Professional Documents
Culture Documents
By
Xiaodong Zhang
November 2002
Daylighting Performance of Tubular Solar Light Pipes:
Measurement, Modelling and Validation
a
9.1nl,
-la M,%Z V9 Cý9
)ý,fQa Iyu
ii
ABSTRACT
The innovation of natural daylighting light pipe took place more than twenty yearsago. Sincethen its
daylighting performancehasbeenreportedin a number of studies.To date,however, no mathematical
method that includes the effect of straight-run and bendswithin light pipes hasbeen madeavailable.
Therefore,a generalmathematicalmodel for light pipes is desirableto assessand predict its daylighting
performance. Furthermore,such a generalmodel can enable the assessment of light pipe system's
efficiency and potential in energy saving.
A modified form of daylight factor, Daylight PenetrationFactor (DPF), hasbeen introducedto build a
sophisticatedmodel that takesaccountof the effect of both internal and external environmentalfactors,
and light pipe configuration. Measurementsand mathematicalmodelling activities aimed at predicting
the daylighting performanceof light pipes with various configurationsunder all weatherconditions in
the UK were undertaken.A generaldaylighting performancemodel, namely DPF model, for light pipes
was developedand validated.The model enablesestimationof daylight provision of the light pipes
with a high degreeof accuracy,i.e. R2values of 0.95 and 0.97 for regressionbetweenpredictedand
measuredilluminance were respectivelyobtainedfor the abovemodel.
The DPF model usesthe most routinely measuredradiation data,i.e. the global illuminance as input.
Consideringthat in real applications,light pipes installed in a particular building may not receivethe
full amount of global illuminance as measuredby local meteorologicaloffice. This may be due to
partial shadingof the light pipe top collector dome.Therefore,to enablethe application of the DPF
model in practical exercisesfundamentalwork on sky diffuse illuminance measurementshave been
undertaken.
One of the main achievementsof this work is the mathematicalmethod developedto evaluatethe
daylighting performanceof light pipes. T'heother main achievementof this work is the development
and validation of the DPF models for predicting light pipes' daylighting performance.
iv
DECLARATION
I hereby declarethat the work presentedin this thesiswas solely carried out by myself
at Napier University, Edinburgh, exceptwhere due acknowledgementis madeand
that it has not beensubmittedfor any other degree.
DATE
iii
ACKNOWLEDGEMENTS
This work would not havebeenpossiblewithout the financial support from Napier University, and I
offer my gratitude to all involved in funding this project.
To my friends and colleaguesat Napier University, I warmly acknowledgeyour friendship and support:
Abdulaziz Imtithal, Dr Ana Booth, Dr Baolei Han, Dr Neil Hey, Dr Tom Grassie,Bill Campbell, Bill
Young, FatemaFairooz, GeorgePringle, Ian Campbell,Kevin McCann, MuhammadAsif, Randall
Claywell and Stewart ScotlandHill. I thank ProfessorA Kudish (Israel) and AssociateProfessorJ Lam
(Hong Kong) for providing data for mathematicalmodelling. I thank Dr Gabriel Lopez for his
friendship and advice in the eveningbefore my viva. I must give credit to Miss Amina Muneer for
lending me her room, from time to time, as a laboratory for light pipe performancemeasurements.
Credit is due to David Jenkins(MonodraughtTCS) for developingLux plot basedon DPF model.
I'd like to thank my parents,Shuying Gong and Xinhua Zhang, for supportingme the past twenty-eight
years.Without your love and education,I could not achieveanything.
V
CONTENTS
TITLE I
ABSTRACT IV
ACKNOWLEDGEMENTS V
CONTENTS VI
LIST OF FIGURES xi
NOMENCLATURE xviii
1. INTRODUCTION I
REFERENCES 13
vi
2.5 SUMMARY 29
REFERENCES 30
3. PREVIOUS WORK 44
3.1 APPLICATIONS OF LIGHT PIPES 45
SYSTEMS 51
3.1.6 THE USE OF LIGHT PIPE SYSTEMS IN DEEP PLAN BUILDINGS 52
3.1.7 A SOLID LIGHT GUIDE SYSTEM: AIR-CLAD OPTICAL ROD 53
3.2 WORKING MECHANISM OF LIGHT PIPES 53
3.2.1 UNDERSTANDING OF THE DAYLIGHT TRANSMISSION WITHIN LIGHT PIPES 54
3.2.2 CRITICAL APPRAISAL 55
3.3 RESEARCH METHODS FOR EVALUATING THE EFFICIENCY OF LIGHT PIPES 55
3.3.1 THE EFFICIENCIES OF LIGHT PIPE SOLAR COLLECTORS AND DIFFUSERS 56
3.3.2 THE EFFICIENCY OF LIGHT PIPE TUBE 57
3.3.3 THE EFFICIENCY OF LIGHT PIPE 13ENDS 58
3.3.4 CRITICAL APPRAISAL 59
3.4 THE DESIGN OF LIGHT PIPES 61
3.4.1 CARTER'S DESIGN CHARTS FOR LIGHT PIPES 62
3.4.2 DESIGN METHOD BASED ON COEFFICIENTS OF UTILIZATION 62
3.4.3 CRITICAL APPRAISAL 63
3.5 SUMMARY 64
REFERENCES 66
4. RELEVANT THEORIES 70
4.1 DAYLIGHT PENETRATIONFACTOR 71
vii
4.3.2 MUNEER'S MODEL 76
ROOM 92
viii
6.3 STRAIGHT LIGHT PIPEDPF MODEL (S-DPF) 110
REFERENCES 156
166
8.2 VALIDATION OF S-DPF MODEL USING MERCHISTON TEST ROOM DATA 168
8.2.1 CASE-BY-CASE ASSESSMENTOF THE S-DPF MODEL UNDER NON-HEAVY-
ix
8.2.2 PERFORMANCE ASSESSMENT OF THE S-DPF MODEL UNDER HEAVY-
CONDITIONS 173
8.3 COMPARISON OF DPF MODELS AGAINST RESULTS DUE TO INDEPENDENT
RESEARCHES 175
8.4 THE DESIGN TOOLS 177
8.5 LIMITATIONS OF THE DPF MODELS 178
REFERENCES 180
APPENDIX 1: THE METHOD FOR CALCULATING THE VIEW FACTORS FROM DIFFERENTIAL AREAS TO
215
SPHERICAL SEGMENTS BY NARAGHI
x
LIST OF FIGURES
14
Figure 1.1 World energy supply scenario....................................................................
Figure 1.2 UK Energy Flows 1998,dti, Departmentof Trade and Industry, 2000..... 15
Figure 1.3 ManchesterAirport departurelounge with solar chandeliers.....................16
16
Figure 1.4 Beam sunlight in a side-lit room ................................................................
Figure 1.5Undergrounddaylighting at the SpaceCentrebuilding, University of
Minnesota,Minneapolis, USA 17
.............................................................................
17
Figure 1.6 Conceptof light shelves.............................................................................
Figure 1.7 Generic forms of atrium buildings ............................................................. 18
Figure 1.8 A schematicof light pipe system................................................................ 18
Figure 2.1 Invention by Hanneborgfrom Norway in 1900: apparatusfor transmitting
overcastsky kt=0.2 69
.
.............................................................................................
xi
Figure 4.1 Sunlight of intensity I and elevation ccdescendinga 2-D straight light pipe
84
..............................................................................................................................
Figure 4.2 A projectedview along the axis of a straight tubular light pipe tube, with
light entering at distancex from the axis followed by its incident on the internal
straight light pipe and (b) for light pipe with two bends(D=H=1.2m) .............136
xii
Figure 6.12 Effect of aspect ratio on DPF, (a) for straight light pipe and (b) for light
light pipes with one, two, three and four bends (530mm-diameter, (X,=45',
diffusers 138
..............................................................................................................
Figure 6.15 The comparison of an old opal against a clear diffuser (transparency
Figure 6.16 The comparison of an old opal against a clear diffuser (reflectivity
Figure 6.17 Ring pattern and "pools of light" phenomenon seen in Bahrain light pipe
Figure 7.1 Comparison of the range of diffuse irradiance correction factor given by
Drummond's method and actual results 157
.............................................................
Figure 7.2 Scatter plot of true diffuse irradiance versus corrected irradiance using the
proposed (top) and Drummond's model (bottom) -Bracknell data ................... 158
Figure 7.3 Proposed model's error histogram under all-sky (a), heavy overcast (0 < kt
< 0.2) (b), part-overcast (0.2 < kt: 5 0.6) (c) and clear-sky (0.6 < kt <1) (d)
kt: 5 0.2) (b), part-overcast (0.2 < kt: 5 0.6) (c) and clear-sky (0.6 < kt <1) (d)
proposed (top) and Drummond's model (bottom) Beer Sheva data 161
- ...............
Figure 7.6 Proposed model's error histogram under all-sky (a), heavy overcast (0 < kt
< 0.2) (b), part-overcast (0.2 < kt:!ý 0.6) (c) and clear-sky (0.6 < kt <1) (d)
xiii
Figure 8.2 Examples of characteristics shown by unsatisfactory residuals behaviour
181
......................................................................................................................
Figure 8.3 The residualplot of the differencebetweenpredicted and measuredinternal
illuminances (Y-axis) versuspredictedvalues(X-axis) for S-DPF model, Unit:
lux 182
......................................................................................................................
Figure 8.4 Histogram of absolute error due to S-DPF model, Unit for X-axis: lux.. 182
Figure 8.5 The residual plot of the difference between predicted and measured internal
illuminances (Y-axis, Unit: lux) versus solar altitude (X-axis, Unit: degree) for S-
DPF model 183
.........................................................................................................
Figure 8.6 The residualplot of the differencebetweenpredictedand measuredinternal
illuminances (Y-axis, Unit: lux) versus sky clearness index (X-axis) for S-DPF
183
model..................................................................................................................
Figure 8.7 Scatter plot of predicted internal illuminance against measured internal
illuminance (light pipe 0.21m in diameter and 0.61m in length), Unit: lux ...... 184
Figure 8.8 Scatter plot of predicted internal illuminance against measured internal
illuminance (light pipe 0.2 1rn in diameter and 1.22m in length), Unit: lux 184
......
Figure 8.9 Scatter plot of predicted internal illuminance against measured internal
illuminance (light pipe 0.33m in diameter and 0.61m in length), Unit: lux 185
......
Figure 8.10 Scatter plot of predicted internal illuminance (Y-axis) against measured
internal illuminance (X-axis), (light pipe 0.33m in diameter and 1.22m in length),
x1v
Figure 8.16 Validation: scatter plot of measured internal illuminance (X-axis) versus
sky, 0.21m in diameter and 1.22m in length), unit: lux ..................................... 188
Figure 8.17 Validation: scatter plot of measured internal illuminance (X-axis) versus
sky, 0.33m in diameter and 1.22m in length), unit: lux ..................................... 189
Figure 8.19 Validation: scatter plot of measured internal illuminance (X-axis) versus
sky, 0.42m in diameter and 0.61m in length), unit: lux ..................................... 190
Figure 8.20 Validation: scatter plot of measured internal illuminance (X-axis) versus
predicted values (Y-axis) due to S-DPF model (Merchiston data, all weather),
unit: lux .............................................................................................................. 191
Figure 8.23 Scannedcopy of page 15 of report by Loncour et al [4] 192
........................
Figure 8.24 Scannedcopy of page 16 of report by Loncour et al [4] 193
........................
Figure 8.25 Scannedcopy of Figure 5 of Carter [5] 194
..................................................
Figure 8.26 Scannedcopy of Figure 6 of Carter [5] 194
..................................................
Figure 8.27 Comparativeplots showing close conformancebetweenLiverpool
xv
LIST OF TABLES
Table 1.1 World fossil fuel reserves to production ratio, years (BP 1999 statistics)... 19
Table 5.1 Designs of light pipes that were monitored in Craighouse campus, Napier
University 102
...........................................................................................................
Table 5.2 Designsof light pipes that were monitored in Merchiston Campus,Napier
University 102
...........................................................................................................
Table 5.3 Internal daylight distribution testson light pipes installed in Craighous
Campus,Napier University 102
................................................................................
Table 6.1Coefficients usedin Eqs. 6.5 & 6.6 ............................................................ 142
Table 6.2 Validation results for the model representedby Eq. 6.5............................142
Table 6.3 Validation results for the model representedby Eq. 6.6............................142
Table 6.4 Coefficients usedin Eqs. 6.9 - 6.13 142
..........................................................
Table 6.5 Error statisticsfor Eqs. 6.9 - 6.13.............................................................. 143
Table 6.6 Relationshipbetweena 'conventional' window daylight factor and an
equivalentdaylight factor for a light pipe 143
..........................................................
Table 7.1 Mean Bias Error (MBE), Root Mean SquareError (RMSE) and Percentage
Average Deviation (PAD) comparisonof the proposedmodel versus
Drummond's method,Bracknell 164
........................................................................
Table 7.2 Histogram percentageerror analysisresults,Bracknell. Figures given below
are the number of data points in eachcategory.................................................. 164
Table 7.3 Statistical comparisonof the two models under discussion- Bracknell data.
164
............................................................................................................................
Table 7.4 Mean Bias Error (MBE), Root Mean SquareError (RMSE) and Percentage
Average Deviation (PAD) comparisonof the proposedmodel versus
Drummond's method,Beer Sheva 165
.....................................................................
Table 7.5 Histogram percentageerror analysisresults,Beer Sheva.Figures given
below are the number of datapoints in eachcategory 165
.......................................
Table 7.6 Statistical comparisonof the two models under discussion- Beer Sheva
data 165
. ....................................................................................................................
Table 8.1 Statisticalresults for the performanceof S-DPF model under heavy-
199
overcastsky conditions (kt:ý0.2) ........................................................................
Table 8.2 Internal illuminance distribution (0.33m-diameterlight pipe, 18thDee 2000)
200
............................................................................................................................
xvi
Table 8.3 Summary of the Internal illuminance distribution test results (Craighouse
xvii
NOMENCLATURE
L, Zenith luminance(Cd/m2)
L, o Zenith luminanceof a sky patch of a altitude of 0 (Cd/m2)
in Coefficient usedin Eqs. 6.9 - 6.13
Xviii
N Number of bends
P The proportion of the sky area obscured by the shadow band
P (X, A given location at which illuminance is to be estimated
Y,Z)
R Radius of light pipe tube (m)
t Hour angle (radian)
xix
1. INTRODUCTION
Energy is an essentialcomponentfor all activity and is required for the production of all goods and the
at a faster rate than ever before. All forms of fossil fuels have their respectivecycle of domination and
depletion phase(Fig. 1.1). According to the energystatisticspublishedby BP [11, the world's resource
to production ratio is projected to be 62 years,as shown in Table 1.1. Therefore,the current scenarioof
energy.
from conventionalpower plants [2]. The InternationalPanelfor Climate Changehas arguedthat the
world is definitely getting warmer. In England,the 1990sexperiencedfour out of the five warmest
yearsin a 340-yearrecord, with 1999being the wan-nestyear ever recorded.Global warming and its
the environment.Many projections of the world's future climate show more intenserainfalls or
the 1990s,the number of significant natural catastrophessuch as floods and stormsrose three-fold, and
the whole human society to explore more efficient and environmentally friendly energyresources.
Presentresearchon the innovative daylighting device light pipe that utilizes solar energy is a new effort
Nowadays,energythat takesthe form of electricity and fuel hasbecomethe backboneof our society.
However, during the causeof converting primary fuels into final use energy,a substantialpart of
energy is lost. Figure 1.2 showsthe energy flows in the UK for the year of 1998 [3]. It is shown that in
which electricity accountsfor 31.7 million tonnes.Within the 31.7 million tonnesof electricity, 25.2
million tonnes by
were generated nuclearpower station, and the other 6.5 million tonneswere
is about 37 per cent [4]; while for conventionalsteamstations,the ratio of fuel used for electricity
generationto electricity generatedis about 12 per cent. Therefore,the overall energyefficiency for
electricity generationis poor with the utility ratio lower than 25 per cent.
The use of electricity is not only inefficient but also environmentally detrimental.The combustionof
fossil fuels is responsiblefor the majority of emissionsof carbondioxide C02, sulphur dioxide S02 and
nitrogen oxides N02 [4]. CarbonDioxide is the main greenhousegas,which may contribute to global
warming and climate change.Sulphur dioxide and oxides of nitrogen are the main causeof acid rain.
According to the UK Energy Sector Indicators,in the UK emissionOf C02 from power station is the
largestsingle source,which accountsfor 26.7% of the total emissions.As to S02 and N02,
resources,but also protectsthe natural environmentthat our human society relies on.
In the United Kingdom, electrical lighting accountsfor an estimated5% of the total primary energy
consumedper annum.Office buildings in the UK may consumeup to 60% of their total energy in the
form of electric lighting [5]. The total amountof electricity consumedby domestichousehold
appliancesincreasedby 85% between 1970and 1998 [4]. According to the UK Energy Sector
2
Indicators, for a typical domestichousehold,lighting accountsfor more than 23% of total electricity
savingsof 20% to 40% are attainablefor office buildings that utilise daylight effectively. Benefits and
effect of lowering cooling and heating energyconsumptionin a building [6]. Another significant
benefit of using daylight is it is totally free and clean, which makesit one of the most cost-effective
stadiumsand so on. Thesebuildings are sheltersfor humansto provide them safety and comforts.
enablesa visual contactto the outsideworld. Daylight is also required to enhancethe appearanceof an
interior and its contentsby admitting areasof light and shadethat give shapeand detail for objects.The
Sunlight variation affects many activities of being on the earth.Humansare also affected to seasonal
and daily variation of sunlight. Daylight that is brought into interior spacetells people the changeof
shown that daylight has an important bearing on the humanbrain's chemistry. Light entering via eyes
stimulatedthe nerve centreswithin the brain, which controls daily rhythms and moods.Further
researchhas also establisheda link betweenhuman exposureto light and Serotonin,a substance
enoughamount of daylight, people can positively adjust the production of Serotoninso as to tune their
physical and spiritual condition better. Therefore,daylight affectsmulti-aspectsof human life and has
3
1.4 DAYLIGHTING IN BUILDINGS
The sourceof daylighting is the sun. The sun is a huge nuclear reactorthat hasbeen continuously
emitting solar radiation for around4.5 billion years.It shouldremain more or lessas it is for another
5.5 billion years [7]. The term solar radiation refers to the energyemanatingfrom the sun. Daylight is
the part of energy containedwithin the visible part of the solar radiation spectrum.Human eye is the
organ by which humanbeing perceive daylight. It receiveslight that is omitted or reflected by objects,
and convertsthe light rays into signalsthat can be recognizedby the brain, which producesthe
Human eye's sensitivity to spectralcolour varies from violet to red, which is correspondingto the solar
radiation spectrumof 0.39 to 0.78 pm. Although daylight coversonly a narrow band of the whole solar
radiation spectrum,it has a fundamentaland significant bearing on humanbeing's life on the earth.
On the earth, daylight has two components,one is the direct beam (sunlight) from the sun, and the
other is the sky-light. The latter one is a result of the scatteringof sunlight within the ambient
atmosphereof the earth. For a building, the more light it admitted,the better its habitantscan see.
However, the daylight quantity shouldbe controlled to be in a range.Too much daylight can be
field of view. Solar heat gain is anotherkiller of building design.The use of large glazedareain an
attempt to admit more daylight can causesolar overheating.However, comparedto sunlight, sky-light
is cooler, gentler, diffuse and more visible with a significantly smaller infrared component.Thus, in
situationswhere daylight is desiredwith minimal solar heat gain, sky-light can provide the best quality
of daylight.
Comparedto artificial light, daylight is of better quality. The good colour renderingqualities of natural
light helps to reduceeyestrain.Another benefit of using daylight is it improves the health of human
body. Lack of daylight exposurefor a long time hasbeen found to be the causeof mood swings and
4
significant bearing on maintaining and improving occupants' health condition. To admit as much as
possible daylight into buildings and meanwhile to minimize the glare and solar gain level is an
Window is the mostly widely used daylighting device in buildings. The main function of a window has
beendescribedas, "to provide an outsideview and to permit light to penetratethe interior of a building
in such quantity and with such distribution that it provides satisfactoryinterior lighting results" (8].
work where thereare windows and that the exclusion of daylight leadsto a senseof deprivation and a
lack of well being. However, in practisethe designand application of window is limited by various
One handicapof using window in buildings is that the window's performanceis greatly constrainedby
the external natural and man-madeobstructions.As a solution, larger sized windows are usedto
extra solar gain or heat loss, discomfort and disability glare, strong noise, and concernsfor safety as
glazedwindow with low-emissivity coatings,gas cavity fillings and insulating frame has greatly
There are also circumstanceswhere windows cannotbe used as daylighting devices.For examples,for
interior room within large buildings or corridors betweenrooms where daylight cannotreach,windows
However, the designof skylight can be a difficult task. When skylight admits sky diffuse light into
5
buildings, it also allows sunlight penetrateinto interior spaces.Sunlight transferredthrough skylight
can causeheatand visual discomfort. Therefore,decisionshave to be made on the size and orientation
of skylight with regard to the balancebetweenthe admissionof sufficient daylight and exclusion of
excessivesunlight.
Both windows and skylights can be classified as conventional passive daylighting devices. A main
feature of passive solar devices is they use the form and fabric of buildings to admit, store and
distribute solar energy for heating and lighting without additional energy input and consumption. A
good "one-off' design of passive solar device can provide both environmental and economical benefits.
However, a major difficulty in designing windows and skylights is the design of control system or
mechanism that can ensure the maximum use of desirable daylight, and minimize the induction of solar
and vision discomforts. People realise conventional windows have two major drawbacks. First, the
most concentrated form of natural light, namely sunlight, is effectively useless as a working illuminant,
and secondly in a deep building without rooflights daylighting is restricted to areas near the window
[9]. Tberefore, in the last twenty years, effort has been made to develop innovative daylighting systems
that can improve the distribution of daylight in a space, and to control direct sunlight so that it can be
Generally there are five types of innovative daylighting devicesincluding mirrors, prismatic glazing,
light shelves,atrium and light pipes. There are also various occasionswhere unique daylighting designs
are applied to serveboth lighting and other purpose.For instance,Figure 1.3 showsthe solar
In daylighting design,mirrors are mainly usedto collect, reflect and redirect direct sunlight. A simple
mirrored louver systemis shown in Fig. 1.4.This systemintendsto make use of direct sunlight in a
controlled way so as to improve the daylight distribution within an interior area.There are also more
complicatedmirror systemsthat can lead sunlight into areasdeepwithin buildings. Figure 1.5 showsa
6
heliostat system that is used in a university building in Minneapolis. The system beams the sunlight
through a lens into an undergroundlaboratory,saving electrical lighting and cooling costs [10].
Prismatic glazing can be usedto alter the direction of incoming daylight. When beamtraversesa prism,
its path is turned through 90" or smaller angle.This refraction phenomenonenablesprismatic glazing
to be usedas a substitutefor mirrored louvres systemshown in Fig. 1.4.There are also sunlight-
excluding prism systemsthat reject sunlight throughoutthe year, but admit diffuse light from higher in
the sky. The diffuse light is then directedto the ceiling so asto provide a controlled and comfortable
daylit interior.
Light shelveswork on the principle of reflection to redirect sunlight and sky light. The light shelvesare
window and a clerestorywindow. Both interior light shelvesand exterior light shelvescan be used.
Figure 1.6 showsan interior light shelvebeing placedbetweena view window and a clerestory
window. Daylight from the clerestorywindow is reflectedto the ceiling and then further diffused more
evenly into the deepareaof the room. When thereis no direct sunlight, light shelvescan help to
the UK aswell [10]. Atrium is rather an innovative building designconceptmore than an daylighting
Light pipe is an innovative design to direct daylight into deep areas within buildings where daylight
cannot reach. The system is crudely analogous to the fibre optic cable, the difference being that the
latter device has much higher transmission efficiency. A schematic of light pipe system is shown in
Fig. 1.8. Daylight is gathered via a polycarbonate dome at roof level and then transmitted downwards
to interior spaces within buildings. The internal surface of the light pipe is coated with a highly
reflective mirror finish material (typically with a reflectance in excess of 0.95) that helps in achieving a
7
reasonable illuminance indoors when daylight is introduced via a light diffuser. The light-reflecting
Like other innovative daylighting devices,light pipe utilizes sunlight that accountsfor two thirds of
global illuminance in a clear day. Both sky daylight componentsare gatheredby light pipe and then
transmittedand diffused into interior whereneedto be day lit. Becauseof its main structureas a well-
sealedtube, light pipe has addedpotential advantagein reducing excessivesolar gain. Sincepiped
daylight emits off only from light pipe diffuser, the output daylight is easierto control than other
directly upon where needto be lit, so as to achievea good internal daylight distribution.
The combineduse of windows and light pipes can reduceglare and further improve the balanceof
daylighting within a room. In daylighting, therehas always beena dilemma for designers.It is that on
one hand large exposureof sky through window is not encouragedbecauseit can causevision
discomfort, while on the other hand a high daylight factor is always preferableand important for
highlight and shadowit may cause.However, by applying light pipes aboveproblems can be solved.
By introducing redirectedand diffused daylighting into deepareaof a room, glare from windows is
reducedand daylighting is of a better uniformity. For skylight designers,the attemptto admit more sky
into buildings. As a comparison,light pipes transmit sunlight and sky diffuse light by multi-reflect
mechanism;thereforethe output daylight is much more uniform and diffused than that by a skylight.
Another potential advantagethat light pipe possessesis that it can be used in multi-storey buildings,
As addressedbefore, the applicationsof light pipes in buildings can bring multi-fold benefit. Compare
to other innovative daylighting devices,light pipe seemsalso have potential advantagesin terms of
visual and thermal performanceand applicability. Moreover, the fact that light pipes have been
commonly used in the USA and Australia implies its further developmentin the United Kingdom and
Europe.However, for most UK and Europeandaylighting designers,light pipe is still a relatively new
8
concept[ 11]. Researchthat focus on the evaluationof light pipe's daylighting performanceis therefore
The difficulties in identifying all decisivefactors that affect the performance,and in quantitatively
weighing the contributions of thesedecisive factors are the main barriers to appraisinglight pipe's
efficiency. The complexity in the mechanismby which light pipes transmit daylight makesit difficult
to appraisethe performanceby meansof physical modelling. On the other hand,prior to this research,
the lack of light pipe performancedata that include both sufficient environmentaland geometricaldata
The overall objective of this work is to provide a generalmathematicalmodel for the prediction of light
(1) To conducta literature review of works in the field in order to provide a basic grounding in the
prediction of performanceof light pipe systemsof various designsand under different weather
conditions.
(2) To developa logical method of building mathematicalmodel for predicting the daylighting
performanceof light pipes. Although the complexity in the working mechanismof light pipe makesit
difficult to appraiseits performance,to the aid of mathematicalsoftware and the computing ability by
9
mathematicalmodel can have a limited applicability. However, more importantly the study aims to
producea valid researchmethod, and a framework under which further developmentin daylighting
aim, as a principle guidesthe implement of the presentresearchfrom the day one to the end, and is
therefore embodiedthrough out this report. The work conductedand the conclusionsdrawn to satisfy
performance of light pipes both in real application and in test rooms. Extensive and detailed
geometrical and environmental data, namely the light pipe configuration information, internal
illuminance and external weather conditions, sun's position and so on need to be gathered.
Measurements on the performance of light pipes in separate test rooms are to be under-taken so as to
enable the validation of the proposed model using independent data. This work is presented in Chapter
5.
(4) To analyse the working mechanism of light pipes. By analysing the measured data and physical
reasoning, build the analytical framework relating the input and output of the proposed model, and
identify the most decisive factors that affect light pipe's daylighting performance. To determine the
"best" mathematical expression to describe the daylighting transmission pattern of light pipe, choose
suitable solar radiation data as input and, choose the "best" internal daylight distribution model to
describe the output of the proposed model. To validate the proposed model using independent data by
irradianceare the three most important quantities.Global horizontal illuminance (E,, ) can be easily
measurementson E, 9 and Evdare widely available aroundthe world by local Met Offices. In real light
pipe applications,treesand buildings that obscurepart of sky diffuse light and sometimessun beam
reducethe input illuminance to the light pipe systems.Under suchconditions, to apply the proposed
10
model, total available illuminance needto be estimatedbasedon E,, and Evddata available from local
Met Offices.
Napier University CIE first-class solar radiation station is currently using the shadowband device to
provide the horizontal diffuse irradiance data.However, it is well known that the shadowband that is
used to block the sunshinealso shadessomediffuse irradianceas well. Hence it is necessaryto correct
method to correct the shadowband diffuse data is the Drummond method. However, Drununond
shall be developed.Work carried out for this aim and conclusiondrawn are reportedin Chapter7.
(6) In recentyears,solar light pipes have firmly set their foot within the UK market place [ 15].
However, till this researchno designtool for light pipe practices,which is basedon sophisticated
performanceby designersin the UK is neither clear nor more valid than limited experience,if there is.
Moreover, becausethe lack of model that can predict the perforinanceof light pipes, the assessment
of
the potential contribution of daylighting using due product to energysaving is rather crude and
empirical. Researchon identifying the potential barriers to exploit daylight in Britain [10] concluded
estimationof electricity saving due to light pipe applicationsunder the UK climate, which can provide
(7) During the courseof presentresearch,the commercialdevelopmentof light pipe in the UK hasbeen
quite rapid. One main progressin improving light pipe's daylighting performanceis the application of
translucenttype of diffuser. Therefore it would be worth to investigatethe effect of using new type of
11
(8) Daylight factor hasbeenacceptedas an industry standardfor window design,However, for
innovative daylighting devicesand someof the new designs,especiallythosethat utilize not only sky-
light but also sunlight, to dateno generalmethod is availableto assesstheir daylighting performance.
Basedon the conceptof light pipe daylight penetrationfactor, DPF, introduced in presentstudy along
Merit (FoM) to the purposeof introducing a generalmethod for assessingall daylighting devices.Work
12
REFERENCES
5. Muneer, T. (1997) Solar radiation & daylight modelsfor the energyefficient designofbuildings,
Architectural Press,Oxford
6. Zhang, X and Muneer, T. (2000) A MathematicalModel for the Performanceof Light pipes Lighting
7. htlp://www. hao.ucar.edu/public/education/general/general.
html
Building ResearchEstablishment
10. Crisp, V. H. S. and Littlefair, P. J., Cooper,I. and McKennan G. (1988) Daylighting as a Passive
11. Muneer, T., Abodahab,N., Weir, G., Kubie, J. (2000) Windowsin Buildings thermal, acoustic
-
12. Shao,L. (1988) Mirror lightpipes: Daylighting performancein real buildings Lighting Research
13. Yohannes,1. (2001) Evaluation of the PetformanceofLight pipes Usedin Offices, PhD Thesis,
14. Oakley, G., Riffat, S. B. and Shao,L. (1999) Daylight performanceof lightpipes Proceedingsof the
159-174
15. Zhang, X. and Muneer, T. (2002) A Design Guide for PerformanceAssessmentof Solar Light
13
World Energy Supply Scenario
10 Wood k coal Hydrogen
CL 80
Non-sustainable Increasingly
fuel sustainable
(D 60 -
C I fuel
a)
76 1
0
40 -
0
20-
Hydrocarbons
CL (Gases)
1850 ii0o 1950 2000 2050 2100
14
II-
BQý)
L
ir I
irn
II >Iii
}
Ir
f 118
It
1 fIH
I
4
V
M, oil
Xm
"0 1
I Lý
S. LWOdWI CNV NOuonCIOWd snON30IONI
R. Y.
r
Cie
'-fl. 9m
R-f
J
Od.
II
IL
d. p. - k. -Qý
tle c
111off"
D, ff. se light
M,, m, ed 'Oorn
louv'sa
Direct
sunlight
cannot
strike
occ. va"T
16
Figure 1.5 Underground daylighting at the Space Centre building, University of Minnesota,
Minneapolis, USA
17
Two-sided atrium Itwo open sides)
18
Table 1.1 World fossil fuel reserves to production ratio, years (BP 1999 statistics)
North America 11
S& Central America 65
Europe 18
Africa 100
Asia Pacific 42
World 62
19
2. SOLAR LIGHT PIPE AS AN INNOVATIVE DAYLIGHTING DEVICE
The idea of piping light from a remote sourceto an interior spacefor illumination purposeappeared
about 120 yearsago. Originally designedto distribute central electronic light into buildings, light pipe
is now being adoptedand applied world-widely for both artificial and natural daylighting purposes.
Commercially available light pipe systemsthat employ daylight as lighting sourcehave beenusedin
Australia, America, and Canada.Within Britain alone there are now a number of companiesthat are
profitably trading daylighting light pipe, or called "solar light pipe" productsand with an enviable
growth rate. With the increasinguse of solar light pipes, more attention is being paid to their
production of the prior developmentof light pipe, and at the sametime aims at pushing a further
the historical developmentof light pipe systems,so asto presentthe position and significanceof
presentstudy.
The earliestformal record that can be found on light pipe innovation may be the patentthat was
registeredin the United Statesby William Wheeler from the MassachusettsStatein September188 1.
This registeredpatent initially reffied the idea of light pipe as an apparatusfor lighting dwellings and
at a central or single place, from which the light generatedis transmittedand distributed to and
dispersionof light [I]. It was indicatedby the inventor that before this invention, therehad been
designsof transmitting beams,or said parallel rays, through enclosedpassagesand clear spaceto
produce secondarylights from one luminary sourcefor lighting purposes.Although the inventor
outlined that light pipe is adaptableto any light sources,his designseemednot being able to utilize
20
Nowadays,becausethe long distancesover which light pipe can operate,it hasbeenregardedas
"perhapsthe most technologically exciting innovative daylighting systems"[2]. Light pipe's advantage
of being able to bring light into core areaswithin buildings hasbeennoticed long time ago. Figure 2.1
collector" placed on top of the roof and beamedthrough a vertical "light-conveyer" to the cellar, where
a light distributor is placed for lighting the room. Another near exampleof the samesort is the light
pipe systemin the University of Minnesota, as shown in Fig. 1.5. Sunlight is collected by a heliostat on
the roof and beamedthrough lensesto a working place 110 ft below the ground.Above two light pipe
optical devices,namely mirrors and lensesto collect and transmit sunlight. The common featureof
Heliostatslight pipe systemsis the use of device that can track the sun at times. Due to the costly
expenseand complex control required by the sun-trackdevice, this sort of light pipe systemhas not
beenwidely used.
Having realized the drawbackof "Heliostats light pipe" system,during the last five or six decades,
focusedon improving the transmittanceof the "tube" part of light pipe systems,to avoid the
dependenceupon sun-trackingdevices,mirrors and lenses.The most basic form of light pipe tube is
simply an empty shaft with ordinary mirrors on the walls, along which a collimated beam of light can
travel [4]. Reflective metal tubeswas proposedbut any off-axis light has to undergomultiple
reflections, with the result that only closely collimated beamsdo not becomeattenuatedafter a few
metreswithin the tube [2]. Fibre optic bundlescan have very good transmissioncharacteristics[5], but
hollow acrylic prismatic light tube that has a transmittanceof between0.95 and 0.97. At the sametime,
different approacheshavebeentaken to improve the light transmittanceof light tube. One exampleis
the application of metallic coating and finishing techniqueto the inner surfaceof light tube. Another
notable progressin this respectis the useof PET (Polyethyleneterephthalate)film. The appearanceof
21
In 1990s, innovations of light pipes by different inventors emerged in endlessly. Sutton patented a
"tubular skylight" in 1994 in the USA [9], as shown in Fig. 2.2. According to Sutton, the system
comprises a tubular body, a first transparent cover and a second transparent cover. The material of the
tubular body is either metal, fibre or plastics, and has a finish which is a highly reflective polish or
coating, as found on "l 150 alloy aluminium", electroplating, anodising or metalized plastic film. As a
device to enhance the performance of the said system, a reflector is located within light-pen-neable
chamber and extending above the roof, to collect more sunlight incident into the system.
In 1996however, Bixby pointed out that aboveSutton's invention "requires the use of a reflector
positioned along the path of the sun, the use of an aboveroof reflector blocks a significant portion of
the sunlight which would otherwiseenter the systemand illuminate the building if the reflector was not
present".Bixby registeredhis invention of tubular skylight for natural light illumination of residential
and commercialbuildings in 1996 [10]. Figure 2.3 showsthe Bixby tubular skylight system.Bixby's
systemcomprisesa highly reflective tubular body positioned in the spacebetweena building's roof and
The developmentof solar light pipe hasbeenan on-going process.The latest developmenton
improving the collimation of internal daylight deliveredby light pipes hasbeenreportedby O'Neil
[II]. Comparedto previous arts of light pipe system,O'Neil's tubular skylight useslight tube of cone
near the radiant energy-collectingaperture.By this new design,the collecteddaylight can be collimated
and delivered to the desiredareaof the room directly beneaththe luminaire. According to O'Neil, "as
advantagesover the prior art, including better solar energycollection, higher throughput optical
22
precise positional control of the interior illumination". However, O'Neil did not give sufficient and
detailed information on the type and material of the luminaire being used in his design. As an integral
part of light pipe system, the design and property of the energy-delivering aperture, or called luminaire,
or diffuser by some literature, has been found having important effect on the daylighting performance
of the system [12,13]. Therefore, in practical applications, the design of light pipe systems needs to
An English company, Monodraught, have been manufacturing and trading light pipe systems
successfully within Europe. By the year of 2000, more than 3000 Monodraught light pipe systems have
been installed throughout the UK in comparison with the figure of about 50 for the year 1997.
Monodraught light pipe systems adopt diffusers of various property and design according to different
needs in real practices. A typical Monodraught light pipe system is shown in Fig. 2.5. Daylight is
gathered via a polycarbonate dome at roof level and then transmitted downwards to interior spaces
within buildings. The light-reflecting tube is adaptable to incorporate any bends around building
structural components. The internal surface of the light tube is laminated with Reflectalite 600, a
silverised PET film [14]. The reflective silver layer is sandwiched between the PET film and the
aluminiurn base, which helps in achieving a reasonable illuminance indoors when daylight is
There are various kinds of diffusers that can be employedin Monodraughtlight pipe system,including
dome opal, dome clear,recessedopal and recessedclear diffusers. The recesseddiffusers are more
effective in keepingout dust and preventingheat loss.Opal diffusers are of better diffusive property,
and henceenablean even spreadof daylight within the interior, while clear diffusers possessa better
uniform daylighting is required,the former kind of diffuser hasbeenwidely used.For application like
open spacein deep-planbuildings and corridors where the brightnessbecomespriority, the latter kind
of diffuser is more suitable.The developmentof light pipe diffuser is an on-going process.News from
Monodraughtreports that a new type of diffuser of diamond shapeis being developedand will be
23
As a conclusion, the development of solar light pipe system has been continuing ever since 100 years
ago. Although historically there have been light pipe system of various designs, they generally have
three main components, namely the daylight collecting device, daylight transmitting device and
daylight emitting device. Accordingly, the development of light pipe art has been focusing on these
three components. It has been realized that the improvements on each part of light pipe system will
upgrade the total daylighting performance of the system. However, to date no general method has been
made available to evaluate and compare the performance of various light pipe systems. One major aim
of present study is to establish a generalized standard by which light pipe systems can be compared
against each other, or even against other daylighting devices like windows. With the aid of such a
proposed generalized standard, it is hoped that the development of light pipe family can be more
The term of light pipe hasbeenlong usedto namethe family of non-imagedevicesthat can transmit
light from either artificial or natural light sourceto building interiors for lighting purpose.Becausethe
developmentand application of light pipes to modem buildings havebeennew and fast, so far no
describethe contentof the family. It is possiblehowever, as a first step,to divide the huge light pipe
family into threegroups.The threegroupsare first, light pipes using artificial lighting as light source;
second,light pipes using externaldaylight as light sourceand the third, light pipes using both artificial
light and externaldaylight as light sources.The light pipe group that usesexternal daylight as light
sourceis also widely called solar light pipe since the ultimate sourceof daylight is the sun. As
Within the group of solar light pipe, further classificationcan be defined. The CIE hasset up a
committee TC 3-38 to establishan international standardas a framework for guiding and regulating the
developmentof solar light pipe systems.One major issueon the standardizationidentified by the TC 3-
38 is what constitutesa solar light pipe system.The answerto this questionhas beengiven as,there are
three major types defined by their collecting methods- active zenithal (e.g. Heliostat systems),passive
24
zenithal (e.g. commercially available Solatube,Sunscope,Sunpipeetc systems)and horizontal (e.g.
anidolic ceilings). Among abovethree types, the passivezenithal tubular light pipe has becomethe
most widely usedsolar light pipe system.The family map of light pipe systemis now shown in
Fig. 2.6. Presentresearchfocuseson the daylighting performanceof passivesolar zenithal tubular light
pipe systems.In the following body of this thesis,if not specified,the name of "light pipe" refers to
A passivesolar zenithal tubular light pipe systemcan be defined as a tubular vertical light guidance
purpose.A schemeof two typical passivesolar zenithal tubular light pipe systemsis shown in Fig. 2.7.
Typical passivesolar zenithal tubular light pipe systemhas three components,namely fixed daylight
collection device, tubular light guide that employsinternal reflection mechanismfor daylight
Passivesolar zenithal tubular light pipe is the most commercially available light pipe systemin the UK
market place. Light pipes provided by different tradersor manufacturersare to various extent slightly
different from eachother. For example,a light pipe companymay have their own-patentedtechniquein
fixing the light tube on or within a roof structure,or in preventing leakageof rain. However, in terms of
sealing components.The threemain componentsare daylight collector, light pipe tube and diffuser.
Daylight collector is fitted at the top end of the light tube,usually on the roof, which actsas a semi-lens
to collect daylight and as a cap to prevent the ingressof water and dust. At the bottom end of the pipe
fitted the diffuser, usually to the ceiling to allow the distribution of the daylight into the interior room
25
2.3.1 The daylight collector
Daylight collector is a clear dome that is mountedon the outsideof the roof 'Me dome is manufactured
from cleanpolycarbonatethat removesundesirableultra violet (UV) light and sealsthe light pipe
againstingressof dust and rain. The domeusually meet the requirementsof fire resistanceand due to
The pipe is constructedfrom a metal tube, typically of alloy aluminiurn material. The inner surfaceof
Internal Reflective) prismatic optic film. Figure 2.8, for instanceshowsa Monodraught(UK) light tube
that usesReflectalite 600, a silverised PET film as its internal reflective coat. The reflective silver layer
The diffuser takesthe form of a white polycarbonatedomemountedon the ceiling inside the room to
be illuminated. 'Me material of diffuser vary in its property of transparency,so as to meet different
needsfor light distribution within the room. For exampleopal diffusers, which are madeof semi-
The shapeof a diffuser can be flat, convex, and concave.Monodraught(UK) has recently developeda
new range of diffusers that are manufacturedfrom clear polycarbonatewith a crystal effect finish. The
developmentof a new type of diffuser of diamond shapeis also underway.Figure 2.9 showsthe
appearancesof different types of difftiser, and Fig. 2.10 demonstratesthe different daylighting effects
26
2.3.4 Sealing components
The application of light pipe systemsrequiresthe installation of sealing componentsto the purposeof
achieving high heat resistanceand preventing solar gain. The sealof the light tube is also important to
keep dust, noise, insects,rain and snow out of the building interiors. A brushednylon gasketat the top
of the pipe preventscondensation,dust and rain from entering the systembut still allows the airs to
expandand contract when subjectto solar gain. The closedcell gasketat ceiling level sealsthe daylight
collector and all the vertical joints are sealedwith silicon and alurninium tape.
An explodedview of a typical light pipe system,by Monodraught(UK) is shown in Fig. 2.11, all the
Passivesolar tubular light pipes are designedto collect light from both the sky and the sun. The two
multiple reflection of sunlight and skylight through the reflecting tube. Daylight then reachesthe inner
surfaceof the light pipe diffuser wherein a refraction followed by a light-scatteringtakesplace before it
transmissionand distribution within buildings. Ile complexity of light pipe's working mechanism
The first aspectis the complicatedoptical processthat takesplace within light tubesand diffusers.
Initially daylight collectedby dome enterslight pipe, then a mixed multi-reflection of sunlight and sky
diffuse illuminances occurswhen daylight is transmittedthrough and diffused within the light tube.
27
After that, a refraction phenomenonhappensin light pipe diffuser when diffused sunlight and sky
diffuse illuminances are further diffused and finally scatteredinto the interior space.
The secondaspectis the complexity of externaldaylight as the input to the light pipe system.
Throughout the year, the quantity and proportion of sunlight and sky diffuse radiation are not constant.
When sunlight is available, the changeof sun's position causesthe variation of the incident angle at
which the sunrayenterslight tube. This implicatesthat the pattern sunlight travel through light tube is
continuously changingwhen sunlight is available(Fig. 2.13). When the sky is overcastor when clouds
block the sun, skylight becomesthe major externaldaylight source.However, it is well known that the
luminancedistribution of the sky is not unifon-n.The sky vault can be divided into small patchesas
shown in Fig. 2.14. Each sky patch has its own position and brightness;so the transmissionof the sky
position of sun, the clarity of the sky, the position of random clouds and so on. Therefore,the process
of sky-light entering light tube and travelling within the tube is a highly dynamic process.Figure 2.14
showsthe sky illuminance emitted from one sky patch entersand travels within the light tube.
The third aspectis the designof light pipes. Light pipes vary in their geometric configurations,
including the length, the width and the numberof bendsincorporatedin a light pipe system.Light tubes
applying different internal coating have different internal reflectance.For a given externalenvironment
and at a given point of time, light pipes of different configurationsand having different internal coating
position, the crossareaof a light pipe, affectsthe light pipe's external illuminance admittance.Since the
daylight illuminance is transmittedby meansof internal reflection within the light tube, and the
number of reflections required for a ray of light to descendthe light tube and the tube's reflectance.
The higher the internal reflectanceof the light tube, the higher is the systern'sdaylight transmittance.
28
The lessthe number of reflection required to descentthe entire light tube, the better would be the
system'sperformance.
2.5 SUMMARY
The idea of piping light from a central artificial light sourceto building interiors for lighting purpose
emergedmore than one century ago. However, the developmentof guiding daylight into buildings for
daylighting purposeis new. In the last two decades,the breakthroughin terms of new reflective
available light pipe is the passivesolar zenithal tubular light pipe. Presentstudy focuseson the
identifying all decisive factors that affect the performance,and in quantitatively weighing the
Presentstudy aims to conquerthis barrier so as to push further the developmentof light pipe systems.
29
REFERENCES
4. Whitehead,L. A., Brown, D. N. and Nodwell, R A. (1984) A new device for distributing
5. Fraas,L. M., Pyle, W. R. and Ryason,P. R. (1983) Concentratedand piped sunlight for indoor
6. Smart,M. and Ballinger, J. (1976) Tracking mirror beamsunlighting for interior spaces,Solar
7. Whitehead,L. A., Nodwell, R. A. and Curzon, L. (1982) New efficient light guide for interior
8. Whitehead,L. A., Scott, J. E., Lee, B. and York, B. (1986) Large-scalecore daylighting by meansof
Cable Associates,416-419
10. Joseph,A. and Bixby (1996) U.S. PatentNo. 5,546,712,United StatesPatentand Trademark
htlp://www. uspto.gov
12. Edmonds,1.R., Moore G. I., Smith G. B. and Swift P. D. (1995) Daylighting enhancementwith
13. Carter,D. J. (2002) The measuredand predictedperformanceof passivesolar light pipe systems
30
No. 6.68,404. Patented.Feb. 19,1901.
0. B. H, HANNEBORG.
APPARATUS FOR TRANSMITTING SUNLIGHT TO BASEMENTSDR.OTHERSTORIES.
(No Model.
)
I
_____________________________________
-.
J17WtJZtäP.
,I
31
U. S. Patent Mar. 31,1992 Sheet 10 of 13 5,099,622
F""
Af
/* Z7
co
e /Z-5
A9
FIC 18
32
U. S. Patent Aug. 20,1996 Sheet I of 3 5,546,712
21
FIG. 1 20
12
18
42
11
26 25 14
19 71
42
13
41 -ý10
15
19
50 ý5
46
26
47
43
100
51
45
48 49
35 3 44
28
28 --- 34
30 32
33
U. S. Patent Apr. 2,2002 Sheet 2 of 6 US 6,363,667B2
Fig. 2A Fig. 2B
Prior Art Improved
10 10
12 11 12
g
R-T
15
16
, J-M.
- 'V,
800 :
-- 0. ) --\ 0
w
14
13
Figure 2.4 Comparison of O'Neil's cone shaped light tube against conventional column light tube
34
Figure 2.5Typical Monodraught light pipes
Light Pipe
Artificial light Daylight source light pipe Light pipe using both
source light pipe daylight and artificial
Figure 2.7 Schematic diagram of passive solar zenithal tubular light pipe
36
--- --,- . -Icclaille ()0()IE-1, tilm
F '1
38
r, *4*N)
%
MPOM,
Figure 2.10 Demonstration of the different daylighting effects by a clear diffuser and an opal
diffuser
39
Components For Flat Roof Applicafton
Brushed tiylm
Brushed trap
cmdensation trap
tiylm cmdensation
and breathable gaskeL
and breadiable gaskcL
f
Hig, impact
Hig,iiiri ipactABS(;cA[ar widi
ABSc(Alar fixingIlugs.
with fixing ugs.
Mite compressible
ceilingsf-al.
Wl screwsandkings pruvided).
- state tmf -
WhIte compressiblecedRigwAl,
Figure 2.11 An exploded view of a typical light pipe system by Monodraught (UK)
40
Total extemal illuminance = sky diffuse + sun illuminance
--A
Figure 2.12 A general process of daylight collection, transmission and distribution through light
pipes
Sun's position I Sun's position 2
0 0
\\
Liý
Lighttube
Di ýr
A sky patc
Figure 2.14 Sky vault divided into small patches with respective luminance
43
3. PREVIOUS WORK
The literature review focused on the following topics and region of work that directly relevant to the
presentstudy:
The daylighting performanceof passivesolar tubular light pipes as natural daylighting device has been
transmittanceof light pipe's main componentsincluding solar collector, tube and diffuser and
Experimental studies carried out by Shao [I] and Yohannes [2], reported the performance of light pipes
of various designs under changing weather conditions: sunny, overcast, heavily overcast, and overcast
with rain. Abundant illuminance of up to 450lux was reported in test rooms, and intemal/extemal
illuminance ratio variance ftom 0.1% to 1% was reported. Oakley's study [3] on 330-mm diameter light
pipes of differing lengths showed that the illuminance could be as high as 15381ux. An average
work have focusedon the estimationof the light pipe tube'stransmittanceunder laboratory conditions.
Zastrow et al did theoreticalwork on the transmissionof mirrored light pipe tube [4]. Swift et al [5]
gave an integral equationinvolving light pipe tube's parameters,reflectivity, aspectratio and the angle
of incidenceof the incident radiation. Edmondset al [6] expressedthe transmissionof light pipe tube as
a function of light pipe tube parameters,reflectivity and aspectratio and solar altitude.
44
Studiesundertakenat the University of Technology, Sydney [6] investigatedthe enhanceddaylighting
illumination that can be achievedwith a vertical light pipe decreasesrapidly as the elevation of the
incident light decreases.Ile outputsof light pipes dependstrongly on sun's position and diffuser
design.It was concludedthat the daylighting of small rooms via light pipe may be enhancedby
deflecting low-elevation light more directly through the light pipe using laser-cutlight deflecting
panels.
Work undertakenby Carter [7] put forward a designmethodintendedfor use either freestandingor in
entering the systemis a function of unknown sky conditions,Carter's basedhis work [7] on the
assumptionof a "design sky", which restrictedthe applicability of this method in real practises.
'ne daylighting performanceof passivesolar tubular light pipes hasbeenmonitored in both real and
laboratorial environments.In the UK, literatureson the researchin this respecthave mainly been
sitesare the Pearl Centrein Peterborough,the Gurney Surgeryin Norwich, the RivermeadCourt in
Marlow and the Guildford College in London. G. Oakley et al [3] reportedthe performanceof six light
pipes in three different applications,namely a workshop, a residentiallanding and a small office. The
locations of the applicationsare the BeaconEnergy Ltd, the West BeaconFarm and the BeaconEnergy
45
3.1.1 Applications of light pipes in residential and office buildings
Measurementson light pipe's daylighting performancein real buildings by Shao[I], Yohannes[2] and
Oakley'sstudy [3] have presentedgeneraland broad pictures of applicationsof the device.Positive and
qualitative conclusionshave beendrawn on the overall efficiency of the device and the benefitsof
pipes plays an important role in the application of the device, Shaoet al concludedthat light pipes "can
be effective devicesfor introducing natural daylight into buildings, provided they are designedwith
care". Oakley et al concludedthat light pipes "are proficient devicesfor introducing daylight into
buildings", and "the most effective light pipes are straight, short oneswith low aspectratios and
consequentlylarger diameterlight pipes would probably be more effective". Theseimplicated that the
light pipe itself was a complex systemand required extensiveresearchto provide detailedand
It was found that the performanceof light pipe systemwas not only dependenton the geometrical
have effect on the performanceof light pipes. Shaoet al and Oakley et al measuredthe daylighting
performancesof light pipes under different weatherconditions and usedthe ratio of internal total
illuminance to external total illuminance as an indicator of the performance.It was found that the ratio
seemedto vary with different weatherconditions and solar altitudes.Shaoet al found that when the sky
explainedas "probably due to most of light ftom an overcastsky originating from aroundthe zenith
and undergoingfewer reflections/absorptionsas it travels down the light pipe". However,both weather
condition and solar altitude are not independent.Thereforeno certain conclusionson the effect of
weathercondition, and the effect of solar altitude on the daylighting performancecan be given based
conditions and solar altitude, and the performanceof light pipes, vast databasethat can provide
necessaryinformation on both light pipe itself and its ambient external environmentis needed.
46
Furthermore,by examining the study by Shaoet al and Oakley et al, it was found that the internal
illuminance distribution plays an indispensablerole in the evaluationof the performanceof the device.
Shaofound "the conceptof daylight factor is not appropriatehere asthe sky was clear and dosenot
conventionaldaylight factor, the ratio betweeninternal illuminance and externalilluminance was used
to indicatethe performanceof light pipes. However, it must be pointed out the ratio is dependanton the
points of interest.For a given light pipe systemunder a given externaland internal environmental
conditions, the ratio for different points within the room, which havevarious vertical and horizontal
distancesfrom the centreof light pipe diffusers have different values for the ratio. Therefore,as an
indicator of light pipe daylighting performance,the ratio shall be defined as a dimensionalvariety that
Useful information on light pipe designhasbeengiven by Shaoet al and Oakley et al. Qualitative
designlight pipe systemsin real applicationsusing only basic principals and qualitative relationships
might not be successful.Designersneedto know the output of a certain light pipe under certain
externaland internal conditions with certain confidence.Most of the measurementson light pipes in
real buildings carried out by Shaoet al and Oakley et al monitored the total output of a group of light
pipes, where eachindividual light pipe had its own configuration and waspositioned in different
places.It shall also be bom.in mind that the output of light pipes measuredin real applicationscould be
affectedby the shadingof interior layout, for examplesthe shadeof furniture and reflection from walls.
performanceof light pipes. In considerationof the presentwork, it was thought that the mathematical
47
3.1.2 Large-scale core daylighting by light pipe
Canada.The illuminated areaconsistedof an 186m2core region of the top floor of the building. The
systemincluded five main components,namely, sun tracking device, focusing mirrors, roof apertures
and prism light guide input elbows,prism light guidesand electrical lights. According to the schemeof
the application given by the authors,therewere eight identical sun-trackingand light distribution
1.22metersin diametercould illuminate 23m2 of office spacewell in excessof 1280lux.It was also
beam solar radiation from a solar collection systemto a utilization system.According to McCluney, a
systemof suchcan be for the daylighting of the core interior spacesof buildings, spacesthat are far
removed from outsidewalls or the roof and are thereforenot amenableto conventionaldaylighting
with sidelight or toplights. McCluney studiedthe effect of absorptionby water on the color of light and
the colorimetry of water-filtered light. McCluney reportedthat the distancewhich daylight can be
appearedto be 6-8m. The distancefor a significant drop in the color-renderingindex of the daylight
transmittedby the systemwas reportedas about I Om.It was found that the visible transmittanceof 5
and 10m long pipes was about 28% and 19%. It was concludedthat water-filled light pipes could be
panels(LCP). Edmondspointed out "although light pipes are effective sourcesor natural illumination
for deepareaswithin buildings, the daylighting performanceof the device decreasesrapidly as the
elevationof the incident light decreases".It was found that the combination of laser-cutlight deflecting
48
panelswith light pipes enhancedthe illumination performancefor all anglesof elevation below about
producedby laminating laser cut acrylic sheetbetweensheetsof glass [101.It was found that the large
enhancementwas obtainedin winter months, when sunlight is incident at low elevation all day. For a
light pipe 0.3m in diameterand 7.2m in length but without LCP, the transmittanceof it falls below 0.1
for sun elevationbelow about45'. While as a comparison,when such a light pipe is coupledwith a
LCP device, the transmittanceabove0.1 extendsto sun elevation as low as 10". It was therefore
concludedthat the enhancementat low elevationsextendsthe time during which light pipes were
useful [6].
Edmondset at [ 11] also gave the theory of using extractorsand emitters in light pipe systemwithin
large commercialbuildings. 'I'lie author discussedthe application of hollow light pipe for the
distribution of sunlight from the roof or fagadeinto the deepinterior of a building. It was found this
application was suited to climateswith a high direct sun component.However, to use sunlight in
buildings more effectively, a device at the faqadeto direct sunlight into a light pipe and devicesto
extract light from the light pipe and distribute the light into interior shall be included in the system.
Devices for the extraction of light are given the nameof "extractors" and devicesfor light distribution
and emitters for light distribution from hollow light pipes. It was concludedthat it was feasibleto
artificial light.
3.1.4 Light pipe system using both natural and artificial light
Aizenberg et al [ 12] reported the application of slit light pipes in the Chkalovskaya Moscow subway
station. The central hall of the station was illuminated by a rectilinear (60m long) structure suspended
under the oval dome of the main hall. The structure consisted of cylindrical multi-slot light pipes
49
alternating with 12 cubes.According to the above-mentionedauthor, the use of hollow light pipes to
illuminate the subway stationnot only permitted new architecturalsolutions,but also yielded good
economicindices of the system.The number of lighting points in the all was reducedby a factor of 30
and the extent of the electrical grid was reducedby a factor of three.Aizenberg et al statedthat an
important advantageof lighting systemsbasedon light pipe "might be further reduction in maintenance
costsas the light sourcesmay be easily accessedto maintenancepersonnelwithout the needfor tall
laddersor cherry pickers". The operationalexperienceof over 12 years confirmed the high reliability of
was the first station where the problem of light-system operationhad beencompletely solved, since
"accessto the input units for lamp replacementand modification of the electrical systempresentsnot
time". According to Aizenberg et al [12] the application of similar artificial lighting systemthat used
light guide was also found in Moscow ring high way. This showsthe wide applicability of artificial
Aizenberg et al [13] proposedand studieda solar and artificial light pipe systemcalled "Heliobus" used
recreationhalls at the centreof eachfloor. The Heliobus systemwas built to light theserecreation
element,a unit containing electric light sources,a vertical hollow light pipe and a diffuser. According
to the above-mentionedauthors,the working mechanismof the systemis: solar light rays capturedby
the light collector are directedthrough the intermediatedevice (by its reflecting surfaces)into the light
pipe and then diffused into areasto be lit. In caseof lack of daylight, a luminous flux of electric light
Testsconducedby Aizenberg showedthat the systemprovided good daylighting even under a cloudy
sky when the sun was absent.It was concludedby afore mentionedauthorsthat the systemtotally met
50
3.1.5 Integration of light pipe daylighting and natural ventilation systems
Shaoand Riffat [14] investigatedthe techniqueof integrating light pipe natural daylighting and natural
ventilation using passivestack system.The systemhas a bi-layer structure.In the middle of the
structure is the light pipe, lined with a highly reflective material. The light pipe is surroundedby an
annular spacethat actsas a stack for the exhaustair. The systemcan be further combinedwith a heat
performanceenhancementof light pipe due to LCP (Laser-cutPanel) was also examined.It was found
light transmittedthrough the light pipe was higher on clear/sunnydays than on overcastdays,but the
uniformity of the illuminance in the test chamberwas greateron overcast/cloudydays. Solar infrared
transmissionwas found to be 1.5%on a cloudy day comparedwith 1.4%on a sunny day. Light pipe
transmittedmarginally less infrared radiation than visible light. It was shown that by fitting a LCP to a
light pipe, higher level of daylight transmittancecould be achievedfor certainperiods of the day,
dependingon the orientation of the LCP. It was found that an eight air changesper hour were achieved
with the natural stack effect. The authorsconcludedthat the most suitablebuilding types for the
Oakely [ 15] also studiedthe integratedsystemof light pipe and ventilation stack. The systemhas two
channels.The central channelis a light pipe that guidesdaylight into occupiedspace;while the outer
performanceof two light pipes with diameterof 2.15m and 2.20m in length in outdoor environmental
ratesand natural stack ventilation. Resultsgiven by the author suggesteda good correlationbetween
solar altitude, internal illuminance valuesand the external illuminance values.It was shown that by
fitting a LCP panel to a light pipe, much higher levels of daylight could be achieved.It was found that
the typical air changerate through the passivestack was about eight air changesper hour in winter, and
the temperatureinside the test chamberwas controlled stableat approximately20'C. The authors
concludedthat light pipes were proficient devicesfor introducing daylight into buildings and reducing
costs.
51
Smith et al [ 16] reportedthe evaluationof dichroic material for enhancinglight pipe daylighting and
natural ventilation in an integratedsystem.It was statedby the authorsthat, "the integration of natural
techniquemore attractive to users". It was found that by constructinglight pipe using dichroic
materials,the infrared part of the solar radiation was allowed to be transmittedto the ventilation stack,
while the visible light was guided by the light pipe into interior space.The heatgain to the interior can.
thereforebe reducedand the thermal stack effect strengthened.The transmittanceof a dichroic light
pipe was found to be similar to that of a light pipe with a 95% specularreflectance.The infrared
natural ventilation flow by up to 14%. It was concludedby the authorsthat the dichroic material
removedapproximatelyhalf of the solar heat in daylight, and thus provided natural daylight without
overheating.
Daylighting possessesgood potential for applications in area such as tropics where the sky is luminous.
However, because the depth of penetration of daylight through windows into deep plan buildings is
limited, the full utilization of daylight is restricted. It was realized that the limitation of windows could
be complemented by purpose designed light pipe systems. Surapong et al [17) developed a technique
for the use of sunlight through light pipe system in multi-storey buildings. Measurements on the
internal illuminance and temperature were undertaken and the results were compared against calculated
values. Surapong et al used elementary ray-tracing method in modelling internal daylight illuminance
and employed analogous model for calculating heat transfer. According to the computed and measured
internal illuminance data given by the authors, Mean Bias Error (MBE), Root Mean Square Error
(RMSE) and Percentage Averaged Deviation (PAD) were derived. It was noted that the averaged
measured internal illuminance value was as high as 1400lux. The MBE was found to be 308lux, which
was about 22% of the averaged measured internal illuminance. Corresponding figures for RMSE were
found as 808lux and 57%. The PAD was found to be 35%. Based on the comparison between
measured
and computed data, the afore-mentioned authors also pointed out that there might be a higher level of
uniform reflection than assumed. The authors concluded that the analysis method developed by them
52
A casestudy on the use of light pipes for deepplan office buildings was carried out in Malaysia by
Hansenet al [ 18]. The casestudy examineddevelopmentwork on the application of light pipe system
for a high rise building in Kuala Lumpur, Malaysia. Light pipes were used in the high-rise building. For
eachfloor, four light pipes coupledwith LCP (Laser-cutPanel)were used to improve the daylighting
performance.The method of scalemodelling was used to simulate the daylighting performanceof light
pipe. In predicting the daylighting performanceof light pipe, elementarytheory was used.The
calculatedresults were comparedwith measuredvalues for various solar elevations.It was found that
light pipes coupled with LCP had a good performancefor deepplan buildings. The authorsconcluded
that the use of light pipes "will increasethe passivezone in the building and hencereduceenergy
consumptionby the use of daylight". It was suggested,however, that further study neededto be done in
order to obtain a better distribution of the out coming light from the light pipe system.
Researchersin Nottingham University [ 19] proposeda novel "light rods" systemthat can be usedto
transportlight ftom the exterior of a building into its interior spaces.Theserods were designed
sufficiently small so that it canbe fitted into most existing buildings without structural modification.
systemrelies on the principal of total internal reflection and makesuse of the transmittanceproperties
light pipe in terms of daylight transmittance.It was found that when external illuminance was greater
than 35000lux light pipe (aspectratio = 4) performedbetter than light rod (aspectratio = 24), while
when external illuminance was lower than 35000lux, the latter device performedbetter than the former
one. It was concludedthat light rod "transmit at a similar efficiency to a standardlight pipe that has an
aspectratio 6 times smaller", and "the small diameterof the device gives it the potential to provide
The working mechanism of passive solar light pipes is most widely described and understood as
"daylight is gather by the light pipe dome, and then transmitted through the reflective light pipe tube by
53
multi-reflection, followed by being distributed by light pipe diffuser into internal spacewithin
buildings". Above descriptionalthough rough and simple, is correct, giving a generalpicture of how
daylight is transmittedthrough light pipes. However, since daylight has two important components,
namely the sunlight and sky-light which are of great difference in their property, with the quantity and
The understandingof the complex processthat sunlight and sky-light travels through light pipes forms
parallel light (or sunlight) within light pipe tubes [1,4,5,6]. Sky diffuse light is the secondmost
important componentof daylight, but the contribution of this componentto the daylighting
The quantity and proportion of the two componentsof daylight vary with the changingof sky
conditions. When the sky is clear, horizontal diffuse illuminance can be ashigh as 1/3 or more of
horizontal global illuminance; while when the sky is overcasthorizontal diffuse illuminance can be
equalto the global illumiance; and when the sky is part-overcast,the proportion of diffuse illumiarince
to global illuminance varies dymatically but can reachas high as 50%. The transmissionof sky difftise
illuminance can have important influence on the overall efficiency of light pipe systems.
Study on the transmissionof sky diffuse light within light pipes was carried out by BBRI (Belgian
Building ResearchInstitute) [20]. Measurementon the diffuse illuminance decreasein a 330 mm light
pipe tube along the length of it was undertakenunder overcastsky conditions. It was reportedthat
diffuse illuminance transmittedby a light pipe had two components,namely the direct view part and
the reflected part. It was shown that the direct view of sky decreasedvery quickly from 100%to I%
after I meter, while the reflectedpart decreased29% per meter.Theseimplied that the physical
processesof the transmissionsof the two daylight componentswithin light pipes were different.
54
3.2.2 Critical appraisal
On light pipe system'stransmittanceof daylight, most efforts have beenput on the theoretical
calculation of light pipe tube's transmittanceof parallel artificial or sunlight [1,4,5,6). Zastrow et al
[4] related the transmittanceof a mirror light pipe to its surfacereflectance,the angle betweenthe
incident light and the light pipe's axis, the length of the light pipe tube, and the diameterof the light
transmittanceof light pipe tube, and the configurationsand the incident angle of parallel light.
However, the application of abovemethodsto the designof light pipes in real practisemay not be
abovemethodsare not applicable,since the input into the light pipe is only diffuse illuminance and
BBRI's work [20] on the evaluationof the loss of diffuse illuminance along light pipe tube revealed
that the transmissionof this daylight componentis complex and different from that of the sunlight
is that the sky illuminance distribution is not uniform. Figure 3.1 graphically presentsthe radianceor
luminancedistributions that have beenwidely reportedin literatures [2 1]. Figure 3.2 showsthe
different sky illuminance distribution under different weatherconditions.Thus, the transmittanceof sky
diffuse illuminance may vary under different weatherconditions,which implies that the 29% per
efficiency drop can only be applied to real light pipe designwith limited confidence.It is therefore
realized that, to produce a mathematicalmodel that can describethe transmissionof sky diffuse
illuminance within light pipes, measurementson light pipes of various configurationsunder all weather
Pervious work on the efficiency of light pipes focused on the evaluation of transmittance of the three
main components of light pipes namely the solar collector (or called dome), the light pipe tube and the
55
diffuser. Different researchershave taken different approachestowards the purposeof assessingthe
efficiency of the components.The proceduresthat were taken and the resultsthat were given by these
Light pipe solar collector is the first part of the light pipe system that daylight passes through. The
efficiency of solar collector influences the performance of the whole light pipe system. The most
widely used and commercially available light pipe solar collectors are the clear acrylic dome and the
clear polycarbonate dome. The evaluations of solar collectors have been focusing on this particular
type of dome. Pervious works done on this subject have been reported by BBRI (Belgian Building
BBRI carried out measurements to test the solar energetic properties of solar collectors (by
Monodraught (UK) Ltd) of various materials. The theoretical values for normal light transmittance of
clear acrylic and polycarbonate domes were reported as 84.3% and 81.3% respectively. Experiment
aimed to determinate the hemispherical transmittance of solar collectors (by Monodraught Ltd) was
also undertaken. Two luxmeters, one put under a solar collector, and the other outside of it serving as a
reference meter were employed. Readings from the former luxmeter was compared to the latter one to
obtain the efficiency ratio. The value of the ratio amounted to 80% for solar collectors. It was therefore
found that, "the use of the normal values instead of the hemisphere values does not lead to a different
result".
The theoretical values for normal light transmittanceof light pipe diffusers of various diameterswere
also measuredby BBRI. The testeddiffusers were Monodraughtopal dome diffusers of diameters330
mm, 450 mm and 530 mm, and the transmittancesfor the diffusers were reportedas 53%, 50% and
54% respectively.Although Monodraught opal dome diffusers were of convex shape,no experiments
56
Above values obtainedby BBRI were then comparedagainstthe valuesgiven by manufacturersof
quite idealistic and lead to a performancethat is almost twice as good as the real performance".
Carter [7] studiedthe relationship betweenthe aspectratio of light pipe (with solar collector and
diffuser) and the transmittanceof light pipe. In his work, Carter used dataprovided by Love et al [22]
aspectratio for light pipes without solar collector or diffuser under overcastsky conditions. To employ
the databy Love et al, Carterusedthe values of 88% as the hemisphericaltransmittanceof solar
collector, and 60% for diffusers to correct the valuesgiven by Love et al.
The earliestwork on the transmissionof miff ored light pipe tube seemsto be the theoreticalcalculation
given by Zastrow et al [4]. Zastrow et al gave a simple formulation to relate the transmittanceof a
mirror light pipe tube to its surfacereflectance(p), the anglebetweenthe incident light and the tube's
axis (0), the length of the tube (L), and the effective diameterof the tube (d,ff). It is
T=p / deff
L* tanO
(3.1)
where T is the transmittance,i. e. the ratio of the amountof light transmittedby light pipe tube to the
incident light.
Swift et al [5] gave an equationto describethe transmissionof collimating light within mirrored light
f=o
1,7r) [S2 /o )1/2 ]Rint(p tanO/s) fl
T= (4 - S2 X
-(1 - R)[p tan Ols - int(p tan Ols)]Ids (3.2)
where T is the transmittance of the light pipe tube, i. e. the ratio of the amount of transmitted light to
that of incident light, R is the reflectance of the interior surface of the light pipe tube, p is the aspect
57
ratio of the light pipe tube, 0 is the anglebetweenthe incident light and the light pipe tube axis, int is
rangeof aspectratio were measured.The light sourcewas a HeNe laser whosebeamwas expandedto
be collimated with a diameterof 50 mm. The transmissionof light pipe tubesof the aspectratio of 2,4,
6,8 and 10 was measuredas a function of the angle of incidenceof the collimated beam.The anglesof
The light transmissionperformanceof light pipe tube was also given by Edmondset al [6],
where P is the power transmissionof light through light pipe tube, 1()is the extraterrestrialintensity
(1353 W/M2) T,,. is the transmissionof the atmosphere,0 is the elevation of sunlight, R is the radius
,
of light pipe tube, p is the reflectanceof the tube, L is the length of the tube, and i is the angle on which
comparedthe ratio of internal to external illuminance, which were due to two light pipes, one straight
and the other elbowed with threebends.It was reportedthat, the internal to externalratio for the
elbowed light pipe ("light pipe A") averagedat approximately0.3%, and 0.5% for the straight light
was found as 1.78,light pipe E was on average43% better than light pipe A and thereforeA losed 14%
light output per bend. It was pointed out that the figure of 14% contradictedMonodraught's figure of
58
The effect of light pipe tube bendson the efficiency of light pipes was also testedby Love et al [22].
Bends of test anglesof 30,60 and 90 degreeswere tested.Internal illuminances were measuredwith
collected for eachof the tests,and the datawere usedto calculateaveragetransmittanceof the elbow
joint at eachof the test angles.It was reportedthat the transmittancevalues for 30 degreeslight pipe
BBRI [20] carried out extensive measurements on the efficiency of light pipe solar collectors of various
materials and shapes. Measurements were also undertaken by BBRI to compare the transmittances of
solar collectors under hemisphere ambient illuminance and normal incidence illuminance conditions.
Carter [7] referred and used the transmission values for light pipe solar collector and diffusers given by
Love et al [22]. However, these values were not found in Love et al's work. Therefore, it is more likely
that the findings of BBRI [20] are more accurate and can be used with higher confidence than the latter
one.
It was not explainedin BBRI's report [20] why no comparisonbetweenthe transmittancesof diffusers
under real and laboratorial testing conditions was undertaken.This may be due to the highly dynamic
and complicatedoutput of illuminance transmittedby light pipe tube, which makesthe comparison
difficult to implement. It hasbeenaddressedin Section2.4 that the working mechanismof light pipe
light pipe systemdoesnot necessarilylead to an applicableresult. Light pipe diffuser is an integral part
componentsas an entirety.
59
Efficiencies of light pipe tube
Zastrow et al [4] gave a simply equation to describe the transmittance of light pipe tube. However, as
pointed out by Swift et al [5], "the reliance of the transmission on the three parameters p (aspect ratio),
R (reflectance) and 0 (the incidence anlge) does not allow a straightforward statement about the
conditions under which the theory of Zastrow et al is a good approximation". It was further commented
by Swift that in the main however Zastrow et al's model "is valid under the conditions of low p, high R
and low 0", and "for calculations of the transmitted spectrum and colour coordinates it is anticipated
that the theory of Zastrow and Wittwer will give incorrect values".
theoretical calculation are sensitiveto the choice of reflectivity, with changesin R of 0.001 resulting in
noticeabledifference in the calculatedcurve." It was noticed that the reflectivity of the reflective sheet
3M Silverlux was found to be 0.972, which is about 0.005 higher than the fitted value. Although Swift
et al pointed out that due to dust and aging, the reflectanceof the sheetusedin light pipe tubesmight be
expectedto be slightly lower; this apposesthe questionon the accuracyof Eq. 3.2. Equation 3.2 is not
Equation 3.3 given by Edmondsenabledthe relative comparisonof the transmittedpower by light pipe
tubesunder different sun elevation angle,aspectratio and reflectanceconditions. However, Eq. 3.3 is
112/R,
basedon a key weighting-factor (R2-X2) which was usedwhen averagingover rays equally spaced
acrossthe tube. A main weaknessof Edmondset al's Eq. 3.3 is it is not validated by comparison
aspectratio. However, it was reportedthat the plots were madeby normalising to the power transmitted
when the sun elevation equals90 degrees.This implies that although the plot showsthe effects of sun
60
elevation, surfacereflectanceand aspectratio, it cannotbe used directly in real light pipe design,
Oakley et al comparedthe illuminances due to two light pipes one with and the other without bends,
and basedon a rough calculation,gave 86% as the transmittanceof one light pipe tube bend. However,
the illuminances were not measuredseparatelyfor the two light pipes, and consideringthe changingof
external environmentand the non-uniform interior layout, it is difficult to quantify the reliability of the
Love et al's study is more sophisticatedand provided the evaluationof the efficiency of bendsin more
details. It was realizedby Love et al however, that the geometryof light pipe (SunScope)systemwith
elbow joints was too complicatedto allow for any specific conclusions.Nevertheless,Love et al
pointed out that the transmittanceof a single elbow joints is about 10 percentlower than the
transmittanceof a single straight sectionof 330 mm light pipe (SunScope).It is noticeablethat the
transmittanceof 80% given by the author. It was also concludedthat the transmittanceof a light bend
was found to be affectedby sky conditions, solar altitude and the geometricalconfiguration of the
bend. This implies that the light transmissionwithin light pipe bendscan be complex, and it may be
more practical to assessbends' performanceas one integral part of the whole light pipe system.
The designof light pipes shouldbaseon methodsthat can predict the performanceof the whole light
pipe systemunder all weatherconditions. Designersneedto know the values of illuminance on certain
working plane or at other designatedpoints due to light pipes, Till this project, no mathematicalmethod
that takes accountof the effect of external and internal environmentalfactors and light pipe
61
3.4.1 Carter's design charts for light pipes
efficiency and aspectratio for 25 pipes, variously of 330,450 and 530 imn diameterwas investigated.
Pipe input was calculatedas a function of externalhorizontal illuminance and pipe cross-sectionarea.
Pipe output was either measuredor calculatedusing protometric integrator.A designchart was given
by Carter to show the attenuationof light output with aspectratio under overcastskies for vertical pipe
systemsincluding a clear light capturedome and an opal diffuser. Carter used Love et al's resultsas the
Experimentswere undertakento measurethe luminous intensity in the vertical plane for the quadrant0
- 90 degrees.
Measurements
were undertakenfor two 330 mm diameter light pipes, one 610 mm in
length and the other 1220mm. Readingswere taken (not continuously) through the months from
Septemberto November. It was noted however, that the sky conditions for the testingperiod were
predominatelyovercastor cloudy with externalhorizontal illuminance only exceeding25 000 lux for
basedon the work of Tregenza[24]. This method treatsthe light pipe's apertureas a fixture and
assumesthat light pipes useperfectly diffuse emitter. The methodwas developedto estimatequickly
the number of light pipes neededto light a room at a given level of lux. It was found by the author that
CU dependedon the geometric form of the light pipes (aspectratio, areaof admittance),the material
the room, the room's surfacereflectance.It was concludedthat the number of light pipe could thus be
expressedas a function of CU, designilluminance, the areaof the working plane and light loss factors
(LLF). It was pointed out that the method was by the time still in its developmentstage,and it could be
62
3.4.3 Critical appraisal
Carter's designtool was testedby comparingthe calculatedinternal illuminances due to light pipes
However, it is noticed that the measuredinternal illuminanceswere mostly lessthan 30 lux. The
maximum external illuminance for the validations were found to be 14 000 lux. 'Me maximum internal
to external illuminance ratio was found to be 0.225%, obtained I rn under light pipe diffuser. This
implies that the designtool given by Carter was only validatedunder heavily overcastsky conditions.
In real applications,light pipes deliver more light in clear or part-overcastsky conditions. Shaoet al's
study [11reportedthat in caseswhere light pipes with moderateaspectratios were installed, good
to overcastsky conditions. '11iisis due to the fact that the main designchartsgiven by Carter were
Edmondset al [61 showedthat, under clear sky conditionsthe transmissionof light pipes is affectedby
solar altitude. According to Edmondset al, for a light pipe of an aspectratio of 3, the variation of
presenteddifferent curvesto describethe transmissionas a function of solar altitude for different aspect
ratio.
It hasbeenaddressedin Section2.4 that the transmissionof sky diffuse illuminance and sunlight
for overcastsky conditions when sky diffuse illuminance is the dominant componentof externalglobal
the number of light pipes that are neededto meet a certain lighting requirements.However, to put the
method in to practical use, the relationship betweenCU and various geometricaland climatic factors
63
3.5 SUMMARY
In this chapter, previous works on daylighting perfonnance of light pipes in real buildings, working
mechanismof light pipes, transmissionefficiency of light pipes and the designguidancefor light pipe
Previousworks on daylighting performanceof light pipe showedthat light pipe is a complex system
performanceof light pipe systemnot only dependson the geometricalconfigurationsof the system,but
reveal the relationship betweenvarious internal and external factors, and the performanceof light
pipes, vast databasethat can provide necessaryinformation on both light pipe itself and its ambient
factor, the ratio betweeninternal illuminance and external illuminance shall be usedto measurethe
daylighting performanceof light pipes.As the indicator of light pipe daylighting performance,the ratio
shall be defined as a dimensionalvariety that can describethe internal illuminance distribution due to
the light pipes. Due to the draw backsof measurementsof light pipes performancein real applications,
evaluationmodel.
The efficiency of light pipe tube shouldbe investigatedunder real sky conditions.Theoretical
equationsgiven by Zastrow et al [4], Swift et al [51 and Edmonds[6] do not take accountof diffused
light and due to their respectdrawbacks,cannotbe used in real applications.Light pipe diffuser is an
integral part of light pipe system;thereforeit may be more practical to assessits performancewith
other componentsas an entirety. It was also found that the transmittanceof a light bend was affectedby
sky conditions, solar altitude and the geometricalconfiguration of the bend. This implies that the light
transmissionwithin light pipe bendscan be complex, and it may be more practical to assessbends'
64
The designtool given by Carterwas only validatedunder heavy overcastsky conditions. It is made
conditions. The applicability of Carter's designtool, which is mainly for overcastsky conditions when
automatically extendedto other sky conditions.Design tools should enabledesignersto know the
values of illuminance on certain working plane due to light pipe installation under all weather
conditions, especiallywhen the sky is clear and the systemdelivers more daylight into buildings. As a
65
REFERENCES
1. Shao,L. (1988) Mirror lightpipes: Daylighting performancein real buildings Lighting Researchand
3. Oakley, G., Riffat, S. B. and Shao,L. (1999) Daylight performanceof lightpipes Proceedingsof the
159-174
4. Zastrow, A. and Wittwer, V. (1987) Daylighting with mirror lightpipes and with fluorescentplanar
5. Swift, P. D. and Smith, G. B. (1995) Cylindrical mirror light pipes Solar Energy Material and Solar
8. Whitehead,L. A. (1987) Large scalecore daylighting by meansof a light pipe, Sunworld, 11(l), 14-
15
9. McCluney, R. (1990) Color-renderingof daylight from water-filled light pipes, Solar Energy
Material, 21,191-206
10. Edmonds,I. R. (1992) Performanceof lasercut light deflecting panelsin daylighting applications,
11. Edmonds,1.R., Reppel,J., Jardine,P. (1997) Extractorsand emitters for light distribution from
12. Aizenberg, J. B., Aleshina, N. A., Pyatigorskii, V. M. (1996) Arched hollow light guidesat the
13. Aizenberg, J. B., Buob, W., Signer, R., Korobko, A. A., Pyatigorsky, V. M. (1996) Solar and
artificial lighting of a schoolbuilding with hollow light guide system"Heliobus", Light and
66
---4
14. Shao,L. and Riffat, S. B. (2000) Daylighting using light pipes and its integration with solar heating
15. Oakely, G., Smith, S. J., Shao,L. (2000) Triple Save- the investigationand monitoring of a
16. Smith, S. J., Elmualim, A. A., Riffat, S. B., Shao,L. (1999) Evaluation of dichroic material for
17. Surapong,C., Siriwat, C. and Liu, R. (2000) Daylighting through light pipe in the topics, Solar
18. Hansen,V. G., Edmonds,I., Hyde, R. (2002), Proceedingsof the International ExpertsMeeting on
19 Callow, J. M., Shao,L. (2002), Air-clad optical rod daylighting system,internal report from
20. Loncour, X., Schouwenaars,S., L'heureux, D., Soenen,S., Voordecker,P., Wouters,P. (2000)
ResearchInstitute
21. Muneer, T., Fairooz, F. and Zhang, X. (2001) Sky Luminance and RadianceDistributions -A
Daylighting
22. Love, J. A., Arch, D., Eng, P. and Dratnal P. (1995) Photometriccomparisonof mirror light pipe
Canada
67
/\
VA U
Horizon Horizon
darkening brightening
effect effect
Figure 3.1 Representation of the sky luminance distribution under non-overcast conditions
68
ý2-5*
0"
(a) (b)
c;n-
5. ?2.5*
0"
(c) (d)
Figure 3.2 Vector presentation of Bahrain sky radiance distribution: (a) clear sky kt=0.7, (b)
part-overcast sky kt=0.5, (c) thin overcast sky kt=0.35 and (d) heavy overcast sky kt=0.2
69
4. RELEVANT THEORIES
Passivetubular solar light pipe is a daylighting device that transmitsexternal daylight into buildings.
To evaluateand predict the daylighting performanceof light pipes, an explicit term that can specify its
standardin daylighting design.However, becauselight pipes utilize both sky diffuse and sunlight
radiation, the method of daylight factor is not suitableto measurethe performanceof light pipes. With
the developmentof various innovative daylighting devices,it is necessaryto use new conceptthat
measuresdevices' perforinancein delivering both sky diffuse light and sunlight. Daylight penetration
factor (DPF) is henceintroducedin this study to specify, and as an index to evaluatethe performance
of light pipes.
The introduction of the conceptof DPF forms the basison which the daylighting performanceof light
pipes can be modelled.The proposedDPF modelsare built to predict the daylight delivery efficiency
and the internal distribution by light pipe systemsof various configurationsunder all weather
conditions.The core task of the proposedDPF model is to predict transmissionof daylight (sky difftise
light + sunlight) through light pipe systems.Light pipes use external daylight as input and produce
diffused daylight as output. However, the input and output for light pipes are complex, thereforethe
proposedDPF model should incorporateother modelsso as to accountfor the complexity of input and
As shown in Section3.2, daylighting performanceof light pipes varies with the changingof external
incorporatedinto the DPF model. Thus the input of light pipe systemscan be described.Light pipes use
diffusers to producediffused light and spreadit into buildings. The internal illuminance distribution
resulting from light pipe's output dependson the amountof the daylight delivered by light pipe and the
propertiesof the diffusers. The proposedDPF model aims to predict the daylight level at any points of
70
4.1 DAYLIGHT PENETRATION FACTOR
(DF). DF is defined as the ratio of the internal illuminance to the external diffuse illuminance available
components,namely the sky component,the externally reflected componentand the internally reflected
components.The sky componentis the ratio of illuminance at any given point that is receivedfrom a
illuminance receivedafter reflections from externaland internal surfacesto the horizontal illuminance
BecauseDF cannotspecify the transmissionof sunlight into buildings, light pipe daylight penetration
factor (DPF) is introduced.DPF is defined as the ratio of the internal illuminance (due to light pipe
component.When light pipe's efficiencies in transmitting sky diffuse light and sunlight are identical,
DPF is a product of externalglobal horizontal illuminance and the transmittanceof light pipe (Eq. 4.1).
where Ej,, nýjis the internal illuminance at a given point and Eg is the total external illuminance. In
The definitions of DPFdiff.. and DPFbe. are shown in Eqs.4.3 and 4.4 respectively.
DPFdiffuse =E /E (4.3)
intemal-diffuse vd
DPFbc= =E (E (4.4)
intemal-beam/ vg -E vd)
71
where Evdis the external horizontal diffiise illuminance, E the internal illuminance at a given
intunal-diffuse
point due to the sky diffuse light transmittedby light pipe, E the
intemal-beam internal illuminance at a
Using DPF, the internal illuminance at a given point with coordinates(x, y, z) due to a given light pipe
Eextemal (4.5)
intcmal (x, y, z) = DPF(,,,
y,,) x
the internal illuminance at the given point P(x,y,,), and DPF(x,y,,) is the light pipe
is
where E intemal (x, y, z)
factors into the proposedDPF model, the effectsof thesefactorson light pipe daylighting performance
can be taken into account.Geometricfactorsthat may determinethe performanceof light pipe are the
configurationsof the systemincluding the length and diameterof the light pipe tube and so on. The
irradiance.The split betweenthesetwo componentscan be dictatedby the numerical value of the sky
clearnessindex k, [I]. Sky clearnessindex (defined as the ratio of global to the extra-terrestrial
factor, other factors such as solar altitude ((x,) that is usually usedto index the position of the sun on its
track through the sky canopycan also have effect on light pipe daylighting performance.'Me internal
illuminance distribution can also be affectedby the internal geometricfactors such asthe position of
the point of interest(due to the inversesquarelaw and cosinelaw). Other factors such as the room
shape,room dimensionsand the internal reflection within a room lit by light pipes will also affect the
internal illuminance distribution. However, becausepresentlythe main aim of the project is to evaluate
light pipe's daylighting performance,internal reflection is discussedin later part of the thesis(Section
(F,), external environmentalfactors (F.) and internal distribution geometric factors (F),
72
DPF f( FF Fc, Fi) (4.6)
y,,) ,,
4.2 TRANSMISSION OF SKY DIFFUSE LIGHT AND SUNLIGHT WITHIN LIGHT PIPES
It hasbeen explainedin Section3.2 that light pipe transmitssky diffuse light and sunlight in different
This
mechanism. is mainly due to the different natureof sky diffuse and beam irradiance.Sky diffuse
irradianceis from all angular directionswithin the 271solid anglerange while sunlight is parallel with
its direction dependenton the position of the sun. Thereforeit is logic to supposethat the
transmittancesof sky diffuse and beamilluminance through light pipe systemcan be different. The
theoreticalmethodsfor calculating the transmittancesof sky diffuse and beam illuminance through
transmittanceis a function of the number of reflections required for a ray of light to descendthe pipe
its
and reflectance(Fig. 4.1). If sunlight of intensity I and elevation ct is incident on a light pipe tube of
only for a two-dimension light pipe tube. In real applications, light pipes are three-dimensional. For a
three-dimensionlight pipe tube, most light is not incident normally to the reflection surfaceof the tube
Figure 4.2 showsa projectedview along the axis of a straight tubular light pipe tube. Light incident on
the apertureat a distancex from the axis is incident on the reflecting surfaceat angle i. Then the
If the reflectance of the reflecting surface is p then the transmission of a ray is pý4.The energy entering
the aperture in the internal between x and x+ Ax is proportional to Ax (R 2_X2)". Therefore the
73
transmissionof the light pipe tube for sunlight (defined as the ratio of output to input beamirradiance)
can be obtainedas:
f,, [(R 2_ X2
1/2
IR ])OL/McositanOd
x (4.8)
The transmissionof sky difftise light is more complex. However, by dividing the sky vault into a series
of patches,the calculation can be carried out. Figure 4.3 showsan isotropic sky vault divided into 21
patches.Presumethe radiancedistribution is uniform and for a given sky patch, its azimuth angle is y
and altitude cc,then the horizontal diffuse illuminance (AIL) entering light pipe tube due to the sky
where doc and dy are the spanof the given sky patch in altitude and azimuth dimensions.
The total horizontal diffuse illuminance (Evd)input into light pipe tube can be obtainedby either
summingup the contributions of all relevant sky patchesor by measurements.Say IL, is the horizontal
diffuse illuminance entering light pipe tube due to a given sky patch, and Tn is the transmissionof IL,,
within the light pipe tube, then the total transmissionof the sky difftise illuminance (Tdiffus,
) can be
obtainedas:
21
Tdiffuse Evd
ý(E IL,, T,,) / (4.10)
n=l
in which IL,, can be obtainedusing Eq. 4.9 and T,, can be calculatedusing Eq. 4.8. Above calculation
on Tdiff,,,,in Equation is basedon the consumptionthat the sky luminancedistribution is uniform, i.e.
an isotropic model is used to provide the input illuminance value for light pipes. However, it is well
known that the sky illuminance distribution is not uniform. Therefore, to calculate Tdim,,
with a higher
74
accuracyrequiresthe application of zenith luminancemodelsthat can describeunisotropic sky
illuminance within light pipe tube, sky luminancedistribution shouldbe madeknown first. Sincethe
models that can describethe luminancedistribution acrossthe sky vault to the theoreticalcalculation.
Zenith luminance,L,,, describesthe luminous intensity of the circular part that is on the top of the sky
vault. Using zenith luminancemodels,the luminanceof other parts of the sky can be directly related to
L,,. This has two important bearing on the designof light pipes. Firstly, in real applicationsa light pipe
can be installed in sucha condition that its solar energycollector can only have a view of part of sky.
The use of zenith luminancemodels enableslighting designersto know the estimatedsky diffuse
illuminance availableto light pipe as input, so as to predict the internal illuminances due to the light
pipe. Secondly,the transmissionof the illuminance from different parts of sky that have varying
be defined as the sky conditions that the sun is completely blocked by cloud. It hasbeen found by
measurementsthat when the sky is overcast,the zenith part of the sky vault tendsto be brighter than
other part of the sky whose altitude is ct. The brightnessof sky patchesincreaseswhen its (x increases.
The ratio of the luminanceof any sky patch, L, ()to zenith luminance,L,,, was found to be a function of
luminancedistribution index, b, and the altitude of the sky patch (x.Moon and Spencer[2] described
aboverelationshipas:
75
L, / L,, = (1+b sin a) /(I + b) (4.11)
O
Moon and Spencerfound that for overcastskies, the best value for b was 2. CommissionInternational
Muneer [3] hasusedEq. 4.11 to establisha relationship betweenslope diffuse irradianceand horizontal
D,8/D = cos2(,612) + [2b I z(3 + 2b)][sin, 6-, 6cosg -; rsin 2(,6/2)] (4.12)
where Dp is the hourly diffuse irradiance for a sloped surface,D is the hourly horizontal diffuse
overcastand non-overcast
conditions. Basedon his study for Bracknell, Muneer found that for a shaded
surface(facing away from sun) the 'best'value of b was 5.73; for a sunlit surfaceunder overcastsky
Applying the b value of 2 to Eq. 4.11, it can be seenthat under overcastsky conditions,the ratio of the
luminanceof the zenithal part of sky to that of the horizontal part is 3. Applying the b value of -0.62
for southernsky vault under non-overcastsky conditions, the ratio of the luminanceof the zenithal part
of sky to that of the horizontal part is 0.38. Shaoet al [4] have reportedthe phenomenonthat under
overcastsky the efficiency of light pipe seemsto be higher than that under clear sky conditions.This
may be explainedbasedon abovebrief analysis.When the sky is overcast,the direct view of the
zenithal part of the sky, which has the highest luminance,is transmittedby light pipe with high
efficiency. While under clear sky conditions, luminanceemitting from the brightest part of sky is
transmittedby light pipe with lower efficiency due to the multiple reflections, resulting in lower overall
transmittanceof daylight.
76
4.3.3 Perez all-sky model
The Perezet al model [5] computesL,, defined as the ratio betweenthe sky luminanceat the
consideredpoint Lp and the luminanceof an arbitrary referencepoint, as a function of the zenith angle
of the consideredpoint and the angle betweenthe consideredpoint and the position of the sun. The
where 4 is the zenith angle of the consideredpoint, and y the angle betweenthe consideredpoint and
the position of the sun. The parametera, b, c, d and e are adjustablecoefficients, functions of insolation
zenith angle of sun, the isotropic correction factor, the sky's clearnessindex and the sky's brightness
index [6].
Basedon Perezmodel, Kittler, Darula and Perezproposeda new range of standardskies.This is a set
conditions.
The equationfor relative sky luminancedistribution to the zenith luminance is defined as a function of
ý, the zenith angle of the consideredpoint and y, the anglebetweenthe consideredpoint and the
L,, f(7)9(ý)
g (4.14)
_
L,,., f (Z, )9(0)
77
in which the function f is the scatteringindex that relatesthe luminanceat a point to its distancefrom
the sun.
ý
f(o =I+ c[exp(do - exp(d)r/ 2)] +e COS2
To determinethe internal illuminance distribution due to light pipe using the proposedDPF model, the
model must be able to predict the illuminance receivedby a given point from light pipe diffuser. The
theoreticalcalculation of a given point due to a certain light sourceoften involves two laws, which are
the Inverse SquareLaw and the CosineLaw. According to the property of the light source,the
procedureof applying Inverse SquareLaw and Lambert'sCosineLaw differs. Particularly, when the
output of light pipe is assumedto be uniform diffuse light, the approachof "radiative view factors" can
also be used to determinethe illuminance receivedat a given point from light pipe.
any direction from an elementof a perfectly diffusing surfacevaries asthe cosineof the anglebetween
that direction and the normal vector of the surface.As a consequence,the luminanceof that surfaceis
the sameregardlessof the viewing angle.It is further illustrated by Fig. 4.4 that the intensity of
radiation along a direction that has angle 0 with the normal to a radiation-emitting surfaceis,
where I,, is the intensity of radiation in normal direction. The intensity of radiation is defined as the rate
of emitted energy from unit surface area through unit solid angle.
78
4.4.2 The Inverse Square Law
Distant lights appear fainter than nearby lights of the same intrinsic brightness. However, the true
brightness of an object (its luminosity), must be distinguished from its apparent brightness. The term of
luminosity is defined as the amount of energy radiated per second by an object. Ile intensity of light
observed from a source of constant intrinsic luminosity falls off as the square of the distance from the
object. This is known as the inverse square law for light intensity. Figure 4.5 illustrates the inverse
square law. The entire light through the I square-foot first area goes through the second one, which is
100 times larger; hence the light intensity per square foot is 100 times smaller in the second area. 'Me
The view factor betweenany two surfacesis defined as that fraction of the radiative energyleaving an
emitting surfacethat is interceptedby the receiving surface.The method for calculating the view
factors from differential areasto sphericalsegmentswas given by Naraghi [7]. Figure 4.6 illustrates
calculation of view factor for the given point usesdifferent formulae as shownbelow by Eqs.4.18
(Appendix 1).
performanceof passivesolar tubular light pipe under all weatherconditions will be developedin this
methods.In theory, data can be manipulatedto produceresults from which almost any conclusioncan
carried out using various statisticalprocedures.A summaryof the statistical tools usedthroughout this
work is presentedforthwith.
79
4.5.1 Mean bias error (MBE)
The meanbias error (MBE) of a model indicatesthe model's ability to replicate data.In anotherword,
the MBE tells whether or not a given model tendsto overestimateor underestimatethe measureddata.
'"(Eestimated- Eactuad
Mean Bias Error, MBE = (4.19)
no. of observatims
A lower value of MBE indicatesbetter model performance.A0 value MBE indicatesoptimal model
Root mean square error (RMSE) is widely used to indicate the degree of scatter of the computed data
Eestimated- Eactual )2
Root Mean Square Error, RMSE -T( (4.20)
no. of observations
where Esthnated
is the estimated internal illuminance due to light pipe, Eactual
the measured internal
80
and Eactual
The greaterthe deviation betweenEestimated
,
the greaterthe sum of the squareof the
difference will be. A low RMSE is indicative of better model performance.A larger RMSE indicates
greaterdispersionof data.
As addressed in Section 4.5.1, in some cases MBE can be misinterpreted. For example, for two
mathematical models A and B, when both model have the same MBE, the performance of them can be
different. This is because that the magnitude of the data that the two models handle can be different. If
the data that model A handles is in average several times to that of model B, then the perfon-nance of
model A is actually better than model B. The value of MBE being equal to 0 does not necessarily
indicate a good performance model. Suppose a model predicts largely overestimated value for half of
the observations and gives heavily underestimated value for the other half, the MBE will also be found
to be zero which can lead to erroneous conclusions such as the model is performing exceptionally well.
I feslimaled-Eaclual II Eeslimaled)
Percentage Average Deviation, PAD= -P(100*
no. of observatims
4.5.4 Slope and the value of the coefficient of determination of predicted versus measured
illuminance
data,the use of scatterplot of calculatedversusactual data,and the slope of the fitted trend line and the
value of the coefficient of determinationare advisedto provide insight into the performanceof models.
The slope value is indicative of the validity of the model under test. The degreeof validity increasesas
the slope approaches unity. In the case of present study, a slope value lower than I implies that the
model tends to underestimate internal illuminance values due to light pipe, and the converse applies for
81
overestimation.However, the slope value cannotbe usedsolely to evaluatethe performanceof a
model. A model that producelarge scattercan producea unity slope trend line, although the
performanceof the model is actually not well. Therefore,the combineduse of slope and the coefficient
performanceis examinedin two ways. Firstly, the histogramprovides a check regardingthe proportion
of datapoints that fall within specific rangeof percentageerror and secondly,it allows an examination
82
REFERENCES
1. Lopez, G., Batlles, F. J., Martinez, M., Perez, M. and Tovar, J. (1998) Estimation of hourly beam
solar fiom measured global solar irradiance in Spain, Proceedings of the World Renewable Energy
Conference, 1998
2. Moon, P., and Spencer, D. E. (1942) Illumination from a non-uniform sky Trans. Illum. Eng. Soc.,
37,707-725
3. Muneer, T. (1990) Solar Radiation Model for Europe Building Serv. Eng. Res. Technol., 11(4), 153-
163
4. Shao, L. (1988) Mirror lightpipes: Daylighting performance in real buildings Lighting Research and
Technology, 30 (1), 37 - 44
5. Perez, R., Seals, R. and Michalsky, J. (1992) Modelling skylight angular luminance distribution from
routine irradiance measurements, IESNA transmission of the 1992 IESNA annual conference
6. Muneer, T. and Zhang, X. (2002) A new method for correcting shadow band diffuse irradiance data,
83
ot
......... ......
Figure 4.1 Sunlight of intensity I and elevation ccdescending a 2-D straight light pipe
Figure 4.2 A projected view along the axis of a straight tubular light pipe tube, with light
entering at distance x from the axis followed by its incident on the internal surface at projected
angle I and travels distance d between reflections
84
Figure 4.3 The uniform overcast sky vault devided into 21 patches
Ii
85
asl2r
86
ing point
Figure 4.6 The calculation of view factors between a point and a hemisphere surface
87
5. MEASUREMENTS AND DATA PROCESSING
To enablethe modelling of the daylighting performanceof light pipes, data on the internal illuminances
obtained.The proposedDPF model (Eq. 4.6) expressedthe daylighting performanceof light pipes as a
distribution geometric factors (Fi). It hasbeenaddressedin 3.5 that to build the proposedmathematical
Other factors rather than F., F, and Fi factors,which can affect the daylighting performanceof light
pipes may include the blocked sky view and shadedsunlight causedby surroundingnatural or artificial
objects,internal reflection related factors (e.g. the colour of the ceiling and walls), the shadingcaused
by interior furniture layout and so on. Once a basic and generaldaylighting performancemodel of light
pipes has beengenerated,the effects of abovefactors can be incorporatedinto or addedto the model.
of wide applicability.
Before the commenceof large scaleand long term measurements,a trial test on single light pipe was
undertakento preliminarily identify the key factorsthat affect the performanceof light pipes. This trial
test was undertakenin a real building room - the Currie test room. After the key factorsbeing
The developmentof light pipe systemhasbeen an on-going process.More monitoring was therefore
applicability and adaptability of the proposedDPF model. The latest developedlight pipes
Effects of different light pipe diffusers on internal illuminance distribution were investigatedand
88
5.1 NAPIER SOLAR STATION
Napier University solar station is locatedon the roof the main building at Merchiston campus,Napier
the south-westof Edinburgh.In a5 km radius aroundthe campus,80% of land comprisesof urban and
suburbanhousing and offices whilst the remainderis a mixture of farmland and natural features[1].
Napier University solar station was built as a part of the international daylight measurement
latitude and longitude of the solar station are 55.95N and 3.20'W respectively.The height of the solar
The following instrumentation was installed at Napier University solar station. Illuminances are taken
Illuminances
89
Shadowband specifications
Napier University solar station employs a Kipp & Zonen CM 121 shadowring with the German
eachmeasuringday and diffuse illuminance correction factor for the obstruction of the shadering has
Figure 5.1 showsthe views of the Napier University solar station.Figure 5.1(a) showsa close-upview
of the installation and the relative position of the instrumentation.The sky diffuse illuminance sensor
and shadowband are shown in the backgroundand the horizontal global and vertical illuminances
sensorsare shown in foreground.Figure 5.1(b) representsa macro view of the stationwith the global
horizontal and vertical illuminance sensorsand sky diffuse illuminance sensorand shadowband
One Kipp & Zonen pyranometerCM1 1 global irradiancesensorwas used to measurethe global
incident energyto the light pipe installed in Currie test room. The Kipp and Zonen CM1 1 global
pV/Wnf2. The CM1 I global irradiancesensorhad been calibratedin November 1999by the UK
Meteorological Office.
Six sensorsof two types were usedto measurethe internal illuminancesdue to light pipes. Three
sensorsof SeleniumPhoto-ElectricCells Type-B were suppliedby Megatron Ltd, London. The active
areaand the sensitivity of Magetron Type-B photoelectric sensorsare 3.5 cm2and 0.245 [LA/lux.
Magetron Type-B photoelectricsensorsare of a dimensionof 42MM diameter and 20mm high with
integral cable 2m long. The advantageof seleniumphotovoltaic cells over other cells is that their
responseis very close to that of the humaneye; this makesthem particularly suitablefor use in light
someother photocells,and so they are not usedas solar cells [2]. The three Magetron Type-B
90
photoelectric sensorsemployedin this study use amplifier that requiresexternal 4.5 Volts AC-DC
power-supply.
The other three sensors were supplied by Kipp & Zonen [3]; the actual component supplied was a Lux
Lite illuminance meter, 54mm diameter x 35mm high with integral cable 3m long. Ile Lux Lite
consists of a photodiode, a filter, a diffuser and a housing. A resistance shunts the photodiode; this is
done to generate a voltage output. The photodiode, the filter and the diffuser on top determine the
spectral specifications. The diffuser ensures a field of view of 180 degrees, and that angular
characteristics fulfil the so-called cosine response. The sensitivities of the three Lux Lite sensors are
given as 10.32 pV/Iux (±l%), 9.68pV/Iux (±I%) and 10.11pV/Iux (±I%) respectively. The ranges of
from a variety and member of sensors,making them suitablefor a wide range of tasksin many
To enablethe measurementof internal illuminance achievedby various light pipes with bends,one
the crosssection of the light pipe tube (end section).The adjustablestandconsistedof two components,
a commonly availablephotographer'stripod and a table. The flexibility offered by the tripod madethe
slope and the distanceof the plane where illuminance sensorswere mountedadjustable.The table was
use, the table was painted with a matt black finish to eliminate any light reflection into areaswithin the
internal daylight distribution due to light pipes. This standwas madesimilar to the one used in
91
Craighousetest room, but in a smaller dimensionto be adaptableto Merchiston test room that is
A trial measurementwas first carried out to determinethe decisive factors that directly affect the
performanceof light pipe. The obtaineddatawere to be analysedto reveal the influencesof Fg,F, and
Fj factors on light pipe DPFs.A simple DPF model on the performanceof a single straight light pipe
can thus be built upon the dataobtainedin the trial measurement.The performanceof this simple
model was used as a pre-checkof the sufficiency and validity of the large-scalemeasurementto be
roof is free of obstructions.The light pipe is allocatedin a child's bedroom on the top floor. The
bedroom areais a rectangular-shapedspaceof 3.7 x 2.1 x 2.3 m (length x breadth x height). Figure 5.2
showsthe dimensionsof the room and the measurementschemeundertaken.The room has four walls,
all of them coveredwith yellowish-white wallpaper.A small window on wall I (Fig. 5.2 refers) has a
westernaspect.The sizesof the window's two panesare 32 x 71 CM2and 26 x 65 cm,; and the height
of the window's sight line (sill level) is 126cmfrom the floor. Near a comer of the room, there is a door
on wall 2, which is oppositeto wall 1. Two piecesof yellow wooden furniture are locatedalong wall 3.
In anothercomer and along wall 4, a bed is placedunder the diffuser of the light pipe.
The light pipe on test was installed on a north-facing roof, which has a slope of 29* (Fig. 5.3). The
length of the light pipe is 121cm (measuredfrom the top edgeto the bottom rim of the mirror pipe).
The horizontal distancesfrom the diffuser centreto wall 1,2,3 and 4 are 225cm, 146cm, 13lcm and
79cm respectively.
From l't -7 th of May and from 26 th-31" of May 200 1, monitoring was carried out to record the
external global illuminance and internal illuminance data. For this experiment, the window was
92
coveredwith a thick dark cotton towel, and the door was sealedto prevent any light entering from
outdoors.One Kipp and Zonen CM II global irradiancesensorwas fixed on the ridge of the roof to
measurethe global incident energyof the light pipe under test. Three Megatron indoor illuminance
sensorswere employedto record the internal illuminance in different positions of the room. These
sensorswere fixed horizontally on threestands,all of height = 0.7m; and there was no obstruction
External global irradiancedataand internal illuminance datawere recordedby a Grant (Squirrel) data-
The measurement carried out included external global irradiance, internal illuminance and distances
from survey points to light pipe diffuser, with corresponding date and time recorded. Sky clearness
index and solar altitude were calculated using the algorithm provided by Muneer [5]. To ensure the
database reliability, data were discarded for solar altitude less than 10', or for k, ý! 1. Out of the 23490
data points, 21147 data were selected to develop and validate the proposed model.
Within the opengroundsthe test room was locatedon flat ground with an open southernaspect.The
test room was 3.Omlong x 2.4m wide with a height of 2.5m. Inside the test room all four sidesand the
roof had hardboardcladding nailed to the frame. All edgesand the door surroundswere coveredwith
equipmentinside the room was protected from the effects of weather.Ile room was equippedwith
power supply unit to support a PC, data-loggerand multiple indoor illuminance sensors.
93
To enablethe performancecomparisonbetweendifferent light pipes, the test room was designedfor
easeof installation of light pipes with various designs.Initially, a total of eight light pipes with varied
configurations(straight runs plus thosewith multiple bends)were suppliedby the project co-sponsor.
The diametersof theselight pipes were 21Omm,330mm, 450mm and 530mm. All the light pipes use
2000. Daylighting performanceof 15 light pipes of various configurationsunder all sky conditions
were carried out during thesefour months.Details of all testscompletedare listed in Table 5.1. The
light pipe diameter,length, numberof bendsand length of bends,if any, were noted. Illuminance
sensorswere arrangedon floor or on standswith their respectivedistancesto the centreof light pipe
The desiccantof the Kipp & Zonen CM II pyranometerhad beencheckedbefore and during the
to solar time.
After the measurements, data obtained were processed and quality controlled. External illuminance, kt
and ot, were calculated using the algorithm provided by Muneer [5]. To ensure the database reliability,
data were discarded for cc,:ý 10', or for kt ý! 1. A total of 65 270 data points for all light pipes
94
5.5 DAYLIGHTING PERFORMANCE MONITORING UNDER ALL SKY CONDITIONS -
MERCHISTON TEST ROOM
A new test room was built upon the roof of the main building ("Main building" roof) at Merchiston
campusof Napier University. The test room was locatedin the central part of the roof and has a
completeopen view to the southernsky, and 95 per centsof the whole sky hemisphere.The test room
was 2m wide and 2m long with a height of 2m. Inside the test room all four sidesand the roof had
hardboardcladding nailed to the frame. All edgesand the door surroundswere coveredwith heavy
paper tapeto sealthe inside spacefrom ingressof daylight. Internal illuminance sensorsand data-
logger equipmentinside the room were protectedfrom the effects of externalenvironment.The room
was not equippedwith power supply, becauseGrant data-loggerusessix AA1.5V batteriesand three
Kipp & Zonon Lux Lite indoor illuminance sensorsdo not require power supply.
The test room was designedfor easeof installation of light pipes with various designs,so as to to
with varied diameterswere installed in the test room. The diametersof theselight pipes were 21Omm,
330mm, 450mm and 530mm.All the light pipes use clear polycarbonatedomes,610mm long silverised
Continuous performance monitoring was undertaken over a period from 28ý' May to 29h August, 200 1.
Daylighting performance of 10 light pipes of various configurations under all sky conditions were
carried out during these three months. Details of all tests completed are listed in Table 5.2. The light
pipe diameter, length, number of bends and length of bends, if any, were noted. Illuminance sensors
were arranged on floor or on stands with their respective distances to the centre of light pipe diffuser
recorded. The relative position of each testing point where each illuminance sensor seats to the light
pipe being measured was also recorded. External horizontal global iffadiance data from the
pyranometer (Kipp & Zonen CM 11) and internal illuminance data from three photoelectric sensors
(Kipp & Zonon Lux Lite) were sampled every ten seconds and the averaged minute-by-minute data
95
External global horizontal illuminance datawere measuredat Napier University CIE IDMP solar
station.Napier University solar station is establishedon the "South workshop" roof at Merchiston
campus,Napier University. Both the test room on the "Main building" roof and the solar station on the
"South workshop" roof have open view to the sky hemisphere.The horizontal distancebetweenthe two
roofs is about 150 m. Napier University CIE IDMP solar station usessolar time as systemtime. The
systemtime of the Grant Squirrel data-loggerusedto record internal illuminances due to light pipes
and (x, were calculatedusing the algorithm provided by Muneer [5]. To ensurethe databasereliability,
datawere discardedfor cc,:5 10*, or for k, ý: 1. A total of 22 270 datapoints for all light pipes
room to investigatethe internal daylight distribution due to light pipe with different configuration and
type of diffuser. Eight testswere carried out in the measurements.Table 5.3 showsthe compositionof
the tests.
Before the commenceof the test, all the light pipes (MonodraughtSunpipe)installed in the Craighouse
test room had their diffusers coveredwith black heavy plastic sheetand securedwith adhensivetape.
The cover was removed from the light pipe to be testedand the test equipmentwas placeddirectly
The diametersof the light pipes were noted and six sensors(including three Kipp & Zonen Lux Lite
sensorsand three Megatron Type-B sensors)were arrangedin an array being equispacedon the same
diameter.The six sensorswere fixed upon a table sustainedby a tripod. The table was manufactured
from a piece of high-density polystyrene25mm thick and sky-blue in colour. Prior to use, the table was
painted with a matt black finish to eliminate any light reflection into areaswithin the sensors'range.it
96
was important to have the planewhere sensorswere fixed perpendicularto the light pipe axis to ensure
that all sensorswere at an equaldistancefrom the diffuser, the adjustabletripod was usedto ensurethis
To study the effect of light pipe diffuser type on light pipe daylighting performance,a 330 min
diameterlight pipe provided by Monodraughtwas installed in the Currie test room. The object of these
testswas to comparethe performanceof the opal diffuser againstflat designclear diffuser. The opal
diffuser takesthe form of a white polycarbonateconvex shapedome,which diffuses the light evenly
into the interior space.The flat designclear diffuser is madeof translucentplastic material with crystal
effect finish and flat end.The latter type of diffuser is usedto improve the daylight transmissionof
For three days from 28"' to 30'hMay 2001, the daylighting performanceof a 330 nun diameterlight
pipe with flat diffuser was monitored.The externalglobal irradianceand internal illuminancesat three
minute basis.The three indoor Megatron illuminance sensorswere installed correspondingto the
measurementscarried out on the light pipe with the older opal diffuser (seeSection5.3). A total of 4
300 datapoints were thus obtained.Daylight penetrationfactor DPF, defined as the ratio of internal
illuminance to external illuminance for flat diffuser were calculatedfor all datapoints.
97
REFERENCES
1. Kinghorn, D. (1999) Solar illuminance modelsbasedon other meteorological data Ph.D. Thesis,
Napier University, 64
5. Muneer, T., Abodahab,N., Weir, G. and Kubie, J. Windowsin buildings thermal, acoustic, visual
98
(a)
(b)
99
- --
f2l% [ -]
***44%w
DOOR; %,j SUN-PIPE DIFFUSER; SENSOR; WINDOW; 414 CENTRE OF THE WINDOW
Figure 5.2 Schematic of the Currie test room and light pipe system
100
Kipp & Zonen CMI 1 Solarimeter
Mirrored light-pipe
Transparent
collector
Light diffuser
101
Table 5.1 Designs of light pipes that were monitored in Craighouse campus, Napier University
Design Diameter, mm Type Length, nun Bendsnumber
1 210 straight 60 N/A
2 210 straight 120 N/A
3 210 elbowed 60 1
4 210 elbowed 60 2
5 330 straight 60 N/A
6 330 straight 121 N/A
7 330 elbowed 60 1
8 330 elbowed 60 2
9 330 elbowed 60 3
10 330 elbowed 60 4
11 420 straight 120 N/A
12 420 straight 60 N/A
13 530 straight 60 N/A
14 530 straight 1201 N/A
15 530 elbowed 601 1
Table 5.2 Designs of light pipes that were monitored in Merchiston Campus, Napier University
Table 5.3 Internal daylight distribution tests on light pipes installed in Craighous Campus,
Napier University
Design Diameter, mm Type Length, cm Diffuser type
1 2 10 straight 60 Clear diffuser
2 330 straight 60 Opal diffuser
3 330 straight 60 Clear diffuser
41 420 straight 60 Opal diffuser
5 420 straight 120 Opal diffuser
6 420 straight 120 Clear diffuser
7 530 straight 60 Opal diffuser
8 5301straight 601 Clear diffuser
102
6. MATHEMATICAL MODELLING
DPF (Daylight PenetrationFactor) is straightforwardand canbe divided into three main blocks:
(1) Parameterisation, to identity the most decisive factors that affect the daylighting performance
Figure 6.1 shows the system structure of the development of the proposed mathematical models and the
Above researchstrategytreatsa light pipe systemas a black box and doesnot care the physical model
experimentaldatabase.
The major advantageof this strategyis that it is simple, practical and highly
stagedtactic was adoptedto examinethe strategy.The main aims of the two stagesare outlined below:
straight light pipe will be developed.After that, numerical error analysiswill be applied to verify the
validity of the straight light pipe model, DPF. The analysisresultsobtainedin this stagewill determine
the sufficiency and the necessityof the proposedstrategy,and hencewill decidewhether the research
led by this strategyshould go onto the secondstage.Further work to improve the performanceof DPF
model and to extent the model to all light pipes can then lead the researchinto its secondstage.
103
(2) Stage11.The secondstageis a logically extendingand deepeningof the researchcarried out
bias identified in stageI. Advanced solar radiation theoriesand techniqueswill be applied in this stage
of research.It is intendedthat the completion of this stageof researchwill further improve the overall
accuracyand reliability of the proposedmodel. Finally, validation of and modification to the general
transmittanceof the system.Light pipes are designedto collect light from both the sky and the sun.
Sincethe external environmentalfactors suchas sky clarity, sky-diffuse radiancedistribution and sun's
changescontinuously.
The determinationof light pipe systemconfiguration aimed at achieving a given internal illuminance
under given sky conditions would be the task of light pipe designers.It may be logical to assumethat
length, diameter,number of bends,angle of bend and the diffuser type, and (b) the abovementioned
be desirable.
The conceptof light pipe Daylight PenetrationFactor (DPF) hasbeen explainedin Chapter4, Section
4.1. DPF, (),, ) is defined as the ratio of a given point's internal illuminance to the total external
y.,
illuminance for a given point with coordinates(x, y, z),
I ()A
where E internal
(x.y,z)is the internal illuminance at the given point P(,, E is the total external
y,,), external
illuminance and DPF(,,, ) is the light pipe Daylight Penetration Factor for the given point.
y,,
Light pipes are daylighting devicesthat utilise both sunlight and skylight. It hasbeenreportedthat
sun's position has an effect on light pipe's perfortnance[I]. When the sky is clear, the solar altitude ct,
influencesthe transmittanceof sunlight within the light pipes. However, when the sky is overcastor
major componentof externalglobal illuminance. Hencein the presentstudy the transmittanceof light
pipes was initially consideredto be a function of cc,and the sky clearnessindex k, (defined as the ratio
According to the research strategy addressed in Section 6, DPF modelling for light pipes were
undertakenin two steps.The first step is to develop a DPF, model basedon the performancemodelling
where D is the distancefrom light pipe diffuser centreto a given position P(,,,y, ). Eq. 6.2 is only
for
applicable straight light pipe designsand thereforecannotserve as a generalDPF model for light
build the model and anotherone year's independentlymeasureddatato validate the model. The
building and validation of DPF models are reportedin the following sections.
105
6.2 THE DPF OF A STRAIGHT LIGHT PIPE
It may be shown from first principles [2] that the illuminance receivedat any given point from an
elementalareaof a luminous sourceof a finite size is proportional to the product of the luminous
intensity of the elementalarea(1), the respectivecosinesof the anglesof emittance(0,) and incidence
(0) and the inversesquareof the distance(D) betweenthe elementalareaand the point of incidence
(seeEq. 6.3).
Cos 0, Cos Oi
IlluminanceocID2 (6.3)
Thereforeto obtain a preciseestimateof the luminanceenvironmentof the test room under discussion,
one has to integrateEq. 6.3 point-by-point within the room, taking accountof the variation of luminous
for designpurposes,in particular by an industry that is in a stageof infancy. In view of the above
where DPF ý (.,.y., ) is the light pipe daylight penetration factor at the point P (,,,
y,, ), due to the particular
straight light pipe in test, A is a parameter that is dependent on sun's position and sky clarity, D is the
distance from point P (,,, )to light pipe diffuser. Note that the integral calculation suggested via Eq.
106
Figure 6.2 showsthe scatterplot of measuredDPF againstD. Although the generaltrend seemsto be
,
in order, the percentagedeviation for any given distanceof the daylight penetrationfactor may be very
large. Figure 6.2 thereforesuggeststhat other non-geometryfactors may be involved. Presentlythe role
To explore the relationship between DPF and cc,,data sets recorded by sensors at a distance of 155cm
,
for different time frames were selected. A total of 258 data points were employed to plot DPF as a
,
function of solar altitude. Figure 6.3 shows this scatter plot. It is evident that for any given cc,a large
scatter is generated, e.g. the percentage deviation of DPF is 120% when cc,::ý;41.5'. This behaviour
suggests that weather parameters additionally affect the daylight penetration factor. A further
demonstration of the influence of weather condition's on DPF . is shown by plotting DPF , against sky
clearness index kt for 48.5': 5 (x, <49.5'. Figure 6.4 shows such a scatter plot for D=155cm. A strong
It was demonstrated in the above section that DPF, is influenced by D, c(, and k,. Presently, a model
DPF, = (aoo+ aoI* cc,+ a02* lXs2+ a03* k, + a04* k, 2+ ao5* cc, *
2*k, 2)
a06 * (Xs + a07 * (Xs *kt2+ aog *a *kt /D2 (6.5)
S2
Equation 6.5 was further simplified and its validity evaluated.The simplified model is:
DPF, = (ajo + all * (x, + a12* (x, 2+ a13* k, + a14*k, 2+ a15* ot, * k,) /D2 (6.6)
Values for the coefficients used in Eqs. 6.5 and 6.6 are listed in Table 6.1.
107
Equations6.5 and 6.6 presenta genericprocedureto associateDPF to the two widely known
parameters,namely cc,and k, that influence daylight receipt. The equationscover the entire range of k,,
i. e. from heavy overcastwhen kt ;L-0.2 to clear skieswhen kt > 0.6. Further precision may be obtained
Model validation was undertakenvia estimationof Mean Bias Error (MBE), Root Mean SquareError
X(Eetimated - Eactuad
Mean Bias Error, MBE =
no. of observatims
ý:(Eesiýated - Eactual)2
Root Mean Square Error, RMSE = :
no. of 6 bservatio ns
Equation 6.7 is usedto obtain estimatedinternal illuminance E (x, for a given position P (,,, ),
estimated
y,z) y,,
(lumen/Watt). KGis computedfrom a formulation provided by Muneer and Kinghorn (Eq. 6.8) [3]:
Based on above procedure using Eq. 6.5 calculated internal illuminance were plotted and regressed
against corresponding measured data to determine the model's performance. Results of this comparison
108
are shown in Fig. 6.5. The slope of the fitted trend line is 0.973, and the coefficient of determinationR2
is 0.929. The MBE was found to be -2lux and RMSE 27lux which is 5% of the maximum value of
was noted as 21%. Calculatedinternal illuminance data were once again plotted againstmeasured
values for 11 different positions within the test room. 'D' for thesepositions range from 154cmto
212cm. Figure 6.6 showsthis plot for D= 194cm.Table 6.2 showsthe slope of the regressedline and
Similar validation procedureswere applied to Eq. 6.6 to examineits performance.In this casethe slope
the fitted trend line found to be 0.970, and the coefficient of determinationW was 0.930. The
of was
MBE was -2lux and the RMSE 28lux, which is 5% of the maximum value of measuredilluminance
(512lux) and 20% of meanilluminance (138lux). Percentageaveragedeviation was once again found
to be 21%. Thus, the simple model representedby Eq. 6.6 has comparableperformanceto the more
involved model (Eq. 6.5). Table 6.3 showsthe validation resultsfor 7 positions at which internal
6.2.3 Summary
undertaken.Experimental analysison the effects of different factors that influence the daylighting
determined.A DPF, model for a particular light pipe hasbeenestablished.Two articles, one on the
DPF, model and the other on light pipe performancemonitoring have beenpublishedon thejournal of
ConferenceProceeding[4,5].
It has beenconcludedthat the internal illuminance at a given point in a room where light pipes are the
only lighting systemcan be estimatedusing Eqs. 6.5 and 6.6. Reasonablygood agreementhas been
109
percentage average deviation have been found to be 27lux, -2lux and 21% respectively. The slope of
the best fitted trend line was noted as 0.97 and the coefficient detennination R2was found to be 0.93.
and modelling procedure will be applied to all types of light pipes available to obtain a general
mathematicalmodel.
Sincethe DPF, model hasbeenverified, DPF modelling for all straight light pipes were carried out. It
was thought that an appropriatemodel for straight light pipes with differing geometricalconfigurations
should take accountof the contribution of light pipe's length and diameteras well as a, and kt. The
proposedgeneralisedformulation that includes 12 coefficients (ao- alo and in) is given by Eq. 6.9,
k12 as 2
S- DPF(.,, = (a, + a, k, + a, a, + a, k, a, + a,,kl2as + a, k, tr s2+a,
Y,:) ) (6.9)
(a7 +a, A, +ag cot a, +a,, A, 2 2
P co'a)R (HID)' ID
R is the radius of the light pipe, p the light pipe surfacereflectance,Ap the aspectratio (defined as the
ratio of light pipe length to diameter)and H the vertical height of light pipe diffuser abovethe working
plane. Values for the coefficients usedin Eq. 6.9 are given in Table 6.4.
It may easily be shown that DPF(,,, ) is proportional to the light pipe's sectionalareaand this accounts
y,,
for the inclusion of the R2term in Eq. 6.9. It may be recalled that the inclusion of (x, and k, and D2
What follows now is a justification of the remainderof the terms included within Eq. 6.9. As a daylight
reflecting device, the light pipe's transmissionis affectedby the reflectanceof its interior coating
material. Basedon physical reasoningpresentedin Section2.4, it is possibleto show that a light pipe's
transmittanceis a function of the number of reflections required for a ray of light to descendthe pipe
and its reflectance(Fig. 6.7). The higher the reflectanceof the pipe's interior surface(p), the higher the
110
light pipe's daylight transmittance.Edmondset aL [1] reportedthat for sunlight of any given intensity,
the number of reflections required is proportional to A,, and cot(x,,where Ap is aspectratio defined as
the ratio of light pipe length to its diameterand cc,the solar altitude. Therefore,a linear function that
occurring within the light pipe. This explains the secondfactor in Eq. 6.9,
(a7+asAp+ag cot a, +a, OAP cot a, )
namelyp
A considerationof the mannerof the spreadof daylight by light pipe diffusers within interior spaces
led to the adoption of the last factor in Eq. 6.9, i. e. (H I D)' ID2, where 1:ý rn: ý2. For a given point in
space,to obtain its illuminance resulting from a point or finite light source,the inversesquarelaw and
the cosine law must hold. However, if a light pipe diffuser is to be consideredas a point light source,
the cosinelaw shouldbe applied only once andhencem=1; while on the other hand if the diffuser
could be treatedas a finite light source,then the cosinelaw must be applied twice, once for the
emanationof light and the other for its incidence(or projection) on the horizontal plane. This requires
in to be equalto 2. Thus a compromisewas achievedby using a variable in. The "JID term
Using the Craighousedataset, Eq. 6.9 was fitted. Good agreementbetweenthe measureddataand
calculatedvalues was obtainedand in this respectFig. 6.8 showsthe scatterplot of the predicted
internal illuminance againstmeasureddata.The trend line was found to have a slope of 0.98 and the
Model validation was also undertakenby estimationof the Mean Bias Error (MBE), the Root Mean
SquareError (RMSE) and the PercentageAverageDeviation (PAD). The MBE was found to be -I lux.
The RMSE was 27lux which was 2% of the maximum illuminance (I 187lux) and 15% of the mean
illuminance (177lux). The PAD was found to be 12%.Resultsare summarizedin Table 6.5.
To facilitate the use of the S-DPF model, Eq. 6.9 was further simplified and its validation was
undertaken by estimation of MBE, RMSE and PAD. The simplified S-DPF model is,
III
(a3 +a, Ap +as Cot a, +a64p Cot a, )
S- DPF(ý,,, = (a, + a,k, + a, tr.,),o R'(HID)'ID'
) )
Values for the coefficients used in Eq. 6.10 are listed in Table 6.4. Calculatedinternal illuminance
The slope of the best-fit trend line is 0.98, and the coefficient of determinationR2is 0.95. The MBE,
RMSE and PAD for the simplified S-DPF model were found to be -31ux, 29lux and 13%respectively
(Table 6.5).
To predict the DPF(,,, ) of elbowed light pipe systems (30-degree bends), the energy loss due to each
y,,
bend has to be considered. The length of each bend (Lb) and the number of bends (N) have been
accounted for by the use of an equivalent-length factor (fl,,,) and the energy-loss factor (/i,,,j within Eq.
6.11. Lb andfl,,, are subsumed within a modified expression for Ap,. The factor used in
Eq. 6.11 accounts for the overall transmittance efficiency of the N bend(s),
Ap,
where = (L + fl,, Lb) / 2R andL is thelengthof straightlightpipe,fl,, theequivalent-length
factor, Lb the sum of the linear lengthsof all bendsand R the radius of light pipe.fl,,,, is the energy-loss
factor for each30-degreebend, most commonly usedin a light pipe system.Values for the coefficients
The performance of the E-DPF model was then evaluated using MBE, RMSE and PAD. The MBE was
found to be -2lux. The RMSE was 23lux which was 2% of the maximum ilIuminance (931lux) and
112
22% of the meanilluminance (1031ux).The PAD of the estimatedinternal illuminance was found to be
25%. A scatter plot of the calculated against measured internal illuminance is shown in Fig. 6.9. The
slope of the trend line was found to be 0.97 and the coefficient of determinationwas noted as 0.97.
Eq. 6.11 was further simplified and the simplified E-DPF is given below,
Values of the coefficients used in Eq. 6.12 are shown in Table 6.4. The MBE, RMSE and PAD for the
simplified E-DPF model are-31ux, 25lux and 24% respectively.The calculatedinternal illuminance
due to Eq. 6.12 were regressedagainstmeasureddata.The slope of the best-fit trend line was found to
be 0.97 and the coefficient of determinationwas noted as 0.97. Resultsare shown in Table 6.5.
light pipes using the factor (H / D)' /D2. The "best-fit" value for parameterm was obtainedas 1.3.To
validate the sufficiency of the expression(H / D)' /D2 in describingthe light pipe internal illuminance
The view factor betweenany two surfacesis defined as that fraction of the radiative energyleaving an
emitting surfacethat is interceptedby the receiving surface.A V-DPF model is defined herein as a DPF
model that usesview factor method to describeinternal illuminance distribution within room. The
proposedview factor model V-DPF of a straight light pipe is shown in Eq. 6.13, where F, (,,y,,) is the
view factor betweenthe ceiling mounted light diffuser and the given point P(,,,
y,,) .
113
The method for calculating the view factors from differential areasto sphericalsegmentsis given by
Naraghi [7]. The entire formulation has been provided in Section 4.4.3. However, only case numbers
Using the Craighousetest room databasefor all straight light pipes, the view factor F, (.,,,,) for eachdata
point has beencalculatedand the value of the coefficients in Eq. 6.13 have beenre-fitted. Fig. 6.10
measureddata.The RMSE and MBE for the V-DPF model were found to be 52lux and -9lux,
correspondingto 27lux and -1lux for the V-DPF model (Eq. 6.13). The slope of the trend line for a plot
The correspondingfigures for the V-DPF model (Eq. 6.9) were 0.98 and 0.95. Resultsare shown in
Table 6.5.
In the modelling process, a number of environmental and geometrical factors that influence the
daylight transmissionof light pipes have beenidentified. In this sectiona parametricanalysisof those
factors is presented.
It was shown abovethat cc,affects the DPF by altering the number of inter-reflectionswithin light
transmittance.However, when kt changes,the extent to which a, affects DPF also changes.Figure 6.11
showsthe variation of S-DPF and E-DPF againsta, for varying weatherconditions. It was noted that
114
6.6.2 Effect of R and L
For any given (x, and k,, the squareof the radius of a light pipe, R2affects the light pipe's external
The DPF model showsthat light pipe's performancestrongly dependson D, namely the distance
betweenlight pipe diffuser centreto the given point P(,,, and H, the vertical distancebetweenthe
y,,)
diffuser centreto the working plane. In Eqs. 6.9,6.10,6.11 and 6.12, the best-fit value for the
coefficient m was found to be 1.3. This indicatesthat in real applications,light pipes work more like a
point light sourcerather than a sourceof a finite area.The internal illuminance achievedby a light pipe
In Eqs. 6.11 and 6.12, the best fitted value forf,.. was found to be 0.20, which meansthat each30-
degreebend looses20% of energyduring daylight transmission.For a light pipe with N bends,the total
increasesthe value of the aspectratio and thus it reducesthe daylight transmittance.Figure 6.13 shows
number of bends.
The internal surfaces(ceiling, roof and floor) of the Craighousetest room were painted white which
has a high reflectance,the assessmentof the contribution of internal reflected illuminance is therefore
any other information regardingthe inter-reflection of diffuse as well as beam illuminance within
rooms and also bearing in mind the fact that light pipes are primarily providers of diffuse illuminance,
115
the internal reflection characteristicsof light pipes were treatedas being of similar order as windows.
Therefore, for the Craighousetest room, the contribution of internal-reflectedilluminance to the total
illuminance that is receivedat survey points can be qualitatively assessedby comparing it to that of a
The surfacereflectanceof the test room hasbeenmatchedusing the CIBSE Lighting Guide [8] and was
found to be 0.74. By abbreviatedanalysis,the contribution of internal reflection to the light pipe DPF
of largestdiameter(0.53m) light pipe was obtainedas only 5% of the total DPF using the chart given
reflection to the total internal illuminance measuredon eachsurvey point in Craighousetest room will
not exceedthe level of 5%. In summary,for the presentDPF model, the effect of internal reflection is
of a low order.
During 2001-2002two new types of light pipe diffuser have beenintroducedinto the market, namely
out due to its short life span.Figure 6.14 presentsthe daylighting performanceof two light pipe
diffuser designs:old opal and clear diffusers. The DPFs that were achievedby the light pipe with a flat
clear diffuser were higher than thoseachievedby an old opal diffuser. The performancegain ratio
rangesfrom 2 (8am, 3V May 2001) to 4 (2pm, 29h May 2001), and the averagevalue hasbeen found
to be 2.9. This factor is herein called the Diffuser Factor (fD). By applying the averagefD factor to the
internal illuminancesvaluesgiven by Eqs. 6.9,6.10,6.11 and 6.12, the internal illuminancesthat can
Due to the irregular knurling finishing of the clear diffuser, it producesapparentlydifferent internal
daylight distributing property from that of an old opal diffuser. Figures 6.15 and 6.16 show the
difference betweenthe two types of diffusers in terms of their reflectivity and transparencyproperties.
The pictures presentedin Figs. 6.15 and 6.16 were taken in Edinburgh, Currie at about 3pm on 280'
April 2002. Figure 6.15(a) showsthat bright sunlight penetratesthe clear diffuser creating a glare
effect, while the sunlight goesthrough old opal diffuser appearsto be much weaker and dimmer.
116
Another obvious phenomenonis that opal diffuser producesmuch more uniform diffused light than
clear diffuser. The latter one producesmulti-directional and irregularly distributed light. Figure 6.16
showsthe reflected light from of a point sourcedue to the two types of diffuser.
Due to the improved transparencyproperty of the new clear diffuser, the transmissionof daylight by
light pipe hasbeen improved. As shown in Fig. 6.14, the total amountof light delivered by light pipe
with new clear difftiser is much more than that deliveredby old opal diffuser light pipe. However, at
the sametime, it was found that the uniformity of the internal light due to former kind of diffuser
seemedto be poorer than the latter one.This becomesobvious when the light pipes are applied in areas
where sunshineis abundantlyavailable.Figure 6.17 showsthat certain kind of light ring patternwas
observedwhen light pipes with new clear diffusers were installed in Bahrain. Picturesin Fig. 6.17
show that in noontime in Bahrain, clear diffuser light pipe seemto delivery pools of light insteadof
uniformly distributed internal light. This "pools of light" pattern is so strong that it causescertain
discomfortssuch as glare and undesiredlight ring on the ground. The problem was eventuallyresolved
by replacing the new clear diffusers with old opal diffusers. Occupantsare satisfiedwith both the
amount of daylight that old opal diffuser light pipe delivers and the way the pipe distributed the internal
daylight.
The Dubai project henceshowsthat, for siteswhere sunlight are abundant,the application of light pipe
with old opal diffuser can not only bring sufficient daylight into buildings but also ensurethe
uniformity of the internal daylight distribution, which is very desirable.However, for siteswhere the
weatheris dominantly part-overcastor medium overcastsky, the influence of light pipe diffusers to the
revealedthat the position and size of the ring of light due to light pipe varies during the day. By simply
physical analysis,it was found that the presentationof the ring pattern of light due to clear diffuser
light pipe is dependenton sky condition, externalsunlight intensity, the geometrical factors of light
pipe tubes,the optical property of light pipe diffusers and the sun's position. Although by applying the
Diffuser Factor (fD) of value of 2.9 to DPF models for old opal diffuser light pipes, an averaged
evaluation of the performanceof clear diffuser light pipes can be obtained,consideringthe complex
117
decisive factors that affect the performance,the evaluationcan only be regardedas a rough estimation.
The Daylight Factor (DF) is the ratio of the internal illuminance to the external diffuse illuminance,
available simultaneously. The Daylight Factor is divided into three components,the direct skylight
(sky component),SC, the externally reflected component,ERC, and the internally reflected component,
IRC. It must be bome in mind that by far the SC is the dominant componentof DF. In presentanalysis
only the SC is consideredas the contributing factor for the make-up of DF.
The CIBSE Lighting Code suggeststhat when the averageDF exceeds5 per cent in a building, which
is usedmainly during the day, electricity consumptionfor lighting shouldbe too small to justify
averageDF is between2 and 5 per cent, the electric lighting shouldbe plannedto take full advantageof
provide the generallighting. When the averageDF is below 2 per cent, supplementaryelectric lighting
will be neededalmostpermanently.
DPF is the ratio of the internal illuminance to the corresponding total external illuminance. Hence when
comparing the performance of windows and light pipes the above fundamental difference ought to be
home in mind, i. e. in the main, the former exploits only the sky-diffuse, whereas the latter exploits total
(sky-diffuse + solar beam) illuminance. In this analysis an attempt is made to relate the window DF to
its counterpart, i. e. an equivalent DF for light pipes with the view to compare the performance of the
For UK sites,Muneer and Saluja [12] and Muneer [13] havepresentedthe relationship betweenthe
observedhourly diffuse ratio (sky-diffuse to total (global) horizontal illuminance) data.A validated
118
obvious that as the sky clarity reduceswith increasingcloud cover, the difftise ratio would tend to
become 1. Under such conditions a light pipe with an areaequal to that of a vertical window would
receiveat least double amountof illuminance due to its exposureto the whole sky hemisphere.As the
sky clarity increases,the light pipe would gain in performanceon two counts- whole sky exposureplus
receipt of beam illuminance. Note that with increasingsky clarity the difftise ratio decreasesand hence
vertical windows would only receivea fraction of a light pipe illuminance. Table 6.6 showsthe ratio of
the internal illuminance achievedby a light pipe with a DPF of 1% to that achievedby a window with a
Due to their very nature light pipes have exposureto the whole sky hemisphere(27Esteradiansolid
angle) and thus have an advantageover their vertical counterparts,i. e. windows within vertical facades.
The latter can, at best, have exposureto Tcsteradiansolid angle.Furthermore,windows are most often
not in receipt of sunlight, whereasthe light pipe can exploit both componentsof daylight - sunlight and
skylight. Table 6.6 is basedon the assumptionthat the window and the light pipe have equalareas.In
Prior to the analysisof cost and value analysisof light pipes, it is essentialto give definitions of 'cost'
and 'value' in this context. Cost is the price paid for a thing, object, service or utility. Value is the
worth, desirability of a thing, object, serviceor utility or the qualities on which thesedepend.In most
casesvalue can at best be estimatedin an approximatemanner.On the other hand, cost is a much more
accuratemeasure.In this sectiona discussionon the cost and value of daylight deliveredby light pipes
is presented.The material on both cost and value ought to be taken as an indicative assessmentowing
to the infancy of this technology.Using light pipes in buildings as daylighting devicescan bring multi-
fold benefit. The application of light pipes producesgood value in terms of energyconservation,
119
6.8.1 Energy conservation
in 1.8 million homes- one in every 13 in the UK - risk being inundatedby rising seasand
Britain [15].
0 An economiccost comparisonis to be carried out for light pipes and electrical lighting for an
Condition:
Location: London
Number of fluorescentlamps=5
Environmental impacts
120
Additional Envirom-nentallevy on electricity use = 15%
Light pipes
Required light pipe dimensions(2-off): 1.2m long with 0.45m diameter [20]
Annual C02 Penalty = E8.3 (see UK Emissions Trading Scheme, Section 6.8.3)
Total ;zýE170/year
Conclusions:
Cost comparison:The costsfor CaseI (only electrical lighting) and CaseII (electrical
lighting with light pipe) are at the samelevel with a minute difference of E4/year;
Emission saved per hour due to the use of light pipes (Case 11):
121
C02 emission = 177 g
6.8.2 Health
Researchhas shown that daylight has an important bearing on the humanbrain chemistry.Light
enteringvia eyesstimulatesthe nerve centreswithin brain that controls daily rhythms and moods.
" People prefer environments with daylight conditions [2 1], and may recover from operations
and illness more quickly in environments which are dayIit, and which afford an exterior view
[22].
" Buildings with low daylight factor create environments with homogenous lighting, having
little contrast and holding limited interest for the occupant, whereas buildings with high
daylight factor transmit more quality daylight, creating conditions likened to those found
" Typically people who are exposed total daylight levels of greater than 2000 lux for only 90
minutes eachday [24] show positive mood symptoms.Light exposureis important to the inner
time keeping of humans.Through evolution, man has adaptedto rhythms such as body
temperatureto provide him with explicit knowledgeof externaltime [25]. The loss of this
" It was concludedthat light is an important factor in the wellbeing of building occupantsand
" Researchwas carried out to reveal the relationshipbetweenSAD and light levels. Seasonal
bright artificial light of higher intensity than is usually found in the home or in the work place.
122
An open trial of bright environmentallight reversedmany of thesesymptomsand improved
" The averagepersonreceiving more than 1000lux from natural daylight for lessthan one hour
per day, is not receiving sufficient levels to maintain optimal mood. A typical office worker
could spend50% or more of their time in environmentsof 0.1 - 100 lux [29).
which are pleasingto the eye, comfortablefor the occupant,and which do not limit work
productivity [22].
" Daylight, a full-spectrum light source,most closely matchesthe visual responsethat, through
evolution, humanshave cometo comparewith all other light. Daylight provides continually
resultedin better attendance- 3.5 fewer days-absenta year. Studentsin ftill-spectrum rooms
had better dental records- nine times lesstooth decay;studentsin full-spectrum classrooms
grew more - more than 20-mm in two years;daylit libraries had significantly lessnoise; a
to productivity becausethey ignore the requirementsof thosewho work in them. Office layout
can be improved by consideringchangesto office furniture, lighting and noise control [3 1].
studentsin classroomswith the most daylight showeda 21% improvementin learning rates
bias analysisfound no assignmentbias that might have skewedthe original results.3) A grade
level analysisfound that the daylighting effect doesnot vary by grade.4) An absenteeism
123
analysisfound that physical classroomcharacteristics(daylighting, operablewindows, air
findings may have important implications for the designof schoolsand other buildings [32].
sale.A number of mechanismsthat can explain the effect of daylighting on salehave beenput
study on the effect of daylighting to students'performancein school was also carried out the
illumination levels; ii) improved visibility due to improved light quality, iii) Improved health,
the costsassociatedwith tubular daylighting and electrical lighting devices.It would be desirableto
translatethe itemisedvalue of daylight into costsso that a 'complete' picture may emerge.Towards
this end the environmentalimpact of using electric light havebeenexpressedin monetarycost via the
by UK government[35]. However, not all of the associatedvalue of daylight can easily be converted
into cost data.For example,only rough cost estimatesmay be provided for the health and productivity
presentdocumentto 'go all the way'. The authorsfeel that such value-to-costconversionsare best left
for the judgement of the reader.The analysispresentedare however conduciveto the researchon the
6.9 SUMMARY
Mathematical modelling activities aiming at predicting the dayligbting performance achievable by light
pipes with various configurations under all weather conditions have been undertaken. Two DPF
models, one for straight light pipes and the other for elbowed light pipes were built. The models enable
124
estimationof daylight provision of the light pipes with a high degreeof accuracy,i. e. R2values of 0.95
abovemodels.The maximum Mean Bias Error (MBE) and Root Mean SquareError (RMSE) were -
125___
REFERENCES
1. Edmonds,I. R., Moore, G. I., Smith, G. B. and Swift, P. D. (1995) Daylighting enhancementwith
35
2. Siegel,R. and Howell, J. R. (1981) Thermalradiation heat transfer, McGraw Hill, New York
3. Muneer, T., Abodahab,N., Weir, G. and Kubie, J. (2000) Windowsin buildings thermal, acoustic,
4. Zhang, X. and Muneer, T. (2000) A MathematicalModel for the Performanceof Light-pipes, Light
Researchand Technology,32(3)
5. Zhang, X. and Muneer, T. (2000) Light pipes for Daylight PenetrationIndoors, Proceedingsof
32-3
8. The Society of Light and Lighting (2001) CIBSELighting Guide 11: SurfaceReflectanceand
9. Muneer, T. (1997) Solar radiation & daylight modelsfor the energyefficient designof buildings,
Architectural Press,Oxford
Building ResearchInstitute
11. Muneer, T. (2002) Brief report on the performanceof Napier's Daylight PenetrationFactor (DPF)
12. Muneer, T. and Saluja, G. S. (1986) Correlation betweenhourly diffuse and global irradiancefor
13. Muneer, T. (1987) Hourly diffuse and global solar irradiation, further correlation Building Serv.
14. Zeguers,J. D. M. and Jacobs,M. J. M. (1997) Energy saving and the perception of visual comfort,
ProceedingsofLux Europa
126
15. The Independent(2002), The Independent15 September2002 No.658, UK
17. Glenigan(1998) Griffiths Electrical Installations Price Book, GleniganCost Information Service,
Bournemouth,UK
conference,234-239
22. Loe, D. and Davidson,P. (1996) A holistic approachto lighting design,Energy Management,
Sep/Oct 16-18
23. Cawthome,D. (1991) Buildings, lighting and the biological clock, The Martin Centrefor
24. Savides,T. J., Messin, C., Senger,C. (1986) Natural light exposureof young adults,Physical
Behavior, 38,571-574
U.S.A
26. Steemers,K. (1993) The role of lighting in the environmentalperformanceof buildings Facilities,
11(5)
27. Plympton, P., Conway, S., Epstein,K. (2000) Daylighting in schools:Improving student
of the Syndromeand it's Public Health Implications, American Journal ofPublic Health, 77(l), 57-60
29. Espiritu, R. C. (1994) Low illumination experiencedby SanDeigo adults: associationwith atypical
127
30. ADE - Alberta Departmentof Education(1992) Study into the effects of Light on Children of
ht!p://www. huvco.com/studies.htm
33. HeschongMahone Group (I 999b) Skylighting and retail sales:An investigation into the
128
CIN
.m "a
W t4 ý4
4ý "o -0
2 E
cu (D
= 9.1
.=
q c 10 0
En
C/)
cd
0
U C'n cn
3M
,0EZ U
5 (D r.
cu
U
ci
r.
*E - E5
- v ý
"C
m
-2 r. 0
2 0. ýý 12 -0 E
.0 -0
-
-
"Ci n4
, ( ý) eu ýc5
U2
cn = (1)
>
Ld
=t.
En Z ;6
CD
rn
rn
w ui
cri
rA CZ.(1
CIS ce
r. M ce
Cli cu
;3
(M
E
22
ce cn
V)
cl
Cd
CL4
cn Z
:=I u
(n
"zi
Z
rn
=cu. cu
- -0 2 ljz
a)
c2.
"Ci E: Cd b = 0 .
-ýg 0.) CD- lý " ZS
- C,2
-ýý :i ý
CD JZm
- -g0
Ln 5 ý- Ln-r,
0 cu
=E
.
ze
(L)
m
-= ý
;ýi cz
0.6
0.5
0.3
LM 0.2
.
>M,
0.1
0
0 50 100 150 200 250
Distance D, cm
0.6
0.5 4
*
I $ *
$ +
I
0.4 $
I$
*
I I2 1 1
*
. *
1
0.2
En
.
0.1
0 I I II II I
40 41 42 43 44 45 46 47 48 49
Solar altitude, Degree
Figure 6.3 Daylight penetration factor variation as a function of solar altitude (D=155cm)
131
0.60
'Ile
0
0.50 ++
4a
0 *+*u0++
0.40 it gv
0
0.30
0.20
0.10
0.004-
0.00 0.05 0.10 0.16 0.20 0.25 0.30 0.36 Lj. -tw I
Figure 6.4 Daylight penetration factor as a function of sky clearness index, k, (D = 155cm; solar
altitude = 49±0.5, light pipe sees only Northern part of sky vault)
600
600
cu
400
300
200
100
0
0 100 200 300 400 500 600
Measured internal illuminance, lux
Figure 6.5 Plot of calculated internal illuminance against measured internal illuminance for all
distances
132
500
400
300
L- 200
.
0 100
133
Figure 6.8 Scatter plot of calculated due to Eq. 6.9 (Y-axis) against measured internal
illuminance (X-axis) for straight light pipes (units = lux)
Figure 6.9 Scatter plot of calculated due to Eq. 6.11 (Y-axis) against measured internal
illuminance (X-axis) for elbowed light pipes (units = lux)
134
Figure 6.10 Scatter plot of calculated due to Eq. 6.13 (Y-axis) against measured internal
illuminance (X-axis) for straight light pipes (units = lux)
135
(a)
Effect of solar altitude on S-DPF
1.8
1.6
14
. kt=Ol
12 .
se . E3 kt=0.2
U: 1.0 kt=0.3
a. A
C? 0.8
3E kt=0.5
0.6 kt=0.7
0.4
0.2
0.0
15 20 25 30 35 40 45 50 55
Solar altitude, degree
1.4 -
1.2 -
1.0 - o kt=O.1
kt=0.2
0.8
kt=0.3
0.6 kt=0.5
U.J kt=0.7
0.4
0.2
0.0
20 25 30 35 40 45 50 55 60
Solar altitude, degree
Figure 6.11 DPF for 420mm-diameter light pipe as a function of a, and kt, (a) for straight light
pipe and (b) for light pipe with two bends (D=H=1.2m)
136
2.0 -
1.8 I-'O-a.
= 30'- a, = 45' --hr- a, = 60'
1.6 -
1.4
;P. 1.2 1
LZ
1.0 -
0.8
0.6-
0.4 -
0.2 -
0.0
24679
Aspect ratio
1.0 -
0.9- a, 30" aS 45' --A-- cc'= 600
i
0.8 -
0.7 i
0.6 -
U: 0.5
CL -
0.4
0.3 -
0.2
0.1 -
0.0 -
36 8 10 12
Aspect ratio
Figure 6.12 Effect of aspect ratio on DPF, (a) for straight light pipe and (b) for light pipe with one
bend (420mm-diameter, D=H=1.2m)
137
Predicted IDPFsfor straight and elbowed light-pipes
25 L
:::
2.0
nobend
1 5 (a one-bend
LL: . - - E3 E3 'S two-bend
a-
x three-bend
1.0 A -four-bend
X
0.5 -
0.0
0.1 0.2 0.3 0.4 0.5
kt
Figure 6.13 Comparison of the predicted DPFs for straight light pipe and elbowed light pipes
with one, two, three and four bends (530mm-diameter, cc,
=45", D=1.2m, H=Im)
100000
10000-
1000-
100
10 11-1.1-
Internalilluminance at 1.76mfromdiffuseron29thMay,2001.
measured
1
789 10 11 12 13 14 15 16 17
Hour
diffuser,lux
Flattranslucent - Domeopaldiffuser,lux -External global,lux
Figure 6.14 Comparison between external and internal illuminance due to different diffusers
138
AA
"4%.
Figure 6.15 The comparison of an old opal against a clear diffuser (transparency property)
139
(;I) I i'1111Id ;' poillf li"'111Solirct, f ellected from the clear diffuser inner surface
(1)) Light Of a point liýýItI -iilýt 1,11,(k dh" II It It, hIII-I lllllý 1' 141-t
Figure 6.16 The comparison of an old opal against a clear diffuser (reflectivity property)
140
6.17 Ring "pools of light" phenomenon seen in Bahrain light pipe project
Figure pattern and
141
Table 6.lCoefficients used in Eqs. 6.5 & 6.6
Table 6.2 Validation results for the model represented by Eq. 6.5
Table 6.3 Validation results for the model represented by Eq. 6.6
Models ao a, a2 a3 a4 a5 a6 a7
a8 ag alo
Eq. 6.9 356.7 -572.4 42.8 0.5 -0.9 137.7 3.5 0.5
-1.2 -10.2 -0.5
Eq. 6.10* 62.5 2.6 136.0 4.3 1.1
-17.2 -0.4
Eq. 6.11 305.0 0.2 0.2 133.8 4.0 7.1
-190.5 -2.9 -5.3 -5.8 -2.1
Eq. 6.12** 192.5 132.4 4.4 8.6
-108.8 -0.3 -2.6
Eq. 6.13 356.7 -572.4 -1.2 -10.2 42.8 0.5 -0.9 137.7 3.5 0.5
-0.5
142
Table 6.5 Error statistics for Eqs. 6.9 - 6.13
Models MBE RMSE RMSE/MaXnicasured X RMSE/MeanmeasuredI PAD Slope -0-
Eq. 6.9 -Ilux 27lux 2% 15% 12% 0.98 0.95
Eq. 6.10* -3lux 29lux 2% 16% 13% 0.98 0.95
Eq. 6.11 -2lux 23lux 2% 22% 25% 0.97 0.97
Eq. 6.12" -3lux 25lux 3% 24% 25% 0.97 0.97
Eq. 6.13 -91ux 52lux 18% 29% 25% 0.94 0.86
The maximum value of measured internal illuminance due to light pipes, lux
xMaXmCasured:
Ncan--,
d: The mean value of measured internal illuminance due to light pipes, lux
Equation6.10: the simplified model of Eq. 6.9
Equation6.12 : the simplified model of Eq. 6.11
Table 6.6 Relationship between a 'conventional' window daylight factor and an equivalent
daylight factor for a light pipe
Sky type Sky ClaritylRatio of external global IDF DPF (%) Ratio of internal
to diffuse illuminance illuminance achieved
by a light pipe to that
by a window
Heavy overcast 0.2 1.00 1 2.0
Thin overcast 0.3 1.06 1 2.1
Part overcast 0.51 1.46ý 1 2.9
Clear 0.7 2.80 1 5.6
143
7. SHADOW BAND DIFFUSE MEASUREMENTS CORRECTION
irradianceare the threemost important quantities.Global horizontal irradiance can be easily measured
applicationslight pipes are often installed on slope roofs where the total input illuminance is not
horizontal global illuminance. For the latter case,true slope illuminance hasto be calculatedso as to
predict the performanceof light pipes. Slope illuminance has two major name
components, direct sun
carried out throughout the world, thereforethe method for calculating slope illuminance from
horizontal illuminance has to be used for light pipe design.In this chapter,a new method of obtaining
more accuratehorizontal diffuse illuminance than the prevailing methodbeing used aroundis reported.
7.1 BACKGROUND
Global (G) and diffuse (D) horizontal irradianceand beamnormal irradiance(B,,) are the three most
plays in a solar systemdesigncannotbe overemphasised.It was reportedby Drummond [1] and others
that the ratio of diffuse componentto the global irradiancecan be ashigh as 0.65. Therefore,to provide
D=G-B,, sina
Therefore, diffuse irradiance can either be calculated from measurement of G and B. or measured
directly. However, in the first instance, the beam normal irradiance measurement device
144
(pyrheliometer)hasto be installed on an equatorialmount with active tracking of the sun'strail. As this
Unfortunately, the shadowband that is usedto block the sunshinealso shadessomediffuse iffadiance
as well. It can lead to the monthly averagederror of horizontal diffuse iffadiance measurementof up to
24% of the true value [1]. Henceit is necessaryto correct the measureddiffuse iffadiance obtained
The inner surfaceof a shadowband is usually paintedblack so as to eliminate any reflected irradiance
from the inner surfaceto the pyranometer.If the sky-diffuse irradiancedistribution is assumedto be
isotropic and the irradiancethat is reflectedby inner surfaceof shadowband negligible, the correction
factor of the diffuse irradiancecan be calculatedby using the simple geometricmethod given by
Drummond [I]. This geometricmethod calculatesthe proportion of the sky areathat is subtended.
by
However, as the real sky-diffuse distribution is not isotropic, an anisotropic factor must be taken into
correction factor varies from 2.9% to 20.9%, whereasthe geometric(isotropic) correction factor varies
only from 5.6% to 14.0%.Painter [3], comparedthe diffuse irradiancemeasuredby both an occulting
disc and a shadowband pyranometer.He concludedthat the anisotropic correction factor has a
significant seasonaldeviation. Eero'sstudy [4] showedthat the deviation of the monthly isotropic
correction factor with respectto total correction factor rangesfrom 1% for Decemberto 7.5% for
August.
undertaken by several investigators during the past decade. Ineichen [5] reported his method of
combining two simple models to determine the total correction factor: one was the isotropic geometric
145
model and the other was basedon the diffuse radiation as a function of solar altitude. Kasten et al[6]
correction factor, namely: the ratio of the diffuse to global radiation, the solar declination and the
componentsky radiation model, showedthat the total correction factor appearedto be a function of the
shadowband geometry,location of the shadowband, and the sky radiation distribution. LeBaron et al.
[8] presenteda model employing the parameterschememethod put forward by Perez.This parameter
schemeis representedas a 256-combinationof the values of the zenith angleof sun, the isotropic
The common characterof the aboveanisotropicmethodsis that they employ different radiation
distribution modelsto obtain the ratio of diffuse irradiancesubtendedby shadowband to the total
diffuse irradiance.However, the use of a complex function and parameterschemewithin such a model
makesit more difficult to apply the model in practise.The purposeof this article is to develop a
relatively simple yet accurateanisotropicmodel that can be applied widely acrossthe global.
In this article, the developmentof a new model basedon the radiancedistribution index formulation
radiancedistribution index and sky clearnessindex, k,, is reported.The model is validatedusing data
and Drummond's isotropic model is illustrated through the use of error histogramsand statistical
7.2.1 Theory
Drummond [I] gave the following formula based on geometrical analysis to calculate the isotropic
146
FD ý 1/0-0
(7.2)
where FD is the correction factor and f is the proportion of diffuse irradiance obstructed by the shadow
band. Note that FDis to be multiplied with the uncorrectedvalue of diffuse irradianceas measuredwith
f= X/T =2(Whi) cos' 5 (to sing sin5 + cosy cos5 sinto) (7.3)
where X is the diffuse irradianceshadedby shadowband,T is the total diffuse irradiance.Eq. 7.3 is
derived from the integration of Eqs. 7.4 and 7.5 below, basedon the assumptionthat the sky-diffuse
radiancedistribution is uniform:
3,5 f I=sunset
X=IW Cos sinadt (7.4)
1=sunrise
T=7c 1 (7.5)
where I is the radiation intensity of the sky. However as real sky-diffuse irradiancedistribution is
anisotropic,to get a more accuratecorrection factor, the I value must be treatedas a variant. Such a
Measurementsof the global and diffuse sky irradiancealong with beamnormal irradiancewere usedto
examinethe validity of Drummond's method [1]. To reveal the inadequacyof Drummond's isotropic
method, comparisonof the correction factor rangeobtainedusing Drummond's method and the range
One complete year's data (1995) for Bracknell (51.4*N, 0.8'W) in England are used to examine
Drummond's method. The data consists of hourly horizontal global, beam normal and diffuse
147
irradiancemeasuredby shadowband. Hourly irradianceon vertical surfacesfacing north, east,south,
west and a tilted surfacefacing south at the local latitude anglewere also available.Thesedata were
control conditions:
G- Bn sina >D
uncorrected;
a> 10 degree
W/M2.A total of 3583 hours datawere thus chosento examineDrummond's correction factor against
F = (G - Bý,sin(y)/I)uncorrected (7.6)
true correction factor rangesftom 1.001 to 1.480.It is thereforeconfirmed that, in the caseof hourly
24%.
148
7.3 NEW MODEL BASED ON SKY-DIFFUSE DISTRIBUTION INDEX
facilitate the present modelling exercise. The model is given in Eq. 7.7,
Lo = Lz (I +b sinO)/ (I + b) (7.7)
,
where LOis the radianceof a given sky patch with an altitude = 0, L, is the zenith radianceand b is the
radiancedistribution index.
where Dp is the hourly diffuse irradiance for a slopedsurface,D is the hourly horizontal diffuse
irradiance,P is the tilt of the slopedsurface.Muneer'smodel also treatsthe shadedand sunlit surfaces
for Bracknell, Muneer found that for a shadedsurface(facing away from sun) the 'best'value of b is
5.73; for a sunlit surfaceunder overcastsky conditionsb=1.68; and for a sunlit surfaceunder non-
Eq. 7.9,
For k, >0.2,
149
2b,
= 0.382- 1.11kt (for southemhalf of sky hemisphere) (7.9a)
g(3 + 2b,)
2b2
= 0.166 + 0.105 kt (for northem half of sky hemisphere) (7.9b)
T(3 + 2b2)
,
A point worth mentioning here is that, strictly speaking the sky radiance distribution is two-
dimensional function of any given sky patch geometry (sky patch altitude and azimuth) and the
,a
position of sun. Eq. 7.9 represents a compromise between simplicity (represented by an isotropic
model) and complexity (represented by a two-dimensional model such as the one described above).
Another feature of Eq. 7.9 is that due to its relative simpler formulation it also tends to be robust
(covering all-sky conditions) and of general applicability, as shall be demonstrated herein via validation
undertaken for disparate sites, i. e. Bracknell, England (51.4'N, 0.8'W) and Beer Sheva, Israel (31.3'N,
34.8'E).
The total sky-diffuse irradiancecan be obtainedthrough a numerical integration basedon Moon and
Spencer'ssky-diffuse distribution model. If the zenith radianceis termed Lz, then the horizontal diffuse
r/2
D=f; (7r sinO cosO)(LO,+ LONO (7.10)
0
radiance emanating from any given patch of the southern half of the sky vault, and L02the
2
D= 7cLzf[ (sinO cosO+blsinO COS2())/(I+bl)dO]+
2
(sinO cosO+b2sinOcos20)/( I+b2)dO]) (7.11)
150
So,
Moon and Spencer[9] showedthat using bl=b2=2 in Eq. 7.12 adequatelydescribesthe radiance
If FAis defined as the diffuse irradiancethat is obscuredby shadowband, then FAmay be calculated
3,5 f I=sunsel
FA: --WCOS Lo sinadt
I= sunrise I
12ý tOsin2cpsin'5 + 2sinto sinT cosy sin5 cos5 + cos2T Cos25 [ to/2 +(sin2to)/4]
Now, using Eqs.7.12 and 7.14, the proposedshadowband diffuse irradiancecorrection factor, Fp, can
be obtained,
Fp = 1/(I-FA / D) (7.15)
7.3.2 Validation
The validity of the proposed model was tested by plotting the corrected hourly diffuse irradiance
against the true diffuse irradiance as obtained via Eq. 7.1. Databases of two different sites, namely:
151
Bracknell, England and Beer Sheva, Israel were used to validate the proposed model. These two sites
The proposedmodel was applied to the culled Bracknell databasewhich was previously used to
(Dcorrcacd)
datawere regressedagainsttrue values of 'measured' diffuse irradiance(D) obtainedvia
Eq. 7.1 to determinethe model's performance.Resultsof this comparisonare shown in Fig. 7.2. The
slope of the fitted trend line is 0.99, and the coefficient of determination,R2,is 0.98. Drummond's
method was also applied to this database.The result is shown in Fig. 7.2. The slope of the fitted trend
line is 0.95, and the coefficient of determinationR2is 0.97. The slope of the fitted trend line is
indicative of the validity of the model under test. The degreeof validity increasesas the slope
To enablefurther insight into the validation of the abovemodelsthe meanbias error (MBE) the root
,
meansquareerror (RMSE) and the PercentageAverage Deviation (PAD) have beenobtained.Results
Y,(D,,,,,,
-D)
Mean Bias Error, MBE = Ied
no.of data points
ýE(D
D)2
Root Mean Square Error, RMSE = .......
no. of data points
E(I 00 *1 D,,,,,,,, I/ D,
d -D orrected
Percentage Average Deviation, PAD =
no. of data points
152
7.3.2.1.2 Histograms
distribution of the percentageerror. A given models'perfori-nanceis examinedin two ways. Firstly, the
histogram provides a check regardingthe proportion of datapoints that fall within specific range of
Histogramsof the proposedmodel for all-sky conditions, and for 0< kt:5 0.2 (heavy overcast),
0.2 < k,: 5 0.6 (part-overcast)and 0.6 < k, <I (clear-sky) are shown in Fig. 7.3 respectively.
Correspondinghistogramsfor Drummond's method are given in Fig. 7.4. The results of theseanalyses
arepresentedin Table 7.2 as the number of datapoints that fall in percentageerror rangesof -3%
I %, -1% - +1%, and +1% - +3%. For the proposedmodel and for all-sky conditions, 70% of
correcteddiffuse value fall in the -3% - +3% percentageerror range,while for Drummond's method
the correspondingfigure is 58%. Under clear-sky condition (0.6 < k, <1), when solar energyhas the
displaysa normal distribution for errors, while Drummond's model is heavily skewed,especiallyunder
clear-sky conditions.
34.8'E) in Israel were used for further evaluationof the modelsunder discussion.The dataconsist of
provided by the Chemical EngineeringDepartmentat Ben-Gurion University. Again, the data setswere
153
Diffuse irradiancedatarespectivelycorrectedvia proposedmodel and Drummond's method,were
The slope of the fitted trend line for proposedmethodis 1.00,and the coefficient of determination,R2
is 0.99. The result obtainedfrom the applicationof Drummond's methodis shown in Fig. 7.5. The
slopeof the fitted trend line in this caseis 0.96, and the coefficient of determinationR2is 0.99. Table
7.3 showsthe MBE, RMSE and PAD statisticsfor Beer Shevadatabase.The proposedmodel produces
7.3.2.2.2Histograms
Figure 7.6 and 7.7 show the histogramsfor different sky conditionsobtainedthrough applicationof the
in Table 7.4. For all-sky conditions,41% of diffuse valuescorrectedby the proposedmethodfall in the
is considerablyskewed,especiallyunder clear-skyconditions.
7.4 SUMMARY
This method is basedon the use of a single diffuse radiancedistribution index, b, introducedafter the
work of Moon and Spencer[9]. Clearnessindex, kt, is usedas a parameterto determinethe value of b,
and b is then usedto obtain the correction factor to accountfor the diffuse irradianceobscuredby
for the caseof Bracknell, England the RMSE MBE and PAD of correcteddiffuse irradianceobtained
,
MBE and PAD due to Drummond's method are 16 W/M2, -5.6 W/M2 and 4% respectively.For the case
Beer Sheva,Israel, the proposedmodel producesan RMSE of 17 W/m2, an MBE of 2W/M2 and a
of
154
PAD of 5%. The correspondingfigures for Drummond's methodare 23 W/mý, -5W/m2and 7%
respectively.
155
REFERENCES
413-436
2. Kudish, A. I. and Ianetz,A. (1993) Analysis of Diffuse RadiationData for Beer Sheva:Measured
Energy, 26,361
IrradianceMeasurementsRenewableEnergy, 17,311-317
5. Ineichen,P., Gremaud,J. M., Guisan,0. and Mermoud, A. (1984) Study of the correctivefactor
involved when measuringthe diffuse solar radiation by using the ring methodSolar Energy, 32,585
7. Siren, K. E. (1987) The shadowband correction for diffuse irradiation basedon a two-component
8. LeBaron, B. A., Michalsky, J. J. and Perez,R. (1990) A simple procedurefor correcting shadow
9. Moon, P. and Spencer,D. E. (1942) Illumination from a non-uniform sky Trans.Illum. Eng. Soc.,
37,707-725
10. Muneer, T. (1990) Solar Radiation Model for EuropeBuilding Serv. Eng. Res. Technol.,11(4), 153-
163
156
I
.0
cu
'a
. c2
cu
Co
(b) Actualresults
Figure 7.1 Comparison of the range of diffuse irradiance correction factor given by Drummond's
method and actual results
157
600
E 500
400
-- -Z
6-
.7-.
A: 300 =, 7 -.
CL)
200
ts 100
0
u
jv
0 100 300 400 500 600- 1
200
600
i:z
500
400
A
.. .. ap A7
300
4
ým 200
lu m
100
0
o-
0 100 200 400
300 500 600
2)
true diffuse irradiance (Wlm
Figure 7.2 Scatter plot of true diffuse irradiance versus corrected irradiance using the proposed
(top) and Drummond's model (bottom) -Bracknell data
158
1
(a)
1
1
(D C)
.
0
CL
0
0.0 2.0 4.0 6.0 8.0 10.0 0.0 1.0 2.0 3.0 4.0 5.0
-10.0 -8.0 -6.0 -4.0 -2.0 -5.0 -4.0 -3.0 -2.0 _1.0
1201
1001
ý5
m -Q
E 80
(I,
C
0
a- 60
CD
40
20
l
-10.0 -8.0 -6.0 -4.0 -2.0 0.0 2.0 4-0 6.0 8.0 10.0
o -10.0 -8.0 -6.0 -4.0 -2.0 0.0 2.0
percentage error, %
4.0 6.0 8.0 10.0
percentage error, %
Figure 7.3 Proposed model's error histogram under all-sky (a), heavy overcast (0 < kt: 5 0.2) (b),
part-overcast (0.2 < kt: 5 0.6) (c) and clear-sky (0.6 < kt <1) (d) conditions - Bracknell data
159
600
(b)
500
400
m
E
2 300
S
CL 0
CL
n
(Z
200
100
0
0.0 2.0 4.0 6.0 8.0 10.0 -5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.
-10.0 -8.0 -6.0 -4.0 -2.0
600
(C) (d)
500
Q)
400
-0
300
CL 0.
m
200
100
20
oL
J.U -t$.U -b.U -4.U -Z.U U.U Z.U 4.U b.U b.U IU.U 0.0 2.0 4.0 6.0 8.0 10.0
-10.0 -8.0 -6.0 -4.0 -2.0
Figure 7.4 Drummond model's error histogram under all-sky (a), heavy overcast (0 < kt: 5 0.2)
(b), part-overcast (0.2 < kt: 5 0.6) (c) and clear-sky (0.6 < kt <1) (d) conditions Bracknell data
-
160
600
* +
ýz 500 4,
E+*
400
a) 300
LA
e
200
ts
U 100 -
o
0 100 200 300 400 500 600
True diffuse irradiance (W/M2)
600
C14 k
E 500
LV
400
V
300 All
M 200
v 100
0 100
200 300 400 500 600
True diffuse irradiance (Mm)
Figure 7.5 Scatter plot of true diffuse irradiance versus corrected irradiance using the proposed
(top) and Drummond's model (bottom) - Beer Sheva data
161
(a) (b)
k)
.0
.o
0
CL
2
ca
'a
-14.0 -10.0 -6.0 -2.0 2.0 6.0 10.0 14.0 -12.0 -8.0 -4.0 0.0 4.0 8.0 12.0
-12.0 -8.0 -4.0 0.0 4.0 8.0 12.0 -10.0 -6.0 -2.0 2.0 6.0 10.0
(c) (d)
ci)
-a
E
(I)
C
0
0
-0 0
101
-12.0 -8.0 -4.0 0.0 4.0 8.0 12.0 -12.0 -8.0 -4.0 0.0 4.0 8.0 12.0
-10.0 -6.0 -2.0 2.0 6.0 10.0 -10.0 -6.0 -2.0 2.0 6.0 10.0
percentageerror, % percentageerror, %
Figure 7.6 Proposed model's error histogram under all-sky (a), heavy overcast (0 < kt: ý 0.2) (b),
part-overcast (0.2 < kt: 5 0.6) (c) and clear-sky (0.6 < kt <1) (d) conditions - Beer Sheva data
162
(a) (b)
a)
m
I O(A .0
E
U)
0
im. 0.
2 C',
cu C',
oL
-14.0
-12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0 -12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0 12.0
10.0
(c) (d)
401
. im
Co
0
CL
a
cc
'a
I oi
oL-12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
12.0 -12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
12.0
percentageerror, % percentageerror, %
Figure 7.7 Drummond model's error histogram under all-sky (a), heavy overcast (0 < kt: 5 0.2)
(b), part-overcast (0.2 < kt: 5 0.6) (c) and clear-sky (0.6 < k, <1) (d) conditions - Beer Sheva data
163
Table 7.1 Mean Bias Error (MBE), Root Mean Square Error (RMSE) and Percentage Average
Deviation (PAD) comparison of the proposed model versus Drummond's method, Bracknell
All-sky conditions 0<k,: 5 0.2 0.2 < k,:5 0.6 0.6 < k, <I
MBE RMSE PAD MBE RMSE PAD MBE RMSE PAD MBE RMSE PAD
(W/mI) (W/mI) (%) (W/ml) (W/ml) (%) (W/mI) (W/mI) (W/ml) (W/ml) (%)
Prop-
osed 12.0 2.8 0.4 1.6 0.7 15.1 3.9 12.4 3.4
Model -0.7 -0.3 -5.2
Drum-
mond 15.7 4.2 0.0 1.4 0.7 16.7 4.9 27.4 10.7
Model -5.6 -4.8 -24.5
Table 7.2 Histogram percentage error analysis results, Bracknell. Figures given below are the
number of data points in each category
All-sky conditions 0<k,: 5 0.2 0.2 < k,:!ý 0.6 0.6 < k, <1
Drummond
Model 374 1081 94 65 503 148 297 429 44 8 3 3
Table 7.3 Statistical comparison of the two models under discussion - Bracknell data.
(W/M2) MBE (W/M2)
Month RMSE Percentage Average Deviation
Proposed Drummond Proposed Drummond Proposed Drummond
Model model Model model model Model
January 4 5 2.0 2.0
-1.5 -1.6
February 6 7 1.7 1.8
-0.6 -1.0
March 9 15 3.2 5.0
-1.8 -6.5
April 12 19 3.0 5.3
-2.5 -10.0
May 18 20 4.3 3.7 5.2
-4.4
june 10 14 2.0 3.8
-0.3 -6.0
July 18 20 3.7 -3.7 3.2 4.8
August 10 19 2.9 6.7
-4.9 -14.0
September 11 16 3.0 4.4
-3.9 -8.0
October 10 13 3.8 4.7
-6.0 -7.6
November 4 4 2.2 2.1
-1.2 -1.3
December 2 2 1.5 1.7
-0.5 -0.6
164
Table 7.4 Mean Bias Error (MBE), Root Mean Square Error (RMSE) and Percentage Average
Deviation (PAD) comparison of the proposed model versus Drummond's method, Beer Sheva
All-sky conditions 0<k, :50.2 0.2 < k,: 5 0.6 0.6<k, < I
NIBE RMSE PAD MBE RMSE PAD MBE RMSE PAD MBE RMSE PAD
W/M2 (W/ml) (W/m) (0/. ) (W/m 2) (w/m2)
(W/m'L (W/m')
.
Prop-
osed 2.1 17.1 5.2 3.2 5.11 3.0 5.0 17.2 4.4 20.3 6.9
Model -0.7
Drum-
mond 23.4 7.3 4.7 8.5 7.0 0.5 18.4 5.2 30.4 9.0
Model -5.3 -15.6
I I I I
Table 7.5 Histogram percentage error analysis results, Beer Sheva. Figures given below are the
number of data points in each category
All-sky conditions 0<k,:: ý 0.2 0.2 < k,: 5 0.6 0.6 < k, <1
1 -3 -1 -1 1 -3 -1 -1 1 -3 -3-4 -1 -1 1 -3
-3-1 -1 -1 -3-4 -3-4
N N N N N N
Proposed
Model 42 110 147 2 22 78 19 56 48 22 32 21
Drummond
Model 39 58 57 2 2 5 20 43 25 18 14 27
Table 7.6 Statistical comparison of the two models under discussion - Beer Sheva data.
165
8. VALIDATION OF DPF MODELS
apply the DPF modelsto independentlymeasureddatain Merchiston test room, and comparethe
The adequacyof a fitted regressionmodel can be carried out by the methodof checkingthe residual
plots due to the model. The residual method appliesgenerally"whenever a linear model is fitted, no
matter how many predictors there are" [I]. The procedureis to produce a graph of the residual e (the
A satisfactoryresidualsplot is one that showsa (more or less)horizontal band of points giving the
impressionof Fig. 8.1. There are many possibleunsatisfactoryplots. Three typical onesappearin
Fig. 8.2. The first of thesethree(the funnel) displaysthe band of residualswidening to the right
showing non-constantvariance.The secondis a downward trend and the third is a curve. For the funnel
type of plot, it indicatesa lack of constantvarianceof the residuals.The corrective measurein this case
is a transformationof the Y variable. A plot of the residualssuch asFig. 8.2(2) indicatesthe absenceof
an independentvariable in the model under examination.If however, a plot such as Fig. 8.2(3) is
Becausethe residualsand the Y,, values are usually correlatedbut the residualsand the Y, are not.
Therefore,ej = (Y. - Yc)j, i=l, 2, n is usually plotted against(Y, ) j. A slope in the ej and (Yji
...,
indicatesthat somethingis wrong. It is possibleto evaluatetest statisticson residuals,but it is often
difficult to know if they are sufficiently deviant to require action. In practical regressionsituations,a
detailed examinationof the corresponding residualsplots is usually far more infonnative, and the plot
166
will almost certainly reveal any violations of assumption serious enough to require corrective action
Ill.
The residualsdue to the S-DPF model (Eq. 6.9) were calculatedfor eachdatapoints from Craighouse
The residualsplot of S versusYc is shown in Fig. 8.3. It can be seenin Fig. 8.3 that the residualplot
,j
showspattern of a horizontal band. This indicatesthat generally speaking,the S-DPFmodel being
examinedis adequate.Figure 8.4 showsthe histogramof the absoluteerror (residual) due to the S-DPF
model. It is shown in Fig. 8.4 that the distribution of absoluteerror is a normal distribution with a mean
value of -2.3 lux. The standarddeviation of the absoluteerror was found to be a=28.31 lux. This
implies that the S-DPFmodel can predict the daylighting performanceof straight light pipes with a
quite high accuracy,namely more than 99 per cent of all predictionsgiven by the model are found
within the error rangeof ±501ux.The meanvalue of measuredinternal illuminance was found to be
177lux, the percentageof prediction due to S-DPFmodel that fall in the error rangeof ±17.7lux (10 per
cent of meanvalue 177lux) was then found to be 75%. Consideringthat the sensitivity of human eyes
to illuminance at the level of 50lux is in a low order, it is thereforeconcludedthat the prediction given
by the S-DPF can be usedwith high confidencein practical light pipe design.
The residualplots of S versusX, and X2 are shown in Figs. 8.5 and 8.6 respectively.Figure 8.5 shows
that the residual due to S-DPF model is generally constant,independentto the changeof solar altitude.
Figure 8.6 showsa similar trend and presentsa profile of horizontal band, which confirms that the S-
DPF model is adequate.For solar energy applications,especiallyfor light pipe daylighting project,
sun's position and sky conditions are the most significant variablesto be considered.Figures 8.5 and
167
8.6 confirm that the S-DPFmodel hasadequatelydescribedthe effect of abovetwo factorsto the
daylighting performanceof light pipes. It is also worthy hereto point out that since S-DPFmodel uses
As addressedin Section5.5 that a total of 22 270 Merchiston test room datapoints for straight
model validation. Merchiston test room datawere further divided into three groupsto revealthe
performanceof the S-DPFmodel under different sky conditions.The whole data setwere divided into
sky conditions(k,>0.2), under heavy overcastsky conditions (k,:ý0.2) and under all-sky condition. The
The S-DPFmodel was applied to sevencasesof straight light pipe configurationsto examineits
performance.The sevencasesare:
Internal illuminance predictions due to S-DPF model for eachof abovesevencaseswere calculated
respectively. Scatter plot of calculated internal iluminances against measured illuminances were then
obtained to assessthe performance of S-DPF model based on independent data. The scatter plots were
168
generatedwithin Excel environment,and after that the best-fit trend lines were generatedby the Excel
to examinethe accuracyof the model. By linear regression,equationsand slopesof the best-fit lines
are presentedbelow.
Case 1: Figure 8.7 shows the scatter plot of predicted internal illuminance against measured internal
illuminance for a light pipe 0.2 1m in diameter and 0.6m in length. The vertical distance between light
pipe diffuser and the working plan was 1.68m. The distances between three testing points and the light
pipe diffuser centre ranged between 1.68 and 2 meters. 'llie sky clearness index varied between 0.2 and
0.8. The slope of the best-fit trend line was found to be 0.96 and the R-Square value was noted as 0.87.
The mean error was found to be Olux and the RMSE was noted as 10 lux. In this case it can be seen that
the prediction of S-DPF is quite good. A RMSE value of 10 lux means that at least about two thirds of
predicted data given by S-DPF is within an error band that is negligible (± I Olux).
illuminance for a light pipe 0.21m in diameterand 1.2min length. The vertical distancebetweenlight
pipe diffuser and the working plan is 1.12m.The distancesbetweenthree testing points and the light
and 0.8. The slope of the best-fit trend line was found to be 0.96 and the R-Squarevalue was noted as
0.96. The meanerror was found to be -4lux and the RMSE was noted as 9 lux. In this caseit can be
seenthat the prediction of S-DPF is also good. A RMSE value of 9 lux meansthat at least about two
thirds of predicteddatagiven by S-DPF is within an error band that is negligible (±I 8lux). The R-
Squarevalue is ashigh as 0.96 which meansthat 96% of datais describedby the S-DPFmodel. It is
found however, that for this case,the S-DPFmodel seemsto slightly underestimatethe internal
illuminance, with a mean error of -4 lux. This might be due to the fact that when the light pipe was
longer, to test the internal illuminance at distancesabout I and 2 meters,sensorswere put further near
to the comer of the test room where internal reflection played a more significant role and led to a
169
Case3: Figure 8.9 showsthe scatterplot of predictedinternal illuminance againstmeasuredinternal
illuminance for a light pipe 0.33m in diameterand 0.6m in length. The vertical distancebetweenlight
pipe diffuser and the working plan was 1.68m.The distancesbetweenthreetestingpoints and the light
pipe diffuser centrerangedbetween1.74and 2.14 meters.The slope of the best-fit trend line was found
to be 1.0 and the R-Squarevalue was noted as 0.88. The MBE was found to be -6lux and the RMSE
was noted as 55 lux. In this case,accordingto the figure of R-Square,MBE and RMSE, which are both
for this case,the averagedmeasuredinternal illuminance was found to be 200lux, while the figures for
CasesI and 2 are 44lux and 51lux respectively.The percentageof RMSE to meanmeasured
illuminance for Cases1,2 and 3 were found 23%, 18%and 27% respectively.Therefore,the valuesof
RMSEs of cases1,2 and 3 arebasically comparableand in the sameorder. It is also notedthat the
averagedexternal illuminance for the third casewas found to be 46835lux, which is 35% higher than
Case1 and 32% higher than Case2. The maximum externalilluminance recordedduring the
measurementfor the third casewas found to be ashigh as I lOklux, and correspondinglythe sky
clearnessindex monitoredduring the day rangedfrom 0.2 to 0.8. The reasonwhy the third case
producesa larger RMSE may thereforebe due to the quite dramaticvarianceof externalconditions.
illuminance for a light pipe 0.33m.in diameterand 1.2m in length. The vertical distancebetweenlight
pipe diffuser and the working plan was 0.9 in. The distancesbetweenthree testingpoints and the light
pipe diffuser centrerangedbetween 1.32and 1.71meters.The slope of the best-fit trend line was found
to be 1.04and the R-Squarevalue was noted as 0.82. The meanerror was found to be llux. and the
RMSE was noted as 55 lux. It is also noted that the averagedexternal illuminance for the third casewas
found to be 47219lux, which is 36% higher than CaseI and 33% higher than Case2. The maximum
external illuminance recordedduring the measurementfor the third casewas found to be as high as
115klux, and correspondinglythe sky clearnessindex monitored during the day rangedfrom 0.2 to 0.8.
The maximum and minimum internal illuminances were found to be 600lux and 20lux respectively.
Case 5: Figure 8.11 shows the scatter plot of predicted internal illuminance against measured internal
illuminance for a light pipe 0.45m in diameter and 0.6m in length. The vertical distance between light
170
pipe diffuser and the working plan was 1.74m. The distances between three testing points and the light
pipe diffuser centre ranged between 1.74 and 2.06 meters. The slope of the best-fit trend line was found
to be 0.93 and the R-Square value was noted as 0.74. The mean error was found to be -25lux, which
was 10% of averaged measured internal illuminance. Corresponding figures for RMSE was noted as
65lux and 26% respectively. In this case it seems that the S-DPF model underestimates the internal
illuminance.
illuminance for a light pipe 0.45m in diameterand 1.2m in length. The vertical distancebetweenlight
pipe diffuser and the working plan was 1.18m. The distancesbetweenthree testingpoints and the light
pipe diffuser centrerangedbetween 1.30and 1.62meters.'Me slope of the best-fit trend line was found
to be 1.00 and the R-Squarevalue was noted as 0.86. The MBE was found to be -17lux and the RMSE
was noted as 75 lux. The MBE was found to be 6% of the averagedinternal illuminance of 289lux, and
illuminance for a light pipe 0.53m in diameterand 0.6m in length. The vertical distancebetweenlight
pipe diffuser and the working plan was 1.18m. The distancesbetweenthree testingpoints and the light
pipe diffuser centrerangedbetween 1.30and 1.62meters.The slope of the best-fit trend line was found
to be 0.95 and the R-Squarevalue was noted as 0.82. The MBE was found to be -33lux and the RMSE
was noted as 106 lux. The MBE was found to be 8% of the averagedinternal illuminance of 403lux,
8.2.2 Performance assessment of the S-DPF model under heavy-overcast sky conditions (kt:50.2)
overcastsky conditions (k,:50.2). S-DPFmodel was applied to predict the internal illuminance values
DPF model for all heavy-overcastsky condition dataand for eachof the sevencasesare shown in Figs.
171
Figure 8.14 showsthat the slopeof the best-fit trend line for all heavy-overcastsky condition scatter
plot was found to be 1.12and the R-squarevalue was 0.94. The RMSE and MBE were found to be
Figure 8.14 shows that under heave-overcast sky conditions, generally speaking the S-DPF model tends
to overestimate the internal illuminance due to light pipes. This can be due to two main reasons.
First, the measurements undertaken in Merchiston test room was one year later than that carried out in
Craighouse test room. The very same light pipes (Monodraught) were tested in the two measurements.
During the two years' time, light pipes have been transported from sites to sites, and during the tests
they were also installed and removed quite frequently according to the test schedules. It is therefore
logical to assume that certain aging effect may have taken place during the course, which can lead to
slight changes of the reflectance of internal surface of light pipe tube. Swift et al [3] pointed out that in
the mathematical modelling of light pipe's transmittance of light, the reflectance of the internal surface
of light pipe tube plays an important role. Swift found that "the theoretical calculations are sensitive to
the choice of reflectivity, with changes in R (reflectivity, author) of 0.001 resulting in noticeable
differences in the calculated curve". This therefore suggests that the method of applying the internal
reflectivity value of 0.95 that was obtained based on Craighouse data, to Merchiston data may produce
The secondpossiblereasonthat can contribute to the overestimateis that under low illuminance levels
the measurementsby Kipp and Zonon sensorsare subjectto more errors. It was found that under heavy
overcastsky conditions, the meanvalue of measuredinternal illuminance was as low as 106lux. The
rangesof the sensorsare all 5±lklux. The sensitivitiesof the three Lux Lite sensorswere given as
10.32[tV/Iux (±I%), 9.68pV/Iux (±I%) and 10.1lgV/Iux (±I%) respectivelyby the manufacturer
(Section 5.2). Thereforeunder heavily overcastsky conditions, the measurementerror due to the
sensorscan reachashigh as ±50lux (equalsto about 1% of the range),which accountsfor almost 50%
172
However, it shall be pointed out that as an innovative device that utilizes sunlight, light pipe contributes
most to the daylighting when the sky is clear and sunlight is available.Unfortunately, when the sky is
heavy-overcast(kt:50.2), light pipes provide a lower level of daylight, though which is sufficient
carried out. Bearing this in mind, it can be seenthat the performanceof S-DPF model under heavy-
To reveal the over all performanceof S-DPFmodel, it is applied to the whole data set from Merchiston
test room. Internal illuminance predictions due to S-DPFmodel for all sevencasespresentedabove in
measuredilluminances was shown in Fig. 8.22.The slope of the best-fit trend line was found to be 1.03
and the R-Squarevalue was noted as 0.90. The MBE was found to be -5lux, the RMSE 76 lux, and the
PAD was found to be 22%. The maximum of the internal illuminance was found to be 1986lux and the
minimum found l0lux. The high R-Squarevalue of 0.90 suggeststhat the S-DPF model built using
slope value of 1.03suggeststhat the S-DPF model can estimatethe meanvaluesof internal
illuminances due to light pipes quite precisely. It is howevernoticed that for Merchiston dataset the
RMSE was found to be 76 lux, which is higher than the RMSE value of 27 due to Craighousedata.
First, the external environmentfor the two testsin Craighouseand Merchiston test rooms were not
absolutelythe same.The Merchiston test room was build on the roof of a 7- storeybuilding with an
almost completeview to the sky (more than 95% of the hemisphere).However, the view angle of the
test room in Craighouseto the sky was not the whole vault. Trees and buildings within a distanceof 20
metersblocked about 20% of the sky although the southernhemispherewas clear. Therefore,the
coefficients developedfor one site may not be the best-fit value for the other. This can be the main
173
The second main reason that may contribute to the up-rise of RMSE may be due to the different
geometry of the two test rooms. To enable an almost complete view of the sky hemisphere, Merchiston
test room was built upon the roof of a 7- storey building within Merchiston campus. However, on the
other hand it was afterwards found that the wind speed on the top of the building was so strong that the
size of the test room had to be minimized so as to meet the safety requirements. The dimension of the
Craighouse test room (3.0 x 2.4 x 2.5 m). Therefore, although the internal surfaces of the two huts
have the some colour (colour of wood), the smaller Merchiston test room caused more significant
internal reflection than that occurred within the Craighouse test room. The application of the S-DPF
model developed based on Craighouse data to the Merchiston data was thus inevitably subject to more
The third possiblefactor that may contribute to the higher RMSE value of 67lux is the asymmetrical
distribution of internal daylight due to light pipes. The S-DPFmodel assumesthat the internal
illuminance distribution on a working plan below the light pipe diffuser is symmetricalto the difftiser
centreand is a function of the vertical and total distancesof one point to the diffuser centre.From 30th
Light pipes of four configurationswere tested,which were light pipe 0.21ra in diameter and 0.6m in
length, 0.33m in diameterand 0.6m in length, 0.45m in diameterand 0.6m in length and 0.45m in
diameterand 1.2min length. Unfortunately it was found that the test for the 0.2Im-diameter light pipe
was faulty.due to the leakageof rainwater into the pipe. The rest of the testsproducedgood data and a
sum of about 900 datapoints were obtained. Resultsare shown in Tables 8.2 and 8.3. Table 8.2 shows
one day's data measuredfor the 0.33m-diameterlight pipe on 18'hDec 2000. It can be seenthat the
standarddeviationsof the five measurements(one Magetron sensorwas found faulty) taken at different
positions surroundingthe centrepoint of diffuser are averagedat a value of 86lux. For the 0.33m-
underneathlight pipe diffuser was found to be 14%.The correspondingfigures for the 0.45m-diameter
light pipes 0.6m and 1.2min length were found to be 12% and 13%respectively(Table 8.3). Therefore
174
the asymmetricaldistribution of internal illuminance althoughnot very significant, has a substantial
effect on the daylighting performanceof light pipe. It is worthy however to point out that in real
even distribution of daylight within the space.For occasionswhere multi-light pipes are installed
In this section the performance of DPF models are evaluated by comparing it against other independent
researchresults.
The first comparisonis madewith the study carried out by Loncour et al [4] on behalf of Monodraught
(a) the actual,measuredlight transmissionwithin straight run of pipe was 48% as comparedto
(c) the straight-lengthenergyloss was found to be 29% with an extra 5% loss for pipe-
junction. Thus for eachmetre pipe-lengthwith a pipe-junction the loss would be 29 +5= 34%. This
To comparethe DPF models againstthe findings by Loncour et al [41, the drop of the daylight
penetrationfactor on one point in room due to the light pipe (with a 2-meter distancefrom the point to
the diffuser centre),as a function of the length of light pipe tube (not including dome and diffuser) has
been calculated.It is found that the per meter DPF drop dependson the solar altitude, sky clearness
index and the aspectratio of light pipe tube. Ile averagevaluesof per meter DPF drop for 0.33m,
0.45m and 0.53m diameter light pipes are 0.42,0.34 and 0.29 respectively.
175
Mus the figures given by DPF model for 0.33-m pipe diameter are comparable to the results given by
Loncour et al [4]. Note that the per cent drop of 42% includes the effect of energy loss through the
diffuser and hence shows slightly higher figure than 34% for a comparable aspect ratio of 3.6 (pipe
diameter 0.33m).
Most scientific literature is basedon a more objective approachfor reporting energyloss in light pipes
and that is to divorce from a per metre loss and insteadreport in terms of loss (or transmission)as a
function of aspectratio (light pipe length to diameterratio). Sucha plot is shown in Fig. 8.25, produced
by a Liverpool basedresearcher,Carter [5]. Note that Carter's loss factor of 60% for an aspectratio of
3.6 is higher thanNapier's result of 42% (averagefigure). Note that Carter's data are within close
proximity to the study undertakenby Love et al (1995) in Canada.Carter has taken a figure of ±10%
difference betweendatasetsas being acceptable.A point of ftirther note in Fig. 8.25 is that there can be
variations (scatter)in the illuminance readings- internal as well as external.Figure 8.26 showsthat
Carter's estimateof energyloss in a 30-degreebend is 20% [5]. This is very close to Napier-based
measurementsthat were found to give this value as 21%. David Jenkins [6] undertook ftirther
comparativework betweenNapier and Liverpool studiesand two suchplots are shown in Fig. 8.27.
Figure 8.28 showsthe light transmissionplot of a light pipe. The point that is worthy of note here is
For short pipes DPF producessimilar resultsto thosefound by Shao[7] and Oakley et al [8]. Note that
Oakley quotesDPF values of 0.48%, 0.38% and 0.18% respectivelyfor aspectratios of 2.1,4.6 and
8.4.
The Napier DPF hasNOT beenevaluatedfor long pipes. Early indications basedon abbreviated
analysissuggestthat a higher loss of energytakesplace within the entrancelength of the light pipe (see
the shapeof the asymptoticcurve producedby Carter, Fig. 8.25). This is analogousto a well-known
phenomenonwithin fluid flow in pipes. Figures 8.29 and 8.30 demonstratethis concept,i. e. two
sourcesof energy loss are shown - entranceto the pipe and developing flow within the starting length.
176
At this stage it is difficult to explain this phenomenon for light flow but further research in this respect
to comparethe performanceof longer light pipes againstthat of shorterpipes could lead to new
discovery.
The proposedmodel S-DPF (Eq. 6.9) enablesthe prediction of the internal illuminancesthat are
achievableby opal diffuser light pipes of various configurationsand under different sky conditions.As
a set of guidelinesfor practical engineeringdesign,Tables 8.4 - 8.9 provide the internal illuminances at
various distancesthat can be achievedby opal diffuser light pipes of varying geometrical
configurationsoperatingunder a set of three weatherconditions for Kew, London [9]. The tablesalso
cover a set of three seasonalvariations: summer,winter, and autumn and spring together.
Basedon E-DPF (Eq. 6.11), it hasbeenfound that light pipe's daylight transmissionreduction factor
due to the use of one 30-degreebend is around 0.2. This factor can be used as a simple guideline for
elbowed light pipe design.It hasalso beennoted that the efficiency loss due to the introduction of
bendsis a function of solar altitude, which meansthat for different weatherconditions and different
times of the year, the actual transmissionreduction factor differs. The more sophisticatedmathematical
model E-DPF (Eq. 6.11) should thereforebe applied wheremore accuratedesignfor elbowedlight
pipes is required.
To predict the achievableinternal illuminance by straight light pipes with flat diffusers, a multiplicative
factor fD is used to obtain the internal illuminancesvaluesindicated in Tables 8.4 - 8.9. For elbowed
light pipes with flat diffusers, combining the use of fo with the factor fl,,, =0.2 may be usedas a
simpler guideline. However, by applying the fD to the resultsgiven by E-DPF model (Eq. 6.11) a more
may be made.
accurateperformanceassessment
Additionally, within the Excel Visual Basic Application environment,lux plot [Courtesy:David
Jenkins,Monodraught] that can visually describethe internal illuminance distribution due to light pipes
has been constructed based on DPF models. The lux plot can be used by building or lighting designers
177
in both of their initial and final stages of design of light pipes. Figure 8.31 shows an example of such
lux plot.
addressthe envelop of the S-DPF and E-DPF models,so as to guide the application of the models in
real designs.
S-DPF and E-DPF modelswere developedand validated using measurementson 17 different light
boundariesin various dimensions,it is believed that the DPF modelshave an envelopetoo. The
envelopestipulatesthe condition under which a given model can perform with its accuracycoherentto
the statistical evaluationof the model. The envelopof the DPF models is decidedby the boundariesof
and physical reasoninghavebeenapplied, which fonris the basis on which DPF models can be
performanceof light pipes with opal diffuser. Geometrically,there are mainly five dimensionsof
boundaries,which are the diameter,length, number of bendsof light pipes, and the vertical and total
distances(seeFig. 8.32) betweena given point and the centreof a light pipe diffuser. On the other
sky clearnessindex (k,) and the global illuminance (E,g). According to Table 8.10 and the limitations of
externalweatherconditions of ct, k, and Evp the envelop of the DPF models can be clarified, as shown
in Table 8.11.
The boundary of DPF models is constrainedby the facility availability of the project. However, the
applicability of DPF modesis not only limited to the configurationsand external conditions given in
Table 8.11. When there is an attemptto apply the DPF modelsto a broaderrangethat is beyond the
178
achievable. But at the same time, it is also worthy here to clarify that the accuracy of the model, where
179
REFERENCES
1. Norman, R., Draper and Smith, H. (199 8) Applied Regression Analysis (Third Edition), Wiley-
2. Muneer, T. (1997) Solar radiation & daylight modelsfor the energy efficient design ofbuildings,
3. Swift, P. D. and Smith, G. B. (1995) Cylindrical mirror light pipes, Solar Energy Material and Solar
ResearchInstitute
6. Muneer, T., Zhang, X. and Jenkins,D. (2002) Brief report on the performanceof Napier's Daylight
Training CompanyProgramMeeting
7. Shao,L. (1988) Mirror lightpipes: Daylighting performancein real buildings Lighting Researchand
Technology,30 (1), 37 - 44
8. Oakley, G., Riffat, S. B. and Shao,L. (1999) Daylight performanceof lightpipes Proceedingsof the
159-174
180
-Y,,
......................
.......................................... Xi or Y,,
............
Yc
-Y.
Xi or Y,,
...............
................. 10
..........
.............
Yo
(2)
Yo
(3)
181
Figure 8.3 The residual plot of the difference between predicted and measured internal
illuminances (Y-axis) versus predicted values (X-axis) for S-DPF model, Unit: lux
14000
12000
10000.
8000
6000
4000
Figure 8.4 Histogram of absolute error due to S-DPF model, Unit for X-axis: lux, Y-axis: number
of data points
182
Figure 8.5 The residual plot of the difference between predicted and measured internal
illuminances (Y-axis, Unit: lux) versus solar altitude (X-axis, Unit: degree) for S-DPF model
Figure 8.6 The residual plot of the difference between predicted and measured internal
illuminances (Y-axis, Unit: lux) versus sky clearnessindex (X-axis) for S-DPF model
183
Figure 8.7 Scatter plot of predicted internal illuminance against measured internal illuminance
(light pipe 0.21m in diameter and 0.61m in length), Unit: lux
6UU -
250 -y0.9572X
200 - 0.9641
150 - A,
100
-
50
0
0 50 100 150 200 250 30C
Figure 8.8 Scatter plot of predicted internal illuminance against measured internal illuminance
(light pipe 0.21m in diameter and 1.22m in length), Unit: lux
184
r%r%
m
OUU
y 1.0215x
600 -R0.8759
400 -
200 -
0
0 200 400 600 800
Figure 8.9 Scatter plot of predicted internal illuminance against measured internal illuminance
(light pipe 0.33m in diameter and 0.61m in length), Unit: lux
800
y=1.0426x
600 -W=0.8235 OP
400
200 - -
+IV'
0
0 200 400 600 800
Figure 8.10 Scatter plot of predicted internal illuminance (Y-axis) against measured internal
illuminance (X-axis), (light pipe 0.33m in diameter and 1.22m in length), Unit: lux
185
-7r%r%
I uu
600
500
400
300
200
100
0
0 100 200 400 500 600 700
300
Figure 8.11 Scatter plot of predicted internal illuminance against measured internal illuminance
(light pipe 0.45m in diameter and 0.61m in length), Unit: lux
1500
y=0.9987x
1200
R2= 0.8643
4V
900
600
4.
4
300
0
0
300 600 900 1200 1500
Figure 8.12 Scatter plot of predicted internal illuminance against measured internal illuminance
(light pipe 0.45m in diameter and 1.22m in length), Unit: lux
186
1600
1200
800
400
0
0 400 800 1200 1600
Figure 8.13 Scatter plot of predicted internal illuminance against measured internal illuminance
(light pipe 0.53m in diameter and 0.61m in length), Unit: lux
600
Y=1.1 199X
500 -W=0.9378
400 -
300 - *+
+
200 -+
100 -
0-
0 100 200 300 400 500 600
Figure 8.14 Validation: scatter plot of measured (X-axis) internal illuminance versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, heavy-overcast sky), Unit: lux
187
8(
40 -
20 -
0 --
0 Z-u 4U bu öu
Figure 8.15 Validation: scatter plot of measured internal illuminance (X-axis) versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, heavy-overcast sky, 0.21m in diameter and
0.61m in length), unit: lux
100
80 -y2=0.9384x
so -R=0.7203
40 -
20 -+
0
0 20 40 60 80 100
Figure 8.16 Validation: scatter plot of measured internal illuminance (X-axis) versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, heavy-overcast sky, 0.21m in diameter and
1.22m,in length), unit: lux
188
Figure 8.17 Validation: scatter plot of measured internal illuminance (X-axis) versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, heavy-overcast sky, 0.33m in diameter and
0.61m in length), unit: lux
Figure 8.18 Validation: scatter plot of measured internal illuminance (X-axis) versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, heavy-overcast sky, 0.33m in diameter and
1.22m in length), unit: lux
189
Figure 8.19 Validation: scatter plot of measured internal illuminance (X-axis) versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, heavy-overcast sky, 0.42m in diameter and
0.61m in length), unit: lux
Figure 8.20 Validation: scatter plot of measured internal illuminance (X-axis) versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, heavy-overcast sky, 0.42m in diameter and
1.22m in length), unit: lux
190
Figure 8.21 Validation: scatter plot of measured internal illuminance (X-axis) versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, heavy-overcast sky, 0.53m in diameter and
0.61m in length), unit: lux
Figure 8.22 Validation: scatter plot of measured internal illuminance (X-axis) versus predicted
values (Y-axis) due to S-DPF model (Merchiston data, all weather), unit: lux
191
3.3 Optical efficiency
The optical system can be split up into three parts:
" The transparent top dome
" The highly reflective aluminium ducts
9 The enddiffuser
Theoretical Measured
Sunpipe Sunpipe 2 Sunpipe 2
Light transmission duct 74% 93% 48%
Global light transmission 33% 42% 22%
Table 3: Light transmission through the Sunpipe system
For the calculationof the light transmissionof the global systemthe top domeandthe diffuser
aretakeninto account.
For the theoreticalvalues the normal light transmittancewere used properties(measuredat
BBRI, resultsare given in Appendix A). Theseamountto about 50 % for the diffusersand
about90'D/o for the top domes. Thesewere chosenbecausethey are typically the valuesthat
would be included in product specifications(although not yet included in the product data
sheet).
Renuirk:
Thesemanufacturerspecifiedvaluesseemto be quite idealistic and lead to a performancethat
is almost twice as good as the real performance.Again, more prudence in product
would be recommendable.
specifications
by Loncour et al 141
8.23 Scanned copy of page 15 of report
Figure
192
3.3.2 Detailed analysis of light decrease in the aluminium duct
To get a better insight in how the light flux diminishes in the duct, a detailed measurement
campaign was set-up. The decreaseof light flux through the duct was determined by on-site
measurementsin Sunpipe 2. The illuminance in the duct was measuredalong the length of
the duct and comparedwith the referenceluxmeter outdoors, under an overscastsky.
The set-up (Figure 22) is comparableto a daylight factor measurement.Figure 23 shows how
much light is left at every point of the duct with respectto the duct entrance.
The effect of the direct view of the
sky can be determined. As can be
seen it decreasesvery quickly from
0.0M 100 %to only I %after I meter.
----
(),25m The remainder of the light is the
interrcflected part, which increases
50ni
-------- __V, Flu\ýctcr steeply in the beginning of the duct
inside Sunpipe
---- ---
0,75m and after 0.75 rn decreases slowly
ReferenceILLxmctLroutside with the absorption of light by the
duct. An extra light loss of about
44553WIM In 5% is noticed between 1.25 and 1.5
1. . 7" L. I rn, right at the point where two
junctions were made.
j, - The average decreaseper meter of
-------------
'Jk( duct, basedon the values where the
direct part is almost completely
- ------ 4- gone (starting from 75 cm), Is in
a '10 this case 29 %, meaning that we
have a light transmission per meter
of 71 %. This is a lot worse than
Figure 22: Measurement of light decreAse in Sunpipe the stated 3% per meter.
100% fý
60%
5os
-1 30% .
20% -
10% .
os
0 05 1 1.5 2 2.5
LeNth (-)
i theoretical -- measured relative - direct part - reflected part!
16 by Loncour et al 141
Figure 8.24 Scanned copy of page of report
193
IM -
90
80
70
60
50
eu
ui 40
30
20 ------- ...............
10
0
0123456789 10
Aspect rallo
Figure 5 (a) Graph of pipe efficiency against aspect ratio. (b) Graph of pipe efficiency against aspect ratio
120
100
80
60
40
20
0
0 20 40 60 so 100
Bond angle
194
Comparisons of predictions and measurements for 330mm Sunpipe, 0.6m in length,
10/8/99
University of Liverpool results for Eh range 9900-52200lx
300
250
200
hL
z
150
1A,
100
p
50
L
11
120 125 13 0 135 14ýO 145 150
Time of day, h
500
450 mý
, a. 3, a M. a Z.
400
350
300
-4 250
E
200
ISO
50
AS -- Predicted corresponding to external luxes recorded in experiment
0
12.5 13 13.5 14 14.5 15 155
Tlý of day, h
Figure 8.27 Comparative plots showing close conformance between Liverpool measurements and
Napier's DPF model estimates
195
100000
............
ioooo
1ON
100
10
A "A AAA
'b '5 '5 'b 't, 'b 'b :.5 '5
e "3 C) 4S
4ý1 (P Q11, z" 134 Q0 cqý co (Zý zq Nczý ý- -IS 10 113 NQ, NIZ, NIS NO NI: IN ýN
Figure 8.28 Plot showing measured external and internal illuminance at Napier University, unit:
lux
FIGURE 8.25
Entrance [Iow condi-
tions and loss coeffi-
cient (Refs. 28. -19).
(a) Reentrant. K,
OX (b) Sharped
cdged, K, - 0.5,
wi Sll(-,Iitl I rounded.
K, -7 0.2 (see Fig.
8.27). (d) Well
rounded, K, = 0.04
(see Fig. 9.27).
196
Figure 8.30 Loss of flow energy during the 'starting-length' within the conduit
Lý] File Edit View Insert FQrmat Jools Data Window Help
X Gb lftl 43 70%
,w 10 wBz 11 IF IF aMW%, *68;'8 1iF i.ir ! ). .&ý
-ý
198
(01%
2A
'Ei ýD - - \tD (=)
r- rn - - C, ý,0 -
ýc 00 cq rn cq rn kn
rn rn
>Z "0 r-
< rq rn 0
<D
vi
10
Ici
pe
AA=,
cl
E
00 cý C% 0%
Icýri 'lý
CD -N lý
CD NZ
42 Kr)
cý
r-
c;
r-
cý c; c5 2 2
-d
C) - 'C '0 "C MJ 0
"C 'CJ
r. = Z' CZ g'. r.
:4.9
Rý.Eý L 2 Rý
gz .!
im. ný $Z n. .
5 A.- ý5Z ý5 15Z
.: .
00
- cq r, ) zr V) %M r- vlý \.0 r-
u lu u &) iL) (D (D
E-
Table 8.2 Sample data of Internal illuminance distribution (0.33m-diameter light pipe, 18th Dec
2000)
Date Time Sl (LUX) S2 (LUX) S3 (LUX) S4 (LUX) S6 (LUX) Average lux SDV, lux
18-Dec-00 9: 22: 12 119 82 82 84 110 95 18
18-Dec-00 9: 23: 12 127 88 86 92 120 103 19
18-Dec-00 9: 24: 12 138 93 91 100 127 110 21
18-Dec-00 9: 25: 12 144 100 98 106 132 116 21
1B-Dec-00 9: 26: 12 148 104 101 113 138 121 21
1B-Dec-00 9: 27: 12 151 106 106 118 140 124 21
18-Dec-00 9: 28: 12 155 ill 110 116 139 126 20
18-Dec-00 9: 29: 12 156 115 115 109 135 126 19
18-Dec-00 9: 30: 12 155 115 115 108 137 126 20
18-Dec-00 9: 31: 12 155 114 109 113 142 127 21
18-Dec-00 9: 32: 12 161 113 109 115 141 128 22
18-Dec-00 9: 33: 12 167 120 117 116 142 132 22
18-Dec-00 9: 34: 12 174 130 125 124 150 141 21
18-Dec-00 9: 35: 12 183 135 130 134 162 149 23
18-Dec-00 9: 36: 12 194 139 135 137 171 155 26
18-Dec-00 9: 37: 12 204 145 144 143 177 163 27
18-Dec-00 9: 38: 12 213 152 145 157 184 170 28
18-Dec-00 9: 39: 12 224 156 149 168 193 178 31
18-Dec-00 9: 40: 12 239 165 158 174 206 188 34
18-Dec-00 9: 41: 12 255 177 171 189 225 203 36
18-Dec-00 9: 42: 12 272 191 186 211 245 221 37
18-Dec-00 9: 43: 12 292 208 203 230 260 239 37
18-Dec-00 9: 44: 12 306 221 221 250 275 255 37
18-Dec-00 9: 45: 12 320 234 238 270 293 271 37
18-Dec-00 9: 46: 12 336 245 251 280 305 283 38
18-Dec-00 9: 47: 12 350 255 258 290 322 295 41
18-Dec-00 9: 48: 12 370 264 263 305 337 308 47
18-Dec-00 9: 49: 12 391 281 279 315 344 322 47
18-Dec-00 9: 50: 12 410 304 303 329 360 341 45
18-Dec-00 9: 51: 12 422 317 321 362 391 363 45
18-Dec-00 9: 52: 12 443 331 340 404 422 388 50
18-Dec-00 9: 53: 12 463 345 361 433 447 410 53
18-Dec-00 9: 54: 12 490 363 379 447 468 429 56
18-Dec-00 9: 55: 12 524 389 394 444 473 445 57
18-Dec-00 9: 56: 12 545 415 407 435 468 454 56
18-Dec-00 9: 57: 12 547 424 415 438 477 460 54
18-Dec-00 9: 58: 12 554 418 413 453 494 466 59
18-Dec-00 9: 59: 12 561 411 415 470 505 472 63
18-Dec-00 10: 00: 12 568 422 434 489 518 486 60
18-Dec-00 10: 01: 12 582 438 453 507 533 503 59
18-Dec-00 10: 02: 12 600 447 466 517 549 516 62
18-Dec-00 10: 03: 12 609 451 469 521 560 522 65
18-Dec-00 10: 04: 12 609 449 463 523 565 522 67
18-Dec-00 10: 05: 12 614 446 457 525 571 523 72
18-Dec-00 10: 06: 12 629 449 455 534 580 529 78
18-Dec-00 10: 07: 12 640 457 460 553 594 541 81
18-Dec-00 10: 08: 12 653 469 476 590 619 561 84
200
Table 8.3 Summary of the Internal illuminance distribution test results (Craighouse test room)
201
CIA
2
rq CD V) CD (D 0 tA vi CD 0 V) (Z Wý
e rq rn 0., CD ýo W) C> ;ý cli V) %M C> CD rn C> 0CD
(Z
"CJ e r4 - rn - - (11 00 \ýo VI - 00 t
CD V) CD - CD vi 0 C> vi CD kn CD CD kn kn 0 CD
11 le kn rn r, c> 0
r- ,:V')> CN
0 "' :: > CD <" \O
ýQ le rq V') r) cýI -
00 *0 trý
r, 4 - 00
rn
E
"000
CD
In c> vý v"ý C) kn vi vý CD CD vl CD kr) V'l CD CD CD CD C> CD
11 11 r-i rq 1- c> r- oý, cý -0 rn c-i \m oo c> cý vi (Z rn rq
Z
e.) -
Ei V-ý CD V) V) vi v-i 0 0 CD - - V')
.
CL r, 1 - cý
E = C,4 - 00 \,O -
-
CD CD CD CD kn (D vi vi CD v) 2 cý (n V) CD
.0
IT
I "CJ11i t-
0-
-e rn r4 v) rq %Z -
ýt i rq i VI i fli i - ýt
Ir
clý -
rq
rl i -
00 V'l
C> rq
rn rn
-
CD
"Ci
11 e vl vý 0 c5 cý (Z cý c2 0 cý vi kn 0 CD V'l -
-zj - 00 le - ýo rn r- le ri V) rn - r, ) rq - rq
c
"CJ (N cý rq %Z 00 ýo rn r- e C,1 VI rn
N
. r.
0 0
4.1 en
-2
(U rn CD V) vi vi , 0 , ýz 0 , CD
* 11 11 Cq -e c-i cý r- ' C> r14 ' ýt <=> vlý CD 0 (D CD CD CD
ýý
-0-
ý_o
"CJ r) rn - cl) -
ýa1 V)
-
CD
- V) -
l i
u
ce
CD
c> Irl l',', Irl
'n 111
1
11 0 ol o vi n 0 Cý, c: > C> (D =ý CD
"0 - ýo rn V) r1) rq rn (q - rq - V). - V') V') v»i v', Irl
Z- -
CD
't CD vi c> V-ý vi V-ý V-ý tn vi tA CD vi
ý -e> = - 00 e 00 ýt rq e rq - r-i - - - - (r) vi
10
Ici
U P. C > U. P'L'3,
> U im. CD >
CD 0ü 0.
>
o U. 0. >
CDuu >
'im.CD
ce
ce
E-
'i I: C: )
c:,
22
CD
CD
CN
=i
C>
CD
rn
C: )
CD
ll)ýo
ý2 - .c,
rn
, 7: 1 - - r- - r- kn r- tn Rzr W) IT m ýT m N M r4 CN
n
rWl E
(= C) W-) tn
-'ý -e- llý - - - - 00 - 00 ýo 00 "o 'IT IC W'Trf) "It M Cq
9-
C'4 5
00 o
%, r- 1.0 IT 1-n 1:
ýt 1 Cý>
cl) en C)
rn C) Cli
N tt*)rqC) W)
- C>
C14 ti) C>
0
.0
u 11
r
. "o C, 11)
" - n n 2 2 2 2
1111tn V') V') VI vi tn V') tf)
tn tn tf) C) o
:r m C-4 en C,4 -
VI C> tn
eq -
1 I
-= c> c:, %n -10 in
= 0M C)
E5 1
cd m (=
11 11 W) 0 tn C) C) C) vi 0 tn tr) tn C)
%0 tn M "t rn C'4 C*4 N - - - -
"a - tr) r) vi tn tn tn tn
v tn
ýC-
co
V') VI kl-j
C13
E C)
(D eq
10
al 0 0 m C*
in (D Q) C4
> > > cli > ol > >
0 0 2 0
$ý
cts t: , M t: (U cl 1) M t: Cl
M > > >. > >
ti .2 C': .2 'd
(D 0 1U 12ý0 U a, 10 ul pýl o UI P. 10
0 C) C>
C) Cý C) C)
C) ý2 ýr C, IC
.4 \0 C14 en en
E
Eý
C14 C) C) 0 it') ý:, ý= VI Vn
Cýl ý10 kn en C14 ý= 00 CD - ýr 1 - - 0 V'I C: ) (D 0
,a N - 00 C14 - r, - - VI - 00 :t - ýo rn C*, tn M
101 I
m 00 00
=(=
0 m - 01, C4 - r- kil
'IT cq - en
0
(n
W) C) trl C) I tfl C) 0 tri C)
I 0
l
o 0 W) kn
cq V) I Cý
11 r- W) 0 en C) 0 tn 0 0 0 0 CD V-) 0 W') CD 0 In
C* "tý Cý tn - r- 'IT CN W) ff) t- 14'T CN Irl r') - -tT C14 -
ýa -
0
Q)
r-L E Lrl C) 0 In
1 11 'o a, o 'o "T 0 (=) - 0 IT 0 ti,
Z: 0
- 00 110
00'IT 0
'1', "1 , ('4 ýo en 00 'IT C14
- - Clq
I I
. I I I I I I I
5f
. cli
C's I
c
(= - - -I- -
0C:,
00w)00C) 00 tn
"V clý - eq - 4
C. cl) C14 - "0 00 'IT - ýc M
.
-9-
00
-
I
I
t6l
,IV
0
Q
cqs
rý
C> "
tn
CD
en t-
0 C)
-T N
0 tf .I
'IT C11
tt .I W)
r) C'j
t4l)
C-4 - - - - tn
9
2
cl
ti tri C> (0 C)
0 10 C,4 - ý,o - r-- ýr CONkn en "o en (14 'IT C11 - M N -
C)
M
cf)
M 0 Eý'In 0 0
0 V)
C7, ýo (= Nr kn C) t'll C) 0 in tn W) III
ý::
- 11
11 c, N - 00 C14 m ýo cq C14 'IT cli
1
cl til
en (11 cq tn
kr,Irl 11)
11r,
u
0
'41 0
C) 0 1 0 ,-11 -
11
1
r- C,4 C, In - tn kn kn in tn
"T Cl ýr _ _j - -
4
E0 E
o0.00- ID
- C)C14
11 11
-41) C) <ý
1,
w
kf-) -41 C> tn
-9 - liz Cý W) M W) rn - rn - -
>
ý
- - ....
C,
+
rtý - -W -w -EA -En -ry)
Cd Cd cl as 03
L) L)
u U 0 U a) in
U5 w rp 'n
> m > ca > cl > cts > C13 > cli
? L) 0 8 0 u u (? u 0
;ý W w
i cl m cl
ctl C, cd Ll
t's m
, ,ý u pm, .0
U oc,
ý 0>0 ýE0>0 p-
ý U)
0> 0>
PC P
. 0> ý 0>
C) C) C) C,
C) C) C) C>
N
tn vi kn C) CD C, 11) V) CD - CD CD ttl C)
W) V) vi C)
c C4
ý 7' cDce)
W,-t,- = 'o 2 C',
*-r,"'. "I'"D"'"'
Cý ý,c en r- tri C-4 'D --I- 4 kn ýzr cq - (w) (14
kt) C)
11
C% tn tt) CD
eq tn Ln W) C) (D (0 tr) C) t4l) C) VI vi CD W)
rn ýo r- ON C14 CD C14 CN It t- C) C% ýo N kf)
Ch 'IT r- M oo \0 m r- N 'IT kf) M
C,4 ttl o C) CD
11 r- Cý vi - "T 0 \0 - VII C11 tn 0 C) tn 0 0 C) til
1 "tv Cý4 C7%eq r- kn - 00 It ON 110 en I-- tn C14
CL C) C) V) 0 0 0 0
.7 ao
- C) tn ý= C:, V'l C)
C\ IýT C) 'T m m C,
Nr rl) m cq cq 00 VI
C)
m cq
C) C)
m C)W)
>)-9- lu1100 W)
t-
vi
r- Cj en
tn
r- 0
en
C14 ý-c *11, N 'IT rl)
ýc
- en r4 -
Cý(oC>ON
\.O t1r) C4 r-Cý
cli
-
0
R
C) 'IT
eq -
C)
r-
"o
cl C) W)tf) in I
0 11
"al
rn
cf)
m4
C,
c) \o 0 \0 - V, ) -
tn
0
00
o c, tn 0 o c,
'4Zt 00 kn en ýo zT C14
C.
0
0
rn
E
C) V-) C) CD kn
00
C) 0
1:
1
I. 11 11 V- c) IC M
(D
cn -
V-1
re) 110 CD
tn
ýc
tn
- W) C) C) 0 W) CD
'T en m C14 - N Do 00 00 kr) r)
U
<D o c, a
w
en, N, C,4 - - - VI . . . . . .
U
Itz
"0
03 V3 m cl 0
lu w U G) (1) CU (U
> 0 > cts >. c* > Cd > C's > cl
19 > 4ý U o U ' o 0
>
0
u lpýl
o 0 10
im. 1
I
C) C) C)
\.o ý2 C14 cn en
110
CD
cli
0 0 0 'r, 'r,
"C3 rl- V'i ON tn en \X) 'IT en tn en C14 zT C14 en ('I
0.
m El
X It) C) c) o c) C) (=, tn
C>
C)
C)
V'I C)
C> t/l
W) c, 0 0
I
V') kn
lzr - It \0 \0 C14 VI) tn
-
rf) N - C14 - - CN - 00 - CN \0 - r- tn (01, V') en
oI vi o (ý C) C) vi o t'l t) tr) o c) C) CD V -) 0
r- 't m In m C14 en C14 - cq - - C-4 - - - ý in
IV I I I I I I I I I I I I
(. I
0
o
cq
7
1
Cý
11 C)
cý rq o It C> C) C) tn C) C) In 1=1 ""I C: l 'In kr, 'r,
10 ell 00 CN \C \c :r r- "r en kn en C14 m cq
1 -1
-9- 1 I
(3.)
41
0
.4
oIV')
0 "tj Oo V') cn V, ) en cq M CN - N - - - - Ln - it) tf)
N
C's r) CD tn
11 11 - CD kn tn kn 0 0 0 0 (=) CD t4r) 0 11) 0 kn C>
as 10 - t-- It r- 'IT M tr) M C-4 rn N - N - - kn
-e-
Ei
-E
E -11 kn vi (ý kf)
" rZ C,4 - - C)
- tn 0
- kn krl 11) Irl 11) 11) 1,1, Irl Ir, I, ) tl,
ei
C-4 (=) u
u. 11 11 kn W) V) 0
'jý W) 0 0
ý -e- '0
cl rl) cq - Cq - - - V-) kf) tn tf) kn 11) kn kn W) if) V11
C13 m cc cl Cd cl
u u Q
U) V, M)
> > cl > m > m > 'I > cl
I? U. "I (? u "' 0 u " 0 u "' 0
Cd as m ., m C's
> m 0> > cl >
P-4 0 P. 0
POL! P. C) PC,.,
<li
m
ce C>
2-2 ": r c: l
8- C14 en
,0
cli
E
C4
5
0 r- 2 CD r, 1 - C> (D t CD CD 0
- <=>
10 Ch
V') rn - - rl 00 r13 ý
10 . - (n - (DO ýt V
'In
r, )
CD 5 D 2 CD vi CD (D kr) kr) V) V) vi CD CD CD CD
11 r) . 00 ri r_ %0 (n rn ýc C> CD O\ ýo m rn CD
X vl (D e vi tf) CD V) CD kn V) CD CD CD CD vi CD CD V) kn
c> oo CD rn 0 -e kn ým cý cq m kn - rn rg rq oo vi
in r, ) oo ýt cq \C rn - V'l rq - ýt rn - cý
CD
rq
> CD CD er) C> CD CD kn kn CD CD vl
1 11 rq %M ýc e C> CD VI e kn Cl, - V) mt CD tf)
e 00 rn
"0
"0
r. "U - \. O t) oo ýt rq kn rn C, 4 e r-1 - rn - - C, 1 - ,
"0
75 lt
c:> cý
Z v' L-- vli (= rý V) vl C> 0
i Pý t cý 1vi - 1.0 rn
C>
n n r,V, n
00 ýt (14 kf) rn rq ýr
1 1
rq
r, ) CD CD V) V) CD kr)
11 11
0 0 ro) (Z r- V) CD - (Z) vi CD V) Kri wi CD CD
m 'Im N - t- - Cý V) - e rn 00 ýt r, 1 kn fl) r*4
-e- rn cq -
CD vi v) v) v) c 4n CD CD
"CJ vi kn V) V) V) V) (A vi V)
2 CD CD vý 0 (2
- ci 00 e t, 1 ýt r, 1 - rq - - - - v .ý vl V -1 kn vý 'n
.
Ln CD (D [Z) 0 0 0 , CD vi CD 0
ý -0- IZJ - ýo rn \M t N rl r-ý - (, q - - vi - tr) tf)
10
w cu m cu 0 ce
0.
44 > cu > re > cu > tu > m > m
u , 0 u ' 0 0 u 0 2 %.9 2
ce (U > > ce
U >
c c r, C» 4'. O L) '. (D u p. 0 P. CD
iz. '
C:) CD CD
.0 c> CD (Z
0 (L) CD ým
E-4 j
ýc rr2 22 rn rn
-
Table 8.10 Measurement Settings
208
9. CONCLUSIONS AND FURTHER WORK
The presentresearchset out to meet specific objective, iternisedin Chapter 1. The overall aim of the
performance.Basedon two years' measurementson the daylighting performanceof light pipes, large
Two years' dataon light pipe daylighting performanceunder all weatherconditions were
the most significant influence on the performanceof light pipes. It was found that the
daylighting performanceof light pipes is affectedby light pipe configuration factors, external
weatherconditions and the position of the point of interest.Light pipe configuration factors
include the aspectratio and sectionareaof light pipe tube, the reflectanceof the internal
surfaceof light pipe tube, number of bends,degreeof bend and types of light pipe dome and
diffuser. External weathercondition includes global horizontal illuminance, solar attitude and
sky clearnessindex. The performanceof light pipe was also found affectedby the vertical and
horizontal distancesbetweenthe point of interestand the centreof the light pipe diffuser.
factors.
209
Light pipes are daylighting devices that utilise both sunlight and skylight. This is in contrast to
traditional daylighting devices such as windows, which owing to their design and orientation
limitations utilise only sky-diffuse and reflected illuminance. Therefore, a new concept of
Daylight Penetration Factor (DPF) has been introduced in present research to specify the
performance of light pipes. DPF is the ratio of the internal illuminance to the corresponding
total external illuminance at a given point. The introduction of DPF into daylighting research
enables the performance comparison between innovative and traditional daylighting devices.
iv. Two mathematical models, S-DPF for straight light pipes and E-DPF for elbowed light pipes
have been built to assessthe performance of straight light pipes and elbowed light pipes.
Compared to S-DPF model, E-DPF model uses an extra term to account for the effects of light
pipe bend. The term is a function of bends number N, and when N is given the value of 0, E-
DPF collapses to the exact form of S-DPF. S-DPF and E-DPF models predict the internal
illuminances due to light pipes under all weather conditions. Since the DPF models use non
wide applicable. The DPF models were incorporated into Excel spreadsheets that facilitate the
use of the DPF models. The DPF models are based on physical analysis of the working
mechanism of light pipes and are adjustable by changing the value of coefficients employed in
the model. The DPF model method therefore provides a flexible framework that enables a
By carrying out more measurement, fitting and validation procedures that were developed in
present research, the accuracy of the DPF models can be continuously improved.
V. In addition to traditional error evaluation techniques,i. e. MBE and RMSE, PAD, slope of
conductedto further investigatethe robustnessof the proposedmodels. The DPF models were
210
data shows that the models can predict the performance of light pipes under all weather
vi. It was found that the type of light pipe diffuser heavily affects the daylighting perforinanceof
light pipes. Comparisonbetweenlight pipes with opal and clearsdiffuser showedthat the
amount of light delivered by the latter one was significantly more than that by the former one.
This is due to the improved transparencyof the diffuser, which allows a better daylight
daylight is improved by using clear diffusers, the internal distribution of the penetrated
daylight is lessuniform than that due to opal diffuse light pipes. This shortcomingbecomes
apparentwhen the external illuminance is high (higher than 30klux), that is certainpattern of
daylight. DPF models given in presentresearchwere developedfor opal diffuse light pipes. A
Diffuser Factor (fD) hasbeen introducedto link the DPF modelsto the performanceof clear
takesthe form of six designtablesthat provide the internal illuminances that can be achieved
adaptedto aid the designof elbowed light pipe. Additionally, within the Excel Visual Basic
of their initial and final stagesof designof light pipes. The application of the lux plot in the
initial stageof designgives a generalpicture of how many and what kinds of light pipes are
neededfor a certain designpurpose.While the application of the lux plot at the final stageof
211
viii. A new anisotropicmodel to correct diffuse irradiancemeasuredby shadowbandhasbeen
show that for the caseof Bracknell, Englandthe RMSE, MBE and PAD of correcteddiffuse
CorrespondingRMSE, MBE and PAD due to Drummond's method are 16 W/M2,-5.6 W/mý
RMSE of 17 W/m2,an MBE of Mrný and a PAD of 5%. The correspondingfigures for
ix. The S-DPF model assumesthat the internal illuminance distribution on a working plan below
the light pipe diffuser is symmetricalto the diffuser centreand is a function of the vertical and
Craighousetest room to test this assumption(Section 5.6). It was found that the asymmetrical
the daylighting performanceof light pipe. It is worthy however to point out that in real
pipes were installed symmetrically to the room layout, the effect of asymmetricaldistribution
more accurate.
In order that future researchfollow in a logical mannerfrom the presentwork, some suggestionsfor
212
A 3-D light pipe DPF model
The DPF models predict the daylighting performance of light pipes based on a validated
observation that the internal distribution of daylighting due to light pipes is approximately
symmetrical to the axis of light pipe tube. This can however, be pointed out as a weak link in
the chain of calculating the performance of light pipe system. DPF models predict not only the
daylight transmittance of light pipe system, but also the internal illuminance level at given
points. The linkage between above two functions of DPF models is the method of describing
internal illuminance distribution. It has been found that the internal distribution of daylight due
to light pipe is not always strictly symmetrical to the axis of light pipe tube. The deviation of
the internal illuminances for points at different positions but having the same distances to the
light pipe diffuser centre can be as high as 15%. This implies that the unsymmetrical internal
illuminance distribution may be a major factor causing system bias of the DPF models. It is
therefore suggested that more geometrical factors that describes the relative position of a given
point to light pipe diffuser should be included into the DPF models, so as to take into account
the effect of unsymmetrical internal illuminance distribution. One proposal that may be
adopted would be to establish a 3-D coordinate system that can describe the geometrical
relationship between the sun's position (soIar altitude and azimuth), the light pipe tube's
position and a given point's position. Once the 3-D DPF model that incorporates above-
illuminance distribution should be undertaken and the data obtained can then be used to
determinethe formal form of the 3-1) DPF model. After that, consideringpractical issuessuch
DPF model.
traditional daylighting device,but owing to their designand orientation limitations utilise only
213
as light pipe utilise both sunlight and skylight. However, for innovative daylighting devices, to
concept of light pipe daylight penetration factor, DPF, a reference method may be adopted for
an agreed standard to assessall daylighting devices. In this respect the author would like to
propose new tasks that may be included for further work. The first task is to design a method
of a Light Pipe Figure of Merit (L-FoM) that would be the ratio of illuminance achieved by
any given light pipe to illuminance due to a "reference" light pipe of prescribed dimensions.
The second task is to design a Light Pipe to Window Figure of Merit (LW-FoM) that would be
the ratio of illuminance achieved by any light pipe to illuminance due to a "reference" window
of prescribed dimensions. The above L-FoM and LW-FoM would be evaluated under
prescribed sky conditions and within the 3-D coordinate system proposed above (section i).
214
APPENDIX 1: THE METHOD FOR CALCULATING THE VIEW FACTORS
FROM DIFFERENTIAL AREAS TO SPHERICAL SEGMENTS BY
NARAGHI
I-LH dh
I: D> and -I<H<L (whereD= H= L=
v , ,
1-0
VD 2 2)L2 VD2 2H 2-
(D 2+ H2)H2 2+H2 )L
-1+2HL-(D2 +H _ -1+ H H-(D
+2 -+2 [sin-'(
D2+ H2 i VH 222
(D2 +H 2)2 + (D - 1)(D +H 2)
2) h]
H(I -D2-H -H2
)]+2sin-'(ý
-sin-I( TH 2+
(D 2- 1)(D2 +H 2) D
(4.18a)
I-LH
H: D< and -I<H<L
4-1--L?
V(D'+H2-1)(1-H2)
I 1-ýý H + H2 -1
F, (tan Cos-1(4D2
U, H2 -1 D'+H2 D
IT (D 2+ H2)2
(4.18b)
215
I-LH
III: when L>0D> and H< -1
7 =77
I-LH I
when L<0 D> 71 --e and <H<-l
L
LH -I
when D> 4-1:ý-e and H<1
L
I I(I
I I-LH 2(1- 2Lý 2_H2 2HL) + D2 + H2 - 2HL + 2DVI Lz)(DVI L2 + LH)
FA (Cos-'( -D+ - - -1
+ tan-'(
2; r V(l 2 (1 N(I+D 22+H iF)(DTI
DII L2 +D2+H2 L2) 2HL 2DIIi Lý +I- LH)
- -2HL)2 -4D _ - - - -
I-LH
IV: when L>0 D<
r- and H< -I
-41-e
I -LH I
when L<0 D<
41--p an <H<-l
L
F, -A, ý-H3
u,
(D 2 +H 2)2 (4.18d)
LH-1
V: when D< T, and H<I
L
--ýP
I- 2Lý D2 2HL
Fu - -H2+
" 4 2; r V(I+D 2 +H2 2(l L2)
-2HL)2 -4D _
(4.18e)
216
APPENDIX 11: LIST OF PUBLICATIONS
4. A New Method for Correction ShadowBand Diffuse IrradianceData, T Muneer and X Zhang,
Buildings", Singapore,2002
review.
7. Cost and Value Analysis of Piped Daylight, T Muneer and X Zhang, submittedto Commission
8. Accuracy Analysis of Solar Position Algorithms Used within CIBSE Guides 'A' and 'J', CIBSE,
217