Year 2009 PhD Thesis Development and validation of methods used to compute time values of indoor daylight illuminances Presented in front of: University of Ljubljana Faculty of Electrical Engineering L’Institut National des Sciences Appliquées de LYON Formation doctorale: Génie Civil Ecole doctorale: M.E.G.A. By: Matej Bernard KOBAV Presented on January 30, 2009 Reviewer Reviewer Thesis supervisor Thesis supervisor Member Member prof. dr. Peter Žunko prof. dr. Pierre Ineichen prof. dr. Grega Bizjak prof. dr. Dominique Dumortier prof. dr. Marc Fontoynont prof. dr. Stanislav Kovaèiè Research laboratories: Laboratoire des Sciences de l'Habitat, DGCB, Ecole Nationale des Travaux Publics de l'Etat, rue Maurice Audin, 69120 Vaulx-en-velin, France. Laboratory of the Lighting and Photometry, Faculty of Electrical Engineering, University of Ljubljana, Tržaška 25, SI-1000 Ljubljana, Slovenia Dedicated to Manja, Ema and Matija Acknowledgments Acknowledgments This PhD thesis is a result of my post-graduate study at Faculty of Electrical Engineering and very good collaboration with Laboratoire des Sciences de l’Habitat, Ecole Nationale des Travaux Publics de l’Etat, Vaulx-en-Velin. I wish to thank my two mentors for their support, patience and spent time. Their gentle but firm direction has been most appreciated. Dr. Grega Bizjak was particularly helping in guiding me toward the finish. Dr. Dominique Dumortier had always the right answers to all my questions and doubts. He was more that just a mentor when I was staying in Lyon and he lent me a working table in his office when I was at ENTPE. Since working conditions are always an important issue, I would like to thank also my mates in the office. Boštjan, Tomaž, Ambrož and Uroš, thank you. I would also like to thank Manja for being with me, giving me all the needed support and love and giving birth to two beautiful children, who fill me up with energy and love. Finally I would like to thank my parents and brother and sister for giving me support to continue with my postgraduate study. I Acknowledgments II Table of contents Table of contents ACKNOWLEDGMENTS ...................................................................................... I TABLE OF CONTENTS .................................................................................... III LIST OF ABBREVIATIONS AND SYMBOLS ........................................................ 1 ABSTRACT OF THE THESIS ............................................................................... 5 RESUME ........................................................................................................... 9 POVZETEK ...................................................................................................... 13 INTRODUCTION ............................................................................................. 17 1 DESCRIPTION OF THE SKY SCANNER ....................................................... 21 1.1 Sky scanner and measurements ............................................................ 21 1.2 Calibration of the sky scanner ............................................................... 23 1.2.1 Calibration of the sky scanner under artificial sky ....................................... 23 1.2.1.1 Zenith luminance ............................................................................... 24 1.2.1.2 Horizontal and vertical illuminances ..................................................... 26 1.2.1.3 Checking the Li-cor sensors ................................................................ 28 1.2.2 Calibration of the sky scanner under real sky ............................................. 30 1.2.2.1 Calculated vertical illuminances and measured data from IDMP station... 30 1.3 Setting up the sky scanner .................................................................... 33 1.3.1 Description of the IDMP station. . .............................................................. 33 1.3.2 Correction of orientation of vertical illuminance sensors .............................. 34 1.3.3 Correction of orientation of the sky scanner ............................................... 39 2 ANALYSIS OF THE SKY SCANNER DATA ................................................... 41 2.1 Description of the visualization ............................................................. 41 2.2 Defining CIE sky type based on CIE standard ........................................ 43 2.2.1 Determination of the gradation group ....................................................... 45 2.2.2 Determination of the indicatrix group ........................................................ 53 2.2.3 Calculation of the CIE sky type with gradation and indicatrix ....................... 60 2.3 Calculation of the CIE sky type based on Lz/Dv ratio ............................ 61 III Table of contents 2.3.1 Calculation of the diffuse horizontal illuminance (Dv) .................................. 65 2.3.2 Calculation of the vertical illuminances ...................................................... 68 2.4 Calculation of the CIE sky type with Tregenza statistical method ......... 69 2.5 Results of measurements ...................................................................... 70 3 USING DIGITAL CAMERA AS A SKY SCANNER .......................................... 75 3.1 Introduction ........................................................................................... 75 3.2 Transforming sky images to .05D files ................................................... 75 3.3 Linking sky scans and images from digital camera ................................ 77 3.3.1 Flipping and rotating an image ................................................................. 77 3.3.2 Flipping and rotating a sky scanner grid .................................................... 80 3.4 Comparing results derived with sky scanner and digital images ........... 81 3.5 Database of sky luminance measurements ........................................... 84 3.6 Conclusion on the use of digital images ................................................ 87 4 SKY LUMINANCE MODELS ........................................................................ 89 4.1 All weather model for sky luminance distribution - Perez ..................... 89 4.2 ASRC - CIE ............................................................................................. 92 4.3 Igawa All sky model ............................................................................... 92 4.4 Comparison of sky luminance models .................................................... 96 4.4.1 Results of comparison - all cases ............................................................ 101 4.4.2 Results of comparison dependent on CIE sky type .................................... 105 4.4.3 Results of comparison ............................................................................ 109 4.4.4 Results of comparison dependent on sky types and sun altitudes .............. 111 4.4.4.1 Sun altitudes 5 - 200 ........................................................................ 112 4.4.4.2 Sun altitudes 20 - 350 ....................................................................... 114 4.4.4.3 Sun altitudes 35 - 500 ....................................................................... 115 4.4.4.4 Sun altitudes over 500 ...................................................................... 117 4.4.4.5 Problems of Igawa model for intermediate skies ................................. 120 4.5 Conclusion on sky luminance models .................................................. 126 5 MEASUREMENTS IN SCALE MODEL ........................................................ 129 5.1 Introduction ......................................................................................... 129 5.2 Description of the model and method .................................................. 129 5.2.1 Illuminance and luminance measurements in scale model ......................... 132 IV Table of contents 5.2.1.1 Illuminance ..................................................................................... 132 5.2.1.2 Indoor luminances ........................................................................... 132 5.2.1.3 Outdoor luminances ......................................................................... 132 5.2.2 View from a scale model ........................................................................ 134 5.3 Calculation of directional daylight factors ........................................... 136 5.4 Calculation of illuminance values in scale model ................................. 144 5.4.1 Calculation of luminance values of sky elements ...................................... 144 5.4.2 Calculation of illuminance values in scale model with daylight factors ......... 146 5.4.2.1 Daylight factors with DIALux ............................................................. 146 5.4.2.2 Daylight factors with illuminance ....................................................... 147 5.4.3 Calculation of illuminance values in scale model from luminance values from sky scanner ........................................................................................... 148 5.5 Mean bias error between measured and calculated illuminance values in scale model .................................................................................................. 148 5.6 Database of scale model measurements ............................................. 149 5.7 Results of the scale model measurements .......................................... 152 5.8 Conclusions for scale model measurements ........................................ 156 6 CONCLUSION .......................................................................................... 157 CONTRIBUTIONS OF THE THESIS TO THE SCIENCE .................................... 159 STATEMENT OF THE AUTHORSHIP .............................................................. 161 REFERENCES ................................................................................................ 163 v Table of contents VI List of abbreviation and symbols List of abbreviations and symbols List of abbreviations IDMP CCD CIE DDF DF ENTPE FE LASH MA MBD MBE RMS RMSD SLR (camera) International Daylight Measurement Programme Charge-coupled device Commission Internationale de l'Eclairage (International Commission on Illumination) Directional daylight factor Daylight factor Ecole Nationale des Travaux Publics de l’Etat Faculty of Electrical engineering of University of Ljubljana Laboratoire des Sciences de l’Habitat Mean absolute Mean bias difference Mean bias error Root mean square Root mean square difference Single-lens reflex (camera) List of symbols used in chapter 1 fC Calibration factor of sky scanner (110.6), k Factor we defined based on measurements under artificial sky (1.65) SS L Luminance of a sky element in .05D file List of symbols additionally used in chapter 2 a,b,c,d,e Parameters defining gradation and indicatrix groups in Standard CIE S 011/E:2003 /ISO 15469.2004 Az A90 B,C,D,E Angular difference between azimuth of the sun and azimuth of the sky element, Difference in azimuth of ideal L90 sky element and solar meridian Standard parameters defining Lz/Dv curves 1 List of abbreviation and symbols D Diffuse horizontal illuminance V Eh Horizontal illuminance from sky scanner measurement EhCIEx Horizontal illuminance for CIE sky type x ÔE contribution to horizontal illuminance from a single sky patch contribution to vertical illuminance from a single sky patch f(x) Indicatrix function La Luminance of a sky element La ' Measured luminance of a sky element La_rel Relative sky element luminance Lp Luminance of a sky patch Lp Normalized luminance of a sky patch LpCIEx Luminance of a sky patch for CIE sky type x LpCIEx Normalized luminance of a sky patch for CIE sky type x LZ Zenith luminance L Z Measured zenith luminance Normalizing luminance, luminance of sky element with angular distance 900 towards sun Measured normalizing luminance L(90) l(%) Luminance of sky element with angular distance x towards sun l(%) Measured luminance of sky element with angular distance x towards sun nX>l5 Number of patches with scattering angle larger than 15 degrees rmsCIEx RMS error between measured values and CIE sky type x Z Zenith angle of a sky element ZS Solar zenith angle aS Solar azimuth a Azimuth of a sky element y Elevation angle of a sky element yS Solar altitude A or Angular difference in azimuth between sky patch and normal of the plane 2 List of abbreviation and symbols cp(Z) X da Altitude of the almucantar (imaginary circle on the celestial sphere, parallel to the horizon. The circle connects sky elements with same altitudes) Gradation function Scattering angle Solid angle of a sky patch List of symbols additionally used in chapter 4 ap , bp , cp , d p , ep a' , b', c' , d' , e' Ce Cle Ces Eed Ee Ee E eg K L Igawa pIgawa LpSS mbd Se m mbd rdIgawaCIEx rmsd Igawa Igawa rmsd IgawaCIEx rmsd Coefficients in Perez All weather model Coefficients in Igawa model Cloud ratio Cloudless index Standard cloud ratio Horizontal diffuse irradiance Global irradiance Normal incident direct irradiance Extraterrestrial direct normal irradiance Clear sky index Luminance of a patch in Igawa model, Luminance of a patch measured with sky scanner. Mean bias difference between Igawa model and sky scanner measurement, Standard global irradiance Optical air mass Mean bias difference between Igawa model and sky scanner measurement without elements on first almucantar and without element with ?<150 Relative difference for Igawa luminance model for CIE sky type X RMSD between Igawa model and sky scanner measurement RMSD for Igawa luminance model for CIE sky type X RMSD for best luminance model for CIE sky type X 3 List of abbreviation and symbols rmsd Igawa Si s A RMSD between Igawa model and sky scanner measurement and without elements on first almucantar and without element with ?<150 Sky index Sky brightness Sky clearness List of symbols additionally used in chapter 5 DDF Ap DDQ Ap DF E E Ap E AIntp E AExtp E diff E E sky13-A1 MBE sky13 RMSE sky13 Z Z p p ?Sp ?S Directional daylight factor related to a sky element for a point A in the model Directional daylight quotient related to a sky element p for a point A in the model Daylight factor for a given point Total illuminance at a given point Illuminance at a given point contributed by a sky element Indoor illuminance at a given point A, produced by sky element p Outdoor (unobstructed) illuminance at a given point A, produced by sky element p Diffuse illuminance on unobstructed horizontal plane Measured illuminance with Licor sensor in point A1 Calculated illuminance from 13 sky zone luminances in point A1 Mean bias error for sky modelled with 13 zones Root mean square error for sky modelled with 13 sky zones Incidence angle between surface in point and sky element Zenith angle of a sky element Solid angle of a sky element Solid angle of a sky element seen from indoor 4 Abstract of the thesis Abstract of the thesis The presented doctoral thesis concentrates on daylight availability calculations with an emphasis on models of the sky luminance distribution. The main goal of the thesis is to recommend the most appropriate sky luminance model used to calculate time values of indoor daylight illuminances and to define the needed accuracy of the sky luminance model, by the number of zones used in the model. Both conclusions are based on measurements. The first chapter introduces the equipment needed to perform sky luminance distribution measurements. The most important device is a sky scanner. The sky scanner that was used with our research was produced by company EKO and is capable to measure sky luminance distribution with accuracy of 145 sky zones in approximate 3 minutes. Before realization of the measurements the calibration of the sky scanner was carried out. Calibration of the sky scanner was carried out with different methods, firstly under artificial sky. Under artificial sky, zenith luminance and horizontal and vertical illuminances were checked. Second calibration was performed under real sky. In this case we compared calculated horizontal illuminance with illuminances measured on IDMP station. After the sky scanner was calibrated and set up on IDMP station it was possible to correct the orientation of the IDMP vertical sensors for illuminance, since they were not perfectly in line with correct azimuths. In the second chapter, we describe how to analyse luminance data gained with sky scanner. In this part also the CIE S 011/2003 standard is explained with an emphasis on gradation and indicatrix function. Determination of the gradation group is based on calculation of luminance ratio between two elements on every almucantar (imaginary circle on the celestial sphere, parallel to the horizon. The circle connects sky elements with same altitudes) with scattering angle as close as possible to solar zenith angle and zenith luminance. Determination of the indicatrix group is also based on luminance ratio. In this case in the numerator is luminance of an element on certain almucantar and in denominator we can find the normalizing luminance. Normalizing luminance is a luminance of a part of the sky with angular distance 90 degrees towards sun. There are 36 combinations of gradation and indicatrix groups and only 15 of them are listed in the CIE standard. Since some of combinations are really rare, we provide the table how to include 36 combinations in 15 standard CIE sky types. CIE sky type can be also defined with ratio between zenith luminance and diffuse horizontal illuminance. Later on in chapter two, we explain how to calculate diffuse illuminance from sky scan and how to derive CIE sky type from before mentioned ratio. Another method, which is also explained in this chapter, provides CIE sky type based on statistical method. The mentioned method is based on statistical approach. Luminances gained with sky scanner and the ones in the luminance model, described with a standard, are normalised with diffuse horizontal illuminance and compared with each other. If comparison is made between measurements and all 15 CIE sky types, the one with smallest discrepancy can be found. 5 Abstract of the thesis At the end of chapter two results are gathered and frequencies of CIE sky types derived with different methods are shown. Chapter three introduces a digital camera with fish-eye lens as a sky scanner. Our goal was to compare sky luminance values captured with sky scanner and the ones with digital camera and fish-eye. Since luminance values gained with sky scanner are gathered in a text file with sophisticated structure, also the luminances gained with digital camera were with help of Photolux software written in such a file. Both files (for the same time period) were processed in the same manner for 160 cases. As the luminance data were processed it was possible to compare significant sky type characteristics. When gained data were compared, we checked horizontal diffuse illuminance, zenith illuminance, Lz/Dv ratio, gradation group, indicatrix group, sky type based on Lz/Dv ratio, sky type based on statistical method and also all 145 luminance values were compared with each other. In this chapter a reader can find described also the whole procedure how to transform fish-eye image into luminance map and how to extract 145 luminances of the CIE sky partitioning, which is also used by the sky scanner. At the end of the chapter we conclude that a digital camera can be easily (with some limitations) used as a sky scanner. In chapter four, a description of different sky luminance models is given. The emphasis is on the following models: Perez all weather model, ASRC-CIE model and Igawa model. For the times, when we had sky scanner measurements (13.006 cases), sky luminance distributions were modelled with all three models. The modelled luminances are calculated with model's equations and irradiance data measured by IDMP station. In this chapter a problem of high luminance values of the solar corona is exposed. Neither sky luminance models nor sky scanner are capable to model or measure (truthfully) high luminance values of sky elements with low scattering angle towards sun. For this reason in our comparison, all sky elements with scattering angle smaller that 150 were eliminated. This decision is based on calculated mean bias error for all sky elements and only for those with scattering angle higher that 150. Later on, a comparison between three models and sky scanner measurements is introduced. Comparison was carried out with calculation of MB error and RMS error between each modelled and measured sky element. In similar way also CIE sky types were compared. A comparison was carried out for two methods used to derive sky scanner sky type; the one based on gradation and indicatrix group and the statistical one and for two sets of luminance values; for all sky elements without the ones with scattering angle less then 150 and the for all sky elements without the ones with scattering angle less then 150 and the ones on first almucantar. In continuation of chapter four, we present quality of luminance models depending on the sky type and on sun altitude. Conclusion on all comparisons in this chapter is that Igawa model is best for extremely cloudy sky types, Perez All weather model gives best results with intermediate sky types and ASRC-CIE model is in most cases in between upper two models, but always close to the best model. In the fifth chapter, we are talking about scale model measurements. Scale model is fully described and also all measuring problems are listed. In this chapter we introduce sky partitioning into different number of zones (212, 145, 97, 26 and 13) and we also introduce directional daylight factors (DDF) and traditionally used daylight factor. 6 Abstract of the thesis Illuminances in the model were measured with illuminance sensors and later compared with calculated values. The calculated values of illuminances were calculated with directional daylight factors and averaged luminance of the sky zone and with daylight factors and diffuse illuminance measured by IDMP station. In continuation, we describe also the database of scale model measurements (available as an attachment on a CDrom) and the results. Results show us that the number of sky zones has practically no influence to MBE and after that we can conclude that 13 zones are enough to model sky luminance if we are considering only sky vault without direct sun. 7 Abstract of the thesis 8 Résumé Résumé Cette these de doctorat traite des calculs de disponibilité de la lumiere du jour a l’intérieur des bâtiments en s’intéressant particulierement aux modeles permettant de décrire la distribution des luminances du ciel. Son objectif principal est de recommander les modeles les plus appropriés au calcul des valeurs instantanées des éclairements lumineux intérieurs et de définir la finesse du maillage utilisé pour représenter les variations de luminance sur la voute céleste. Le premier chapitre présente le dispositif expérimental utilisé pour mesurer la distribution des luminances du ciel sur le site de l’ENTPE. L’appareil le plus important est le scanner de ciel. Celui que nous avons utilisé est de la marque EKO. Il est capable de mesurer les luminances de 145 points de la voute céleste en 3 minutes. Nous avons d’abord étalonné le scanner. Dans un premier temps, nous avons placé le scanner sous le ciel artificiel de l’ENTPE. Cette source de lumiere parfaitement contrôlée nous a permis de nous assurer de la cohérence des mesures effectuées par le scanner et des valeurs calculées a partir des mesures : luminances, luminance du zénith, éclairement horizontal, éclairements verticaux… Dans un deuxieme temps, le scanner de ciel a été installé sur le lieu de la station de mesure de la lumiere de l’ENTPE. Nous avons vérifié que les cellules de mesure des éclairements verticaux de la station étaient correctement orientées. Nous avons ensuite orienté le scanner puis nous l’avons étalonné de maniere a ce que toutes les informations calculables a partir de ses mesures de luminance soient cohérentes avec les mesures de la station. Le scanner de ciel a été programmé pour effectuer des mesures automatiques des luminances du ciel toutes les 10 minutes pendant un peu moins d’une année. Dans le deuxieme chapitre, nous décrivons comment les mesures de luminances provenant du scanner de ciel ont été analysées. Nous présentons la classification des ciels standardisée par la Commission Internationale de l’Eclairage (CIE) en 2003. Nous expliquons le principe des fonctions de gradation et de diffusion utilisées par la classification et nous montrons comment obtenir ces fonctions a partir des mesures. La détermination du groupe de gradation est basée sur la variation en fonction de l’angle au zénith, du rapport entre les luminances de deux points de mesure ayant le meme angle au zénith et un angle avec le soleil, le plus proche possible de l’angle au zénith du soleil. La détermination du groupe de diffusion est basée sur la variation en fonction de l’angle de diffusion (angle entre le soleil et un point du ciel) du rapport entre la luminance d’un point du ciel et la luminance moyenne des points du ciel situés a 90 degrés du soleil. La combinaison des 6 groupes de gradation et des 6 groupes de diffusion conduit a un total de 36 types de ciels dont 15 seulement font partie du standard CIE. Nous proposons donc une table d’équivalence. Certains auteurs ont proposé d’autres méthodes ne nécessitant pas les luminances de la voute céleste pour déterminer les types de ciel CIE. Nous les avons aussi testées. L’une est basée sur le rapport entre la luminance du zénith et l’éclairement horizontal diffus. L’autre est basée sur une approche statistique comparant la distribution des luminances mesurée avec la distribution des luminances calculée a partir de modeles n’utilisant que des 9 Résumé éclairements horizontaux. La fin de ce chapitre permet de comparer toutes ces méthodes sur la base des fréquences de ciels CIE obtenus pendant la période de mesures. Le chapitre 3 nous permet de présenter et de valider une nouvelle méthode de mesure des luminances du ciel basée sur l’utilisation d’un appareil photo numérique étalonné en luminances. Le systeme appelé PHOTOLUX et développé a l’ENTPE, avait jusqu’ici été utilisé pour l’évaluation du confort visuel dans les ambiances intérieures. Notre objectif était de l’utiliser a l’extérieur en parallele avec le scanner de ciel et de vérifier la qualité des informations qu’il était capable de produire. Nous présentons d’abord le systeme et sa mise en place sur la station. L’appareil est équipé d’un objectif fish-eye a 180° qui lui permet de voir l’intégralité de la voute céleste. Comme pour le scanner de ciel, sa mise en place s’est accompagnée d’une mise en cohérence avec les mesures de la station. Nous avons ensuite utilisé le systeme pendant des journées bien précises. Au total, nous avons effectué 160 mesures. La carte des luminances générée par PHOTOLUX (plus de 400000 pixels) a été utilisée pour calculer les éclairements lumineux horizontaux et verticaux mesurés sur la station. Pour pouvoir comparer ce systeme avec le scanner de ciel, nous avons calculé les luminances moyennes des 145 zones vues par le scanner. Nous avons aussi utilisé le systeme pour produire des informations liées a la classification des ciels de la CIE. Chacune des 160 mesures est présentée dans une fiche constituant une base de données utile pour comprendre la classification des ciels. La fin du chapitre nous permet de conclure que le systeme fournit des informations beaucoup plus riches qu’un scanner de ciel (photo et distribution tres fine des luminances) malgré quelques limitations liées a l’impact du soleil direct. Dans le chapitre 4, nous décrivons différents modeles de luminances de ciel. Nous nous intéressons aux modeles les plus récents ou les plus utilisés : « tout temps Perez », « ASRC-CIE Perez » et « Igawa ». Nous comparons leurs performances sur la base des 13000 mesures effectuées avec le scanner de ciel. Pour tenir compte du fait que ni les modeles, ni le scanner de ciel ne peuvent déterminer avec précision les luminances tres élevées de la zone circumsolaire, nous n’avons pas pris en compte les luminances correspondant aux zones dont l’angle par rapport au soleil était inférieur a 15°. Nous nous sommes intéressés a l’écart moyen et l’écart quadratique moyen entre les luminances des modeles et les luminances mesurées. Nous nous sommes aussi intéressés a la capacité de chacun des modeles a restituer les ciels types de la CIE. Nous présentons la performance de chacun des modeles en fonction du type de ciel et de la hauteur du soleil. Meme si le modele « Igawa » est le plus performant pour la modélisation des ciels couverts. Meme si le « tout temps Perez » est le plus performant pour la modélisation des ciels intermédiaires. Le modele « ASRC-CIE Perez » est souvent proche des deux autres dans les catégories citées précédemment et globalement le meilleur. Le cinquieme chapitre s’intéresse au calcul des éclairements intérieurs et notamment a l’influence de la finesse du maillage utilisé pour représenter les variations de luminance sur la voute céleste. Pour tester les calculs, nous avons mis en place un dispositif expérimental original s’appuyant sur une maquette placée a l’extérieur. La maquette représente un bureau avec une ouverture en façade. Les parois de la maquette ont été 10 Résumé peintes en noir pour supprimer au maximum les réflexions multiples. Nous nous sommes donc intéressés au calcul de la composante directe de la lumiere du jour résultant d’un ciel de distribution de luminance quelconque. Nous avons mesuré les éclairements lumineux sur les parois de la maquette (sol et murs). Nous avons utilisé deux systemes PHOTOLUX (donc deux appareils photos) : l’un pour produire la carte des luminances du ciel vus par l’ouverture, l’autre la carte des luminances des parois de la maquette. Nous avons calculé les éclairements lumineux directs correspondant aux points de mesure dans la maquette en utilisant la notion de facteur de lumiere du jour directionnel (DDF). Ce facteur permet de décrire la contribution d’une zone du ciel a l’éclairement d’un point de la piece. Nous l’avons calculé analytiquement pour chacun des points de mesure dans la maquette et pour chacune de zones du ciel d’un maillage variant de 13 a 212 zones. Nous les avons ensuite combinés avec les luminances moyennes de chaque zone du ciel pour déterminer l’éclairement a l’intérieur de la maquette aux différents points de mesure. Les mesures et les résultats des calculs sont documentées dans des fiches qui facilitent leur interprétation. La comparaison entre les mesures et les calculs montre que la méthode des facteurs de lumiere du jour directionnels fournit des meilleurs résultats que la méthode classique du facteur de lumiere du jour. Pour la configuration correspondant a notre maquette, un nombre minimum de 13 zones semble suffisant pour prendre en compte la diversité des distributions de luminances des ciels rencontrés dans la réalité 11 Résumé 12 Povzetek Povzetek Prièujoèa doktorska disertacija se osredotoèa na izraèun razpoložljive dnevne svetlobe s poudarkom na modelih svetlosti neba. Prvi cilj naloge je predlagati najbolj primeren model svetlosti neba za izraèun trenutnih vrednosti osvetljenosti z dnevno svetlobo v notranjih prostorih. Drugi cilj naloge pa je doloèitev potrebne natanènosti modela svetlosti neba. Natanènost modela svetlosti neba je doloèena s številom uporabljenih elementov v modelu. Oba zakljuèka sta osnovana na meritvah, ki so bile opravljene v Lyonu. V prvem poglavju je opisana oprema, ki je potrebna za opravljanje meritev porazdelitve svetlosti neba. Najbolj pomembna naprava je vsekakor merilnik porazdelitev svetlosti neba (sky scanner). Merilnik, ki smo ga uporabili pri meritvah, je znamke EKO in je sposoben meritve porazdelitve svetlosti neba v 145 toèkah v èasu približno treh minut. Pred opravljanjem meritev je bilo potrebno merilnik tudi kalibrirati. Kalibracija je izvedena z uporabo razliènih metod. Prva kalibracija je bila izvedena v simulatorju dnevne svetlobe. Pri tej kalibraciji smo preverjali meritve svetlosti zenita in izraèunane vrednosti vertikalnih in horizontalne osvetljenosti. Druga kalibracija je bila izvedena pod resniènim nebom. V tem primeru smo primerjali izraèunane vrednosti vertikalnih in horizontalne osvetljenosti z vrednostmi, ki so bile izmerjene z merilniki postaje IDMP (International Daylight Measuring Program), ki so namešèeni že od leta 1992. Ko je bil merilnik porazdelitve svetlosti neba umerjen in namešèen na IDMP postaji, smo lahko popravili usmerjenosti merilnikov vertikalnih osvetljenosti IDMP postaje. Ti merilniki namreè niso imeli popolnoma ustrezne usmerjenosti. V drugem poglavju je opisan postopek analize meritev porazdelitve svetlosti neba. V tem delu je opisan naèin doloèitve CIE tipa neba in tudi CIE standard S 011/2003 s poudarkom na funkcijah postopnosti (gradation) in razpršenosti (indicatrix). Doloèitev skupine postopnosti temelji na razmerju svetlosti dveh elementov na vsakem almukantarju (Krog na nebesni polobli, ki je vzporeden s horizontom. Krog povezuje toèke z enako kotno višino.) in svetlosti zenita. Elementa na vsakem almukantarju morate imeti kotno razdaljo do sonca èim bliže zenitnemu kotu sonca. Tako dobimo na vsaki strani sonènega poldnevnika en element, ki ustreza zahtevi. Prav tako tudi doloèitev skupine razpršenosti temelji na razmerju dveh svetlosti. V tem primeru je v števcu svetlost posameznega elementa na almukantarju, v imenovalcu pa se nahaja normalna svetlost almukantarja. Normalna svetlost almukantarja je svetlost elementa neba na almukantarju, ki ima kotno razdaljo 900 do sonca. Ker imamo šest skupin postopnosti in šest skupin razpršenosti, imamo skupno 36 kombinacij. Nekatere od kombinacij so izjemno redke in s pomoèjo tabele, ki je v nalogi podana, lahko vseh 36 kombinacij uvrstimo med 15 osnovnih tipov, kolikor jih je tudi v CIE standardu. CIE tip neba se lahko doloèi iz razmerja svetlosti zenita (Lz) in difuzne horizontalne osvetljenosti (Dv). V nadaljevanju drugega poglavja je opisan postopek izraèuna horizontalne osvetljenosti iz meritev porazdelitve svetlosti neba in postopek doloèitve CIE tipa neba iz omenjenega razmerja. 13 Povzetek V tem poglavju je opisan tudi postopek doloèitve CIE tipa neba s pomoèjo statistiène metode. Izmerjene vrednosti svetlosti elementov neba se normirajo z difuzno horizontalno osvetljenostjo. Na enak naèin se normirajo tudi vrednosti vseh 15 tipov neba, ki so opisani s standardom. Nato primerjamo izmerjene normirane vrednosti z normiranimi izraèunanimi vrednostmi za vseh 15 tipov. Tip neba, pri katerem je odstopanje najmanjše, je rezultat metode. Na koncu poglavja so podane tabele frekvenc pojavljanj posameznih tipov neba izraèunanih z razliènimi metodami. V tretjem poglavju predstavimo digitalni fotoaparat s širokokotnim objektivom kot merilnik porazdelitve svetlosti. Cilj, ki smo ga imeli, je primerjava porazdelitve svetlosti neba, ki jo dobimo s pomoèjo digitalnega fotoaparata s širokokotnim objektivom in porazdelitve, ki jo izmeri merilnik porazdelitve svetlosti neba. Meritve, izmerjene z merilnikom so zapisane v tekstovni datoteki s specifièno obliko zapisa, zato smo morali podatke dobljene z digitalnim fotoaparatom in obdelane s pomoèjo programske opreme Photolux, zapisati v datoteko z enako obliko zapisa. Obe datoteki (za isto èasovno obdobje) smo analizirali na enak naèin za vseh 160 primerov, kolikor smo imeli digitalnih fotografij. Z analizo vrednosti svetlosti elementov neba smo lahko primerjali tudi znaèilne lastnosti neba. Ko so bili vsi podatki zbrani, smo med seboj primerjali horizontalno difuzno osvetljenost, svetlost zenita, razmerje Lz/Dv, skupino postopnosti in razpršenosti, tip neba doloèen z razmerjem Lz/Dv in tip neba doloèen s statistièno metodo. Na koncu smo med seboj primerjali tudi pare vseh 145 svetlosti elementov. Na koncu poglavja zakljuèujemo, da lahko digitalni fotoaparat s širokokotnim objektivom uporabimo kot merilnik svetlosti, pri tem pa moramo upoštevati nekatere omejitve. V èetrtem poglavju so opisani modeli porazdelitve svetlosti neba. Poudarek je na modelih, ki se najveè uporabljajo: Perez-ov model za vse tipe neba, ASRC-CIE model in Igawa model. Za vse èase (13.006 meritev), ko je deloval merilnik svetlosti neba (april 2005 – december 2005) smo izdelali porazdelitve svetlosti neba za vse tri modele. Vrednosti svetlosti so bile izraèunane na podlagi modelov in meritev obsevanosti na IDMP postaji. V tem poglavju je opisan tudi problem visokih vrednosti svetlosti sonca in sonèeve korone. Niti modeli porazdelitve svetlosti neba niti merilnik porazdelitve svetlosti neba ne morejo modelirati oz. izmeriti visokih nivojev svetlosti elementov neba, ki imajo majhno kotno razdaljo do sonca. Prav zaradi tega razloga smo pri analizi podatkov izloèili vse elemente neba, ki imajo kotno razdaljo do sonca manjšo od 150. Ta odloèitev temelji na izraèunu povpreène razlike svetlosti elementov neba z upoštevanjem vseh elementov neba in le tistih s kotno razdaljo do sonca veèjo od 150. V nadaljevanju je predstavljena primerjava izraèunanih vrednosti svetlosti elementov neba z modeli in vrednosti izmerjenih z merilnikom porazdelitve. Primerjava je bila izvedena z izraèunom povpreène vrednosti razlike in efektivne vrednosti razlike med posameznim modelom in izmerjeno vrednostjo. Na podoben naèin smo primerjali tudi CIE tip neba. Primerjava je bila izvedena za dve metodi doloèitve CIE tipa neba, za doloèitev na podlagi postopnosti in razpršenosti in za doloèitev na podlagi statistiène metode. Primerjava smo izvedeli na dveh skupnah podatkov o vrednostih svetlosti; za vse vrednosti in le za tiste, katerih elementi imajo kotno razdaljo do sonca veèjo od 150 in višino veèjo od 60. Prvi elementi so bili izvzeti zaradi sonèeve korone, drugi pa zaradi 14 Povzetek ovir-stavb, ki so v okolici IDMP postaje. V nadaljevanju èetrtega poglavja je predstavljena kakovost modelov svetlosti neba v odvisnosti od tipa neba in višine sonca. Na koncu poglavja so podani zakljuèki, da je Igawa model najboljši zelo oblaène tipe neba, Perezev model je najboljši za srednje tipe neba, ASRC-CIE model neba pa je v povpreèju najboljši model. ta model je namreè v veèini primerov po kakovosti med prej omenjenima, vedno pa blizu najboljšega modela. Peto poglavje opisuje meritve na modelu prostora in analizo le-teh. Model prostora je natanèno opisan, prav tako pa so omenjeni tudi problemi, ki so nastali ob meritvah. Na tem mestu predstavimo tudi delitev neba na razlièno število elementov (212, 145, 97, 26 in 13), faktor dnevne svetlobe in direktni faktor dnevne svetlobe. Osvetljenosti v modelu so bile merjene z merilniki osvetljenosti, nato pa so bile te vrednosti primerjane z modeliranimi. Modelirane vrednosti so bile izraèunane na podlagi direktnega faktorja dnevne svetlobe in srednje vrednosti svetlosti elementa neba na podlagi faktorja dnevne svetlobe in izmerjenih vrednosti osvetljenosti na IDMP postaji. V nadaljevanju je opisana tudi baza podatkov, ki so bili izmerjeni z meritvami na modelu prostora. Rezultati meritev in primerjave kažejo, da natanènost modela (število elementov neba) praktièno ne vpliva na velikost napake pri izraèunu osvetljenosti. Tako zakljuèujemo, da je pri izraèunih, kjer ne upoštevamo direktne sonène svetlobe, dovolj delitev neba na 13 elementov. 15 Povzetek 16 Introduction Introduction Our life is closely connected to daylight. It has a bearing on our schedule, way of life and biological processes in the body. In fact, we cannot even imagine our life without daylight. Daylight constitutes a renewable source of energy. Consequently, we would like to exploit it to the greatest extent possible. If we strive for its rational exploitation, we need to provide for at least an approximate forecast. Thus, all over the world different methods for calculating the daylight contribution to the indoor illuminances are being developed. The sky is the main contributor to the indoor illuminances, whereas the smaller portion is contributed directly by the sun. The prerequisite for the calculation of indoor daylight illuminances is the knowledge of the distribution of sky luminances, which in turn enables the calculation of the indoor illuminances. First mathematical descriptions of the distribution of sky luminances can be traced as far back as beginning of 20th century, yet their focus was only on a cloudy sky [11, 12]. Complexity of the problem was the reason why the majority of researchers limited their work at the beginning to two basic sky types, i.e. clear and cloudy. It was only later that intermediate types were described as well. Researchers formulated mathematical equations for the description of sky luminances on the basis of local measurements. Most equations are consequently adapted to local conditions or sky types. 2003 saw the adoption of the standard by CIE - Commission Internationale de l'Eclairage (International Commission on Illumination). The standard [16] gives mathematical description of the distribution of luminances for 15 different sky types, five for a cloudy sky, five for a partly cloudy and five for a clear sky. These 15 sky types now constitute an instrument allowing for the worldwide description of the most frequent sky types. Various methods can be employed for the calculation of the daylight contribution to indoor illuminances on the basis of the recorded distribution of sky luminances. The most frequent method of calculation is the daylight factor. The calculation by way of computer simulations of the room conditions has been gaining ground recently. In 2003, our Laboratory of Lighting and Photometry developed a simulation procedure [17] whereby the calculation of indoor daylight illuminances employs photometry of the window through which the daylight comes into a room. Description of the problem The basis for the method developed at the Faculty of Electrical Engineering in Ljubljana is the sky division into 77 zones. The luminances of certain zones provide the basis for subsequent calculation of photometric data giving luminances for certain directions of the window through which the light comes into the premises. Similar methods have also been developed in other labs. ENTPE-LASH developed a method founded on sky division into 13 zones. It is presented in [1] and is also available at the website www.satel-light.com. The method is based on the calculation of the contribution which the individual sky zone makes to the illuminances of a certain point in a room. The method is also called Directional Daylight Factor – DDF. The methods used for the 17 Introduction calculation of the daylight contribution have unfortunately not yet been adequately verified, the primary reason being the lack of comparable measurements. The use of different software (e.g. Photolux enabling the measurement of distribution of sky luminances and the drawing of the luminance maps on the basis of the photographs taken by the digital camera and wide-angle lens) and measurements on a model hold promises for certain progress in the field. Rationale of the contribution to the science In 2003, CIE issued a standard which describes the distribution of sky luminances in different weather conditions. The standard provides for universal equations enabling the calculation of luminances of the sky element in an arbitrary direction and gives the coefficient table used for the determination of 15 different sky types, ranging from cloudy to clear skies. Not long ago after the standard had been issued, it became evident that the frequency of emergence of certain sky types at a certain location also needed to be determined so that the described model would be used correctly. Hence, the purpose of this doctoral thesis is to develop an adequate method which shall make this possible. Tables denoting the frequency of emergence of certain sky types at certain locations would be compiled in the easiest way by employing measurements of the distribution of sky luminances. Such measurements, however, are extremely rare since only 17 measurement stations in the world make use of the "sky scanner" for the measurements of sky luminances. There is not enough data on luminance distribution or, what is more, there is none for most of the locations. Therefore, the method in question shall be based on more comprehensive and accessible measurements. Satellite images obtained from geostationary satellites shall be used. The satellites provide permanent coverage of the surface of the Earth, with their resolution reaching the accuracy of up to 5 km. Each image of the Earth surface is made every 30 minutes, which allows for a great amount of data for the purpose of the analysis. Most of this data for Europe and Africa is to be accessed via the Internet. The accessibility of the data has led to the decision to develop a new method. This method shall allow us to determine the frequency of emergence of CIE sky types at a specific location on the basis of global and diffuse horizontal irradiances. We shall be able to carry out calculations to determine time values of indoor daylight illuminances by way of devised tables and known algorithms. Method description We shall carry out the method by comparing the measurements of sky luminance distribution, the measurements of spatial illuminance by way of the model, and the illuminance calculation by way of various simulation programmes. The measurements shall be carried out at ENTPE (Ecole Nationale des Travaux Publics de l’Etat) Vaulx-en-Velin, France, which is home to one of IDMP (International Daylight Measurement Programme) locations. Our efforts shall (simultaneously) focus on the following: 18 Introduction - measurement of sky luminance distribution by way of the sky scanner (a special measuring device); - measurement of zenith luminance (10 – spatial angle) by way of a luminance measuring device on a stand; - measurement of sky luminance distribution by way of a digital camera, wide-angle lens and Photolux program package, and the drawing of sky luminance maps; - measurement of luminance distribution within the scale model by way of a digital camera, wide-angle lens and Photolux program package, as well as the drawing of maps for wall luminance within the model; - measurements of illuminance in the model by way of Li-cor sensors. When processing the data, all additional data collected through calculations within the IDMP locations shall be taken into consideration. The corresponding sky type shall be determined for each scan of sky luminance distribution (obtained through sky scanner) according to CIE standards. As for classifying individual sky luminance scans, Kittler’s method shall be used. Furthermore, we shall also determine the corresponding sky type for both the sky scan and the measurements carried out at the IDMP location, which shall be measured at the same time as the scan of sky luminance distribution. In this case, the Perez method shall be used for determining the sky type. The method is instrumental in determining sky clearness and sky brightness and also the corresponding sky type on the basis of the measurements obtained from IDMP. The results of the measurements and both methods shall be duly compared. On the basis of the results, a new method for determining CIE sky types from satellites scans shall be devised. In the second part, we shall carry out computer simulation of illuminances in the model by way of various methods. As far as the calculation is concerned, the illuminances in the scale model shall be calculated with different sky luminance models and with daylight quotients. Input data for sky luminance models shall consist of the maps developed by a digital camera and PhotoLux programme. Computer simulations results shall be compared to the measurements on a model. Focus shall be placed on comparing the values of illuminances. This will allow for evaluating the applicability of the method as regards determining CIE sky type when calculating indoor daylight illuminances. 19 Introduction 20 Description of the sky scanner 1 Description of the sky scanner The sky scanner measures the spatial distribution of the luminance or radiance of the entire sky automatically. 1.1 Sky scanner and measurements For measuring purposes a sky scanner (Fig. 1.1) was kindly borrowed from Kyushu University (Japan) for one year (January 2005 - January 2006). Sky scanner was made by Japanese company EKO Instruments CO., LTD and it was already used for measurements in Japan. Its measurement head is mounted on a two axis turning table, the luminance sensor is a SI-photodiode with a V(?) filter. An amplifier and automatic temperature compensation is assembled in the sensor head. Sky scanner came together with acquisition system (I/O controller and portable computer) and power supply since the input voltage sky scanner acquisition system and computer is 110 V. Sky scanner measures luminance and radiance of 145 patches in sky hemisphere (Fig. 1.2), following the CIE recommendations [44]. The solid angle of each measured sky patch is 110. Measured value represents the average value of luminance and radiance in the solid angle of 110. Measurement takes place on eight almucantars (imaginary circle on the celestial sphere, parallel to the horizon. The circle connects sky elements with same altitudes) - altitudes (60, 180, 300, 420, 540, 660, 780, 900). Number of measurements at different altitudes is changing and depends on the size of virtual bend at that altitude (Table 1.1). Table 1.1. Number of measurements according to the altitude of the point Altitude (0) -Almucantar No. of measurements 6 30 18 30 30 24 42 24 54 18 66 12 78 6 90 1 SUM 145 Fig. 1.1. Sky scanner 21 Description of the sky scanner Fig. 1.2. 145 sky patches Measurement takes place every 10 minutes from sunrise to sunset. Time needed to perform one measurement of the sky luminance distribution is a little bit more than 3 minutes, which can cause problems under dynamic weather conditions, since a single moving cloud can cover different patches of sky in a single measurement. The sky scanner has two measuring heads, one for luminance and the other for radiance. After one measurement is finished, the computer creates two files with data, the first one is for luminance and the second one is for radiance. In our thesis, we used only data for luminance, but we stored also the one for the radiance. All files are in ASCII format and they include data from the scanner, date and time of the beginning and the end of the measurement, inserted constants and location of the measurement. (Fig. 1.3). Fig. 1.3. Example of the data file. 22 Description of the sky scanner Luminance data are written in a file .05D (in year 2005), where luminances of all 145 sky patches are listed one after another without any punctuation mark (Fig. 1.3 and Fig. 1.4). Luminances in a file have number format of X.XXX and they are not given in absolute values, but they are divided by constant k (Fig. 1.4). Factor k can be calculated with next equation: L = k-L L = l06/fC-kSS-LF ^k = i06/fC-k 1.1) C SS Where: L fC k SS L luminance of a sky patch, calibration factor of sky scanner (110.6), factor we defined based on calibration under artificial sky (1.65) Chapter 1.2) luminance in .05D file. (see Calculated value of the factor k is 14.918,63. File .05D 0.0000.0900.0820.0730.0730.07 60.0990. X Factor k 14 .918,6 3 Luminance of a sky zone [cd/m 2 ] r i i = i 1 r 13 421 12 23M10 89 108 9 11 33 12 38 ... Fig. 1.4. Calculation of luminances of sky patches from a .05D file 1.2 Calibration of the sky scanner The sky scanner was last used in Japan in 2000/2001 and it has not been calibrated since then. And so we had to calibrate it before performing any measurements. First calibration was done under artificial sky and second under real sky conditions. 1.2.1 Calibration of the sky scanner under artificial sky For calibration of sky scanner under artificial sky we used the artificial sky of ENTPE (Vaulx-en-Velin). This is a uniform sky with adjustable luminance. 23 Description of the sky scanner 1.2.1.1 Zenith luminance As a first calibration, we decided to compare the "zenith luminance" of the sky scanner with luminance of the artificial sky. Zenith luminance is the 145th patch, which is measured by the sky scanner. Luminance value of this patch was compared with value measured with Minolta handheld luminance meter LS-100 with 10 aperture and with the value derived with digital camera and software Photolux. The sky scanner was placed in the middle of a room with artificial sky and it was lifted to a height where luminance and radiance heads were facing mirror on side walls (Fig. 1.5, Fig. 1.6). The scanner was placed also in horizontal position with water level. When the scanner was placed in right position, the door was closed and the automatic measurement was started. Fig. 1.5. Diagram of the sky scanner under Fig. 1.6. Picture of the sky scanner under artificial sky artificial sky With the same luminance level, a picture with calibrated digital camera (Nikon Coolpix 5000) with fish eye lens was taken. Digital camera was placed on tripod in the middle of a room with artificial sky and it was lifted to a level, where top of the fish-eye lens was in the same height as luminance and radiance heads of the sky scanner. When the digital camera was placed in right position, the door was closed and the picture was taken. With the same luminance level also luminance of the "zenith" was measured with handheld luminance meter. The "zenith" zone was marked on ceiling with transparent adhesive tape, using plumb line and calculating the area which is covered at height 1.7 m with solid angle of 110 (Fig. 1.7) 24 Description of the sky scanner Transparent adhesive tape Fig. 1.7. Digital image of artificial sky with zoomed area of zenith luminance The same procedure was used for eight times with different luminance levels. Measured results are gathered in Table 1.2. The luminances of sky patch number 145, measured with the sky scanner were derived using calibration factor and data from files produced by the scanner. The luminances measured with digital camera and fish-eye lens were obtained with Photolux. Photolux is a software in a system including a digital camera calibrated in terms of luminance and a software which translated the image(s) produced by the camera into a luminance map. The measurements were performed on April 14th, 2005. Table 1.2. Results of "zenith" luminance measurements. Measurement No. (time) Zenith Luminance (EKO) [cd/m2] Zenith Luminance Minolta (cd/m2) Zenith Luminance Photolux (cd/m2) Coefficient Minolta/EKO Coefficient Photolux /EKO M1 (10:20) 651 1070 1080 1.644 1.659 M2 (10:30) 940 1540 1550 1.638 1.649 M3 (10:40) 1573 2580 2612 1.640 1.660 M4 (10:50) 2495 4140 4158 1.659 1.667 M5 (11:00) 2975 4920 4504 1.654 1.514 M6 (11:10) 3400 5620 5062 1.653 1.489 M7 (11:20) 3770 6400 5833 1.698 1.547 M8 (11:30) 3128 5150 4713 1.646 1.507 Average coefficient 1.654 1.576 From the measurements we calculated the coefficients between Minolta luminance meter and sky scanner and Photolux and sky scanner values. The average coefficient between Minolta luminance meter and the scanner is 1.654 and the average coefficient between Photolux and sky scanner is 1.576. 25 Description of the sky scanner As it is seen from the most right column in Table 1.2 deviation of values of coefficient between luminances calculated with Photolux and luminances from sky scanner are more fluctuating than values of coefficient between Minolta luminance meter and sky scanner. Due to this fact we decided to take coefficient 1.65 for further calculations. In this sense, we plotted diagram (Fig. 1.8), where almost perfect match between corrected values of sky scanner luminances and luminances measured with Minolta is seen. Fig. 1.8. Measured luminances under artificial sky 1.2.1.2 Horizontal and vertical illuminances The second calibration was done also under the artificial sky. At this time, we compared illuminances derived from the sky scanner and from the illuminance meter. We compared horizontal and 4 vertical illuminances. Illuminances from sky scanner are calculated by integrating of all luminances for horizontal illuminance (see chapter 2.3.1) and by integrating of only certain luminances for vertical illuminances. Measured illuminances were measured with Li-cor sensors and stored on logger. Li-cor sensors and a logger (Fig. 1.9) is a system capable of reading and storing different values depending on type of measurement we perform. In our case we were measuring illuminances, so illuminance sensors were attached to the logger. We used five illuminance sensors, one for horizontal illuminance and four for vertical illuminances. Sensors were fixed on a specially designed wooden cube (Fig. 1.10), which was painted black and fixed together with black wooden plate to a tripod (Fig. 1.11). 26 Description of the sky scanner Fig. 1.9. Li-cor illuminance sensors and data logger Fig. 1.10. Illuminance sensors fixed on a black wooden cube Fig. 1.11. Illuminance sensors with wooden cube fixed on a plate and on a tripod 27 Description of the sky scanner But before doing the comparison, Li-cor sensors had to be checked (see chapter 1.2.1.3). After the validation of Li-cor sensors we installed the sky scanner under the artificial sky in the same way as we did it when checking zenith luminance. When the scanner was in place we started measurement and when the measurement was finished, the sky scanner was removed from the artificial sky room and on the same position as there was the measuring head of the sky scanner we put the tripod with 5 sensors. One sensor was set up in horizontal position and the other four in four vertical positions (N, E, S, W). Then again the room with artificial sky was closed and from the data logger, which was placed outside, reading of all five illuminances from sensors was done. The same procedure was used for six different settings of illuminance levels under artificial sky. Illuminance values from sky scanner were obtained with integration of all or only certain luminances (for vertical illuminances) and then compared with measurement readings from Li-cor logger. The procedure of computation of horizontal and vertical illuminances from the sky scanner is explained in section 2.3.1 and 2.3.2. For all six different illuminance setups we calculated coefficients between illuminances from Li-cor sensors and illuminances calculated from sky scans. Coefficients were calculated for horizontal and for four vertical illuminances. For each measurement those five coefficients were averaged and values are collected in Table 1.3. Table 1.3. Average Li-cor/sky scanner coefficient for all measurements Measurement Average coefficient Li-cor/Sky scanner M1 1.626458 M2 1.653968 M3 1.697119 M4 1.62945 M5 1.679245 M6 1.627261 Average 1.65225 With averaging the average coefficient at all six measurements we got practically the same coefficient as we got it with luminance validation. So we can conclude that taking the coefficient 1.65 is the right decision 1.2.1.3 Checking the Li-cor sensors Li-cor sensors were last calibrated in July 2002 and they should be calibrated every two year. Instead of an absolute calibration we did a relative calibration and we compared sensors between each other. Comparison was done under artificial sky. All sensors were lied down on a horizontal plane and the artificial sky was switched on. At eight 28 Description of the sky scanner different luminance levels illuminance measurements were done and results are in table Table 1.4. After measurements the average illuminance and deviation of all measurements from average illuminance was calculated for all luminance levels. After that illuminance profiles on Fig. 1.12 were plotted. In the diagram, we can see the measured illuminances from all 5 sensors compared to average illuminance. Table 1.4. Calibration of illuminance sensors Probe El lx E2 lx E3 lx E4 lx E5 lx E6 lx E7 lx E8 lx Mean deviation % A-I1 1522 1977 2860 4059 5975 7120 8560 10150 -0.563 A-I2 1552 2006 2907 4170 6083 7280 8720 10360 1.401 A-I3 1508 1940 2815 4057 5895 7075 8430 9956 -1.786 A-I4 1556 1980 2853 4145 6050 7228 8630 10160 0.421 A-I5 1549 1988 2864 4146 6055 7305 8650 10070 0.528 Average 1537.4 1978.2 2859.8 4115.4 6011.6 7201.6 8598.0 10139.2 -¦—A-I1 *A-I2 ¦A—A-I3 A-I4 *—A-I5 0 2000 4000 6000 8000 10000 12000 Av e rage illuminance (lx ) Fig. 1.12. Measured illuminance under artificial sky. Deviation of illuminances when compared to average illuminance is less then 2 % and from that result, we can conclude, that sensors are coherent and they don't need calibration. 10000 - 8000 6000 4000 2000 29 Description of the sky scanner 1.2.2 Calibration of the sky scanner under real sky 1.2.2.1 Calculated vertical illuminances and measured data from IDMP station The same validation of illuminances as it was done under the artificial sky was also done under real sky conditions. After the installation of the sky scanner we compared four vertical illuminances calculated from sky luminance distribution and illuminances from IDMP station. With comparison it was possible to calculate the coefficient between IDMP measurements and sky scanner calculated values. The coefficient was calculated for each illuminance, but only for times when, there was no direct sun component since the sky scanner can't measure direct sun luminance. This fact is best viewed on next diagrams (Fig. 1.13 - Fig. 1.16), where sky scanner curve can not follow high values obtained from IDMP station for times with direct component of sun. Calculated coefficient was 1.62. Since almost the same coefficient was already calculated with measurements under artificial sky and it is 1.65, we decided to use 1.65 and plot diagrams of all four vertical illuminances for time from July 19 5:10:00 to July 22 14:50:00. Fig. 1.13. Diagram of North vertical illuminance measured on IDMP station and calculated from sky scanner data 30 Description of the sky scanner Fig. 1.14. Diagram of East vertical illuminance measured on IDMP station and calculated from sky scanner data 75 60 45 30 15 -¦ 6:00 12:00 18:00 0:00 6:00 12:00 18:00 0:00 6:00 12:00 18:00 0:00 6:00 12:00 Time Fig. 1.15. Diagram of South vertical illuminance measured on IDMP station and calculated from sky scanner data 0 31 Description of the sky scanner Fig. 1.16. Diagram of West vertical illuminance measured on IDMP station and calculated from sky scanner data 32 Description of the sky scanner 1.3 Setting up the sky scanner 1.3.1 Description of the IDMP station The IDMP (International Daylight Measuring Program) Vaulx-en-Velin station (Fig. 1.17) was set up with the IDMP project in 1992. On the station there are the following measurements (Fig. 1.18): Illuminances: - Global horizontal - Diffuse horizontal - North vertical - East vertical - South vertical - West vertical Irradiances: - Global horizontal - Diffuse horizontal Others: - Zenith luminance - Dry Bulb Temperature - Wind direction - Wind speed - Normal incidence direct solar radiation - UV-A, UV-B Fig. 1.17. Outlook of the IDMP station Vaulx-en-Velin West vertical Illuminance Global Horizontal Irradiance Zenith Luminance Sky scanner UV-A UV-B Diffuse horizontal Illuminance \ ^k ¦¦- - «3 i ' ^ Pyrhelio meter North Vertical Illuminance East Vertical Illuminance Global Horizontal Illuminance ^^H J r South Vertical Illuminance Diffuse Horizontal Irradiance Fig. 1.18. IDMP station measurements (3600 view) 33 Description of the sky scanner Beside these measurements, in April 2005, an EKO sky scanner was added. It was placed on the south east part of the station. 1.3.2 Correction of orientation of vertical illuminance sensors Although the station has been working more then ten years, the actual orientation of vertical illuminance meters was not checked in last two years. The orientation was checked before we did the comparison between vertical illuminances derived from the sky scanner and from the IDMP station. Precise positioning of measuring devices with normal magnetic compass is on the IDMP station almost impossible since the whole station is made of iron tubes. To define correct orientation of vertical illuminance meters, we compared data from IDMP station and data derived from a program with Perez – All weather sky luminance model [14]. The input data for the Perez model were global and diffuse illuminances and position of the Sun (date, time). In the sky luminance model, the whole hemisphere was divided into patches with a grid of 10. For all patches, luminance was calculated and from luminance values four vertical illuminances were calculated. To find the best matching orientation we calculated profiles for rotations from 100 counter-clockwise to 100 clockwise and in some cases up to 120 clockwise for every degree. Suggested corrections were calculated with root mean square (RMS) differences between measurements and values from Perez All weather sky luminance model. Comparison was done under clear sky conditions. For all four vertical illuminances RMS difference was calculated for different orientation setups for 4 cloudless or semi cloudless days (May 26, 27, 31 and July 3, 2005). In next tables (Table 1.5 - Table 1.8) we present diagrams with original setup, diagrams with calculated values if sensors would be rotated for 5 degrees in clockwise and 5 degrees in counter-clockwise and the last diagram is plotted with suggested correction of the direction of vertical sensors. All diagrams in tables (Table 1.5 - Table 1.8) present illuminance values for July 3, 2005. From the results for all four days we concluded that vertical illuminance meters should be rotated for: - North vertical illuminance: 40 counter-clockwise - East vertical illuminance: 50 counter-clockwise - South vertical illuminance: 2.50 counter-clockwise - West vertical illuminance: 120 clockwise 34 Description of the sky scanner Table 1.5. North Global Vertical Illuminance Original setup Rotation: 50 clockwise 30 20 ¦7 5:00 7:00 9:00 11:00 13:00 15:00 17:00 19:00 Tim IDMP Measurement Perez All weather model Result: Worse than original setup. Rotation: 50 counter-clockwise Suggested correction: 40 counter-clockwise 3 r v Li 5:00 7:00 9:00 11:00 13:00 15:00 17:00 19:00 Tim Perez All weather model Perez All weather model Result: Better than original setup Result: Best fit with least RMS difference 25 15 10 5 0 25 20 5 0 35 Description of the sky scanner Table 1.6. East Global Vertical Illuminance Original setup Rotation: 50 clockwise Result: Worse than original setup Rotation: 50 counter-clockwise Suggested correction: 50 counter -clockwise 90 i 60 50 40 30 20 - 10 0 5:00 \ j L \ / X / \ \ " \ ¦•' * : V •• [ f 1 ^^^»^ 7:00 9:00 11:00 Per ez Al l weat her model 13:00 15:00 17:00 19:00 Time Perez All weather model Result: Better than original setup 90 80 70 60 50 40 30 20 10 0 l\ \ i f[ \ \ \ : \ j \ / l / : i~,„...... i ^^^^, 9:00 11:00 13:00 15:00 17:00 19:00 Tim Per ez A l l weat her m odel Perez All weather model Result: Best fit with least RMS difference 80 70 5:00 7:00 36 Description of the sky scanner Table 1.7. South Global Vertical Illuminance Original setup Rotation: 50 clockwise Result: Worse than original setup. Rotation: 50 counter -clockwise Suggested correction: 2.50 counter clockwise 0 5:00 / /' "*\ // \ 1 i • V ^rfll ^^*« ^*-, 7:00 9:00 11:00 13:00 15:00 17:00 19:00 Tim Perez All weather model Perez All weather model Result: Better than original setup Result: Best fit with least RMS difference 50 40 30 20 10 37 Description of the sky scanner Table 1.8. West Global Vertical Illuminance Original setup Rotation: 50 clockwise Result: Better than original setup Rotation: 50 counter-clockwise Suggested correction: 120 clockwise 40 10 /y TE.-- /¦' V- /¦ v /.' \. i v. I- li \ rU' i. .,^-mr---- * 5:00 7:00 9:00 11:00 13:00 15:00 17:00 19:00 Time Perez All weather model Perez All weather model Result: Worse than original setup Result: Best fit with least RMS difference 20 0 38 Description of the sky scanner 1.3.3 Correction of orientation of the sky scanner The sky scanner was positioned on a new aluminium plate, which was fixed on the metallic frame of the IDMP station. The scanner was placed in horizontal position with spirit level. Orientation of the scanner was approximately defined and checked with compass. Setting the sky scanner in the right orientation with only compass is impossible since the sky scanner and also the frame of the IDMP station is made of metal. The influence of surrounding objects to the compass was noticeable up to a distance of 1 meter above metal frame of the station. Approximate orientation was set up with a handheld compass from a distance 1 meter from sky scanner. After the approximate set up, first measurement were performed under real sky. From measurements on a cloudless sky it was possible to check the orientation of the sky scanner. First orientation was not correct (Fig. 1.19). Calculated position of sun (yellow dot) which was calculated from date and time and location, was not on the brightest part of sky, which was measured with sky scanner. If the orientation of the sky scanner would be correct, the calculated position of the sun would be on the brightest part of the sky. After first positioning, sky scanner was rotated for 10 degrees counterclockwise. From image with new setup (Fig. 1.20), we can conclude that repositioning of the sky scanner was successful. Fig. 1.19. Measurements done with original Fig. 1.20. Measurements done with original setup setup with rotation 10 degrees counter- clockwise 39 Description of the sky scanner 40 Analysis of the sky scanner data 2 Analysis of the sky scanner data 2.1 Description of the visualization Data got from sky scanner are in raw format. With a Java program, we did visualisation of the data as shown on Fig. 2.1. First and most important matter was to present the sky luminance distribution in a graphical way. The whole hemisphere was flattened onto 2D circle. Every measured part of the sky is presented as trapezoid with curved top and bottom borders, except the last one, the zenith luminance is presented as a circle. All elements are coloured in shades of grey, where the brightest sky element is coloured in white and the darkest in black. On all elements there are values of luminance of that sky element. If the brightest sky element has a value less than 1100 cd/m2, the values are in cd/m2 otherwise values are in kcd/m2. On the luminance distribution map there is also a position of sun drawn. The position of the sun is calculated from date and time and the location of measurement. Azimuth of ideal L9 Azimuth of ideal L9 Azimuth of ideal L9 sky element on 300 sky element on 180 sky element on 60 almucantar almucantar almucantar Measurement data (location, date, start time, end time) Solar data (azimuth, altitude) Lz/Dv ratio Visualization of sky luminance distribution. Every patch is presented with trapezoid. Luminance of each patch is written on the trapezoid (in cd/m2 or kcd/m2). Intersection of virtual cone with hemisphere. Patch on 300 almucantar with angular distance to sun approximately Zs Position of the sun Solar meridian Fig. 2.1. Visualization of the sky scanner data 41 Analysis of the sky scanner data 2.2 Defining CIE sky type based on CIE standard Calculation of gradation and indicatrix for all sky scans was performed on the basis of the CIE and ISO standard CIE S 011/E:2003 /ISO 15469.2004 "Spatial distribution of daylight – CIE standard general sky" [16]. The standard defines 15 standard skies; 5 overcast, 5 intermediate and 5 clear skies. Skies are defined with next equations and parameters a, b, c, d and e that can be found in Table 2.7. a La-rel ------ L Z f(z)-(p(Z) f(ZS)-(p(0) f(x) = l + c. U<*) 2 + e-cos x (2.1) (2.2) Where: La-rel L lz fix)
{wl2) = \ (2.7) And in zenith:
) (2.10) Equation 2.10 describes relative gradation, which is the fact of decision when searching gradation group. This equation can be applied on measured values on all almucantars and gradation can be calculated by equation 2.11.