Measuring and modelling the influence of weather factors on CSP reflector soiling

Thesis (MEng)--Stellenbosch University, 2016.ENGLISH ABSTRACT: South Africa is among the countries with a plan to reduce GHG emissions of up to 34% by 2020 through investment in renewable energy in order to reduce its base load coal generation. High solar resource and the potential of concentrated solar power (CSP) to address the challenges facing other solar technologies have attracted CSP investment in the country. CSP uses reflector surfaces or mirrors to focus low solar energy radiation from a large field into a small area of high energy concentration. In their working conditions, CSP reflector surfaces are subjected to harsh outdoor environments which drastically degrade their performance. The main objective of the study was to develop and test reflector optical degradation measurement device and use these optical degradation measurements from the device to relate optical losses caused by soiling to weather. A reflector optical degradation assessment device termed as real-time cleanliness monitoring sensor (RCMS) was designed, developed and tested for its ability to measure soiling optical degradation. The device was then utilised to relate weather conditions with optical losses caused by soiling on CSP reflectors. Furthermore, a neural network model was developed to simultaneously relate various weather factors to the optical loss caused by soiling on CSP reflectors. Error analysis and calibration were undertaken for RCMS measurement in order to improve confidence in the data, which was further used in experimental analysis. Two experimental analyses were carried out. The results from the first experimental analysis showed only the variation of cleanliness with weather factors that directly influence the rate of soiling. Wind speed and humidity were observed to degrade cleanliness, while rain lead to reflector cleaning. In the second experimental analysis, factors that directly or indirectly influence cleanliness were statistically analysed using clustering method. The results showed that temperature and direct normal irradiation (DNI) correlate relatively well with cleanliness although they do not directly influence it. The neural network model demonstrated that a combination of weather factors could be used to estimate the optical degradation caused by soiling on CSP reflectors. High coefficient of determination was observed from the neural network model results, as compared to the correlations that considered the relationship between cleanliness and a single weather factor, done in the experimental analysis.AFRIKAANSE OPSOMMING: Suid-Afrika is een van die lande wat oor ’n plan beskik om die vrylating van kweekhuisgas (KHG) met tot 34% teen 2020 te verminder deur in hernubare energie te belê en sy basislas-steenkoolopwekking te verlaag. Die omvangryke sonkraghulpbron en die potensiaal van gekonsentreerde sonkrag (GSK) om die uitdagings wat ander sonkragtegnologieë in die gesig staar te trotseer, lok GSK-belegging vir die land. GSK gebruik reflektoroppervlakke of spieëls om lae sonkraguitstraling vanaf ’n uitgestrekte gebied na ’n klein area met ’n hoë konsentrasie energie te herlei. GSKreflektoroppervlakke word in hul werkende toestand aan strawwe buitelugomgewings onderwerp, wat hul werkverrigting drasties verlaag of degradeer. Die studieoogmerk was om ’n model vir die beraming van hierdie optiese degradasie, wat die gevolg van die bevuiling van GSK-reflektoroppervlakke vanweë weerfaktore is, te oorweeg en te ontwikkel. ’n reflektor- optiese degradasie-assesseringstoestel, of intydse "skoongehalte-monitorsensor (ISMS), is ontwerp, ontwikkel en getoets om die vermoë van die toestel vir die meet van bevuilings- optiese degradasie te bepaal. Die toestel word ook gebruik om die verband tussen weersomstandighede en optiese verliese weens vuil GSKreflektors aan te dui. Daarby is ’n neuralenetwerkmodel ontwikkel wat verskillende weerfaktore gelyktydig met optiese verliese vanweë die bevuiling van GSK-reflektors in verband bring. Foutontleding en kalibrering is vir ISMS-meting gedoen om databetroubaarheid te verhoog en by eksperimentele ontledings te gebruik. Twee eksperimentele ontledings is uitgevoer. Die resultate van die eerste eksperimentele ontleding het slegs die “skoongehalte”-variasie by weersomstandighede wat die bevuilingskoers direk beïnvloed, aangedui. Daar is bevind dat windspoed en vogtigheid skoongehalte degradeer, terwyl reën die reflektors skoon spoel. In die tweede eksperimentele ontleding is faktore wat ’n direkte of indirekte uitwerking op “skoongehalte” het statisties aan die hand van die klustermetode ontleed. Die uitslae toon dat temperatuur en direkte normale irradiasie (DNI) relatief goed met skoongehalte korreleer, ofskoon dit nie ’n direkte uitwerking op skoongehalte het nie. Die neuralenetwerkmodel toon dat ’n kombinasie van weerfaktore gebruik kan word om ’n raming van die optiese degradasie wat deur bevuiling aan die GSK-reflektors aangerig word, te gee. ’n Hoë vasstellingskoëffisiënt is by die neuralenetwerkmodel waargeneem in teenstelling met die korrelasies wat die verband tussen skoongehalte en ’n enkele weerfaktor oorweeg het, soos in die geval van die eksperimentele ontleding