nitmtAdagirein lc c E o nm rapeD tnemt fo lE tce r laci reenignE gni dna tamotuA noi

era ytilauq ruoloc dna ycneicfife ygrenE ycneicfife dna ruoloc ytilauq era aA l t o -

I .ntg efion tes tnatropmi stcepsa ni ecfifo l .gnithgi nI DD

E ygren ycneicfife dna cf h hi ttarpis i ,noitidda ti si tnatropmi ot thgil eht ecfifo 602 derreferp ehT . gni thgi l derreferp ht iw moor iw ht derreferp l thgi gni . ehT derreferp / ecfifo na ni snoitidnoc suonimul snoitidnoc ni na ecfifo 6102 c ruolo ytilauq tnemevorpmi

no tceffe evitisop a etaerc nac tnemnorivne nac etaerc a evitisop tceffe no puR a k aR j aB n i y a devorpmi ot sdael nrut ni taht rekrow ecfifo rekrow taht ni nrut sdael ot devorpmi laicos dna ,ytivitaerc ,ecnamrofrep krow ,ecnamrofrep ,ytivitaerc dna laicos a dn evitcejbus ecnereferp fo ruoivaheb . L DE ecfifo gnithgil tnereffid ruof serolpxe siseht sihT siseht serolpxe ruof tnereffid ulc n cecf gee rf sehcaorppa rof ygrene ycneicfife dna ruoloc T .tneevorp q u a l i t y i pm r o v eme n t . hT e s e a r e gnithgil ,artceps DEL fo noitazimitpo fo DEL ,artceps gnithgil rm da sshoy te-u ,gittfiorter t gni , taeh-euh sisehtopyh dna trams R kapu jaR ayinaB sa iet h .lrncgihil roodni l gnithgi lortnoc . ehT siseht osla E f enrfr v cju senimaxe tcejbus evi secnereferp fo DEL setagitsevni ti ,noitidda nI .gnithgil ecfifo .gnithgil nI ,noitidda ti setagitsevni pog cihe ert f enrfr TC eht TCC secnereferp fo eerht inhte c spuorg E rof )a A da ,n A ,neoruE( naepo , sA i na , dna fA r i )nac f o r DEL .gnithgi l ecfifo l .gnithgi

B I NBS 9-2607-06-259-879 ( p r i n t de ) nith c E cerefer vitc u n eeorm u u c da yn fife y enE r yg e f cfi i ycne dna SSENISUB oc l ruo + auq l i t y nemevo t rpmi dna bus j ce t i ev rp e f e r ecne o f DEL o f ecfi l i hg t i gn B I NBS 2-1607-06-259-879 ( dp f ) E MONOC Y S I NSS - L 9971 - 4394 S I NSS 9971 - 4394 ( p r i n t de ) TRA + S I NSS 9971 - 2494 ( dp f ) NGISED + HCRA I T TCE ERU ytisrevinU otlaA ytisrevinU n engE airteE o loohcS fo tcelE r laci reenignE gni ECNEICS + o aou n n engE airtel f tetrapeD tnemt fo lE tce r laci reenignE gni dna tamotuA noi T E ONHC L GO Y www . a a l t o . f i EVOSSORC RE COD T RO A L +jcgaha*GMFTSH9 COD T RO A L SNOITATRESSID

SNOITATRESSID ytisrevinU otlaA

6102 seires noitacilbup ytisrevinU otlaA ytisrevinU noitacilbup seires SNOITATRESSID LAROTCOD SNOITATRESSID 602/6102

ytilauq ruoloc dna ycneicfife ygrenE ycneicfife dna ruoloc ytilauq evitcejbus dna tnemevorpmi dna evitcejbus gnithgil ecfifo DEL fo ecnereferp fo DEL ecfifo gnithgil

ayinaB jaR kapuR jaR ayinaB

fo rotcoD fo eerged eht rof detelpmoc noitatressid larotcod A larotcod noitatressid detelpmoc rof eht eerged fo rotcoD fo eht fo noissimrep eht htiw ,dednefed eb ot )ygolonhceT( ecneicS )ygolonhceT( ot eb ,dednefed htiw eht noissimrep fo eht cilbup a ta ,gnireenigne lacirtcelE fo loohcS ytisrevinU otlaA ytisrevinU loohcS fo lacirtcelE ,gnireenigne ta a cilbup rebmevoN 11 no loohcs eht fo 2T llah erutcel eht ta dleh noitanimaxe dleh ta eht erutcel llah 2T fo eht loohcs no 11 rebmevoN .noon 21 ta 6102 ta 21 .noon

ytisrevinU otlaA ytisrevinU gnireenignE lacirtcelE fo loohcS fo lacirtcelE gnireenignE oitamotuA dna gnireenignE lacirtcelE fo tnemtrapeD fo lacirtcelE gnireenignE dna n oitamotuA tinU gnithgiL tinU rosseforp gnisivrepuS rosseforp nalniF ,oopsE ,ytisrevinu otlaA ,nenolaH asiiL .forP asiiL ,nenolaH otlaA ,ytisrevinu ,oopsE d nalniF

srosivda sisehT srosivda dnalniF ,oopsE ,ytisrevinU otlaA ,irteT oniE .rD oniE ,irteT otlaA ,ytisrevinU ,oopsE dnalniF nalniF ,oopsE ,ytisrevinU otlaA ,lasuhB domarP .rD domarP ,lasuhB otlaA ,ytisrevinU ,oopsE d nalniF

srenimaxe yranimilerP srenimaxe ynamreG ,tdatsmraD fo ytisrevinU lacinhceT ,hnahK couQ narT .forP narT couQ ,hnahK lacinhceT ytisrevinU fo ,tdatsmraD ynamreG ekruT ,ytisrevinU lacinhceT lubnatsI ,ligyanO nimreS .forP nimreS ,ligyanO lubnatsI lacinhceT ,ytisrevinU y ekruT

tnenoppO sdnalrehteN ehT ,ygolonhceT fo ytisrevinU nevohdniE ,nnamesoR rednaxelA .forP rednaxelA ,nnamesoR nevohdniE ytisrevinU fo ,ygolonhceT ehT sdnalrehteN

seires noitacilbup ytisrevinU otlaA ytisrevinU noitacilbup seires SNOITATRESSID LAROTCOD SNOITATRESSID 602/6102

© kapuR jaR a yinaB

NBSI 9-2607-06-259-879 )detnirp( NBSI 2-1607-06-259-879 )fdp( L-NSSI 4394-9971 NSSI 4394-9971 )detnirp( NSSI 2494-9971 )fdp( :NBSI:NRU/if.nru//:ptth 2-1607-06-259-879

yO aifarginU yO iknisleH 6102

d nalniF

tcartsbA otlaA 67000-IF ,00011 xoB .O.P ,ytisrevinU otlaA ,ytisrevinU .O.P xoB ,00011 67000-IF otlaA if.otlaa.www

rohtuA ayinaB jaR kapuR jaR ayinaB noitatressid larotcod eht fo emaN fo eht larotcod noitatressid ecfifo DEL fo ecnereferp evitcejbus dna tnemevorpmi ytilauq ruoloc dna ycneicfife ygrenE ycneicfife dna ruoloc ytilauq tnemevorpmi dna evitcejbus ecnereferp fo DEL ecfifo gnithgil rehsilbuP loohcS fo lacirtcelE gnireenignE tinU tnemtrapeD fo lacirtcelE gnireenignE dna noitamotuA seireS otlaA ytisrevinU noitacilbup seires LAROTCOD SNOITATRESSID 602/6102 hcraeser fo dleiF fo hcraeser noitanimullI gnireenignE dna lacirtcelE gnidliuB secivreS dettimbus tpircsunaM dettimbus 12 enuJ 6102 etaD fo eht ecnefed 11 rebmevoN 6102 )etad( detnarg hsilbup ot noissimreP ot hsilbup detnarg )etad( 21 tsuguA 6102 egaugnaL hsilgnE hpargonoM elcitrA noitatressid yassE noitatressid

tcartsbA ycneicfife ygrene evorpmi ot sehcaorppa tnereffid erolpxe ot erew siseht siht fo sevitcejbo ehT sevitcejbo fo siht siseht erew ot erolpxe tnereffid sehcaorppa ot evorpmi ygrene ycneicfife ruoF .gnithgil ecfifo DEL fo ecnereferp evitcejbus eht tuo dnfi ot dna ,ytilauq ruoloc dna ruoloc ,ytilauq dna ot dnfi tuo eht evitcejbus ecnereferp fo DEL ecfifo .gnithgil ruoF trams dna ,sisehtopyh taeh-euh ,gnittfiorter gnithgil ,artceps DEL fo noitazimitpo( sehcaorppa noitazimitpo( fo DEL ,artceps gnithgil ,gnittfiorter taeh-euh ,sisehtopyh dna trams .tnemevorpmi ytilauq ruoloc dna ycneicfife ygrene rof deiduts erew )lortnoc gnithgil roodni gnithgil )lortnoc erew deiduts rof ygrene ycneicfife dna ruoloc ytilauq .tnemevorpmi ot detcudnoc erew ,moor ecfifo na ni rehto eht dna shtoob gnithgil ni eno ,stnemirepxe owT ,stnemirepxe eno ni gnithgil shtoob dna eht rehto ni na ecfifo ,moor erew detcudnoc ot .secnereferp evitcejbus enimaxe evitcejbus .secnereferp seiduts ecnereferp evitcejbus dna krow noitalumis ,artceps DEL fo noitazimitpo eht nI nI eht noitazimitpo fo DEL ,artceps noitalumis krow dna evitcejbus ecnereferp seiduts ot 9 htiw dezilaer( artceps DEL xelpmoc derreferp taht detseggus shtoob gnithgil ni detcudnoc ni gnithgil shtoob detseggus taht derreferp xelpmoc DEL artceps dezilaer( htiw 9 ot rof htob dezimitpo eb nac seiduts ecnatpecca resu suoiverp ni dnuof )sDEL fo sepyt tnereffid 11 tnereffid sepyt fo )sDEL dnuof ni suoiverp resu ecnatpecca seiduts nac eb dezimitpo htob rof gnittfiorter gnithgil ehT .thgil eht fo ytilauq ruoloc eht gnicfiircas tuohtiw tsoc dna ycneicfife dna tsoc tuohtiw gnicfiircas eht ruoloc ytilauq fo eht .thgil ehT gnithgil gnittfiorter eht dna ,seiranimul DEL htiw serianimul tnecseroufl 8T gnicalper yb taht dewohs yduts esac yduts dewohs taht yb gnicalper 8T tnecseroufl serianimul htiw DEL ,seiranimul dna eht ,ylevitcepser %86 dna %83 ot pu fo sgnivas ygrene ,smetsys lortnoc tliubni htiw seiranimul DEL seiranimul htiw tliubni lortnoc ,smetsys ygrene sgnivas fo pu ot %83 dna %86 ,ylevitcepser tnemirepxe ehT .noitcafsitas resu dna ytilauq gnithgil fo tnemevorpmi htiw deveihca eb nac eb deveihca htiw tnemevorpmi fo gnithgil ytilauq dna resu .noitcafsitas ehT tnemirepxe serutarepmet ruoloc detalerroc tnereffid gnisu sisehtopyh taeh-euh eht tset ot detcudnoc ot tset eht taeh-euh sisehtopyh gnisu tnereffid detalerroc ruoloc serutarepmet neewteb noitalerroc lareneg a fo ecnetsixe eht rof troppus edivorp ton did thgil etihw fo )sTCC( fo etihw thgil did ton edivorp troppus rof eht ecnetsixe fo a lareneg noitalerroc neewteb erom yllamreht tlef elpoep taht detacidni dna trofmoc lamreht ro noitasnes lamreht dna TCC dna lamreht noitasnes ro lamreht trofmoc dna detacidni taht elpoep tlef yllamreht erom 0026 ro K 0072 fo TCC eht ta naht K 0004 fo TCC eht ta ecalpkrow roodni na ni elbatrofmoc ni na roodni ecalpkrow ta eht TCC fo 0004 K naht ta eht TCC fo 0072 K ro 0026 trofmoc ot ytilauq ruoloc gnitaler seiduts gnithgil morf atad etaroprocni ot ygolodohtem A .K A ygolodohtem ot etaroprocni atad morf gnithgil seiduts gnitaler ruoloc ytilauq ot trofmoc .detartsnomed dna desoporp neeb sah metsys noitamotua gnidliub eht ot ytivitcudorp dna ytivitcudorp ot eht gnidliub noitamotua metsys sah neeb desoporp dna .detartsnomed eerht fo ecnereferp TCC eht dna TCC dna ecnanimulli fo smret ni secnereferp evitcejbus ehT ehT evitcejbus secnereferp ni smret fo ecnanimulli dna TCC dna eht TCC ecnereferp fo eerht na ni detagitsevni erew gnithgil ecfifo DEL rof )nacirfA dna ,naeporuE ,naisA( spuorg cinhte spuorg ,naisA( ,naeporuE dna )nacirfA rof DEL ecfifo gnithgil erew detagitsevni ni na 0003 htiw xl 057 dna ,xl 005 ,xl 003 gninibmoc( snoitautis gnithgil tnereffid eniN .moor ecfifo .moor eniN tnereffid gnithgil snoitautis gninibmoc( 003 ,xl 005 ,xl dna 057 xl htiw 0003 saw K 0004 htiw xl 057 fo noitanibmoc ehT .derapmoc erew )K 0005 dna ,K 0004 ,K 0004 ,K dna 0005 )K erew .derapmoc ehT noitanibmoc fo 057 xl htiw 0004 K saw saw tI .gnithgil ecfifo DEL rof snoitanibmoc rehto eht revo derreferp yltnacfiingis yllacitsitats yltnacfiingis derreferp revo eht rehto snoitanibmoc rof DEL ecfifo .gnithgil tI saw ta gnithgil ecfifo rof K 0003 fo TCC eht revo derreferp saw K 0004 fo TCC eht taht dnuof osla dnuof taht eht TCC fo 0004 K saw derreferp revo eht TCC fo 0003 K rof ecfifo gnithgil ta fo gnileef eht ,ssenthgirb fo noisserpmi eht taht detacidni stluser ehT .xl 057 dna ,xl 005 ,xl 003 ,xl 005 ,xl dna 057 .xl ehT stluser detacidni taht eht noisserpmi fo ,ssenthgirb eht gnileef fo .TCC desaercni htiw sesaercni moor ecfifo ni ssensuoicaps fo noisserpmi eht dna ,noitalumits dna eht noisserpmi fo ssensuoicaps ni ecfifo moor sesaercni htiw desaercni .TCC ,K 0004 fo TCC eht referp puorg naeporuE eht gnithgil ecfifo rof taht dewohs osla yduts ehT yduts osla dewohs taht rof ecfifo gnithgil eht naeporuE puorg referp eht TCC fo 0004 ,K sdneped K 0005 dna K 0004 neewteb ecnereferp eht spuorg nacirfA dna naisA eht htiw dna htiw eht naisA dna nacirfA spuorg eht ecnereferp neewteb 0004 K dna 0005 K sdneped .slevel ecnanimulli nopu ecnanimulli .slevel

sdrowyeK thgil gnittime ,edoid ygrene ,ycneicfife ruoloc ,ytilauq gnithgil ,ecnereferp gnithgil ecfifo ,erutarepmet ruoloc detalerroc ruoloc ,erutarepmet ecfifo gnithgil )detnirp( NBSI )detnirp( 9-2607-06-259-879 NBSI )fdp( 2-1607-06-259-879 L-NSSI 4394-9971 NSSI )detnirp( 4394-9971 NSSI )fdp( 2494-9971 rehsilbup fo noitacoL fo rehsilbup iknisleH noitacoL fo gnitnirp iknisleH raeY 6102 segaP 421 nru :NBSI:NRU/fi.nru//:ptth 2-1607-06-259-879

Preface

The research work in this thesis was carried out at the Lighting Unit, Aalto University School of Electrical Engineering. A Part of the work was carried out in the project “Solid State Lighting for Europe (SSL4EU)” funded by the European Commission in the Seventh Framework Programme. Another part of the work was supported by the project “RYM Indoor Environment” funded by the Finnish Funding Agency for Technology and Innovation (TEKES), Aalto University and several Finnish companies. The work was also partly carried out in the projects “Energy Efficiency with Nanotechnology (EffiNano)” and ”Light Energy” funded by Aalto University Energy Efficiency Research Programme. I acknowledge and thank all of the institutions for their support. I would also like to thank the Aalto ELEC Doctoral School for funding my doctoral studies for the duration of 2 years. I like to sincerely thank my supervisor, Professor Liisa Halonen, for believing in me and giving me opportunity to join in the Lighting Unit for my doctoral studies. I would like to express deepest appreciation to her for all of her help, support, advice, guidance and encouragement throughout the years that I have been working at the Lighting Unit. I would like to express my gratitude to my advisors Dr. Eino Tetri and Dr. Pramod Bhusal for their guidance, constructive criticism and encouragement that has helped me grow as a scientist. I would like to thank the preliminary examiners of this thesis, Professor Tran Quoc Khanh from Technical University of Darmstadt, Germany and Professor Sermin Onaygil from Istanbul Technical University, Turkey for their observations and comments on this thesis. I would also like to thank Professor Alexander Rosemann from Eindhoven University of Technology, the Netherlands for accepting to be my opponent. I am also grateful to Dr. Marjukka Puolakka, Dr. Rajendra Dangol, M.Sc. (Tech.) Jukka Virtanen, M.Sc. (Tech.) Marta Gavioli, Dipl-Ing (FH) Alexander Wilm and Dipl-Phys Elmar Baur for their cooperation in publications which have been included in this thesis. I would like to thank Dr. Mikko Hyvärinen, Dr. Mohammad Shahidul Islam, Dr. Paulo Pinho and M.Sc. (Tech.) Mikko Maksimainen, Leena Väisänen, Esa Kurhinen for their help, discussion and giving their time when needed. I would also like to thank all of my colleagues at the lighting Unit for the positive working atmosphere. My heartfelt gratitude go out to my family: To my parents, who have dedicated their life for me and made me what I am today---thank you for your

5 unconditional love, support and all the sacrifices you have made for me. I believe you both are very happy and proud with my achievements. To my dear wife, Sharmila Basnet: Thank you my love for your great care, support, patience and for always being there for me, in good and bad times, as well as all the sacrifices you have made to complete my doctoral studies. To my lovely twin sons Reyan and Shayan: You are the most important achievement during these doctoral studies and the best compliment that destiny has given me. I would also like to thank my sisters and other family members who have always prayed for my success to the almighty God, and my friends who encouraged and supported me during the whole process.

Helsinki, 5 October 2016 Rupak Raj Baniya

6 Contents

Abstract ...... 3 Preface ...... 5 Contents ...... 7 List of Abbreviations and Symbols ...... 9 List of Publications ...... 11 Introduction ...... 13 1.1 Background ...... 13 1.2 The aim of the work ...... 14 State of the art ...... 15 2.1 Energy efficiency improvement ...... 15 2.2 Metrics to measure the colour quality of light source ...... 17 2.3 Visual effects and preference of illuminance and CCT ...... 19 Energy efficiency and colour quality improvement approaches . 21 3.1 Optimization and simplification of preferred LED spectra ..... 21 3.1.1 Generating the simplified LED spectra by simulation ...... 22 3.2 Lighting retrofitting ...... 24 3.2.1 Methods ...... 24 3.2.2 Results ...... 26 3.3 Impact of CCT on thermal sensation and thermal comfort .... 29 3.3.1 Experimental set-up and method ...... 29 3.3.2 Results and statistical analysis ...... 31 3.4 Smart indoor lighting control ...... 32 3.4.1 Experimental set-up and evaluation methodology ...... 32 3.4.2 Simulation and result ...... 34 3.5 Summary ...... 36 Preference studies for LED lighting ...... 37 4.1 Preference of simplified LED spectra ...... 37 4.1.1 Experimental set-up and method ...... 37 4.1.2 Results and statistical analysis ...... 39

7 Contents

4.2 Preference for LED office lighting ...... 40 4.2.1 Experimental set-up and method ...... 40 4.2.2 Results and statistical analysis ...... 41 4.3 Summary ...... 44 Discussion ...... 45 5.1 Energy efficiency and colour quality improvement approaches ...... 45 5.2 Preference studies for LED lighting ...... 48 Conclusions ...... 50 References ...... 52 Publications ...... 59

8 List of Abbreviations and Symbols

Abbreviations

ANOVA Analysis of Variance

ANSI American National Standards Institute

CCRI Categorical Colour Rendering Index

CCT Correlated Colour Temperature

CIE Commission Internationale de l'Eclairage

CM-LED Colour Mixed LED

CRI Colour Rendering Index

CQS Colour Quality Scale

DALI Digital Addressable Lighting Interface

DOE Department of Energy

FCI Feeling of Contrast Index

FL Fluorescent Lamp

GAI Gamut Area Index

HLC High Level Control

HRI Harmony Rendering Index

IEA International Energy Agency

IES Illuminating Engineering Society

IESNA Illuminating Engineering Society of North America

LED Light Emitting Diode

LER Luminous Efficacy of Radiation

LLC Low Level Control

MCC Macbeth Colour Checker

MCRI Memory Colour Rendering Index

9 List of Abbreviations and symbols

PAM Pulse Amplitude Modulation

PC-LED Phosphor Conversion LED

PSU Power Supply Unit

RCRI Rank Order Based Colour Rendering Index

SPD Spectral Power Distribution

SPSS Statistical Package for the Social Sciences

UGR Unified Glare Rating

Symbols

Duv The distance from the chromaticity coordinate of the test light source to the closet point on the Planckian locus on the CIE (u', 2/3 v') coordinates, with a plus sign for above and a minus sign for below the Planckian locus p-value The probability of statistical significance test

Qa General colour quality scale

Qg Gamut area scale, a CQS supplementary metric

Qp Colour preference scale, a CQS supplementary metric

Ra CIE general colour rendering index

10 List of Publications

This doctoral dissertation consists of a summary and of the following publications which are referred to in the text by their numerals

I. Baniya, R.R; Dangol, R; Bhusal, P; Wilm, A; Baur, E; Puolakka, M; Halonen, L. 2015. User-acceptance studies for simplified light-emitting diode spectra. Lighting Research and Technology, volume 47, number 2, pages 177-191.

II. Gavioli, M; Tetri, E; Baniya, R.R; Halonen, L. 2015. Lighting Retrofitting: Improving Energy Efficiency and Lighting Quality. In: Proceedings of the 28th Session of the CIE, Manchester, United Kingdom, June 28 – July 4. International Commission on Illumination (CIE). Volume 1, part 2, pages 1903-1910.

III. Baniya, R.R; Tetri, E; Virtanen, J; Halonen, L. 2016. The effect of correlated colour temperature of lighting on thermal sensation and thermal comfort in a simulated indoor workplace. Indoor and Built Environment (Accepted for publication on 19/09/2016).

IV. Baniya, R.R; Maksimainen, M; Sierla, S; Pang, C; Yang, C. W; Vyatkin, V. 2014. Smart indoor lighting control: Power, illuminance, and colour quality. In: IEEE 23rd International Symposium on Industrial Electronics (ISIE), Istanbul, Turkey, 1-4 June 2014. IEEE. Pages 1745-1750.

V. Baniya, R.R; Tetri, E; Halonen, L. 2015. A study of preferred illuminance and correlated colour temperature for LED office lighting. Light and Engineering, volume 23, number 3, pages 39-47.

The author played a major role in all stages of the work presented in the thesis. The author was responsible as a main author in Publications I, III, IV and V. In Publication I, the author contributed to the experimental design, generated the simplified LED spectra, performed the measurements, conducted the user tests and analysed the user answers. The author contributed in measurements, questionnaire design and was responsible as a co-author in Publication II.

11 List of Publications

In Publication III, the author contributed to experimental design and measurements, and analysed user answers. The author designed the study and generated the data in Publication IV. In Publication V, the author contributed to the experimental design, conducted the user tests, performed the measurements and analysed the user answers.

12 Introduction

1.1 Background

The energy demand for lighting is rapidly growing due to population and economic growth. By 2030, the global electricity consumption for lighting is expected to grow to around 4 250 TWh, an alarming increase of more than 40% within 25 years [1]. The amount of electricity consumption for lighting in buildings depends on many factors including building type and function, the technology used and climate. In commercial buildings, lighting represents generally about 20% to 40% of electricity demand [2]. The electricity consumption for lighting in European office buildings is about 50% of their total electricity consumption [3]. To meet the increased demand for lighting either more energy needs to be supplied or the energy efficiency of lighting should be improved. Energy efficiency in lighting is a way of managing and restraining the growth of energy consumption for lighting without compromising the users’ lighting need. Lighting systems have been identified with significant improvement potential for energy efficiency [4]. Currently, there is a wide range of more efficient lighting technologies available which offer improvements in lamp and luminaire technology and lighting control. The energy efficiency of a light source is typically characterized by using luminous efficacy, which is the ratio of the luminous flux emitted to the electrical power consumed by the source [5]. The luminous efficacy of a light source depends upon both the conversion efficiency from electrical power to optical power called radiant efficiency and the conversion factor from optical power to luminous flux called the luminous efficacy of radiation (LER). Presently, the white light emitting diode (LED) has become a popular candidate to replace conventional light sources in many lighting applications. The main reason for using white LEDs in many lighting applications is the potential of substantial energy savings. The luminous efficacy of white LEDs has increased significantly from 5 lm/W in 1996 to 150 lm/W in 2010 [6]. In 2013, the luminous efficacy of commercially available LED luminaires reached 131 lm/W [7]. Cree demonstrated an LED concept luminaire with the efficacy of 200 lm/W (CIE CRI = 80, CCT =3000 K) [8] and an LED chip with the efficacy of 303 lm/W at the CCT of 5150 K in 2014 [9]. This indicates that LED technology is still developing fast and further improvements of the luminous efficacy of white LEDs is possible. On the other hand, most of other conventional light

13 Introduction sources are generally considered mature technologies with less opportunity for improved efficiency. Properly controlled lighting in a space minimizes the energy consumption of lighting without compromising user lighting need. Lighting control systems enable providing the right amount of light where it is needed and when it is needed. The occupancy control strategy and the daylight linked control strategy are established energy efficient lighting control strategies. LED lighting offers greater freedom in the tunability or controllability of lighting. The colour quality of light source is an important characteristic to be considered when using a light source for indoor lighting [10]. The colour appearance of the light itself and the ability of emitted light to render objects’ colour are two important properties for describing the colour quality of a light source. In current practice, the correlated colour temperature (CCT) and the CIE colour rendering index (CIE CRI) are metrics used to describe the colour quality of a light source. Lighting, through visual and non-visual pathways in the brain, influences the performance of occupants through several mechanisms such as visual performance, visual ambience, visual comfort, and stimulation [11]. For instance, lighting that improves the visual comfort of occupants’ influences performance because of increased concentration. The aesthetic judgements (assessments of the appearance of the space or the lighting) suitable for the needs of occupants in a space play an essential role in providing quality lighting in the space [12]. The performance of the occupants can increase with the quality of light [13]. The cost of employing people in an office is significantly higher than the cost of operating lighting systems [14], [15]. Therefore, compared to the savings gain through a huge improvement in the energy efficiency of office lighting, the savings gain through a slight improvement in the performance of office workers significantly increases the profitability of an office. Thus, in addition to energy efficiency improvements, it is essential to improve the quality of office lighting as it enhances the performance of office workers.

1.2 The aim of the work

The aim of this work was to improve LED office lighting by investigating different approaches for energy efficiency and colour quality improvement, and subjective preferences. The first objective was to explore approaches for energy efficiency and colour quality improvement in LED office lighting. The approaches under study were: the optimization of LED spectra, lighting retrofitting, hue-heat hypothesis and smart indoor lighting control. The second objective was to find out the subjective preferences for LED office lighting. Subjective preference of LED spectra and subjective preference of LED office lighting in terms of CCT and illuminance were studied. In addition, the CCT preferences of three ethnic groups (European, Asian, and African) were investigated.

14 State of the art

2.1 Energy efficiency improvement

Indoor lighting has significant improvement potential for energy efficiency. Adapting purely technological options such as the use of the most energy efficient light sources and the use of lighting controls in indoor lighting lead to significant energy savings [2]. At the moment, white LEDs are one of the most energy efficient light sources. White LEDs are gradually replacing other conventional light sources in many lighting applications. One of the areas for improvement in the luminous efficacy of white LEDs is to optimize the spectra of white LEDs for higher luminous efficacy of radiation (LER). LER is the conversion efficiency from optical power to luminous flux which represents the amount of useful light in lumens obtained from the spectrum of a light source per optical power [16]. For a monochromatic light source, LER is equal to the eye sensitivity function V (λ) multiplied by 683 lm/W [16]. The theoretical maximum LER of 683 lm/W with monochromatic radiation is achieved at the wavelength of 555 nm [17]. A monochromatic light source, as efficient as it might be, would be unacceptable in indoor lighting because it provides no colour differentiation [17], [18]. One approach for producing white light from LEDs is to mix the light of different single colour LEDs in appropriate proportions (CM-LEDs) and other approach is by down conversion of the emission from blue or UV LED to longer wavelength light using phosphors (PC-LEDs) [10], [17]. The United States Department of Energy (DOE) has projected a luminous efficacy goal of 250 lm/W for LED package with CM-LEDs and of 247 lm/W for PC-LEDs [7]. In CM-LEDs, LER generally decreases with increasing numbers of LEDs [19], [20]. Theoretically, dichromatic white light sources are the most efficient, however they are not suitable for general lighting applications because of poor colour rendering performance [21]. The theoretical maximum LER of white light that could be obtained with a pair of monochromatic emitters is about 420 lm/W for a CCT of 6500 K and more than 500 lm/W for lower CCTs [16]. Žukauskas et al. [19] analysed optimal solid-state lamps composed of two, three, four, and five different LEDs and showed that a dichromatic LED lamp offer high efficacy (430 lm/W) with a general CIE CRI close to zero, while a trichromatic LED lamp offer the LER of 366 lm/W with a CIE CRI value of 85. Ohno [10] analysed various LED models generated through simulation and showed that a white LED spectra with a CIE CRI value of 80 can be modelled with three LEDs that yield an LER of 409 lm/W, however, the spectra exhibits very poor rendering of

15 State of the art red colour (R9 = -90). The author demonstrated that it is possible to achieve an LER of 370 lm/W with three LEDs having very good colour rendering properties (CIE CRI = 89, R9 =65) [10]. Coltrin et al. [22] have shown theoretically that an LED spectra with a CIE CRI of 90 can be modelled using four discrete wavelengths that yield an LER of 408 lm/W. The United States Department of Energy (DOE) has estimated the maximum attainable LER of 394 lm/W for a white LED spectra at the CCT of 2700 K with a CIE CRI of 95 and 356 lm/W for a white LED spectra at the CCT of 5000 K with a CIE CRI of 95 [7]. Accurately designed lighting controls have greater potential for energy savings than an increase in light source efficacies [23]. The European standard EN 15193 [24] has specified corrective factors for the adoption of lighting controls in the calculation of lighting energy. The potential savings gain calculated through calculation is generally higher than the savings gain in actual installations [25]. The energy savings potential of the use of occupancy based lighting control typically ranges from 20% to 75% depending on numerous factors including building type, space occupancy profile, occupant behaviour, sensor selection, and applied time delay setting [26]–[28]. By utilizing a daylight linked lighting control strategy about 30% to 77% energy savings can be achieved [29]–[32]. The potential energy savings with daylight-linked control depend on many factors such as daylight availability, building and window design, site orientation and required illuminance. In corridor spaces with low illuminance requirements and high daylight availability, energy savings as high as 93% can be achieved [33]. Williams et al. [25] performed a comprehensive analysis of energy savings of lighting controls used in commercial building based on the review of 88 scientific papers and case studies related to energy savings from lighting controls. The authors concluded that individual lighting control in actual installations save on average between one quarter and one third of lighting energy, while multiple approaches save nearly 40% of lighting energy. Retrofitting the old lighting systems in existing buildings is another option to improve the energy efficiency of indoor lighting. Di Stefano [34] reported that by replacing existing fluorescent lighting systems (T12) at Melbourne University with electronic ballast, T8 magnetic (32 W), T8 electronic (32 W), and T5 electronic (21 W) systems would result in energy savings of 13.9%, 20.5%, 24.4%, and 64.9% respectively. Mahlia et al. [35] investigated potential energy savings by retrofitting the existing fluorescent lighting systems (T12) with T8 electronic (18/36W), HPT8 electronic (17/32W) and T5 electronic (14/28W) systems in the campus buildings of the University of Malaya. The authors found that the T8 electronic, the HPT8 and the T5 electronic system reduce energy consumption up to 17%, 31% and 40% respectively. Ryckaert et al. [36] found that by replacing T8 fluorescent lamps with retrofitted LED lamps in an office room would bring energy savings up to 70%, however, the mean illuminance level was reduced by 50%. The above mentioned studies [34]–[36] did not evaluate the change in users’ satisfaction due to the lighting retrofit. Another possible area with opportunities to improve energy efficiency in indoor lighting is the possible association between light colour and thermal

16 State of the art perception called the hue-heat hypothesis. Studies to test the hue-heat hypothesis have utilized coloured lights [37]–[40], coloured walls [41], coloured cylinders [42], and coloured goggles [43]. Studies that utilized coloured light to test the hue-heat hypothesis have led to conflicting conclusions. Houghten et al. [37] conducted a study wherein two subjects’ judgements of thermal comfort as well as their oral and skin temperatures were recorded while watching a screen illuminated by red, blue, or white light for seven hours. The results did not show the light colour effects on thermal comfort. Berry [38] performed an experiment in a room illuminated with five different colour (amber, yellow, white, green, and blue) lighting produced through fluorescent tubes and coloured filters, where twenty-five subjects were asked to decide when they felt “uncomfortably warm” as the temperature of room was increased. The author did not find any significant difference between the discomfort indices under the five different coloured lighting conditions. Fanger et al. [39] showed a difference in thermal sensation of 0.4°C depending on the illumination of a room by either blue or red light. Albers et al. [40] found yellow lights being perceived as warmer than blue lights in an aircraft-like environment. Existing research is ambiguous regarding the association between coloured light and thermal perception, and the different ways of manipulating colour and response types employed in previous studies complicate the comparability between studies. The studies [37]–[40] used coloured light which does not have any practical application in most indoor work environments. Revisiting the hue-heat hypothesis with the CCT of white light is required before considering a lighting implementation for energy savings inside the buildings.

2.2 Metrics to measure the colour quality of light source

The colour appearance of the light emitted from a light source and the ability of emitted light to render objects’ colour are two important properties for describing the colour quality of a light source. The colour appearance of the light emitted from a light source is often described by the correlated colour temperature (CCT). Moreover, the distance from the chromaticity coordinates of a light source to the Planckian locus also affects the colour appearance of the light source. The chromaticity coordinate of illumination should preferably be very close to the Planckian locus since greenish or pinkish white light is not acceptable for indoor lighting [10]. In 2008, the American National Standards Institute (ANSI) officially defined Duv as “the distance from the chromaticity coordinate of the test light source to the closet point on the Planckian on the CIE (u', 2/3 v') coordinates, with a plus sign for above and a minus sign for below the Planckian locus” [44]. Studies [45], [46] have found that Duv values also affect the preference for LED spectra. The CIE has defined colour rendering as the “effect of an illuminant on the colour appearance of objects by conscious or subconscious comparison with their colour appearance under a reference illuminant” [47]. In 1965, the CIE recommended the procedure of measuring and specifying the colour rendering of light sources based on a test sample method, which rates the colour rendering

17 State of the art properties of a light source in terms of the CIE colour rendering index (CRI) [47]. Despite its prominence, the CIE CRI became problematic when evaluating the colour rendering performance of white light sources particularly for non-smooth and narrowband sources such as tri-band fluorescent lamp and white LEDs [48]. The CIE has acknowledged the problems and shortcomings with the CIE CRI and established technical committees (TC) several times to tackle the problems and to develop proper metric(s). The CIE TC 1-69 (Color Rendition by White light Sources) was one such committee established in 2006 to investigate new methods for assessing the colour rendition properties of white light sources used for illumination, including solid-state light sources, with the goal of recommending new assessment procedures. Nine different metrics have been proposed to the CIE TC 1-69 [45]: (1) Rank-order based colour rendering index (RCRI), (2) Feeling of contrast index (FCI), (3) CRI-CAM02UCS, (4) Colour quality scale (CQS), (5) Harmony rendering index (HRI), (6) Memory colour rendering index (MCRI), (7) Categorical colour rendering index (CCRI), (8) Gamut area index (GAI) and CIE CRI, and (9) Monte Carlo method of assessment. Guo and Houser [49] investigated the correlation among nine metrics based on simulation and proposed the sum z-score of the nine colour rendering metrics to provide information about the colour rendering capabilities of a light source. Bodrogi et al. [50] found that CRI-CAM02UCS and CQS predicted greater and RCRI predicted generally lower colour rendering scores than the CIE CRI. Rea and Freyssinier-Nova [51] demonstrated that CIE CRI and GAI together can ensure subjective judgements of naturalness, vividness and colour discrimination. Smet et al. [52] evaluated thirteen different metrics and found that MCRI correlated highly with perceived appreciation (preference or attractiveness), while the combination of CIE CRI and GAI correlated highly with perceived naturalness. A study conducted by Ryckaert et al. [53] found that CQS and MCRI values did not show consistency with subjective evaluations. Pousset et al. [54] found that visual subjective rankings for the LED light sources were not in line with the ranking provided by CQS. GAI described perceived attractiveness and CQS described perceived naturalness better than other metrics in a visual experiment conducted by Jost-Boissard et al. [55]. The preferred LED SPDs found in studies by Dangol et al. [45] and Islam et al. [46] were characterized by a high CQS colour preference scale (Qp) and a high CQS gamut area scale (Qg). Smet et al. [56] optimized simulated and real tri- and tetrachromatic LED clusters for LER and MCRI and conducted user acceptance studies with an optimized LED lamp. The colour quality of a light source is a complex term which includes several aspects such as: visual appreciation, colour fidelity, naturalness, vividness, colour harmony, visual clarity, and colour discrimination [57]. One single metric cannot fully encapsulate all the aspects of the colour quality of light sources, providing a complete description of a person’s perception of the colour quality of a light source [49], [51]. Houser et al. [58] analysed twenty-two metrics including those considered by the CIE TC 1-69 and found that one measure of colour fidelity and one measure of relative gamut can provide the

18 State of the art most information about the colour rendering properties of light sources. The CIE realized that only one colour fidelity metric does not assess all the aspects of the colour quality of a light source. In 2012, the CIE established two new TCs: CIE TC 1-90 (Colour Fidelity Index) to recommend a single colour fidelity index and CIE TC 1-91 (New Methods for Evaluating the Colour Quality of White-Light Sources) to develop colour preference metric(s). The CIE technical committees are still working to recommend new and more suitable metric(s) to describe the colour quality of light sources. In 2015, the Illuminating Engineering Society of North America (IESNA) published “IES TM-30-15” for evaluating light source colour rendition. The IES TM-30-15 method includes a fidelity index (Rf), a relative-gamut index (Rg), and colour vector and distortion graphics that visually present information about hue and saturation [59].

2.3 Visual effects and preference of illuminance and CCT

In 1941, when fluorescent lighting had just emerged, Kruithof [60] conducted a study to identify “pleasing” combinations between CCT and illuminance level. The results of the study, popularly known as “Kruithof curve”, suggests that people find high CCTs more pleasant at high illuminance levels and low CCTs more pleasant at low illuminance levels. Later in 1967, Bodmann [61] demonstrated that the Kruithof curve holds true only for illuminance levels greater than 3000 lx. The lamps used in the studies [60], [61] had large differences in their CIE CRI values. In 1990, Boyce and Cuttle [62] and Davis and Ginthner [63] concluded that the CCT of light sources with good colour rendering had little effect on people’s impression of lighting. Manav [64] conducted a study in 2007 on the appraisal of the visual environment of offices using fluorescent lamps (CIE CRI = 80) and found that both colour temperature and illuminance had affected the visual appeal of the office room. In 2009, Vienot et al. [65] using LEDs modules (CIE CRI= 90) in a lighting booth found that subjective sensations of bright, clear, and glaring increases with an increase of CCT and the subjective sensation of pleasant, comfortable, relaxing and warm increases when CCT decreases. Lin et al. [66], in 2012, using LED lighting (CIE CRI > 90) in an office room found that CCT had significant effect on the impression of brightness. In 2013, Islam et al. [46] conducted user acceptance studies in a lighting booth with 21 different LED spectra at 500 lx and reported that CCT had significant effect on visual conditions in the lighting booth. All the studies presented above, except the studies [46], [65], [66] were conducted with fluorescent lamps. There were only eight subjects in the study by Lin et al. [66] and studies by Islam et al. [46] and Vienot et al. [65] were conducted with a restricted illuminated field of view in a lighting booth. Presently, the recommended illuminance level for general offices is 500 lx. The recommended illuminance level for general offices was reduced from 1000 lx in 1972 to 500 lx in 1981 [67]. The illuminance recommendation for offices is not determined on scientific basis alone but also depends on the influence of technology, economics and politics [67]. The recommended illuminance level

19 State of the art might be enough for visual performance but might not be enough for the visual comfort of occupants. Existing research is ambiguous regarding the preferred illuminance level for office lighting. Begemann et al. [68] found that office workers expressed preference for illuminance levels significantly higher than the illuminance standard for offices. Moore et al. [69] reported that occupants preferred lower illuminance levels than the recommended illuminance level for office lighting. Manav [64] found that high illumination level (2000 lx) was preferred over low illuminance level (500 lx) for the impression of comfort, spaciousness, brightness and saturation. In a recent study, Viitanen et al. [70] found that 600 lx was equally pleasant and equally easy for a reading task compared to 1000 lx, and 300 lx was less pleasant and more difficult for a reading task than 600 lx and 1000 lx. In the study [70], subjects preferred an illuminance level about 130 lx higher on average with T5 fluorescent lighting than with LED lighting. A study done by Cockram et al. [71] found that people preferred the CCT of 4300 K over the CCT of 6500 K for office work. The CCT of 4000 K was preferred for comfort and spaciousness and 2700 K was preferred for a relaxed feelings in the office environment in the study conducted by Manav [64]. Park et al. [72] found that 4000 K was the most preferred CCT and 5000 K was the appropriate CCT for office lighting. The authors concluded that the preference of CCT depends upon the purpose of the space and the specific activities to be performed in that space. The illuminance level in the study [72] was maintained at about 183 lx which is much lower than the recommended illuminance level for offices. Wei et al. [73] showed that the CCT of 3500 K was more comfortable and satisfying than the CCT of 5000 K for office lighting. In the study by Lin et al. [66], observers preferred both CCT of 4000 K and 7000 K for comfortable. The study by Islam et al. [46] found that observers preferred the CCT of 4000 K and 6500 K over the CCT of 2700 K at 500 lx. Focused and sustained attention was found to be better under 4300 K than 2700 K and 6500 K in a study conducted by Huang et al. [74]. Few studies have been conducted on lighting preference of people from different cultures. Yano and Hashimoto [75] conducted a study to find the chromaticity coordinates of the preferred complexion of Japanese women under the daylight simulating D65 illuminant, and compared their results with the preferred complexion of Caucasian women found by Sanders [76]. The authors found that Japanese women preferred their complexion to look more reddish than their actual complexion but, to be less saturated than complexion of Caucasian women. People with different skin tones preferred different lamp types in a study conducted by Quellman and Boyce [77]. Liu et al. [78] reported that Chinese subjects choose on average a lower CCT (4300 K) than American subjects (6000 K) while selecting the colour quality of illumination for an unfamiliar painting.

20 Energy efficiency and colour quality improvement approaches

3.1 Optimization and simplification of preferred LED spectra

Presently, white light from light emitting diodes (LEDs) is realized by a mixture of multi-color LEDs in appropriate proportions or by combinations of phosphors excited by blue or UV LED emission [10], [17]. Therefore, there is substantial freedom in the spectral design of white LEDs. By tuning the spectrum of LEDs, it is possible to achieve high luminous efficacy or good colour quality characteristics for a light source. The CIE Colour Rendering Index (CIE CRI) is the most widely used metric to evaluate the colour rendering performance of light sources. Despite its prominence, the CIE CRI is not well suited for white LED light and CIE technical committees’ are working to recommend new and more suitable metric(s) to describe the colour quality of light sources. In the absence of proper and suitable metric(s) to assess the colour quality of light sources, user acceptance studies are important to investigate peoples’ judgment of the colour quality of light sources. Several user acceptance studies [55], [79]–[82] have been conducted to investigate the colour quality of light sources. Dangol et al. [45] and Islam et al. [46] also investigated user preferences for lighting in terms of naturalness, colourfulness and visual appearance in small-scale booths where several different LED spectral power distributions (SPDs) were compared with conventional fluorescent lamp SPDs. The preferred LED SPDs found in above mentioned studies [45], [46] were composed of 9 to 11 different types LEDs and were characterised by high CQS Qp and high CQS Qg values. A LED light source composed of 9 to 11 different types LEDs will probably not be commercially exploitable due to its complexity. The problem of a complex light source like this one is that different LEDs have different junction/ambient temperature behaviour characteristics. Therefore, the light outputs of the different LEDs change unequally with varying temperature. This results in a change of the chromaticity point of light output. This could be compensated for by active measurement via optical sensors but it would result in excessive complexity and higher costs in a situation with larger numbers of LEDs. Therefore, a commercially successful solution requires a significant reduction in the types of LEDs used.

21 Energy efficiency and colour quality improvement approaches

The purpose of this work (Publication I) was to investigate the possibility of generating simplified LED spectra having CQS Qp and CQS Qg values similar to those of the preferred complex LED spectra found in the previous studies [45], [46] without sacrificing the luminous efficacy of radiation (LER).

3.1.1 Generating the simplified LED spectra by simulation One of the preferred LED spectra found in the studies by Dangol et al. [45] and Islam et al. [46] at 4000 K is illustrated in Figure 1, which was realized with nine different types of LEDs. Hereafter, this SPD is called the “Aalto SPD”.

1.00

0.50 radiance

Relative spectral spectral Relative 0.00 380 480 580 680 780

Wavelength [nm]

Figure 1. The preferred LED SPD [45], [46], called the Aalto SPD

Using simulation software (which is proprietary software from OSRAM OS, Regensburg, Germany), a number of simplified LED spectra having CQS Qp and CQS Qg values similar to those of the Aalto SPD were generated. In the simulation software, an InGaN chip with different types of phosphors was used to generate the target SPD. Various coloured LEDs may also be added if the target SPD could not be generated by InGaN chip and phosphors only. Five trial LED spectra generated with the simulation software are presented in Figure 2.

Figure 2. Simulated LED SPDs using Osram simulation software. (a) first trial, (b) second trial, (c) third trial, (d) fourth trial and (e) fifth trial

22 Energy efficiency and colour quality improvement approaches

The colour characteristics of the five trial LED spectra and the Aalto SPD are presented in Table 1.

Table 1. Colour characteristics of the Aalto SPD and different trial LED SPDs. The target sets of colour quality metrics and LER values are shown in bold.

First Second Third Fourth Fifth Aalto trial trial trial trial trial LED LED LED LED LED SPD SPD SPD SPD SPD SPD CCT (K) 4153 3970 4000 3978 4048 4000 Duv -0.0050 -0.0014 -0.0007 -0.0005 -0.0042 -0.0041 CIE CRI (Ra) 80 77 94 85 74 96 R (9-12) 62 57 89 78 66 87 R9 23 -21 92 72 39 95

CQS, (Qa v7.5) 90 89 94 89 84 92 LER (lm/W) 308 275 280 325 327 319 CQS Colour preference 101 102 97 94 98 95 scale, (CQS Qp, v7.5) CQS gamut area scale, 114 117 103 107 118 104 (CQS Qg, v7.5) Feeling of contrast index 144 144 117 125 139 123 (FCI)

Among the five trial LED spectra, the first trial LED SPD and the fourth trial LED SPD are the closest matches to the Aalto SPD in terms of target sets of colour quality metrics. The first trial LED SPD has CQS Qp (First trial SPD = 102, Aalto SPD = 101) and CQS Qg (First trial SPD = 117, Aalto SPD = 114) values closed to those of the Aalto SPD, however the LER value of this SPD (275 lm/W) is lower than that of the Aalto SPD (308 lm/W). The first trial LED SPD was generated with three types of phosphors (8.62 wt. % (weight percent) of phosphor 1, 53.05 wt. % of phosphor 2 and 0.14 wt. % of phosphor 3) on an InGaN chip along with a red monochromatic LED of dominant wavelength 635 nm. Therefore, this SPD might be difficult to manufacture and more expensive. The fourth trial SPD was generated with one type of phosphor (48.81 wt. % of phosphor 5) on an InGaN chip along with an amber LED of dominant wavelength 615 nm. This SPD had a higher LER value (327 lm/W) than that of the Aalto SPD (308 lm/W) and had CQS Qp (Fourth trial SPD = 98, Aalto SPD = 101) and CQS Qg (Fourth trial SPD = 118, Aalto SPD = 114) values close to those of the Aalto SPD. The simulation results suggested that it is possible to generate simplified LED SPDs that have CQS Qp and CQS Qg values similar to those of the preferred complex LED spectra without sacrificing LER. Subjective preferences for a lit environment under both the complex and simplified LED spectra are discussed in chapter 4 (Section 4.1).

23 Energy efficiency and colour quality improvement approaches

3.2 Lighting retrofitting

Old lighting systems in existing buildings can be retrofitted with more efficient luminaires in order to minimize energy consumption for lighting. Additional energy savings can be achieved by installing a lighting control system. However, in retrofitting an old lighting system the installation time and costs increase if additional wiring is needed for the control system. Installing luminaires equipped with inbuilt lighting controls can reduce installation time and costs. Lighting retrofit projects should be accurately designed in order to achieve maximum energy savings. Nevertheless, energy savings are not the only requirement in a lighting retrofit. Any attempt to develop an energy efficient lighting strategy should have as its first priority to guarantee that the quality of the luminous environment is as high as possible [3]. In the case of lighting in workplaces, improving lighting quality is particularly important since the luminous environment may greatly affect users’ activities, productivity, and well-being [83]. Therefore, in addition to energy savings, the judgement of lighting quality from the users’ perspective is also important in lighting retrofit projects. A lighting retrofit case study was conducted to determine energy savings and the improvement of lighting quality and user satisfaction achieved by the installation of new lighting technology and by the use of an active dimming control system (Publication II). Old and new lighting systems performances were measured in terms of energy efficiency, lighting quality and user satisfaction.

3.2.1 Methods Six office rooms in the building of Electrical Engineering at the Aalto University were studied for 6 months (Between September 2014 and February 2015). The original lighting system of each room consisted of four T8 fluorescent luminaires (two luminaires with two lamp slots and two luminaires with three lamp slots) which were more than 40 years old. Nevertheless, only one T8 fluorescent lamp was used in each luminaire. An example of room layout and luminaire distribution is shown in Figure 3.

Figure 3. Layout of a test room

24 Energy efficiency and colour quality improvement approaches

The electrical and photometric properties of the old fluorescent lamp luminaires and the new LED luminaires are shown in Table 2. The properties were measured in an integrating sphere.

Table 2. Characteristics of the old fluorescent lamp luminaires and the new LED luminaires

Luminous Power CCT CIE CRI Efficacy Luminaire Dimming Flux (W) (lm) (K) (Ra) lm/W Old 1 No 49 1831 3027 85 37.4 Old 2 No 49 1983 2986 85 40.5 LED No 40 2755 4111 84 68.9 LED-CTRL Yes 35 3378 4094 84 96.5

In two of the rooms, the old luminaires were retrofitted with new LED luminaires without a control system (LED). In two similar rooms, LED luminaires with inbuilt active dimming control (LED-CTRL) were installed. The remaining two rooms were left with the old lighting system as reference. The inbuilt dimming solution for LED luminaires consists of daylight and presence detection sensors as shown in Figure 4 [84]. The installation procedure of the LED luminaires with active dimming control was exactly the same as the installation of LED luminaires without control system.

Figure 4. Sensor detection areas

After the retrofit, two energy meters were installed in two rooms to measure the energy consumption of the new lighting systems. One of the energy meters measured the energy consumption of the lighting system without dimming control; the other measured the energy consumption of the lighting system with active dimming control. In order to obtain a general understanding of the luminous environment as perceived by the occupants as well as subjects’ preferences and usage patterns, a general preliminary questionnaire was first used. The aim was to record the usage pattern of each room, to get to know the room users, and to understand the general condition of the rooms. The information collected was used to design a second user survey questionnaire to evaluate the users’ opinion and satisfaction. Room occupants had to fill in the same users’ survey twice: once before and once after the retrofitting. The questions in the survey were about the appearance of the lighting system, the appearance of the room and of the lighting environment, the amount of light, glare, naturalness, visibility and lighting preference.

25 Energy efficiency and colour quality improvement approaches

3.2.2 Results

Lighting environment Figure 5 shows Dialux simulations for pre- and post-retrofit lighting conditions. Before the retrofit, the average illuminance was 180…350 lx and after retrofit the average illuminance increased to 520…580 lx. Illuminance uniformity increased from 0.5 to 0.8 and the unified glare rating (UGR) increased from less than 10 to 16…18 due to retrofit. The CIE CRI remained almost the same: 84 before and 83 after retrofitting.

(a) (b)

Figure 5. Dialux simulations for (a) pre-retrofit and (b) post-retrofit lighting conditions. False colour rendering shows illuminance distribution

Calculated energy savings To calculate the energy savings some assumptions were made about the usage patterns of the space. First of all, according to the general questionnaire, users stated they worked in the rooms for an average of 6-8 hours a day. Two people worked in each office, therefore the rooms were considered to be occupied for 8 hours a day by at least one employee. Second, average daylight time usage (tD) and non-daylight time usage (tN) were calculated based on the monthly average of daylight hours per day in Helsinki, Finland. The monthly average of daylight hours was decreased by two hours considering the hour after sunrise and the hour before sunset as non-daylight time hours. The numbers of workdays per month take into account weekends and national holidays. These calculations resulted in 240 workdays/year; the 1920 hours/year for which the rooms were occupied were divided into an annual tD of 1740 hours and an annual tN of 180 hours. The users were asked if they had the habit of switching off the lights if daylight was enough to work with. In four of the rooms, the occupants switched the lights off only during the summer. In the other two rooms, the occupants preferred to work with the blinds shut during whole year. Computing with Dialux daylight availably in Helsinki during the summer, this usage behaviour has been modelled assuming that for 20% of summer days the lights were turned down. In addition, the four old luminaires in each room were divided into two manual control groups that could be separately switched on and off. Users were asked if they used to switch on both groups or only one. Moreover, they were asked whether they were using additional desktop lamps or not. In two rooms, before the retrofit, both of the occupants used desktop lamps of 30 W each. After the retrofit, the occupants declared that the desktop lamps were not needed anymore.

26 Energy efficiency and colour quality improvement approaches

The total installed power takes into account the luminaires per room plus the additional desktop lamps if present. The annual running costs have been computed considering a typical Finnish electricity cost of 12 c€/kWh. Table 3 shows the energy savings in the different rooms.

Table 3. Calculated energy savings

Room Number i336 i337 i338 i433 i434 i435 LED-CTRL LED OLD LED LED-CTRL OLD ROOM USAGE PATTERN Operating hours/day 8 8 8 8 8 8 Number of luminaires used Pre-retrofit 2 2 4 4 4 4 Number of luminaires used Post-retrofit 4 4 4 4 4 4 Desktop lamp Pre-retrofit (W) 0 0 0 60 60 0 Desktop lamp Post-retrofit (W) 0 0 0 0 0 0 Only daylight in Summer (% of time) 0.2 0 0.2 0.2 0.2 0 PRE-RETROFIT Operating hours/ year 1726,4 1920 1726,4 1726,4 1726,4 1920 Total power [kW] 0,098 0,098 0,196 0,256 0,256 0,196 Annual energy consumption [kWh] 169,2 188,2 338,4 442 442 376,3 Annual running cost[€] 20 23 41 53 53 45 LENI [kWh/(m2 year)] 8,55 9,51 17,11 22,34 22,34 19,03 POST-RETROFIT Equivalent hours of max. power/year 1019,1* 1920 1726,4 1019,1* Total power [kW] 0,14 0,16 0,16 0,14 Annual energy consumption [kWh] 142,7 307,2 276,2 142,7 Annual running cost[€] 17 37 33 17 LENI[kWh/(m2 year)] 7,21 15,53 13,96 7,21

ANALYSIS Reference Reference Energy savings [kWh/years] 26 -119 166 299 Energy savings [%] 16 -63 38 68 * these values were computed in Dialux LENI - Lighting Energy Numeric Indicator, Power/Area

Energy savings have been achieved in three of the four rooms that were retrofitted. The magnitude of the savings depended on the amount of energy used before the retrofitting and on the installation of lighting controls. In the rooms where desktop lamps were used and all four old luminaires were in operation 8 hours a day, energy savings were important (38% and 68% respectively). In rooms’ i336 and i337, only two of the four old luminaires were used before retrofitting, while after the retrofitting all four lamps were used. Therefore, the total power used for lighting was increased by the retrofit. However, the installation of the active control system in room i336 reduced the operating hours of the lighting system. Thus, some energy savings were achieved (16%) in spite of the fact that the total power used for lighting was increased by the retrofit.

Measured energy consumption In room i433 (LED) and i434 (LED-CTRL) energy meters were installed to measure the weekly energy consumption for three months. The energy consumption of room i435 (OLD) was computed assuming a daily operating time of luminaires equal to 8 hours and used as reference. Total energy consumption and weekly energy consumption of the new lighting systems are shown in Figure 6 and Figure 7 respectively.

27 Energy efficiency and colour quality improvement approaches

room i435(OLD) room i433 (LED) room i434 (LED-CTRL) 100 80 60 40 KWh 20 0 01.12.2014 01.01.2015 01.02.2015 01.03.201 Date

Figure 6. Total energy consumption of the new lighting system in room i433 (LED) and room i434 (LED-CTRL), room i435 (OLD) is used as reference

room i435(OLD) room i433 (LED) room i434 (LED-CTRL) 10

5 kWh 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Week

Figure 7. Weekly energy consumption of the new lighting system in room i433 (LED) and room i434 (LED-CTRL), room i435 (OLD) is used as reference

Energy saving due to control system The installed power in the rooms was not the same. Therefore, the saving related only to the control system was computed in terms of equivalent hours of maximum power (ratio between the metered kWh and the installed power). Computing the average between week 7 and week 13 (when usage patterns stabilised after Christmas holidays), resulted in the lighting system in room i433 (LED) being used 7 hours and 57 minutes per working day compared to 5 hours and 35 minutes per working day in room i434 (LED-CTRL). This indicates that a lighting system of a certain wattage and equipped with dimming control on average consumes 30% less than the same system without dimming control.

User satisfaction The users’ mean rating for pre-retrofitting and post-retrofitting of lighting is shown in Figure 8. It can be seen from Figure 8 that lighting quality has been improved after the lighting retrofit. All the users preferred the new lighting system.

Figure 8. User mean ratings for pre- and post-retrofit questionnaires

28 Energy efficiency and colour quality improvement approaches

3.3 Impact of CCT on thermal sensation and thermal comfort

The “hue-heat hypothesis” states that light or colour with wavelengths predominantly toward the red end of the visual spectrum is perceived as warmer, while that with wavelengths mainly toward the blue end is perceived as cooler [37], [40], [43]. The amount of wavelengths towards the red or blue end of the visual spectrum of white light can be described by correlated colour temperature (CCT). A low CCT light contains more wavelengths towards the red end of the visual spectrum and looks reddish and a high CCT light contains more wavelengths towards the blue end and looks bluish. Drawing on the hypothesis, light source CCTs could be used for saving energy in buildings if CCT has an effect on the thermal sensation of occupants in a room. For example, room temperatures could be lower at a low CCT during the heating season and less air-conditioning might be required at a high CCT during the cooling season. The objectives of this work (Publication III) were to test the hue-heat hypothesis and to study the impact of light source CCT on occupants’ thermal sensation and thermal comfort.

3.3.1 Experimental set-up and method The study was conducted in a test room (4.20 m long x 3.45 m wide x 2.45 m high). The temperature and air velocity of the room were maintained constant with the help of a ceiling mounted air-conditioning unit. Outside the test room in the corridor, a lighting control booth was built from where the experimenter controlled the lighting system to ensure that the subjects were not aware of the CCT changes. To monitor the room temperature remotely, a digital indoor thermometer (Digitales Innen-Außen-Thermometer, Model 30.1012, TFA Dostmann GmbH & Co., Germany) was used in this study. The thermometer sensor was placed inside the test room on a 60 cm-high stool and the thermometer display unit was placed inside the lighting control booth. This also prevented the investigator’s body heat from interfering with the test subjects’ thermal sensation. The layout of the test room and the lighting control booth is shown in Figure 9.

Figure 9. Layout of the test room

29 Energy efficiency and colour quality improvement approaches

Six round (Power = 27 W) recessed downlight ceiling LED luminaires were installed in the test room. All six luminaires were controlled as a group through a digital addressable lighting interface (DALI). This interface was connected to a computer with lighting control software installed. The computer was placed in the lighting control booth outside the test room. Fourteen pre-set lighting situations combining different CCTs with an illuminance level of about 500 lx were utilized for the study as shown in Table 4.

Table 4. Preset lighting situations used in the study

Illuminance CCT CIE Duv Time in minute No. [lx] [K] CRI [Ra]

1 497 2733 92 -0.0016 0 - 11 (for adaptation and filling in questionnaire) 2 497 3010 93 -0.0018 12 3 513 3255 94 -0.0018 13 4 504 3462 94 -0.0016 14 5 498 3686 94 -0.0015 15 6 494 3805 93 -0.0014 16 7 507 4084 92 -0.0011 17 - 28 (for adaptation and filling in questionnaire) 8 506 4331 91 -0.0010 29 9 509 4717 91 -0.0010 30 10 506 4923 90 -0.0010 31 11 504 5342 90 -0.0014 32 12 504 5590 90 -0.0013 33 13 502 5893 90 -0.0006 34 14 496 6208 90 -0.0007 35 - 46 (for adaptation and filling in questionnaire)

Sixteen human subjects aged between 15 and 57 (mean = 38, standard deviation (SD) = 15) took part in the experiments. All the subjects were tested for colour vision with the Ishihara pseudoisochromatic plates’ screening method and found to have normal colour vision. None of the subjects had a background in the lighting field. The subjects wore the same level of clothing and were exposed on different days to a high room temperature (25°C) and a low room temperature (20°C). The lighting situations were gradually changed in steps from the lighting control booth and the subjects were adapted to low CCT (2700 K), medium CCT (4000 K), and high CCT (6200 K) lighting for 10 minutes. After the adaptation, a sound signal was given to prompt subjects to fill in the questionnaire about their thermal comfort and thermal sensation. The questions asked in the experiments queried the subjects for their thermal sensation and thermal comfort after experiencing the lighting for 10 minutes. The rating scale of the questions was a nine point continuous scale with verbal explanations above the scale: very hot, hot, warm, slightly warm, neutral, slightly cool, cool, cold, and very cold for thermal sensation. For thermal comfort, there were fewer verbal explanations, namely, very comfortable, just comfortable, just uncomfortable, and very uncomfortable.

30 Energy efficiency and colour quality improvement approaches

3.3.2 Results and statistical analysis The observers’ median ratings for thermal sensation are shown in Figure 10 and for thermal comfort in Figure 11. At 20°C, the median thermal sensations are -2.00 (cool) with all CCTs. At 25°C, the median thermal sensations are 1.55 (warm), 0.05 (neutral), and 1.15 (slightly warm) at CCTs of 2700 K, 4000 K, and 6200 K, respectively. The median thermal comforts at 20°C are -1.55 at 2700 K, -1.10 at 4000 K, and -2.00 at 6200 K. At 25°C, the median thermal comforts are 1.05 at 2700 K, 1.65 at 4000 K, and 0.95 at 6200 K.

5 5 4 4 3 * 3 * 2 2 1 1 0 0 -1 -1 -2 -2 -3 -3 Thermal sensation Thermal Thermal sensation Thermal

Thermal sensation Thermal -4 -4 -5 -5 2700 K 4000 K 6200 K 2700 K 4000 K 6200 K CCT CCT (a) (b)

Figure 10. Observers’ median ratings (4.5 = very hot, 0 = neutral, -4.5 = very cold) for thermal sensation at different CCTs (a) at 20°C and (b) at 25°C. X = median value; horizontal markers = 25th and 75th percentiles; bars = min and max values. No * over the bar indicates no significant differences and a * over the bar indicates significant differences among groups. Friedman Test (p < 0.05) and Post hoc Wilcoxon signed-rank test.

5 5 4 4 3 3 2 2 1 1 0 0 -1 -1 -2 -2 -3 Thermal comfort Thermal

Thermal comfort Thermal -3 -4 -4 -5 -5 2700 K 4000 K 6200 K 2700 K 4000 K 6200 K CCT CCT

(a) (b)

Figure 11. Observers’ median ratings (4.5 = very comfortable, 0 = neutral, -4.5 = very uncomfortable) for thermal comfort at different CCTs (a) at 20°C and (b) at 25°C. X = median value; horizontal markers = 25th and 75th percentiles; bars = min and max values. No * over the bar indicates no significant differences among groups. Friedman Test (p < 0.05).

The Friedman test showed that light source CCT had no statistically significant effect on the subjects’ thermal comfort at either room temperatures (at 20°C: p = 0.950 and at 25°C: p = 0.321). Similarly, the light source CCT also had no statistically significant effect on the subjects’ thermal sensation at 20°C (p = 0.219). However, at 25°C there was a statistically significant difference in the subjects’ thermal sensation between 2700 K and 4000 K (p < 0.001). The order of the subjects’ median rating for thermal sensation according to the hue-heat hypothesis should range from the highest at 2700 K to the lowest

31 Energy efficiency and colour quality improvement approaches at 6200 K at both room temperatures. At 20°C, the median thermal sensation was equal for all CCTs (Figure 10 (a)). At 25°C, although the subjects’ median rating for thermal sensation at 2700 K was the highest, the median thermal sensation rating at 6200 K was higher than at 4000 K (Figure 10 (b)) which is not in line with the hue-heat hypothesis. Similarly, the order of the subjects’ median rating for thermal comfort according to the hue-heat hypothesis at 20°C (Figure 11 (a)) should range from the highest at 2700 K to the lowest at 6200 K, and at 25°C (Figure 11 (b)), it should range from the highest at 6200 K to the lowest at 2700 K. In this study, at both room temperatures, the subjects’ median thermal comfort rating was highest at 4000 K lighting indicating that the subjects’ thermal comfort rating was not in accordance with the hue-heat hypothesis. Hence, the results do not indicate the order of the subjects’ median rating for thermal sensation and thermal comfort completely in accordance with the hue-heat hypothesis and fail to provide support for the existence of a general correlation between light source CCT and thermal sensation or thermal comfort. Lighting at 4000 K was rated higher for thermal comfort than lighting at 2700 K or 6200 K at both room temperatures.

3.4 Smart indoor lighting control

A methodology has been proposed to incorporate data from lighting studies relating colour quality to comfort or productivity to an intelligent building automation system (Publication IV). An additional objective was to investigate whether the energy savings from varying CCT keeping the CRI value above the minimum requirement are sufficient enough to be implemented in indoor lighting control.

3.4.1 Experimental set-up and evaluation methodology Different LED spectra were generated with the help of a LED-based SPD Simulator System. The characteristics of generated LED spectra with power consumption are shown in Table 5. The columns Blue, White, Amber and Green in Table 5 are the pulse amplitude modulation (PAM) control values to LEDs.

Table 5. Measurement data and power consumption

Illuminance CIE CRI CCT Power Blue White Amber Green (lx) (Ra) (K) (W) 0 127 48 0 333 91 2904 6,32 4 127 43 0 328 91 3065 6,35 37 0 72 209 331 62 6224 11,98 23 0 91 181 329 54 4098 10,79 8 0 121 149 331 51 2645 9,70 0 164 0 0 331 64 4576 6,48 9 61 64 52 333 82 3294 6,64 14 67 50 52 329 85 4004 6,69 0 63 89 31 331 82 2385 6,18 0 106 58 7 331 90 2844 6,05 21 121 25 15 331 87 4910 6,91

32 Energy efficiency and colour quality improvement approaches

The proposed methodology aims to incorporate data from studies relating colour quality to occupant comfort or productivity. The approach is general and any study relating colour quality to comfort or productivity may be used. In order to demonstrate the approach we used the data from the study by Lin et al. [66]. However, it should be noted that the method is general and data from other studies could be used instead. Figure 12 is adapted from the study [66] by using linear interpolation.

Figure 12. Occupant comfort as a function of CCT. Adapted from Lin et al.[66]

Using the function in Figure 12, the comfort rating for each row in Table 5 was obtained. This rating is one metric of comfort. The CIE general colour rendering index (Ra) is another such metric. Both metrics should be maximized. In Figure 13, each measurement in Table 5 is plotted as a dot based on its comfort rating and Ra values. These two are uncorrelated, but since both should be optimized, the ideal measurements are on the Pareto front shown as a red line.

Figure 13. Correlation between comfort rating based on CCT and Ra. Dots correspond to rows in Table 5

However, Figure 13 does not consider power consumption. Our goal is to maximize both the comfort rating based on CCT and Ra and minimize power consumption. The Pareto optimal settings for maximizing comfort and minimizing power consumption are on the Pareto front shown as a red line in Figure 14.

33 Energy efficiency and colour quality improvement approaches

Figure 14. Comfort metric verses power consumption. Dots correspond to rows in Table 5

Table 6 is an excerpt of Table 5 focusing on the four points on the Pareto front in Figure 14. The user will be presented with an interface for selecting any of these settings (setting 1 has the best power saving and setting 4 has the best quality). Based on the user setting, the system will output the corresponding PAM control values as specified in the columns Blue, White, Amber and Green.

Table 6. An excerpt of Table 5 focusing on the four points on the Pareto Front in Figure 14

Blue White Amber Green Power (W) Quality Setting 0 106 58 7 6,05 280,08 1 0 127 48 0 6,32 285,92 2 4 127 43 0 6,35 293,24 3 14 67 50 52 6,69 313,57 4

3.4.2 Simulation and result The lighting simulation system developed in this work was designed following the component-based approach previously proposed in previous studies [85], [86] and is illustrated in Figure 15. The simulation framework consists of three parts: BuildingModelFB, BACFB and SimulatorFB. BuildingModelFB models environment data such as occupant presence and daylight illuminance in each room of a building. It also simulates the operations of the lighting system which provides status data for the virtual meter implemented in BACFB. BACFB consists of three components: the higher level control (HLC) analyses environment data and assigns appropriate control polices for the lower level control (LLC) which interacts with the lighting system. Once the operation status data from the lighting system is evaluated the virtual meter can produce the energy usage data which will be analysed and visualized in SimulatorFB.

Figure 15. Lighting control simulation and virtual metering system

34 Energy efficiency and colour quality improvement approaches

The building used in the simulation has 5 floors and each floor has 5 rooms with different occupancy data for each room. Figure 16 shows the data for one room.

Figure 16. Occupant profile for one of the rooms in BuildingModelFB

The simulation was run for 24 h twice with the exact same occupant profiles. The first run was done using intelligent control: the settings in Table 6 were chosen so that higher settings (with higher comfort and energy consumption) were chosen when the occupants per room exceeded certain thresholds, and lights were off when there were no occupants. The settings were set separately for each room, and results are shown as “smart lighting” in Figure 17. The second run used the maximum comfort setting 4 in Table 6 whenever there was at least one person in the room; otherwise the lights were off. Results are shown as “maximum comfort” in Figure 17.

Figure 17. Power consumption

The simulation results suggested that some energy savings can be attained by the approach. The presented data is sufficient for demonstrating the methodology, but not sufficient for drawing definitive conclusions on the percentage of power savings that may be attained by the approach.

35 Energy efficiency and colour quality improvement approaches

3.5 Summary

The first study on the optimization of LED spectra suggested that without sacrificing luminous efficacy of radiation, it is possible to generate simplified LED spectra having CQS Qp and CQS Qg values similar to those of the preferred LED spectra found in previous studies [45], [46] which were composed with 9 to 11 different types of LEDs. The lighting retrofitting case study showed that by replacing T8 fluorescent luminaires of existing lighting installations in office rooms with LED luminaires bring up to 38% energy savings, and LED luminaires that have inbuilt control system bring up to 68% energy savings. The savings were achieved with increased lighting quality and user satisfaction. The results of the third study did not provide support for the hue-heat hypothesis. Office lighting at 4000 K was rated higher for thermal comfort than lighting at 2700 K or 6200 K. A methodology to incorporate data from lighting studies relating colour quality to comfort or productivity to an intelligent building automation system was demonstrated in the fourth study.

36 Preference studies for LED lighting

4.1 Preference of simplified LED spectra

The simulation presented in Chapter 3 (Section 3.1) suggested that it is possible to generate simplified LED SPDs which have CQS Qp and CQS Qg values similar to those of the preferred complex LED SPDs found in previous studies [45], [46]. User acceptance studies are needed in order to understand subjective preference of lit environments under the simplified LED SPD compared to the preferred complex LED SPD (Publication I). The best approximation trial SPD found in the simulation could not be used as a test SPD in the user acceptance studies because it was not economically viable to manufacture the same simulated LED just for testing user preferences. Moreover, due to limitations of the LEDs in the available LED panel it was not possible to generate the simplified LED SPD at 4000 K but it could be generated at 2700 K. Therefore, user acceptance studies were conducted at 2700 K in lighting booths with forty observers to investigate people’s preferences for the lit environment under the complex and simplified LED spectra.

4.1.1 Experimental set-up and method One of the preferred LED SPDs at 2700 K found in previous studies [45], [46] was chosen as the reference SPD and it was composed of nine different types of LEDs. The SPDs of the nine different types of LEDs chosen to realize the reference SPD are shown in Figure 18 (a). Two LED SPDs were chosen as test SPDs. The first test SPD (called SPD1) was a simplified LED SPD composed of three different types of LEDs and had CQS Qp and CQS Qg values similar to those of the reference SPD. The SPDs of the three different types of LEDs chosen to realize SPD1 are shown in Figure 18 (b).

Figure 18. Individual LED SPDs chosen to realize (a) the reference SPD and (b) SPD1

37 Preference studies for LED lighting

The second test SPD (called SPDP) was a single warm white LED and had lower values of CQS Qp and CQS Qg but a higher value of CIE CRI than the reference SPD. A lighting booth with two identical sections (each section measuring 1 m high, 0.5 m long and 0.5 m deep) was constructed in a dark room. One section of the booth was illuminated by the reference SPD and the other section by the test SPDs (either SPD1 or SPDP). The spectral power distribution of the reference SPD and the test SPDs is shown in Figure 19. The photometric and colorimetric characteristics of the reference SPD and the test SPDs are shown in Table 7.

1.00 1.00 1.00

0.50 0.50 0.50 radiance radiance radiance Relative spectral Relative spectral 0.00 0.00 Relative spectral 0.00 380 580 780 380 580 780 380 580 780 Wavelength [nm] Wavelength [nm] wavelength [nm] (a) (b) (c)

Figure 19. LED SPDs used in the experiment: (a) reference SPD, (b) SPD1, and (c) SPDP

Table 7. Photometric and colorimetric characteristics of the LED SPDs used in the experiment

Test SPDs Reference SPD SPD1 SPDP CCT 2803 2794 2854 Duv -0.0050 -0.0048 0.0015 CIE CRI 82 86 96 CQS colour preference scale, (CQS Qp, v7.5) 100 98 94 CQS gamut area scale, (CQS Qg, v7.5) 119 114 100 LER (lm/W) 256 322 265

Average horizontal illuminance (lx) 458 446 455

Forty observers aged between 20 and 30 (mean 24, standard deviation (SD) 3) took part in the experiments. Among the 40 observers, 20 were of European ethnicity and 20 were of Asian ethnicity. The test objects in the experiment were natural fruits (a green apple, a tomato, an orange, and a lemon), the observer’s own hand and a Macbeth Colour Checker (MCC) chart. The test objects were selected according to the dimension of the perceived colour quality that was to be assessed. For example, to assess perceived naturalness, familiar natural fruits and observer’s hand skin were selected. Similarly, to assess colourfulness, the MCC chart (red, blue, green and yellow colours) and natural fruits of different colours were selected. The questions used in the study queried the subjects for their perception regarding the naturalness of the fruits and their hands, the colourfulness of the MCC chart colours and the fruits, the brightness, and overall preference of the lit environment inside the booth under the test SPDs in comparison with under the reference SPD. The rating scale of the questions in the questionnaire was a seven-point (-3 to +3) scale. Observers responded on the positive side of the scale if they perceived the lit environment under the test

38 Preference studies for LED lighting

SPD to be better than the reference SPD and on the negative side if they perceive the lit environment under the test SPD to be worse than the reference SPD.

4.1.2 Results and statistical analysis The observers’ mean rating for a particular question under the test SPDs in comparison with the reference SPD is shown in Figure 20.A one sample t-test with a significance level p ≤ 0.05 was performed on questions related to naturalness, colourfulness and brightness to investigate the statistical significance of the observer’s ratings irrespective of ethnicity and gender.

Mean value of SPD1 in comparison with Mean value of SPDP in comparison with reference SPD reference SPD 3 3 2 2 1 1 0 0 -1 -1

Mean rating -2 -2 Mean rating -3 -3

(a) (b)

Figure 20. The result of a comparison evaluation in mean value: (a) between the reference SPD and SPD1 and (b) between the reference SPD and SPDP. Error bars represent ± one standard error of the mean

In the comparison between SPD1 and the reference SPD, the observers’ mean ratings for naturalness of observer’s hands and lemon as well as the brightness of lit the environment under the SPD1 were statistically significantly higher than under the reference SPD. However, the colourfulness of the MCC as well as the colourfulness of objects were rated significantly higher under the reference SPD than under SPD1. There was no statistically significant differences between SPD1 and the reference SPD for the naturalness of apple, tomato and orange. The comparison between SPDP and the reference SPD showed that the observers’ mean rating for the naturalness of the fruits and their own hands under SPDP were lower than under the reference SPD. In addition, the observers’ mean ratings for the brightness of the lit environment and the colourfulness of the MCC and the objects were statistically significantly lower under SPDP compared to the reference SPD. For overall preference of the lit environment inside the booth, a Pearson chi-square test with a significance level p ≤ 0.05 was performed because the comparison evaluation data were collected and converted into frequency form by counting the number of observers who chose the reference SPD or either of the test SPDs. The difference in overall preference for the lit environment inside the booth between SPD1 and the reference SPD was not statistically significant, while the difference between SPDP and the reference SPD was statistically significant. The observers strongly preferred the reference SPD over SPDP.

39 Preference studies for LED lighting

4.2 Preference for LED office lighting

The preferred visual conditions in an office environment can create a state of positive affect that in turn leads to higher performance, creativity, and social behaviour [64], [87], [88]. A study was conducted in an office room illuminated with LED lighting to study the preferred combination of illuminance level and CCT and the preferred CCT by different ethnic groups in office lighting (Publication V).

4.2.1 Experimental set-up and method A test room (length 4.20 m, width 3.50 m and height 2.80 m) was equipped and furnished to provide a workspace for the experiment as shown in Figure 21. The walls of the room were painted white with a surface reflectance of 85 %. The reflectance of the ceiling and the floor were 89 % and 25 % respectively.

(a) (b)

Figure 21. The test room: (a) a view of the test room from the door and (b) the layout of the test room

Nine round (Power = 27 W) downlight recessed ceiling LED luminaires were installed to illuminate the test room. All nine luminaires were controlled as a group through a digital addressable lighting interface (DALI). Nine preset lighting situations combining three illumination levels (300, 500, and 750 lx) and three CCTs (3000, 4000, and 5000 K) were saved for the study. The preset lighting situations used in the study are listed in Table 8 with CIE CRI and Duv values.

Table 8. Preset lighting situations. Presentation order was randomized

Situation Illuminance CCT CIE Duv (No) (lx) (K) CRI 1 300 3000 93 -0.0024 2 300 4000 94 -0.0017 3 300 5000 92 -0.0020 4 500 3000 93 -0.0023 5 500 4000 94 -0.0019 6 500 5000 92 -0.0022 7 750 3000 93 -0.0026 8 750 4000 94 -0.0023 9 750 5000 92 -0.0024

Altogether 53 observers aged between 20 and 53 (mean = 29, standard deviation (SD) = 6) took part in the experiment. Among the 53 observers, 20 were of European ethnicity, 20 were of Asian ethnicity and 13 were of African ethnicity. The questionnaire was developed to investigate various aspects of

40 Preference studies for LED lighting office lighting: pleasantness, brightness, visual comfort, stimulation, the apparent spaciousness of room, the naturalness of colours and overall preference. The ratings scale of the questions was a seven-point scale. The right end of the scale was labelled with terms related to the most positive response and the left with the most negative response. The nine different preset lighting situations were presented to the observer in random order. The lighting situation was turned on before the observer entered the room. The observers performed different office activities (a reading task, a matching task and a typing task) under nine different preset lighting situations and rated the lit environment of the room.

4.2.2 Results and statistical analysis The observers’ mean ratings and standard deviations (SD) for different questions are shown in Table 9.

Table 9. The observers’ mean ratings and standard deviations (SD) for different questions under nine lighting situations

300 300 300 500 500 500 750 750 750 lx, lx, lx, lx, lx, lx, lx, lx, lx, 3000 4000 5000 3000 4000 5000 3000 4000 5000 K K K K K K K K K Mean -0.42 0.15 -0.26 0.15 0.91 0.68 0.66 1.53 1.15 Pleasantness SD 1.29 1.51 1.50 1.28 1.32 1.38 1.43 1.20 1.46 Mean -1.38 -0.91 -1.11 -0.11 0.15 0.60 0.57 1.23 1.70 Brightness SD 0.92 1.18 1.03 1.15 1.12 1.15 1.32 0.95 0.97 Visual Mean -0.25 0.17 -0.34 0.32 0.77 0.77 0.57 1.34 1.02 comfort SD 1.39 1.30 1.52 1.38 1.32 1.19 1.35 1.30 1.46 Mean -1.15 -0.66 -0.57 -0.28 0.32 0.64 0.06 1.02 1.09 Stimulation SD 1.32 1.09 1.20 1.32 1.12 1.30 1.46 1.41 1.32 Amount of Mean -1.51 -1.13 -1.19 -0.45 -0.13 0.25 0.43 0.64 0.89 light SD 0.89 1.00 0.94 0.97 0.88 1.07 1.12 1.02 1.03 Mean -0.77 -0.40 -0.25 -0.09 0.34 0.92 0.42 1.09 1.23 Spaciousness SD 1.19 1.15 1.16 1.11 1.06 1.21 1.18 1.06 1.12 Mean 0.26 0.11 -0.32 0.89 0.43 0.42 1.09 1.15 0.74 Hand SD 1.55 1.35 1.31 1.31 1.38 1.59 1.26 1.50 1.53 MCC Mean 0.19 0.17 0.15 0.83 0.91 0.98 1.13 1.38 1.13 colour SD 1.44 1.35 1.28 1.17 1.32 1.39 1.24 1.26 1.26

Naturalness of Mean 0.08 0.19 0.19 0.79 1.04 0.74 1.02 1.28 0.96 Furniture SD 1.41 1.14 1.37 1.27 1.29 1.42 1.15 1.23 1.40 Overall Mean -1.02 -0.58 -0.79 0.13 0.57 0.57 0.34 1.49 0.91 Preference SD 1.23 1.41 1.56 1.41 1.28 1.35 1.48 1.35 1.51

An analysis of variance (ANOVA) with a significance level of p = 0.05 was performed to investigate the statistical significance of the observers’ ratings for a particular question. A post-hoc analysis using the Duncan procedure was performed to investigate which lighting situations and which CCT observers preferred for the questions for which the differences in observers’ mean ratings were statistically significant.

41 Preference studies for LED lighting

Preferred combination of illuminance level and CCT A summary of the statistical analysis of the observers’ preference among nine lighting situations are presented in Table 10. The combination of 750 lx with 4000 K had the highest observers’ mean rating for the pleasantness of the lit environment, visual comfort, the natural appearance of the observer’s hand, the natural appearance of the MCC’s colours, and the natural appearance of furniture. In addition, the combination of 750 lx with 4000 K had the statistically significantly highest observers’ mean rating for overall preference.

Table 10. Observers’ preference among nine lighting situations

Questionnaires Lighting situation Pleasantness 750 lx, 4000 K Visual comfort 750 lx, 4000 K Hand 750 lx, 4000 K Naturalness of MCC colour 750 lx, 4000 K Furniture 750 lx, 4000 K Overall Preference 750 lx, 4000 K

Lighting situation in regular font was in the group of the highest mean value in the Duncan test and had the highest mean value among other lighting situations in the group Lighting situation in bold font had the statistically significantly highest mean value of all the combinations

Preferred CCT at the illuminance levels of 300 lx, 500 lx, and 750 lx Table 11 shows a summary of the statistical analysis of the observers’ preference of CCT at 300 lx, 500 lx, and 750 lx. CCT of 3000 K was considered significantly less pleasant than 4000 K at both 500 lx and 750 lx. The observers’ mean rating for visual comfort under 4000 K was higher than under 3000 K at 300 lx and 500 lx and statistically significantly higher under 4000 K than under 3000 K at 750 lx. At 750 lx, the CCT of 4000 K had the statistically significantly highest observers’ mean ratings for overall preference.

Table 11. Observers’ preference of CCT at different illuminance levels

Questionnaires 300 lx 500 lx 750 lx Pleasantness 4000 K 4000 K 4000 K 4000 K and Visual comfort 4000 K 4000 K 5000 K Hand 3000 K 3000 K 4000 K

Naturalness of MCC colour 3000 K 5000 K 4000 K 4000 K and Furniture 4000 K 4000 K 5000 K 4000 K and Overall Preference 4000 K 4000 K* 5000 K CCT in regular font had the highest observers’ mean ratings for the particular question at the particular illuminance level but the difference in observers’ ratings were not statistically significant Bold font CCT was in the group of the highest mean value in the Duncan test and had the highest mean value in the group for the particular question at the particular illuminance level Bold font CCT* had the statistically significantly highest mean value of all the CCT

42 Preference studies for LED lighting

CCT preference of Asian, European, and African observers A summary of the statistical analysis for the CCT preference of ethnic groups (Asian, European and African) at 300 lx, 500 lx and 750 lx is presented in Table 12. The European preference for pleasantness of lit environment under 4000 K was significantly higher than under 3000 K at 750 lx, and the Asian preference for pleasantness of lit environment under 5000 K was significantly higher than under 3000 K at 500 lx.

Table 12. CCT preference of ethnic groups at different illuminance levels

Naturalness Visual Overall Illuminance Ethnicity Pleasantness of preference level comfort hand 4000 K and Asian 5000 K 4000 K 3000 K 4000 K 300 lx European 4000 K 4000 K 3000 K 4000 K African 4000 K 4000 K 3000 K 4000 K Asian 5000 K 5000 K 5000 K 5000 K 500 lx European 4000 K 4000 K 3000 K 4000 K African 5000 K 5000 K 3000 K 5000 K Asian 5000 K 4000 K 4000 K 4000 K 750 lx European 4000 K 4000 K 3000 K 4000 K African 4000 K 4000 K 4000 K 4000 K CCT in regular font had the highest observers’ mean ratings for the particular question the particular illuminance level but the difference in observers’ ratings were not statistically significant Bold font CCT was in the group of the highest mean value in the Duncan test and had the highest mean value in the group for the particular question at the particular illuminance level At 300 lx and 750 lx, all three ethnic groups’ mean ratings for visual comfort as well as their overall preference, were highest under 4000 K. However, at 500 lx observers from Asia and Africa had the highest mean rating under 5000 K and Europeans had the highest mean rating under 4000 K for visual comfort and overall preference. The European observers preferred 3000 K for the natural appearance of their hand at all three illuminance levels.

Observers’ ratings for brightness, stimulation, and spaciousness A summary of the statistical analysis of the observers’ mean ratings for brightness, stimulation and spaciousness among nine lighting situations are presented in Table 13.

Table 13. A summary of the statistical analysis of the observers’ mean ratings among nine different lighting situations

Questionnaires Lighting situation Brightness 750 lx, 5000 K Stimulation 750 lx, 5000 K Spaciousness 750 lx, 5000 K

Lighting situation without bold was in the group of the highest mean value in the Duncan test and had the highest mean value among other lighting situations in the group

Bold lighting situation had the statistically significantly highest mean value of all the combinations

43 Preference studies for LED lighting

For brightness, the observers’ mean rating for the combination of 750 lx with 5000 K was statistically significantly highest of all the combinations. The combination was also rated highest of all the combination for stimulation and spaciousness of the room. The summary of statistical analysis of the observers’ mean ratings for brightness, stimulation, and spaciousness at 300 lx, 500 lx, and 750 lx are presented in Table 14. At 500 lx and 750 lx, the observers’ mean rating for brightness was statistically significantly highest under 5000 K. The observers’ mean rating for stimulation and spaciousness of the room was also highest under 5000 K at all three illuminance levels

Table 14. A summary of the statistical analysis of the observers mean ratings of CCT at 300 lx, 500 lx, 750 lx respectively

Questionnaires 300 lx 500 lx 750 lx Brightness 4000 K 5000 K* 5000 K* Stimulation 5000 K 5000 K 5000 K Spaciousness 5000 K 5000 K* 5000 K

CCT without bold had the highest observers mean ratings for a particular question at particular illuminance level but the difference in observers ratings were not statistically significant Bold CCT was in the group of the highest mean value in the Duncan test and had the highest mean value in the group for particular question at particular illuminance level Bold CCT* had statistically significantly highest mean value of all the CCTs

4.3 Summary

The user acceptance studies conducted in the lighting booth demonstrated that the simplified LED being more efficient, cost effective, and simpler can provide colour quality characteristic similar or better than the preferred complex LED SPD. The results also provide guidelines for the design of simple LED SPDs with good colour quality, where attention should be given to the values of the CQS Qp and the CQS Qg, rather than the value of CIE CRI. In the office room experiments, the combination of 750 lx with 4000 K was statistically significantly preferred for LED office lighting. The results also indicated that the impression of brightness, feeling of stimulation and impression of spaciousness in office room increases with a higher CCT. The European group preferred a lit environment under 4000 K for LED office lighting and with the Asian and African groups the preference between 4000 K and 5000 K depended upon illuminance levels.

44 Discussion

5.1 Energy efficiency and colour quality improvement approaches

Four different approaches to improve energy efficiency and the colour quality of office lighting were studied. The first study [I] focused on spectral power distribution (SPD) of LEDs. The purpose of the first study was to investigate the possibility of generating optimized and simplified LED spectra that people would prefer. The purpose of the second study [II] was to investigate whether retrofitting old lighting systems in existing buildings with more efficient luminaires and the luminaires equipped with an inbuilt control system, could result in energy savings and at the same time improve lighting quality and user satisfaction. The third study [III] explored the hue-heat hypothesis by varying the CCT of lighting and its possible energy saving application in indoor workplaces. A methodology for smart indoor lighting control was proposed in the fourth study [IV] which incorporated data relating colour quality to comfort to an intelligent building automation system. In the first study [I], simulation was carried out to find the simplified LED SPDs having CQS Qp and CQS Qg values similar to those of the preferred complex LED spectra found in previous user acceptance studies [45], [46]. Among other colour quality metrics, CQS Qp and CQS Qg values were chosen as target sets of colour quality metrics in the simulation because the preferred LED spectra found in previous user acceptance studies [45], [46] were characterised by high CQS Qp and CQS Qg values. The preferred LED spectra were of a complex nature as they consisted of 9 to 11 different types of LEDs, hence obviously expensive to realize. Moreover, the different LEDs have different junction/ambient temperature behaviour and hence the light outputs of the different LEDs change unequally with varying junction/ambient temperature. This results in a change of the chromaticity point of light output. A commercially successful solution requires simplified LED SPDs. LER is another important aspect to consider when generating LED spectra for general indoor lighting. Both colour quality and LER depend on the spectrum of the light source. White LED spectra for general indoor lighting should be optimized for both colour quality and LER. The first trial SPD configuration in the simulation results was a close match to the Aalto SPD in terms of CQS Qp (First trial SPD = 102, Aalto SPD = 101) and CQS Qg (First trial SPD = 117, Aalto SPD = 114), however its LER value was lower (275 lm/W) than that of the Aalto SPD (308 lm/W). Moreover, the first trial SPD uses three types of phosphors and a red monochromatic LED which might be difficult to manufacture and more

45 Discussion expensive. The fourth trial SPD uses only one type of phosphor and an amber LED. This SPD had higher LER value (327 lm/W) compared to the LER value of the Aalto SPD (308 lm/W). The fourth trial SPD also had CQS Qp (Fourth trial SPD = 98, Aalto SPD = 101) and CQS Qg (Fourth trial SPD = 118, Aalto SPD = 114) values close to those of the Aalto SPD. However, its CIE CRI value was low. The CIE CRI has many deficiencies and CIE Technical Committee 1–62 concluded that it is not well-suited for white LEDs [48]. Moreover, several visual experiments [45], [46], [55], [82] with LED light sources indicate that even LEDs with a low value of CIE CRI can produce visually appealing, vivid and natural light. Therefore, if CIE CRI is not considered, the fourth trial SPD was a simplified LED SPD having CQS Qp and CQS Qg values close to those of the Aalto SPD. Hence, the simulation results show that it is possible to generate simplified LED spectra that have higher LER and also CQS Qp and CQS Qg values similar to those of the preferred complex LED SPDs found in previous user acceptance studies [45], [46]. As electric lighting is one of the major consumers of electricity in office buildings, retrofitting old lighting systems in existing office buildings is one of the options to increase energy efficiency. Presently, there is a wide range of lighting retrofitting technologies on the market that offer improvements in lamp, ballast, luminaire technology and lighting control. The lighting retrofitting case study [II] demonstrated that retrofitting T8 fluorescent luminaires with LED luminaries in an existing office building minimizes the energy consumption and improves lighting quality and user satisfaction. The study also demonstrated that by installing LED luminaires equipped with an inbuilt lighting control system installation time and costs can be reduced and additional energy savings can be achieved. Energy savings due to LED luminaires without dimming were 38% of the pre-retrofit power consumption, while the LED luminaires with active dimming brought 68% savings. The greater savings given by the LED luminaires with active dimming control are due to the fact that the control systems reduce the average daily equivalent hours of peak power of the lighting system. The study indicates that a lighting system with dimming control on average consumes about 30% less than the same lighting system without dimming control. Photometric measurements showed that the illuminance level and illuminance uniformity before the retrofit did not meet European standard [89] requirements. After the retrofit, the illuminance level and illuminance uniformity were increased to above the minimum requirements of the European standard. Although due to the retrofit the UGR values in the office rooms were increased, nevertheless the values were below the upper limit specified in the European standard. The retrofit also increased the user satisfaction of different aspects of lighting and the users preferred the new lighting system. The third study [III] investigated the hue-heat hypothesis with CCTs of white light commonly found in indoor workplaces. This study sought to test the hue-heat hypothesis with white light rather than with coloured light since the latter does not have any practical application in offices. The results of the study did not indicate the order of subjects’ median rating for thermal sensation and

46 Discussion thermal comfort completely in accordance with the hue-heat hypothesis and failed to provide support for the existence of a general correlation between light source CCT and occupants’ thermal sensation or thermal comfort. The results also demonstrated that using CCT to produce a significantly cooler sensation in order to feel thermally more comfortable at high room temperature and vice-versa was not possible. One possible factor for the failure to provide support for the hue-heat hypothesis might be that, contrary to common belief, the light source CCT simply does not affect subjective thermal sensation, or the effect might be too small to have practical significance [90]. For example, Takakura et al. [91] evaluated subjects’ thermal sensation while watching video images of a warm (scenery of a hot desert) or a neutral (a gray screen) or a cool (scenery of snow) landscape found significant differences in the subjects’ core body temperature but did not find significant differences in the subjects’ thermal sensation. This implies that the effect might be insufficient to modify the subject’s judgement of thermal sensation. Ho et al. [92] demonstrated that a blue object is more likely to be judged as warm than a red object, an effect which contradicts the hue-heat hypothesis. In this study such opposite effect was not observed. A second reason might be that the physiological effects of CCT on thermal sensation or thermal comfort might be more pronounced in a colder temperature than in the temperatures employed in this study [93]. However, if the physiological effects of CCT are observable in a colder temperature, it is still not possible to use the temperature in offices and hence would be of no practical consequence. Another reason might be that CCT preference rather than the judgement of thermal sensation or comfort might be acting during the subjects’ assessments. The highest rating for thermal comfort at 4000 K might be due to CCT preference rather than thermal judgement as people prefer the CCT of 4000 K for office lighting [V]. The results of this study indicated that people feel thermally more comfortable in office with lighting at 4000 K than with lighting at 2700 K or 6200 K. In the fourth study [IV], a methodology was proposed to incorporate data from lighting studies relating colour quality to comfort or productivity in an intelligent building automation system. The approach for proposed methodology was general and any studies relating colour quality to comfort or productivity may be used. The simulation results demonstrated that there is energy savings potential due to the proposed approach. The presented data was sufficient for demonstrating the methodology, however, the data were not sufficient for drawing definitive conclusions on the percentage of power savings that may be attained by the approach. Therefore, the contribution of the study was the methodology and not the numerical result which was only presented as proof-of-concept. The function used in the study (Figure 12) could be replaced by other studies relating colour quality to comfort and productivity and the proposed methodology would be otherwise useful.

47 Discussion

5.2 Preference studies for LED lighting

Two experiments were conducted for subjective preference of LED office lighting. The first study [I] was carried out in lighting booths to investigate the subjective preference of a simplified LED SPD (realized with 3 different types of LEDs called SPD1) in comparison with the preferred LED SPD found in studies [45], [46] (realized with 9 different types of LEDs called reference SPD). Both SPDs (simplified and reference) had similar values of CQS Qp (SPD1 =98, reference SPD = 100) and CQS Qg (SPD1 = 114, reference SPD = 119). The CIE CRI of reference SPD was 82 and that of SPD1 was 86. In addition, another single warm white LED (called SPDP) that had a very high CIE CRI (Ra = 96) was also compared with the reference SPD. SPDP had lower values of CQS Qp (SPDP =94, reference SPD = 100) and CQS Qg (SPDP = 100, reference SPD = 119) than the reference SPD. In the comparison evaluation between SPD1 and the reference SPD, the naturalness of lemon and the observers’ hands as well as the brightness of the lit environment inside the booth were better under the SPD1 than under the reference SPD. However, the MCC and other objects looked more colourful under the reference SPD compared to SPD1. The colourfulness value of the most objects calculated with CAM02UCS under the reference SPD was higher than under SPD1 which provides an explanation for the higher observer mean ratings for colourfulness under the reference SPD. Both SPD1 and the reference SPD had similar values of CQS Qp, CQS Qg, and the LER of SPD1 (322 lm/W) was greater than that of the reference SPD (256 lm/W). Moreover, SPD1 was realized with only three different types of LEDs whereas the reference SPD was realized with nine different types of LEDs. Therefore, it can be seen that the simplified SPD being more efficient, cost effective, and simpler can provide similar or better colour quality characteristics compared to the preferred complex SPD. Similarly, in the comparison evaluation between SPDP and the reference SPD, the naturalness of the fruits and the observers’ hands under the reference SPD were better than under the SPDP. This indicates that the high value of CIE CRI (Ra = 96) of SPDP compared to that of the reference SPD (Ra = 82) did not guarantee that SPDP renders the naturalness of the fruits and hands better. The observers’ mean ratings for the brightness of the lit environment under the reference SPD was statistically significantly higher compared to SPDP. This contradicts the claims of Fotios [94] that when two scenes are under lamps of similar CCT, the lamp with higher CIE CRI will appear brighter. The colourfulness of the MCC and the objects, and overall preference for the lit environment were also statistically significantly higher under the reference SPD compared to SPDP. The results indicated that peoples’ preference for LED lighting corresponds better to the CQS Qp and CQS Qg values than the CIE CRI value. The second preference study [V] was conducted in an office room. Among nine different lighting situations (combining 300 lx, 500 lx, and 750 lx with 3000 K, 4000 K, and 5000 K), the combination of 750 lx with 4000 K was statistically significantly preferred for LED office lighting. This combination also had the highest observers’ mean rating for the pleasantness of the lit environment,

48 Discussion visual comfort, the natural appearance of the observers’ hand, the natural appearance of the MCC’s colours, and the natural appearance of furniture. Although, the combination of 750 lx with 3000 K was a higher illuminance level than the combinations of 500 lx with 4000 K or 5000 K, however, the observers’ mean ratings for pleasantness, visual comfort, brightness, stimulation, spaciousness and overall preference for the combination of 750 lx with 3000 K were lower than the combinations of 500 lx with 4000 K or 5000 K. The CCT of 4000 K was found more pleasant than the CCT of 3000 K for LED office lighting at all the illuminance levels (300 lx, 500 lx, and 750 lx). In contrast, Viénot et al. [65] found that a lit environment with LEDs inside a lighting booth under 2700 K was judged more pleasant than 4000 K and 6500 K at 150 lx and 600 lx. The reason for the highest rating for pleasantness under 2700 K at the lowest illuminance (150 lx) or at the highest illuminance (600 lx) in the study [65] might be due to the restricted illuminated field of view in the light booth. The observers’ mean rating for visual comfort under 4000 K was higher than under 3000 K at 300 lx and 500 lx and statistically significantly higher under 4000 K than under 3000 K at 750 lx. This result supports the findings of Lin et al. [66] who found that observers’ mean rating for comfort under 4000 K was higher than under 3000 K at 600 lx. At 500 lx and 750 lx, the observers’ mean rating for the brightness of the room under 5000 K was statistically significantly higher than with 3000 K or 4000 K, which supports the finding of previous studies [65], [66], [95] that the observers’ impression of brightness increased with an increased CCT. The office room appeared most spacious under 5000 K at all three illuminance levels which might be due to the fact that brightness prompts the impression of spaciousness [96]. The statistically significantly higher observers’ mean rating for stimulation under 5000 K at all three illuminance levels found in this study might be due to the observation that mental activity levels are higher under a higher CCT than under a lower CCT [97]. Observers from Europe found the CCT of 4000 K more pleasant than the CCT of 3000 K and 5000 K, and observers from Asia found the CCT of 5000 K more pleasant than the CCT of 3000 K or 4000 K for LED office lighting. At 300 lx and 750 lx, all three ethnic groups’ mean ratings for visual comfort as well as their overall preference were highest under 4000 K. However, at 500 lx observers from Asia and Africa had the highest mean rating under 5000 K and Europeans had the highest mean rating under 4000 K for visual comfort and overall preference. The European observers preferred 3000 K for the naturalness of their hand at all three illuminance levels which is in line with the results of Quellman and Boyce [77] who found that a compact fluorescent lamp with 3000 K was preferred over the 4000 K and 5000 K for naturalness of European hands’ skin. The results of this study showed that Europeans preferred the CCT of 4000 K for LED office lighting. Although both Asians and Africans preferred a higher CCT than 3000 K for LED office lighting the choice between 4000 or 5000 K depended on illuminance levels; both of the groups preferred 4000 K at 300 lx and 750 lx, and 5000 K at 500 lx.

49 Conclusions

The goal of this thesis was to improve LED office lighting by investigating different approaches for energy efficiency and colour quality improvement and subjective preferences. Four approaches, namely, optimization of LED spectra, lighting retrofitting, potential correlation between CCT and thermal sensation, and smart indoor lighting control were studied for energy efficiency and colour quality improvement. User acceptance studies for simplified and preferred complex LED spectra were conducted in lighting booths. The preferred combination of illuminance level and CCT, preferred CCT at different illuminance levels, and preferred CCT of ethnic groups (Asian, European, and African) for LED office lighting were investigated in an office room. The simulation results and the user acceptance studies of LED spectra indicated that it is possible to generate simplified LED spectra that people would prefer, and that would provide colour quality characteristics similar or even better than the preferred complex spectra and thus could be commercially exploitable. The user acceptance studies also suggested that while designing simplified LED spectra attention should be given to the values of reference-based metrics (such as the CQS colour preference scale (Qp)) and area-or-volume-based metrics (such as the CQS gamut area scale (Qg)), rather than the value of CIE CRI. The lighting retrofitting case study showed that by replacing T8 fluorescent luminaires with LED luminaries in an existing office building’s lighting systems contribute significant energy savings, and improve lighting quality and office worker satisfaction. The study also demonstrated that additional energy savings without increase in installation time and costs for a control system can be achieved by replacing the T8 fluorescent luminaires with LED luminaries equipped with inbuilt lighting controls. The results on the impact of light source CCT on thermal sensation and thermal comfort indicated that varying the CCT of office lighting cannot be a tool for reducing the energy consumption for lighting in offices and simultaneously maintaining the thermal comfort of office workers. Office workers feel thermally more comfortable with lighting at the CCT of 4000 K than lighting at 2700 K or 6200 K. A methodology has been proposed and demonstrated to incorporate data from lighting studies relating colour quality to comfort and productivity to a building automation system. The findings for the preferences for LED office lighting indicated that the preferences are not only influenced by illuminance level changes but also by light source CCT changes, and that the proper combination of the illuminance

50 Conclusions level and CCT is important for giving a better visual impression. Office workers prefer to have a higher illuminance level than given in today's indoor lighting standards and the CCT of 4000 K for office lighting. The CCT of 4000 K was found to be visually more comfortable than the CCT of 3000 K for office workers. The impression of brightness, feeling of stimulation and impression of spaciousness in the office room was found to increase with a higher CCT. The CCT of 4000 K was preferred for office lighting by the European group and with the Asian and African groups the preference between 4000 K and 5000 K depended on illuminance levels.

51 References

[1] P. Waide and S. Tanishima, “Light’s labour lost: Policies for energy-efficient lighting,” OECD/IEA, Paris, p. 561, 2006.

[2] M.-C. Dubois and Å. Blomsterberg, “Energy saving potential and strategies for electric lighting in future North European, low energy office buildings: A literature review,” Energy and Buildings, vol. 43, no. 10, pp. 2572–2582, 2011.

[3] L. Halonen, E. Tetri, and P. Bhusal, “Annex 45 - Guidebook on energy efficient electric lighting for buildings,” Aalto University School of Science and Technology, Espoo, p. 376, 2010.

[4] L.-B. Desroches and K. Garbesi, “Max tech and beyond: Maximizing appliance and equipment efficiency by design,” Lawrence Berkeley National Laboratory Report - LBNL-4998E, Berkeley, p. 60, 2011.

[5] CIE S 017/E, “ILV: International lighting vocabulary,” Commission Internationale de L’Eclairage (CIE), Vienna, p. 203, 2011.

[6] Y. Narukawa, M. Ichikawa, D. Sanga, M. Sano, and T. Mukai, “White light emitting diodes with super-high luminous efficacy,” Journal of Physics D: Applied Physics, vol. 43, no. 35, p. 354002, 2010.

[7] U.S. Department of Energy (DOE), “Solid-state lighting research and development multi-year program plan,” p. 104, 2014.

[8] CREE, “Cree shatters efficiency benchmark with first 200-lumen-per-watt LED luminaire,” 2014. [Online]. Available: http://www.cree.com/News-and- Events/Cree-News/Press Releases/2014/January/200-LPW-fixture. [Accessed: 25-May-2016].

[9] CREE, “Cree first to break 300 lumens-per-watt barrier,” 2014. [Online]. Available: http://www.cree.com/News-and-Events/Cree-News/Press Releases/2014/March/300LPW-LED-barrier. [Accessed: 25-May-2016].

[10] Y. Ohno, “Spectral design considerations for white LED color rendering,” Optical Engineering , vol. 44, no. 11, p. 111302, 2005.

[11] H. Juslén and A. Tenner, “Mechanisms involved in enhancing human performance by changing the lighting in the industrial workplace,” International Journal of Industrial Ergonomics, vol. 35, no. 9, pp. 843–855, 2005.

52 References

[12] J. Veitch and G. Newsham, “Determinants of lighting quality I: State of the science,” Journal of the Illuminating Engineering Society , vol. 27, no. 1, pp. 92–106, 1998.

[13] L. Edwards and P. Torcellini, “A literature review of the effects of natural light on building occupants,” National Renewable Energy Laboratory Report - NREL/TP-550-30769, Colorado, p. 58, 2002.

[14] M. Fontoynont, “At daytime, performance and well-being,” in Symposium Healthy Lighting, Eindhoven, 2002.

[15] Lighting design lab, “Lighting & productivity.” [Online]. Available: https://www.lightingdesignlab.com/resources/articles/articles-lighting- productivity/lighting-productivity. [Accessed: 25-May-2016].

[16] E. F. Schubert, Light-Emitting Diodes, Second ed. Cambridge UK: Cambridge University Press, p. 434, 2006.

[17] R. Mirhosseini, M. F. Schubert, S. Chhajed, J. Cho, J. K. Kim, and E. F. Schubert, “Improved color rendering and luminous efficacy in phosphor- converted white light-emitting diodes by use of dual-blue emitting active regions,” Optics Express, vol. 17, no. 13, pp. 10806–10813, 2009.

[18] T. W. Murphy, “Maximum spectral luminous efficacy of white light,” Journal of Applied Physics, vol. 111, p. 104909, 2012.

[19] A. Žukauskas, R. Vaicekauskas, F. Ivanauskas, R. Gaska, and M. S. Shur, “Optimization of white polychromatic semiconductor lamps,” Applied Physics Letters, vol. 80, no. 2, pp. 234–236, 2002.

[20] G. He and H. Yan, “Optimal spectra of the phosphor-coated white LEDs with excellent color rendering property and high luminous efficacy of radiation,” Optics Express, vol. 19, no. 3, pp. 2519–2529, 2011.

[21] Y.-L. Li, J. M. Shah, P.-H. Leung, T. Gessmann, and E. F. Schubert, “Performance characteristics of white light sources consisting of multiple light emitting diodes,” in Proceedings of SPIE, 2004, vol. 5187, pp. 178–184.

[22] M. E. Coltrin, J. Y. Tsao, and Y. Ohno, “Limits on the maximum attainable efficiency for solid-state lighting,” in proceedings of SPIE, 2008, vol. 6841, pp. 1–12.

[23] U.S. Department of Energy (DOE), “Energy efficiency program: test procedure for lighting systems (luminaires),” p. 12, 2011.

[24] EN 15193/AC, “Energy performance of buildings - Energy requirements for lighting,” Brussels: European Committee for Standardization, p. 78, 2011.

[25] A. Williams, B. Atkinson, K. Garbesi, F. Rubinstein, and E. Page, “A meta- analysis of energy savings from lighting controls in commercial buildings,” Lawrence Berkeley National Laboratory Report - LBNL-5095E, 2011.

[26] V. Garg and N. K. Bansal, “Smart occupancy sensors to reduce energy consumption,” Energy and Buildings, vol. 32, no. 1, pp. 81–87, 2000.

[27] X. Guo, D. Tiller, G. Henze, and W. CE, “The performance of occupancy-based lighting control systems: A review,” Lighting Research and Technology, vol. 42, pp. 415–431, 2010.

53 References

[28] J. Granderson and A. Agogino, “Intelligent office lighting: Demand-responsive conditioning and increased user satisfaction,” LEUKOS, vol. 2, no. 3, pp. 185– 198, 2006.

[29] L. Doulos, A. Tsangrassoulis, and F. Topalis, “Quantifying energy savings in daylight responsive systems: The role of dimming electronic ballasts,” Energy and Buildings, vol. 40, no. 1, pp. 36–50, 2008.

[30] E. S. Lee and S. E. Selkowitz, “The New York Times Headquarters daylighting mockup: Monitored performance of the daylighting control system,” Energy and Buildings, vol. 38, no. 7, pp. 914–929, 2006.

[31] D. H. W. Li, T. N. T. Lam, and S. L. Wong, “Lighting and energy performance for an office using high frequency dimming controls,” Energy Conversion and Management, vol. 47, pp. 1133–1145, 2006.

[32] S. Onaygıl and Ö. Güler, “Determination of the energy saving by daylight responsive lighting control systems with an example from Istanbul,” Building and Environment, vol. 38, no. 7, pp. 973–977, 2003.

[33] S. K. H. Chow, D. H. W. Li, E. W. M. Lee, and J. C. Lam, “Analysis and prediction of daylighting and energy performance in atrium spaces using daylight-linked lighting controls,” Applied Energy, vol. 112, pp. 1016–1024, 2013.

[34] J. Di Stefano, “Energy efficiency and the environment: The potential for energy efficient lighting to save energy and reduce carbon dioxide emissions at Melbourne University, Australia,” Energy, vol. 25, no. 9, pp. 823–839, 2000.

[35] T. M. I. Mahlia, H. A. Razak, and M. A. Nursahida, “Life cycle cost analysis and payback period of lighting retrofit at the University of Malaya,” Renewable and Sustainable Energy Reviews , vol. 15, no. 2, pp. 1125–1132, 2011.

[36] W. Ryckaert, I. Roelandts, M. Van Gils, G. Durinck, S. Forment, J. Audenaert, and P. Hanselaer, “Performances of LED linear replacement lamps,” in proceedings of CIE 27th session, 2011, pp. 643–652.

[37] F. C. Houghten, H. T. Olson, and J. J. Suciu, “Sensation of warmth as affected by the color of the environment,” Illuminating Engineering, vol. 35, no. 10, pp. 908–914, 1940.

[38] P. C. Berry, “Effect of colored illumination upon perceived temperature,” Journal of Applied Psychology, vol. 45, no. 4, pp. 248–250, 1961.

[39] P. O. Fanger, N. O. Breum, and E. Jerking, “Can colour and noise influence man’s thermal comfort?” Ergonomics, vol. 20, no. 1, pp. 11–18, 1977.

[40] F. Albers, J. Maier, and C. Marggraf-Micheel, “In search of evidence for the hue- heat hypothesis in the aircraft cabin,” Lighting Research and Technology, vol. 47, no. 4, pp. 483-494, 2015.

[41] T. C. Greene and P. A. Bell, “Additional considerations concerning the effects of ‘warm’ and ‘cool’ wall colours on energy conservation,” Ergonomics, vol. 23, no. 10, pp. 949–954, 1980.

[42] M. F. Mogensen and H. B. English, “The apparent warmth of colors,” The American Journal of Psychology, vol. 37, pp. 427–428, 1926.

54 References

[43] C. A. Bennett and P. Rey, “What’s so hot about red?,” Human Factors: The Journal of the Human Factors and Ergonomics Society, vol. 14, no. 2, pp. 149– 154, 1972.

[44] Y. Ohno, “Practical use and alculation of CCT and Duv,” LEUKOS, vol. 10, no. 1, pp. 47–55, 2014.

[45] R. Dangol, M. Islam, H. Hyvärinen, P. Bhusal, M. Puolakka, and L. Halonen, “Subjective preferences and colour quality metrics of LED light sources,” Lighting Research and Technology, vol. 45, no. 6, pp. 666–688, 2013.

[46] M. Islam, R. Dangol, M. Hyvärinen, P. Bhusal, M. Puolakka, and L. Halonen, “User preferences for LED lighting in terms of light spectrum,” Lighting Research and Technology, vol. 45, no. 6, pp. 641–665, 2013.

[47] CIE 13.3-1995, “Method of measuring and specifying colour rendering properties of light sources,” Vienna: Commission Internationale de L’Eclairage (CIE), p. 20, 1995.

[48] CIE 177:2007, “Colour rendering of white LED light sources,” Vienna: Commission Internationale de l’Eclairage (CIE), p. 14, 2007.

[49] X. Guo and K. W. Houser, “A review of colour rendering indices and their application to commercial light sources,” Lighting Research and Technology, vol. 36, no. 3, pp. 183–199, 2004.

[50] P. Bodrogi, S. Brückner, and T. Q. Khanh, “Ordinal scale based description of colour rendering,” Color Research and Application, vol. 36, no. 4, pp. 272–285, 2011.

[51] M. S. Rea and J. P. Freyssinier-Nova, “Color rendering: A tale of two metrics,” Color Research and Application, vol. 33, no. 3, pp. 192–202, 2008.

[52] K. Smet, W. R. Ryckaert, M. R. Pointer, G. Deconinck, and P. Hanselaer, “Correlation between color quality metric predictions and visual appreciation of light sources,” Optics Express, vol. 19, no. 9, pp. 8151–8166, 2011.

[53] W. R. Ryckaert, K. A. G. Smet, I. A. A. Roelandts, M. Van Gils, and P. Hanselaer, “Linear LED tubes versus fluorescent lamps: An evaluation,” Energy and Buildings, vol. 49, pp. 429–436, 2012.

[54] N. Pousset, G. Obein, and A. Razet, “Visual experiment on LED lighting quality with color quality scale colored samples,” in Proceedings of CIE 2010: Lighting Quality and Energy Efficiency, 2010, pp. 722–729.

[55] S. Jost-Boissard, M. Fontoynont, and J. Blanc-Gonnet, “Perceived lighting quality of LED sources for the presentation of fruit and vegetables,” Journal of Modern Optics, vol. 56, no. 13, pp. 1420–1432, 2009.

[56] K. Smet, W. Ryckaert, M. Pointer, G. Deconinck, and P. Hanselaer, “Optimization of colour quality of LED lighting with reference to memory colours,” Lighting Research and Technology, vol. 44, no. 1, pp. 7–15, 2012.

[57] K. A. G. Smet, W. R. Ryckaert, M. R. Pointer, G. Deconinck, and P. Hanselaer, “A memory colour quality metric for white light sources,” Energy and Buildings, vol. 49, pp. 216–225, 2012.

55 References

[58] K. W. Houser, M. Wei, A. David, M. R. Krames, and X. S. Shen, “Review of measures for light-source color rendition and considerations for a two-measure system for characterizing color rendition,” Optics Express, vol. 21, no. 8, pp. 10393–411, 2013.

[59] IES TM-30-15,"IES method for evaluating light source color rendition,” New York: Illuminating Engineering Society (IES), p. 26, 2015.

[60] A. Kruithof, “Tubular luminescence lamps for general illumination,” Philips Technical Review, vol. 6, no. 3, pp. 65–96, 1941.

[61] H. Bodmann, “Quality of interior lighting based on luminance,” Lighting Research and Technology, vol. 32, no. 1, pp. 22–40, 1967.

[62] P. R. Boyce and C. Cuttle, “Effect of correlated colour temperature on the perception of interiors and colour discrimination performance,” Lighting Research and Technology, vol. 22, no. 1, pp. 19–36, 1990.

[63] R. Davis and D. Ginthner, “Correlated color temperature, illuminance level, and the Kruithof curve,” Journal of the Illuminating Engineering Society, vol. 19, no. 1, pp. 27–38, 1990.

[64] B. Manav, “An experimental study on the appraisal of the visual environment at offices in relation to colour temperature and illuminance,” Building and Environment, vol. 42, no. 2, pp. 979–983, 2007.

[65] F. Viénot, M.-L. Durand, and E. Mahler, “Kruithof’s rule revisited using LED illumination,” Journal of Modern Optics, vol. 56, no. 13, pp. 1433–1446, 2009.

[66] R. F. Lin, C. Chou, and Y. Wang, “Effects of LED Color Temperature on Office Workers,” in proceedings of 2nd Southeast Asian network of Ergonomics Societies Conference (SEANES), 2012, pp. 9–12.

[67] P. R. Boyce, “Illuminance selection based on visual performance—and other fairy stories,” Journal of the Illuminating Engineering Society, vol. 25, no. 2, pp. 41–49, 1996.

[68] S. H. A. Begemann, G. J. van den Beld, and A. D. Tenner, “Daylight, artificial light and people in an office environment, overview of visual and biological responses,” International Journal of Industrial Ergonomics, vol. 20, no. 3, pp. 231–239, 1997.

[69] T. Moore, D. J. Carter, and A. I. Slater, “Long-term patterns of use of occupant controlled office lighting,” Lighting Research and Technology, vol. 35, no. 1, pp. 43–59, 2003.

[70] J. Viitanen, J. Lehtovaara, E. Tetri, and L. Halonen, “User preferences in office lighting: A case study comparing LED and T5 lighting,” LEUKOS, vol. 9, no. 4, pp. 261–290, 2013.

[71] A. Cockram, J. Collins, and F. Langdon, “A study of user preferences for fluorescent lamp colours for daytime and night-time lighting,” Lighting Research and Technology, vol. 2, no. 4, pp. 249–256, 1970.

[72] B.-C. Park, J.-H. Chang, Y.-S. Kim, J.-W. Jeong, and A.-S. Choi, “A study on the subjective response for correlated colour temperature conditions in a specific space," Indoor and Built Environment, vol. 19, no. 6, pp. 623–637, 2010.

56 References

[73] M. Wei, K. Houser, B. Orland, D. Lang, N. Ram, M. Sliwinski, and M. Bose, “Office worker response to fluorescent lamps of different CCT and lumen output,” in Proceedings of 2013 Architectural Engineering Institute, 2013, pp. 554–563.

[74] R.-H. Huang, L. Lee, Y.-A. Chiu, and Y. Sun, “Effects of correlated color temperature on focused and sustained attention under white LED desk lighting,” Color Research and Application, vol. 40, no. 3, pp. 281–286, 2015.

[75] T. Yano and K. Hashimoto, “Preference for Japanese complexion color under illumination,” Color Research and Application, vol. 22, no. 4, pp. 269–274, 1997.

[76] C. L. Sanders, “Colour preferences for natural objects,” Journal of the Illuminating Engineering Society, vol. 54, pp. 452–456, 1959.

[77] E. M. Quellman and P. R. Boyce, “The light source color preferences of people of different skin tones,” Journal of the Illuminating Engineering Society, vol. 31, no. 1, pp. 109–118, 2002.

[78] A. Liu, A. Tuzikas, A. Žukauskas, R. Vaicekauskas, P. Vitta, and M. Shur, “Cultural preferences to color quality of illumination of different artwork objects revealed by a color rendition engine,” IEEE Photonics Journal, vol. 5, no. 4, 2013.

[79] S. Boissard and M. Fontoynont, “Optimization of LED-based light blendings for object presentation,” Color Research and Application, vol. 34, no. 4, pp. 310– 320, 2009.

[80] N. Sándor and J. Schanda, “Visual colour rendering based on colour difference evaluations,” Lighting Research and Technology, vol. 38, no. 3, pp. 225–239, 2006.

[81] K. A. G. Smet, W. R. Ryckaert, M. R. Pointer, G. Deconinck, and P. Hanselaer, “Memory colours and colour quality evaluation of conventional and solid-state lamps,” Optics Express, vol. 18, pp. 26229–26244, 2010.

[82] N. Narendran and L. Deng, “Color rendering properties of LED light sources,” in Proceedings of the SPIE, 2002, Vol. 4776, p. 61-67.

[83] P. R. Boyce, Human factors in lighting, Third ed. CRC Press, p. 667, 2014.

[84] Helvar, “Helvar LED solutions brochure”, 2015. [Online]. Available: http://www.helvar.com/sites/default/files/Led_Short_Form_2015_EN_WE B.pdf. [Accessed: 11-Jan-2016].

[85] V. Vyatkin, H. M. Hanisch, C. Pang, and C. H. Yang, “Closed-loop modelling in future automation system engineering and validation,” IEEE Transactions on System, Man, and Cybernetics- Part C: Applications and Reviews, vol. 39, no. 1, pp. 17–28, 2009.

[86] C. Pang and V. Vyatkin, “Systematic closed-loop modelling in IEC 61499 function blocks: A case study,” in 13th IFAC Symposium on Information Control Problems in Manufacturing (INCOM09), 2009, pp. 199–204.

57 References

[87] R. A. Baron, M. S. Rea, and S.G. Daniels, “Effects of indoor lighting (illuminance and spectral distribution) on the performance of cognitive tasks and interpersonal behaviours: The potential mediating role of positive affect,” Motivation and Emotion, vol. 16, no. 1, pp. 1–33, 1992.

[88] J. A. Veitch and G. R. Newsham, “Preferred luminous conditions in open-plan offices: Research and practice recommendations,” Lighting Research and Technology, vol. 32, no. 4, pp. 199–212, 2000.

[89] EN 12464-1, “Light and lighting – Lighting of work places – Part 1: Indoor work places,” Brussels: European Committee for Standardization, p. 43, 2002.

[90] M. Kulve, L. Schellen, L. J. Schlangen, and W. D. Marken Lichtenbelt, “The influence of light on thermal responses,” Acta Physiologica, vol. 216, no. 2, pp. 163–185, 2015.

[91] J. Takakura, T. Nishimura, D. Choi, Y. Egashira, and S. Watanuki, “Nonthermal sensory input and altered human thermoregulation: Effects of visual information depicting hot or cold environments,” International Journal of Biometeorology, vol. 59, no. 10, pp. 1453–1460, 2015.

[92] H.-N. Ho, D. Iwai, Y. Yoshikawa, J. Watanabe, and S. Nishida, “Combining colour and temperature: A blue object is more likely to be judged as warm than a red object,” Scientific Reports, vol. 4, no. 5527, 2014.

[93] A. Yasukouchi, Y. Yasukouchi, and K. Ishibashi, “Effects of color temperature of fluorescent lamps on body temperature regulation in a moderately cold environment,” Journal of Physiological Anthropology and Applied Human Science, vol. 19, no. 3, pp. 125–134, 2000.

[94] S. A. Fotios, “Lamp colour properties and apparent brightness : A review,” Lighting Research and Technology, vol. 33, no. 3, pp. 163–181, 2001.

[95] J. Ju, D. Chen, and Y. Lin, “Effects of correlated color temperature on spatial brightness perception,” Color Research and Application, vol. 37, no. 6, pp. 450–454, 2012.

[96] M. S. Rea, The IESNA lighting handbook: Reference & application, ninth ed. New York: Illuminating Engineering Society of North America, p. 1000, 2000.

[97] T. Deguchi and M. Sato, “The effect of color temperature of lighting sources on mental activity level,” The Annals of Physiological Anthropology, vol. 11, no. 1, pp. 37–43, 1992.

58 nitmtAdagirein lc c E o nm rapeD tnemt fo lE tce r laci reenignE gni dna tamotuA noi

era ytilauq ruoloc dna ycneicfife ygrenE ycneicfife dna ruoloc ytilauq era aA l t o -

I .ntg efion tes tnatropmi stcepsa ni ecfifo l .gnithgi nI DD

ygrenE ycneicfife dna cf h hi ttarpis i ,noitidda ti si tnatropmi ot thgil eht ecfifo 602 derreferp ehT . gni thgi l derreferp ht iw moor iw ht derreferp l thgi gni . ehT derreferp /

ecfifo na ni snoitidnoc suonimul snoitidnoc ni na ecfifo 6102 ruoloc ytilauq tnemevorpmi

no tceffe evitisop a etaerc nac tnemnorivne nac etaerc a evitisop tceffe no puR a k aR j aB n i y a devorpmi ot sdael nrut ni taht rekrow ecfifo rekrow taht ni nrut sdael ot devorpmi laicos dna ,ytivitaerc ,ecnamrofrep krow ,ecnamrofrep ,ytivitaerc dna laicos dna evitcejbus ecnereferp fo ruoivaheb . DEL ecfifo gnithgil tnereffid ruof serolpxe siseht sihT siseht serolpxe ruof tnereffid ulc n cecf gee rf sehcaorppa rof ygrene ycneicfife dna ruoloc T .tneevorp q u a l i t y i pm r o v eme n t . hT e s e a r e gnithgil ,artceps DEL fo noitazimitpo fo DEL ,artceps gnithgil rm da sshoy te-u ,gittfiorter t gni , taeh-euh sisehtopyh dna trams kapuR jaR ayinaB sa iet h .lrncgihil roodni l gnithgi lortnoc . ehT siseht osla E f enrfr v cju senimaxe tcejbus evi secnereferp fo DEL setagitsevni ti ,noitidda nI .gnithgil ecfifo .gnithgil nI ,noitidda ti setagitsevni pog cihe ert f enrfr TC eht TCC secnereferp fo eerht inhte c spuorg E rof )a A da ,n A ,neoruE( naepo , sA i na , dna fA r i )nac f o r DEL .gnithgi l ecfifo l .gnithgi

B I NBS 9-2607-06-259-879 ( p r i n t de ) nith c E cerefer vitc u n eeorm u u c da yn fife y enE r yg e f cfi i ycne dna SSENISUB oc l ruo + auq l i t y nemevo t rpmi dna bus j ce t i ev rp e f e r ecne o f DEL o f ecfi l i hg t i gn B I NBS 2-1607-06-259-879 ( dp f ) E MONOC Y S I NSS - L 9971 - 4394 S I NSS 9971 - 4394 ( p r i n t de ) TRA + S I NSS 9971 - 2494 ( dp f ) NGISED + HCRA I T TCE ERU ytisrevinU otlaA ytisrevinU n engE airteE o loohcS fo tcelE r laci reenignE gni ECNEICS + o aou n n engE airtel f tetrapeD tnemt fo lE tce r laci reenignE gni dna tamotuA noi T E ONHC L GO Y www . a a l t o . f i EVOSSORC RE COD T RO A L +jcgaha*GMFTSH9 COD T RO A L SNOITATRESSID

SNOITATRESSID ytisrevinU otlaA

6102