SUNScholar will be offline for maintenance from 10:00 SAST on Wednesday the 21st of November 2018.

Development of a ray tracer for concentrating solar power systems

Bode, Sebastian-James (2017-12)

Thesis (MEng)--Stellenbosch University, 2017.

Thesis

ENGLISH ABSTRACT: The world limate system is hanging and recent researh shows that man-made arbon dioxide emissions from proesses such as coal-fired power generation is the most likely cause. In South Afri a the bulk of eletricity generation comes from coal power plants. In addition to the possible environmental damage caused by these plants, South Africa's reliance on coal exposes the country to possible fuel shortages and rising fuel prices. Concentrating Solar Power (CSP), which uses lean, renewable solar energy for electricity generation, an possibly mitigate these risks and reduce South Africa's dependence on coal. However, for CSP to play a significant role in electriciity generation, its Levelised Cost of Electricity (LCOE) must be reduced. This an be achieved through research and development of CSP systems. Numerical software tools, such as ray tracing, play an essential role in design and optimisation of these systems. The aim of this research was to develop, program, and validate a CSP ray tracer which an be used for ongoing research at the Solar Thermal Energy Research Group (STERG) at Stellenbosch University. The ray tracer, which has been named SUNRAY (Stellenbosch UNiversity RAY tra er), was written in C++. It was developed with realisti sunshape and reflection modules, a number of geometric shapes, automatic tracking algorithms, acceleration routines, and a method to simulate actual mirror surface profiles using surface scanned data. SUNRAY has been extensively validated using behavioural tests, comparative tests against previously validated ray tra ers, and through experimental investigation. For the behavioural cases, it was found that SUNRAY was able to resolve the correct solution for every test. In the comparative tests the relative difference for the power and flux distribution between SUNRAY and validated ray tracers was, on average, no more than 0.4% and 3%, respetively. Furthermore, the execution times for the simulations were, in most cases, faster than that of the validated codes. In the experimental validation various mirror shape profiles were tested under different weather conditions. The experimental tests demonstrated that SUNRAY an be used to adequately determine the magnitude of flux and flux distribution on a target. This thesis presents the theory behind the various features, algorithms, and routines of SUNRAY as well as the validation of these features. Two novel algo- rithms are also proposed in this thesis. The first is a method which reduces the number of missed rays by only generating rays above each object in the simulation. The second uses statistical tools to predict the number of rays which are needed in a simulation.

AFRIKAANSE OPSOMMING: Die wêreld se klimaatstelsel is besig om te verander en onlangse navorsing het bewys dat mensgemaakte olstofdioksieduitlaatgasse, wat afkomstig is van die gebruik van steenkool as brandstof vir lektrisiteitsopwekking, waarskynlik die oorsaak daarvan is. Bykomend tot die skade wat deur hierdie aanlegte veroorsaak kan word, stel die feit dat Suid-Afrika tot ‘n groot mate staatmaak op steenkool die land bloot aan ‘n moontlike brandstoftekort en stygende brandstofpryse. Ge-konsentreerde sonkrag (Concentrating Solar Power - CSP), wat skoon, hernubar sonenergie gebruik om elektrisiteit op te wek, kan moontlik hierdie risiko's teen-werk en Suid-Afrika se afhanklikheid van steenkool verminder. Om werklik ‘n beduidende rol in elektrisiteitsopwekking te kan speel, moet die koste van elektri-siteit wat deur CSP opgewek word, egter verminder word. Dit kan bereik word deur CSP-stelsels na te vors en te ontwikkel. Numeriese programmatuurgereed skap, soos straalsporing, speel ‘n noodsaaklike rol in die ontwerp en optimering van hierdie stelsels. Die doel van hierdie navorsing was om ‘n CSP-straalspoorder te ontwikkel, te programmeer en te sertifiseer, wat gebruik kan word in voortgesotte navorsing deur die Sontermiese Energienavorsingsgroep (STERG) by Stellenbosch Universiteit. Dié straalspoorder, wat SUNRAY (Stellenbosch University RAY tracer) genoem word, is in C++ geskryf en ontwikkel met realistiese modelle van son-vorms, refleksiemodules, ‘n aantal geometriese vorms, outomatiese naspooralgo- ritmes, versnellingsroetines en ′n metode om werklike spieëloppervlak-profiele te simuleer deur van geskandeerde data gebruik te maak. SUNRAY is uitgebreid getoets deur middel van gedragstoetse asook vergelykende toetse tussen SUNRAY en kommersiële straalspoorders wat hulself al bewys het. Verder is dit ook getoets deur eksperimentele ondersoeke. Tydens die gedragstoetse is daar bevind dat SUNRAY daartoe in staat is om die korrekte oplossing vir elke toets te vind. In die vergelykende toetse was die relatiewe verskil tussen SUNRAY en die getoet- ste straalspoorders se oplossings vir krag en vloedverspreiding gemiddeld nie meer as onderskeidelik 0.4% en 3% nie. Verder was die uitvoertye vir die simulasies in meeste gevalle vinniger as dié van die gekontroleerde sagteware. In die eksperi- mentele kontroletoetse is verskeie spieëlvorm-profiele tydens verskillende weersom-standighede getoets. Die eksperimentele toetse het daarop gedui dat SUNRAY gebruik kan word om die omvang en verspreiding van vloed op ‘n teiken te bepaal. In hierdie tesis word die teorie onderliggend aan SUNRAY se kenmerke, algoritmes en roetines bespreek en die sertifisering van hierdie aspekte word aangebied. Twee nuwe algoritmes word ook in hierdio tesis voorgestel. Die eerste es 'n metode wat die verlie aan strale verminder deur slegs strale bo elke voorworp in die simulasie te genereer en die tweede is ’n algoritme wat van statistiese metodes gebruik maak om die aantal strale wat in ‘n simulasie benodig word, te voorspel.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/102825
This item appears in the following collections: