry -tegnieke gemeet. Die laboratoriumdata word gebruik om die sensitiwiteit van die SWASHmodel in terme van numeriese verandelikes te toets; met betrekking tot die eksplisiete behoud van momentum, bodemwrywing en verskeie diskretisasie- en interpolasieskemas wat die stabiliteit en akkuraatheid van die numeriese model beïnvloed. Die sensitiwiteit van die model word vergelyk met die laboratoriumgemete datastelle in terme van breekpunt, fase-gemiddelde watervlakke en die fase-gemiddelde horisontale en vertikale snelhede. Die resultate van die sensitiwiteitsontleding is gebruik om die robuustheid van die SWASH-model te toets deur dit te vergelyk met drie tipes SWASHmodelle: 'n geykte golfmodel, 'n ongeykte model (met behulp van standaard of aanbevole veranderlikes) en die hidrostatic front approximation . Die drie tipes word vergelyk met die gemete data en met mekaar in terme van brander indeks, golfopstuwing, fase-gemiddelde watervlakke en die fase-gemiddeld horisontale en vertikale snelhede en turbulensie. Die standaardafwyking is bereken tussen die modeldata en die gemete data ten einde die akkuraatheid van die drie tipes modelle te meet. Dit kan afgelei word dat die SWASH-model in staat is om numeries brandersone prosesse te simuleer wat die ingenieur in sy ontwerpe kan help. Daarbenewens presteer die model baie goed in die meeste gebiede wat getoets is. Die ongeykte SWASH-model kan gebruik word met 'n akkuraatheid wat tot diè van die voorlopige ontwerp-stadium beperk is. Die SWASH-model bied 'n wye verskeidenheid van numeriese skemas en veranderlikes wat vir ykings doeleindes gebruik kan word. Dit oorskadu egter nie die robuustheid en eenvoud waarmee die model kan opgestel en gebruik word nie. Ongeag, die SWASH-model bewys sy aanspraak van robuustheid en akkuraatheid sonder om eksterne ykingsveranderlikes hoef in te voer; natuurlik is die aanspraak geldig op 'n voorlopige ontwerp-vlak. Vir detailontwerp is behoorlike yking van die model nodig voordat die modelresultate kan gebruik word.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/101097
This item appears in the following collections: