Non Imaging Optics Dizertation

UNIVERSITY OF CALIFORNIA, MERCED
Nonimaging Optics and the Radiative Transport Properties of Ideal Illumination
A dissertation submitted in partial satisfaction of

the requirements for the degree of
Doctorate of Philosophy
in
Mechanical Engineering
by
Jonathan James Ferry
Committee in charge:
Professor Gerardo Diaz, Chair of Advisory Committee
Professor Arnold Kim
Professor Sarah Kurtz
Professor Roland Winston, Supervisor
2020
Copyright
Jonathan James Ferry, 2020
All rights reserved.

The dissertation of Jonathan James Ferry, titled, Nonimaging Optics and the Radiative
Transport Properties of Ideal Illumination, is approved, and is acceptable in quality and
form for publication on microlm and electronically:
Professor Arnold Kim Date
Professor Sarah Kurtz Date
Supervisor
Professor Roland Winston Date
Chair
Professor Gerardo Diaz Date
University of California, Merced
2020
iii
To my parents, Dr. Andrea Schauer and Dr. James Ferry.
For the love and support you've shown me, and your unquestioning condence in my ability.
You have both been an example in shaping who I have become.
To Kathryn Fischer, my best friend and the love of my life.

You are the starlight that guides my way.
iv
Contents
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Vita for Jonathan James Ferry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xviii
1 Illumination Optics and the Characterization of Light 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The Basics of Light Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2.1 Quantication of Light . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2.2 Transfer of Light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Optics for Illumination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.1 Imaging Optical Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.2 Nonimaging Optical Systems . . . . . . . . . . . . . . . . . . . . . . . 14
1.4 Characterization Standards and Metrics . . . . . . . . . . . . . . . . . . . . . 17
1.4.1 Color Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.4.2 Geometric Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.5 Structure of this Dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2 The Thermodynamics of Ideal Illumination 24

2.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.1 The Thermodynamics of Illumination . . . . . . . . . . . . . . . . . . 25
2.3.2 Simulation and Validation . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3 Coupling of Nonimaging and Imaging Optics for Uniform Illumination

in Theatrical Spotlighting 34
3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.1 Khöler Illumination . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.2 Étendu Matching from Light Source to Target . . . . . . . . . . . . . . 37
3.4 Simulation and Experimental Validation . . . . . . . . . . . . . . . . . . . . . 39
v
3.4.1 Optics Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4 The Three-Dimensional Asymmetric CPC 49

4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3.1 The Asymmetrical CPC in Two-Dimensions . . . . . . . . . . . . . . . 50
4.3.2 Extrapolation to Three-Dimensions . . . . . . . . . . . . . . . . . . . . 51
4.3.3 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5 Conclusions and Future Directions 62

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.2 Insights into illumination design based on the new gure of merit . . . . . . . 62
5.3 Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.4 Overall Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
References 69
vi
Nomenclature
Units
Candela [cd = lm/sr]. Lux [lx = lm/m2 ]. Steradians [sr].

Joules [J]. Meters [m]. Watts [W = J/s].
Kelvin [K]. Radians [rad].
Lumens [lm]. Seconds [s].
Variables
α Absorption [−] Ee Irradiance [W/m2 ]

αt ACPC aperture tilt angle [rad] Ev Illuminance [lx]
η Optical eciency [−] F View factor [−]
Γ Random number [−] f Focal length [m]
κ Conic constant [−] G Étendue [m2 ]
λ Wavelength [nm] Ie Radiant intensity [W/sr]
Ω Solid angle [sr] Iv Luminous intensity [cd]
φ angle between vector and X-axis L X-axis direction cosine [rad]
[rad] Le Radiance [W/m2 /sr]
Φe Radiant power [W ] Lv Luminance [cd/m2 ]
Φv Luminous ux [lm] LER Luminous ecacy of radiation [lm/W ]
ρ Reectance [−] LES Luminous ecacy of the source [lm/W ]
τ Transmittance [−] M Y-axis direction cosine [rad]
θ angle between vector and Z-axis Mv Luminous exitance [lx]
[rad] N Z-axis direction cosine [rad]
ϕ ACPC design angle [rad] n Index of refraction [−]
A Surface area [m2 ] P Radiative probability [−]
Cg Geometric concentration ratio [−] p Focal point.
Cmax Maximum concentration ratio [−] Qe Radiant energy [J]

D Diameter m Qv Luminous energy [lms]
d Lens thickness parameter [m] R Lens surface radius of curvature [m]
vii
Acronyms
2D Two-dimensional.
3D Three-dimensional.
ACPC Asymmetric compound parabolic concentrator.
CAD Computer aided design.
CCT Correlated color temperature.
CEC Compound elliptical concentrator.
CFL Compact uorescent.
CIE International Commission on Illumination.
CNC Computer numerical control.
CPC Compound parabolic concentrator.
CRI Color rendering index.
ERS Ellipsoidal reector spotlight.
GHG Greenhouse Gas.
LED Light emitting diode.
NIS Nonimaging Spotlight.
PV Photovoltaic.
SPD Spectral power distribution.
viii
List of Figures
1.1 Graphical denition of a solid angle as depicted on a unit hemisphere. Adapted
from Radiative Heat Transfer by Modest [1] . . . . . . . . . . . . . . . . . . . 2
1.2 A ray (red arrow) depicted in Cartesian coordinate system, dened by spher-
ical coordinate system variables θ and φ. . . . . . . . . . . . . . . . . . . . . . 3
1.3 Photopic and Scoptic relative spectral response of the human eye used by the
CIE as visible luminosity functions to convert from radiometric to photometric
properties. Data adapted from CIE TN 003-2015[3] . . . . . . . . . . . . . . . 4
1.4 Examples of the spectral power distribution for various common light sources.
Data provided by the Lamp Spectral Power Distribution Database[6]. . . . . . 5
1.5 Lambertian angular distribution of a light source. . . . . . . . . . . . . . . . . 6
1.6 Diagram of (a) a specular reective surface, and (b) a diuse reective surface. 7
1.7 Diagram depicting (a) Snell's law of refraction for two dierent media, and
(b) a diuse or opaque refractive media. . . . . . . . . . . . . . . . . . . . . . 8
1.8 A camera obscura, commonly known as a pinhole camera. . . . . . . . . . . . 10
1.9 Two simple lens shapes: (a) a convex lens, and (b) a concave lens. Depicted
are the parameters used in the lens makers equations: f , d, R1 , and R2 . . . . 11
1.10 Diagram depicting the eects of (a) spherical aberration, and (b) chromatic
aberration on the focal length of a lens. . . . . . . . . . . . . . . . . . . . . . 11
1.11 Diagram of the deviation of the surface of an aspherical lens from that of a
spherical lens. The surface curvature deviation, or "sag" Z , is a function of
a conic constant, κ, and the spherical radius of curvature, R. . . . . . . . . . 12
1.12 Diagrams of basic reector based imaging optical systems: (a) a circular
reector, (b) an elliptical reector, (c) a parabolic reector . . . . . . . . . . . 13
1.13 Diagram of the geometry of a compound parabolic concentrator. . . . . . . . 14
1.14 Diagram of the geometry of a compound elliptical concentrator. . . . . . . . . 15
1.15 Relative spectral response of human vision for the Erythropic, Chloropic,
and Cyanopic cone cells and the Rhodopic rod cells along with the Melanopic
response plotted against the Photopic response curve. Data adapted from the
CIE TM 003-2015[3] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
o
1.16 The CIE 1931 2 chromaticity diagram showing the location of the spectrum
locus, purple boundary and blackbody locus. Several lines of constant CCT
are shown intersecting the blackbody locus. Figure adapted from [4] . . . . . 18
1.17 Conguration parameters for calculating the view factor between two surfaces
based on eq. 1.26 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.18 Raytracing diagram depicting a bundle of rays emitted from a source. Some
rays are traced and either absorbed, transmitted, or reected by the object
in their path. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
ix
2.1 Diagram of (a) the concepts of ideal, over, and under illumination for a de-
sired illumination area, and (b) an illumination setup with (1) the target
illumination area (red), (2) the aperture of a lighting xture (green), and (3)
the light source plane (blue). . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2 Diagram of the physical signicance of the radiative probabilities P12 (red
dashed line) and P13 (blue dashed line) and how they relate to under illumi-
nation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3 P31 P13 /P12 showing the bounding
vs limits of illumination with the ideal
located at P31 = 1 and P13 /P12 = 1. . . . . . . . . . . . . . . . . . . . . . . . 28
2.4 Diagrams of the ve simple reector proles used in benchmark raytracing
simulations testing the P31 and P13 /P12 gures of merit. (a) tapered/conical,
(b) elliptical, (c) parabolic, (d) CPC, and (e) CEC reectors. Blue depicts the
light source area, green depicts the aperture area, gray depicts the reector
prole, red depicts the ray path, and the dotted arrow shows the beam angle
for each design. Two-dimensional designs were used as shown in the gure,
and three-dimensional designs were developed by revolving the proles about
a central axis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5 Results of the raytracing simulations for the 2D reector proles calculating
P31 and P13 /P12 by varying the target distance L from the aperture. The
rst column shows P31 in dashed lines and P13 /P12 in solid lines with the
blue, green, red, and purple representing the 15, 30, 45, and 60 degree beam
half-angles respectively. Figures a, b, c, d, and e correspond to the reector
designs depicted in Fig. 2.4(a)-(e). The second column shows P31 vs P13 /P12
for each design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.6 Results of P31 vs P13 /P12 for all (a) 2D and (b) 3D reector designs. The
points plotted are the values calculated from raytracing simulations for a
target distance of 1 m from the aperture of each luminary. Colors distinguish
the reector prole and the symbols distinguish each beam half-angle. . . . . 31
2.7 Trends seen in the P31 vs P13 /P12 data as a function of distance from the
aperture to target illumination area for an ideal illumination lighting xture. . 32
3.1 Diagram depicting a traditional Ellipsoidal Reector Spotlight design. . . . . 35

3.2 Diagram depicting the dierence in optical elements of a basic microscope for
(a) critical (Abbe) illumination, and (b) Köhler illumination. . . . . . . . . . 36
3.3 A CPC luminary can be a replacement for Köhler illumination in microscope
and projection optics systems to provide a uniform illumination at the slide
plane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 A Diagram of a two plano-convex lens system designed to project the uni-
formly illuminated aperture of a CPC to a desired target. . . . . . . . . . . . 38
3.5 Validation of Irradiance uniformity for the CPC and lens design proposed in
Table 3.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.6 Irradiance distribution at the aperture of dierent CPC cone designs. From
top to bottom: revolved CPC cone, octagonal CPC cone, octagonal CPC
cone with cylindrical extension. . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.7 3D printed octagonal CPC cone lined with 3M ESR reective coating. Red
paper is used as a visual reference to ensure ideal optical reection. . . . . . . 41
x
3.8 Lens fabrication process. (a) machining of lens blanks, (b) CNC lathe of
curved surface, (c) rough lenses produced on CNC lathe, (d) nal polished
lenses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.9 Experimental benchtop setup of the octagonal CPC, cylindrical extension,
and two lens spotlight assembly. . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.10 Peak illuminance as a function of spotlight throw distance for various industry
spotlights compared to simulated results for the NIS spotlight design. . . . . . 43
3.11 Results of the angular distribution of normalized luminous intensity for (a)
ETC Source Four series spotlights, (b) Altman PHX LED series spotlights,
(C) NIS designed spotlight, and (d) Combined luminous intensity proles for
comparison. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.12 Projected spot on a target 2 m away showing a bubble pattern for the NIS
spotlight (a) simulated color rendered result, and (b) experimental result. . . 45
3.13 Energy usage and cost comparison to run a single 750 W ERS, 250 W LED
ERS, or 70 W NIS vs hours of operation. * Community theater data from Theater
Communications Group for 2018[58] ** Broadway theater data from the Broadway League
for 2018-19[59] + High Performance Lamp (HPL) data from ETC[60] . . . . . . . . . . . 46
4.1 Design parameters and two-dimensional proles for both the (a) CPC and
(b) ACPC nonimaging concentrators. . . . . . . . . . . . . . . . . . . . . . . . 50
4.2 (a) XZ plane showing original 2D design parabolas (solid blue and red) for
an ACPC, (b) XY plane view of the circular aperture and receiver (magenta
and green respectively) of the 3D ACPC revolution, (c) the β plane showing
the new ACPC parabolas (dashed blue and red) rotated by an angle β from
the XZ plane, and (d) Isomorphic view of a 3D ACPC surface generated by
stitching together parabolic proles swept around a center axis. . . . . . . . 51
4.3 A ray (red arrow) depicted in Cartesian coordinate system, dened by spher-
ical coordinate system variables θ and φ. . . . . . . . . . . . . . . . . . . . . . 52
4.4 LM space direction cosine plots of the acceptance/illumination region for (a)
a standard revolved CPC, and (b) the desired shifted space for an 3D ACPC.
The black circle represents the unit circle of 2π steradians and the red circle
represents the direction cosine space for the concentrator. . . . . . . . . . . . 53
4.5 Diagrams of the reector proles for: (a) CPC with bent tube, (b) ACPC, (c)
tilted CPC, and (d) sliced CPC. Each reector design has an aperture tilted
at the same angle, αt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.6 LM space direction cosine simulation for the distribution of light from a stan-
dard 30
◦ CPC. The red circle show the desired acceptance/illumination region. 55
4.7
◦ ◦ ◦
LM space direction cosine plots of four αt angles (9.53 , 21.96 , 37.29 , and
◦
59.55 ) for each of the concentrator designs outlined in Fig. 4.5. The red
circles show the desired acceptance/illumination region for each αt angle. . . . 56
4.8 Results of extreme aperture tilt angle, αt , on the design and performance of
the 3D ACPC. (a) isomorphic rendering of the aring eect in the initial
3D ACPC design, (b) LM space distribution for the 3D ACPC design. The
jagged artifacts at the edge of the aperture in (a) are due to the faceted
structure of the design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
xi
4.9 Sample of results of aperture tilt angle, αt , on the design and performance of
the truncated 3D ACPC. The rst row denes the αt angle, the second row
depicts the LM space distribution, and the nal row shows the isomorphic
view of each truncated 3D ACPC design. . . . . . . . . . . . . . . . . . . . . . 58
4.10 Top-down view of a 3D ACPC rendering depicting the outer and inner aring
eects seen at extreme αt angles. The jagged artifacts at the edge of the
aperture are due to the faceted structure of the design. . . . . . . . . . . . . . 59
4.11 Deviation of the geometric concentration ratio for the truncated 3D ACPC
from the ideal for a 30
o half angle concentrator. Primary reason for the large
deviation is due to 'aring' eect at increased αt angles. . . . . . . . . . . . . 60

4.12 Percentage of light that falls within the target region of LM space as a function
of αt angle for each of the simulated designs. . . . . . . . . . . . . . . . . . . 60
5.1 Diagram of the simulation setup for comparing an ERS to the new NIS spot-
lighting designs. (a) The ERS design setup, and (b) the NIS design setup.
Each design uses the same lens column, target area, and target distance for
a baseline comparison. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.2 Target irradiance distribution for (a) The ERS design setup, and (b) the NIS
design setup. The desired target area is shown as a black circle and the color
gradient represents the irradiance in W/m2 . . . . . . . . . . . . . . . . . . . . 63
5.3 Diagram of the illumination target conguration for (a) a standard CPC, and
(b) an ACPC design. The target illumination area is the same, but shifted
to the right for the ACPC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.4 Target irradiance distributions for (a) a standard 30
o CPC, (b) a 30o CPC
o o o o
sliced at 27.52 , (c) a 30 CPC tilted by 27.52 , and (d) 30 ACPC with
o
aperture tilt αt = 27.52 . The target for each simulation was placed at 3 m
from source with the ideal target area shown as a black circle in each image.
The color gradient represents the irradiance in W/m2 . . . . . . . . . . . . . . 65
5.5 P31 vs P13 /P12 plot comparing each new design presented in this work in
contrast to their existing counterparts in the literature and previous chapters. 66
5.6 A 3D ACPC cast out of polyester resin. The source plane is painted blue for
contrast to visually inspect optical performance. . . . . . . . . . . . . . . . . . 67
xii
List of Tables
1.1 Radiometric properties of light. . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Photometric properties of visible light. . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Eect of κ values on an aspherical lens shape. . . . . . . . . . . . . . . . . . . 12
1.4 CIE test color samples for calculation of CRI. Adapted from [4] . . . . . . . . 19
3.1 Lens parameters for the design of a nonimaging to imaging projection system. 39
3.2 Industry spotlight choices for comparison to the Nonimaging spotlight designs
with published metrics for power consumption, luminous ux, and luminous
ecacy of source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3 Estimated annual market impact for the state of California assuming annual
savings for a 1% and 10% addoption and replacement of existing ERSs with
new NIS xtures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.1 Angular parameters for the 3D ACPC designs simulated. All parameters are
dened in degrees. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
xiii
Acknowledgements
I would rst and foremost like to acknowledge the guidance and support of my thesis adviser,
Dr. Roland Winston. He has shown me great kindness and encouragement to seek the best
of my potential. I would also like to acknowledge the input and expertise of my committee
members, Dr. Gerardo Diaz, Dr. Sarah Kurtz, and Dr. Arnold Kim, who were critical in
helping structure the work that led to this dissertation. Much of this work would also not
have been possible without the support, advise, and encouragement of my colleagues, past
and present: Dr. Bennett Widyolar, Dr. Melissa Ricketts, Dr. Lun Jiang, Jordyn Brinkley,
Yogesh Bhusal, Robyn Lukens, and Ron Durbin. Finally I would like to acknowledge my
family and friends who have stood by my side throughout my graduate education, without
whom I may not have had the mental fortitude to get to where I am today.
Funding for the work presented here was provided for in part by UC Solar, RAF Elec-
tronics Corp., and the CalSEED grant fund. Chapter 2 is published in part in the journal
Optics Express Vol. 28, Issue 2, pp. 1927-1935 (2020). Chapter 3 is published in part in
the SPIE conference proceedings Volume 10758 Nonimaging Optics: Ecient Design for
Illumination and Solar Concentration XV. Chapter 4 is under revision for publication in the
Journal of Photonics for Energy.
xiv
Vita for Jonathan James Ferry
EDUCATION:
Ph.D. Mechanical Engineering 2020
Advisor: Dr. Roland Winston
Areas of Concentration: Heat Transfer, Radiation, Solar Energy, Lighting, and Non-
imaging Optics.

B.S. Mechanical Engineering 2013
PUBLICATIONS:
1. J. Ferry, B. Widyolar, L. Jiang, and R. Winston, Solar thermal wastewater evapo-

ration for brine management and low pressure steam using the XCPC, Appl. Energy,
vol. 265, p. 114746, 2020.
2. J. Ferry, L. Jiang, and R. Winston, "The thermodynamics of ideal illumination: a

novel gure of merit for characterizing illumination eciency," Opt. Express 28(2),
19271935 (2019).
3. B. K. Widyolar, L. Jiang, J. Ferry, J. Brinkley, Y. Bhusal, and R. Winston, Nonimag-

ing solar collectors toward net-zero GHG emission, in Nonimaging Optics: Ecient
Design for Illumination and Solar Concentration XVI, 2019, vol. 11120, pp. 6786.
4. Y. Bhusal, B. Widyolar, L. Jiang, J. Ferry, J. Brinkley, and R. Winston, High

concentration parabolic trough solar collector with novel secondary reector in the
vacuum receiver, in Nonimaging Optics: Ecient Design for Illumination and Solar
Concentration XVI, 2019, vol. 11120, pp. 1527.
5. B. Widyolar, L. Jiang, J. Ferry, R. Winston, A. Kirk, M. Osowski, D. Cygan, and

H. Abbasi, "Theoretical and experimental performance of a two-stage (50X) hybrid
spectrum splitting solar collector tested to 600 °C," Appl. Energy 239, 514525 (2019).
6. B. Widyolar, L. Jiang, J. Ferry, R. Winston, D. Cygan, and H. Abbasi, Experimental

performance of a two-stage (50Ö) parabolic trough collector tested to 650 °C using a
suspended particulate heat transfer uid, Appl. Energy, vol. 240, pp. 436445, 2019.
7. B. Widyolar, L. Jiang, J. Ferry, and R. Winston, "Non-tracking East-West XCPC

solar thermal collector for 200 celsius applications," Appl. Energy 216(November
2017), 521533 (2018).
8. J. Ferry, R. Winston, and R. A. Flasck, Nonimaging optics applications in theater

lighting, in Nonimaging Optics: Ecient Design for Illumination and Solar Concen-
tration XV, 2018, Volume 10758. (Best Conference Presentation 3rd Prize)
9. D. Cygan, A. Kirk, M. Osowski, H. Abbasi, B. K. Widyolar, L. Jiang, J. Ferry,
and R. Winston, "Two-stage 50X hybrid spectrum splitting CSP/CPV collector with
InGaP/GaAs solar cells," 1075806(September 2018), 5 (2018).
xv
10. R. R. Milczarek, J. J. Ferry, F. S. Alleyne, C. W. Olsen, D. A. Olson, and R.
Winston, "Solar thermal drum drying performance of prune and tomato pomaces,"
Food Bioprod. Process. 106, 5364 (2017).
11. M. N. Ricketts, R. Winston, and J. Ferry, "Étendue and Angular Acceptance of the
Asymmetric Compound Parabolic Concentrator," J. Photonics Energy 7(2), 028003
(2017).
12. J. Ferry, M. N. Ricketts, and R. Winston, "Optical performance eects of the mis-
alignment of nonimaging optics solar collectors," 1037907(September 2017), 5 (2017).
13. M. Ricketts, J. Ferry, L. Jiang, and R. Winston, "Nonimaging optics in lighting

to reduce light pollution," Nonimaging Opt. Ec. Des. Illum. Sol. Conc. XIII-
Commemorating 50th Anniv. Nonimaging Opt. 9955(September 2016), 99550G (2016).
14. J. J. Ferry, F. S. Alleyne, R. R. Milczarek, R. Winston, and D. A. Olson, "Eciency

and design analysis of a solar thermal powered at plate dryer," 2016 Am. Soc. Agric.
Biol. Eng. Annu. Int. Meet. ASABE 2016 16 (2016).
15. R. Winston, L. Jiang, M. Abdelhamid, B. K. Widyolar, J. Ferry, D. Cygan, H. Abbasi,

A. Kozlov, A. Kirk, V. Elarde, and M. Osowski, "Nonimaging optics maximizing
exergy for hybrid solar system," Nonimaging Opt. Ec. Des. Illum. Sol. Conc.
XIIICommemorating 50th Anniv. Nonimaging Opt. 9955(September 2016), 99550N
(2016).
PROFESSIONAL EXPERIENCE:
Graduate Student Researcher 2014-Present
Developed the optical design and simulation of a prototype nonimaging theater spot-
light in collaboration with RAF Electronics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2018-Present
Project lead on integrating concentrating solar thermal collectors with waste water
evaporator equipment. Project sponsored by the California Department of Water
Resources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2017-2018
Project lead on integrating concentrating solar thermal collectors with food drum dry-
ing equipment. Project sponsored by the US Department of Agriculture Agriculture
Research Service. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2016-2017
Produced experimental apparatus for vacuum calorimetry measurements to under-
◦
stand heat loss from a 650 C solar thermal collector. . . . . . . . . . . . . . . . . . . . . . . 2015-2016
Helped design and fabricate concentrating solar thermal prototype collectors.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2014-2015
Graduate Student Teaching Assistant 2013-2014
Vibrations and Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spring 2014
Computer Aided Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fall 2013
Heat and Mass Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summer 2013
xvi
AWARDS AND FELLOWSHIPS:
Mechanical Engineering Bobcat Fellowship, University of California, Merced . . . 2018

Dan David Solar Fellowship, University of California, Merced . . . . . . . . . . . . . . . . . . 2017
Mechanical Engineering Bobcat Fellowship, University of California, Merced . . . 2016
xvii
Abstract
Illumination systems have advanced signicantly in the 21st century to develop more en-
ergy ecient lighting, however, many of the optical systems that direct the light from an
illumination source remain archaic in their design principles. This leads to signicant perfor-
mance ineciencies in lighting, even with a highly ecient energy source. With the advent
of nonimaging optics, understanding of ideal light transfer has unlocked new opportunities
to advance the eld of illumination and deliver truly ideal optical systems, which minimize
energy waste and light pollution while maximizing spatial illumination.
This dissertation will present novel work advancing the eld of illumination by outlining
a new gure of merit that characterizes the thermodynamic limits to geometric illumination.
With this gure of merit, any optical design can be quantied by performing a ray tracing
simulation. This gure of merit will help to inform illumination engineers whether their
design is fully illuminating a space or wasting light to the environment.
Utilizing this novel gure of merit, two new optical systems are presented to improve on
existing illumination designs. The rst design, a new theatrical spotlight based on nonimag-
ing optics, oers equivalent light output to existing commercial designs, while decreasing
the energy requirement down to 10-fold. The second nonimaging design focuses on three-
dimensional asymmetric o axis illumination. Building on the two-dimensional design prin-
ciples of the nonimaging asymmetric compound parabolic concentrator (ACPC), a new 3D
version was developed that preserves the asymmetry while maintaining ideal light transfer.
These two new designs were characterized by the novel gure of merit. Both the non-
imaging theatrical spotlight and 3D ACPC showed signicant optical performance benets
over existing designs in literature and commercial production. Future work will be proposed
on how to enhance the usefulness of this novel gure of merit, as well as future applications,
characterization, and production of the new illumination designs developed.
xviii
Chapter 1
Illumination Optics and the Characterization of Light
1.1 Introduction
The availability of light has regulated human activity since before we mastered the ability
to create re. Methods of illumination have expanded since then; with the development of
the candle and the oil lantern, the lament light bulb, compact orescent lighting, and now
modern light emitting diodes. Over the last few centuries humanity has also improved its
understanding of light, with many scientic theories to explain the propagation and qualities
of light. Optical science has also seen much advancement alongside illumination where it is
often desired to direct light from a source to a specic location.
The fundamental material introduced here will serve as a background for the following
chapters in this manuscript. This chapter will discuss some of the optical principles, designs,
and characterization methods used in the illumination eld. An explanation of the qualities
of light and the units to quantify them will be shown. From this foundation methods will
be discussed on how to calculate the transport of light through dierent optical systems.
Next some basic optical designs using some common geometric principles will be explored.
Finally, some standard characterization metrics to quantify illumination quality and optical
performance will be discussed.
1.2 The Basics of Light Transfer

To appreciate how a new understanding of optics can benet illumination, the foundation
and history of illumination optics needs to be discussed. The primary goal of illumination is
to transfer as much of the light from a source to a desired target. Illumination can be dened
in terms of three constituent component groups; the light source, the optical system, and the
target area. A light source is simply a device or system that emits light, an optical system
is anything that bends or shapes that light to travel in a certain path, and the target area is
simply a physical area in three-dimensional space. This section will provide denitions for
how lighting and optical systems are quantied and the methods for light transfer.
1.2.1 Quantication of Light

Light is both an electromagnetic wave and a particle, and thus has the properties and
behavior associated with both waves and particles. The units of measure for light take into
account this dual nature when dening measurable quantities. To begin to understand the
units and properties of light a few geometric terms need to be dened. The rst of these
geometric quantities is the solid angle, Ω. A solid angle, as depicted in Fig. 1.1, is an area
in three-dimensional space projected onto the surface of a unit hemisphere. This projected
area is measured in units of steradians [sr].
1
A nite solid angle can be dened as
cos θj dAj
dΩ = = sin θdθdφ, (1.1)
S2
where dAj is the nite area being projected on the unit hemisphere, θj is the angle formed
between the unit vector, n̂j , normal to the area dAj and the unit vector, ŝ, along a line, S,
from the origin of the unit hemisphere to dAj . Integrating Eq. 1.1 over all angles, the solid
angle for the entire unit hemisphere is 2π sr as shown in Eq. 1.2.
Z 2π Z π/2
Ω= sin θdθdφ = 2π (1.2)
φ=0 θ=0
Figure 1.1: Graphical denition of a solid angle as depicted on a unit hemisphere. Adapted from Radiative
Heat Transfer by Modest [1]
Another important geometric property that is useful in quantifying light and illumination
is the direction cosine[2]. Since light is a particle, it tends to travel in straight lines until
obstructed or deected by other objects in its path. These light paths are referred to as
rays, which are vectors in three-dimensional space. Any ray of light can be dened in terms
of its direction cosines in three-dimensional space as its projections along the X, Y, and
Z axes using spherical coordinate variables θ and φ, as shown in Fig. 1.2. The direction
cosines are then dened mathematically in Eqs. 1.3-1.5 where L, M, and N are the direction
cosines in X, Y, and Z respectively. Direction cosines are a useful tool when characterizing
performance of illumination optics, as will be seen in later chapters.
L = sin θ cos φ, (1.3)
M = sin θ sin φ, (1.4)
N = cos θ. (1.5)
2
Figure 1.2: A ray (red arrow) depicted in Cartesian coordinate system, dened by spherical coordinate
system variables θ and φ.
Units of Measure
The primary units for quantifying light and light transfer are grouped into a category called
radiometric properties. Radiometric properties are based on SI units (such as the Joule,
Watt, etc.) that establish the total energy transferred by light. Table 1.1 shows the terms,
denitions, and units associated with the radiometric properties of light.
Table 1.1: Radiometric properties of light.
Term Symbol Denition Units
Radiant Energy Qe Energy emitted by a light source Joules [J]

Radiant Power Φe Emitted energy/time Watts [W ]
Irradiance Ee Emitted energy//time/area W/m2
Radiant Intensity Ie Emitted energy/time/solid angle W/sr
Radiance Le Emitted energy/time/area/solid angle W/m2 /sr
From these radiometric properties, and the geometric principles discussed above, almost all
of the relevant information about light can be determined for any illumination conguration.
The nal piece of information needed to fully quantify light for illumination is how the
human eye perceives light. Light comes in a variety of wavelengths, λ, each having their
own associated energy levels. These wavelengths produce a spectrum of light which can
be analyzed using radiometric properties. Human vision only detects a small portion of
the light spectrum, which we call visible radiation. The range of wavelengths that make
up visible light are from 350 to 750 nm approximately. The human eye perceives regions of
wavelengths in the visible spectrum dierently based on the biological photoreceptors within
the eye. Because illumination is chiey important to humans, a way of characterizing light
with respect to the human response is important.
3
In 1931 the International Commission on Illumination (CIE) adopted a standard metric
for the human observer to convert radiant properties into a more useful metric for illumina-
tion, which standardized the unit called the Lumen [lm][3, 4]. The standard human observer
determined by the CIE perceives visible light with the response region called the photopic
response which peaks at 555 nm, shown in Fig. 1.3. Under low light environments the eye's
response shifts to the scoptic zone which peaks at 505 nm. From the photopic function
the CIE has determined a maximum conversion from radiative properties to human visual
response of 683 lm/W at 550 nm. This conversion, termed luminous ecacy, along with
the photopic response curve helped develop a new set of photometric properties (Table 1.2),
including the lumen [lm], lux [lx = lm/m2 ], and the candella [cd = lm/sr].
Figure 1.3: Photopic and Scoptic relative spectral response of the human eye used by the CIE as visible
luminosity functions to convert from radiometric to photometric properties. Data adapted from CIE TN
003-2015[3]
Table 1.2: Photometric properties of visible light.
Term Symbol Denition Units
Luminous Energy Qv [lm s]

Luminous Flux Φv Visible light perceived by the eye Lumen [lm]
Luminous Exitance Mv Luminous ux emitted by a surface lx or lm/m2
Illuminance Ev Luminous ux incident on a surface lx or lm/m2
Luminous Intensity Iv Luminous ux/solid angle cd or lm/sr
Luminance Lv Luminous ux/area/solid angle cd/m2
4
Spectral Distribution
When developing illumination technology and optics it is important to consider the wave-
length distribution of the light source being used. The distribution of wavelengths of light
is called a spectrum. Each light source has a dierent spectral output (Fig. 1.4), which
is termed the spectral power distribution (SPD)[4, 5]. Most common light sources produce
light that extends beyond the human visual response curve. Thus the SPD of each light
source can be weighted by the CIE standard observer and integrated to calculate a light
source's luminous ecacy of radiation (LER) (Eq. 1.6). The LER is means for determining
how eectively a light source produces light within the visible spectrum. This should not
be confused with the luminous ecacy of the source (LES) (Eq. 1.7), which is the ratio of
luminous ux to the electrical power required to produce the light.
Luminous Flux [lm] Φv

Luminous Ecacy of Radiation (LER) = = (1.6)
Radiant Power [W] Φe
Luminous Flux [lm] Φv

Luminous Ecacy of the Source (LES) = = (1.7)
Electrical Power [W] W attelectric
Figure 1.4: Examples of the spectral power distribution for various common light sources. Data provided
by the Lamp Spectral Power Distribution Database[6].
5
Angular Distribution
Any surface that emits light produces photons that tend to travel in straight lines until
obstructed or deected from their path. Most light sources will emit photons over a wide
range of angles from the normal of the surface which they are emitted from. This is called
the angular distribution of a light source. An ideal, diusely radiating at source of light
produces photons equally in all directions. This is called a Lambertian distribution, which
looks the same (isotropic) in all directions. The Lambertian distribution follows the cosine
law, which states that the intensity of light incident on a at surface will fall o with the
cosine of the angle (Eq. 1.8). Essentially this means that the light at the center of a target
plane will have a higher illuminance than the light at the edges. An illustration of the
Lambertian distribution is shown in a polar plot in Fig. 1.5.
ILamb = Io ∗ cos θ (1.8)
Figure 1.5: Lambertian angular distribution of a light source.
6
1.2.2 Transfer of Light
Reective Optics
Reective surfaces fall into two categories; specular or diuse. A specularly reective surface
will reect a ray of light at an equal and opposite angle to its incident with respect to the
surface normal (Fig. 1.6(a)). Specular reectors are ideal for use as mirrors in optical
systems where the path of the light needs to be controlled. The second category of reectors
are diuse reectors, which scatter light in all directions (Fig. 1.6(b)). An ideal diuse
reector will follow the same Lambertian's cosine law as dened above. Diuse reectors
can be useful in illumination optics where a "soft" or even distribution of light is desired
rather than directional or collimated light. Most real surfaces exhibit both specular and
diuse reectance.
Figure 1.6: Diagram of (a) a specular reective surface, and (b) a diuse reective surface.
The reectance, ρ, of a surface is dened as
reected part of incoming radiation

ρ= . (1.9)
total incoming radiation
The magnitude of reectance of any surface that is a function of angle and spectrum, called
the spectral bidirectional reectance function; This is dened as the magnitude of reected
light that travels in the direction of the reected ray. Mathematically this is stated as
dLe (λ, θi , φi , θr , φr )
ρ” (λ, θi , φi , θr , φr ) = . (1.10)
Le (λ, θi , φi ) cos θi dΩi
The spectral bidirectional reectance function can become extremely complicated to
calculate for real surfaces, however its use is critical in determining a total average reectance
for a surface. This is done by integrating Eq. 1.10 over all incoming and outgoing solid
angles and wavelengths, producing what is called the total hemispherical reectance[1]. The
total hemispherical reectance is measured empirically for a surface by using an integrating
sphere, which measures the spectral bidirectional reectance over all angles simultaneously.
An integrating sphere measurement produces a total spectral reectance, which can then be
integrated over all wavelengths to calculate the total reectance. Total reectance will have
a value that ranges from 0 to 1, with 0 being no reected light and 1 meaning all incident
light is reected. Integrating spheres can also be used to measure the diuse reectance of
a surface by opening a portion of the sphere to allow the specularly reected light beam to
escape.
7
Refractive Optics
Refractive optics, unlike reectors who bounce light o their surface, will bend light as it
passes through. This bending is caused by changing the speed of light. In a vacuum the
speed of light is c = 2.98 ∗ 108 m/s, but when light enters a dierent media, that speed will
change and thus change the path of the photon. To quantify how much bending will occur
as light enters a new medium Snell's law of refraction is used (Eq. 1.11). Snell's law states
that light will bend as a function of the indexes of refraction of the two media and the angle
of incidence of the light ray with respect to the surface normal. The index of refraction, n,
is a measure of quantifying how light travels through a media.
ni sin θi = no sin θo (1.11)
Like reectors, refractive media transmit light in specular and diuse patterns. A specu-
lar transmitter will follow Snell's law, allowing light to pass through, while a diuse transmit-
ter will scatter the light as it enters the media. Examples of specular and diuse transmission
are shown in Fig. 1.7.
Figure 1.7: Diagram depicting (a) Snell's law of refraction for two dierent media, and (b) a diuse or
opaque refractive media.
Similarly to reectance, the amount of light transmitted by a surface can be expressed as

the ratio of light that passes through a media to the total light that comes into that media
(Eq. 1.12). This ratio is called transmittance, τ. Transmittance is dependent on both
spectrum and angle. Experimentally, the transmittance of a media is determined by using
an integrating sphere and the integrating spectral transmittance over all wavelengths. This
gives a value between 0 and 1, which equate to no light being transmitted to all the light
being transmitted respectively.
transmitted part of incoming radiation

τ= (1.12)
8
Optical Eciency
Both the reection and refraction of light follow the basic rules outlined here. In each case
they are specularly dependent and some of the light can be absorbed by the surface or
media, leading to optical ineciencies. To get some idea of the overall optical eciency
for illumination, some deeper understanding of the concepts of reection, transmission, and
absorption are necessary. Absorption, α, like reection and transmission, is dened as the
fraction of light absorbed by a surface or media compared to the total incoming radiation
(Eq. 1.13). Absorbed radiation is usually converted into heat and thus leads to losses in
light transfer.
absorbed part of incoming radiation

α= (1.13)
All real surfaces and media will have some combination of reection, ρ, transmission, τ , and
absorption, α. Therefore these quantities are related as
ρs + ρd + τ s + τ d + α = 1, (1.14)
where the superscripts s and d represent specular and diuse respectively.

All optical systems for illumination have components with varying optical properties,
none of which are 100% ideal. This is the primary cause of optical ineciency or loss.
The optical eciency, η, of an illumination system (sometimes called target eciency) is
expressed as the ratio of total radiation that reaches the target to the radiation emitted by
the source (Eq. 1.15).
Φtarget
η= (1.15)
Φsource
The optical eciency of a system can be estimated by multiplying the reectance or trans-
mittance of each optical component within an illumination system;
N
Y
ηo = ei , (1.16)
i=1
where ei is the reectance or transmittance of the ith optical element. Equation 1.16 is a
gross over simplication of the optical eciency, but it works as an approximation. Along
with the reectance and transmittance, the geometry of the optics plays a crucial roll in
determining how light is distributed across a target.
9
1.3 Optics for Illumination
1.3.1 Imaging Optical Systems
An imaging optical system is either a reective or refractive optic that preserves the "image"
of the light being transfered. These optics tend to be highly specular to ensure the maximum
throughput and control the direction of the light from source to target. An ideal imaging
optic will preserve all of the information from a light source (the radiance, the direction
cosines, etc.) and project them to the target. The oldest example of an imaging optics
system dates back to around 500 BCE in China[7] with the development of the camera
obscura, or pinhole camera (Fig 1.8). The pinhole camera uses a small hole in a wall or
other barrier to bend the light passing through one side of the hole to form an image on the
other side. The pinhole acts as a refractive lens in this case by constraining the amount of
light that can pass through it.
Figure 1.8: A camera obscura, commonly known as a pinhole camera.
A true refractive lens uses the principles of least time, otherwise known as Fermat's
principle[2], which states that a light ray traveling between two points will take the path of
least time to travel. Fermat's principle is what led to the eventual development of Snell's
law and the Fresnel relationships for the behavior of light rays passing from one medium to
the next. The design of lenses are based on these principles and can be simply dened by
the lens makers equations[8, 9, 10, 11]. The lens makers equations dene the focal length of
a lens based on lens axial thickness, d, and the radius of curvature of the rst and second
surfaces, R1 and R2 respectively. There are two primary types of lens curvature, concave
and convex, seen in Fig 1.9(a) and (b). The dierence between a concave and a convex lens
mathematically is simply a sign change in either R1 or R2 . If the lens thickness is small
in comparison to the surface curvature radii then the thin lens approximation can be used
(Eq. 1.17). Otherwise the thickness of a lens must be factored in to the full lens makers
equation, Eq. 1.18.
Thin Lens:

1 1 1
= (n − 1) − (1.17)
f R1 R2
Thick Lens:

1 1 1 (n − 1)d
= (n − 1) − + (1.18)
f R1 R2 nR1 R2
10
Figure 1.9: Two simple lens shapes: (a) a convex lens, and (b) a concave lens. Depicted are the parameters
used in the lens makers equations: f , d, R 1 , and R2
In reality, lenses with spherical curvature will have inherent ineciencies at focusing an
object. This is primarily due to spherical and chromatic aberrations[9, 11, 2, 8]. Spherical
aberration is due to the spherical curvature of a lens surface, which produces dierent focal
distances between light rays traveling near the edge of a lens and light rays traveling closer
to the optical axis (Fig. 1.10(a)). Chromatic aberration is due to the index of refraction of
a transparent material being dependent on the wavelength of light passing though it, which
produces dierent focal lengths for dierent colors of light (Fig. 1.10(b)).
Figure 1.10: Diagram depicting the eects of (a) spherical aberration, and (b) chromatic aberration on the
focal length of a lens.
11
Both spherical and chromatic aberrations are usually corrected by using multiple lenses,
sometimes of diering indexes of refraction. In some cases spherical aberrations can be
corrected by creating an aspherical surface, which uses a conic constant to adjust the radius
of curvature as it approaches the outer edge of a lens. This adjustment is commonly called
lens sag and is dened by the equation:
r2
z= q (1.19)
2
R(1 + 1 − (1 + κ) Rr 2 )
where z is coordinate along the lens optical axis, r is coordinate along the lens radius, R is
the lens surface radius of curvature, and κ is the conic constant. κ will dictate how a lens
surface will deviate from the spherical radius of curvature as shown in Fig. 1.11. The actual
conic surface deviation generated by κ are described in Table 1.3.
Figure 1.11: Diagram of the deviation of the surface of an aspherical lens from that of a spherical lens.
The surface curvature deviation, or "sag" Z, is a function of a conic constant, κ, and the spherical radius of
curvature, R.
Table 1.3: Eect of κ values on an aspherical lens shape.
κ Conic Section
κ < −1 hyperbola
κ = −1 parabola
−1 < κ < 0 ellipse (prolate spheroid)
κ=0 sphere
κ>0 ellipse (oblate spheroid)
12
Aside from lens optics, there are geometric reector shapes that fall under imaging
optical systems. These imaging reector shapes are based predominantly on conic sections;
the circle, ellipse, and parabola. Light incident on the aperture of these conic sections will
be reected back to a focal point. Inversely, any light emitted from the focal point of a conic
section will be reected out to a target (Fig. 1.12(a)-(c)). These imaging reector proles
are useful when trying to direct light from a point source, which radiates in all directions,
into a single dened direction. Imaging reectors have been used for centuries to distribute
light from candle and oil lanterns as well as in modern lament bulb illumination xtures.
Since most conventional light sources are not ideal point sources, imaging optical systems
produce aberrations in the image of the light source on the target.
Figure 1.12: Diagrams of basic reector based imaging optical systems: (a) a circular reector, (b) an
elliptical reector, (c) a parabolic reector
13
1.3.2 Nonimaging Optical Systems
An anecdotal description of nonimaging optics is to think of an optical system as a bowl of
alphabet soup, and you are challenged to t all the letters of the alphabet in the smallest
bowl possible. In an imaging optical system, you are forced to keep all the letters arranged
from A to Z in alphabetical order, reading correctly from left to right. At a certain point
the letters become so pressed against one another that the soup bowl can get no smaller.
Nonimaging optics on the other hand doesn't care about the order of the alphabet soup
nor whether the letters are upright or stacked on top of one another. In this case the bowl
of alphabet soup can become much smaller than the one conned by the rules of imaging
optics. Nonimaging optics does not preserve the delity of an image but does preserve the
total information an image contains. In other words, nonimaging optics will not preserve
a particular pattern of light from one surface to the next, but it will ensure that all of the
light from one surface will reach another. This simplied description of nonimaging optics
is what makes it so benecial to elds of radiative transport and concentration but can also
make it less intuitive to understand.
Nonimaging optics was rst developed over fty years ago for use in the detection of
erenkov radiation[12]. The design used was later termed the compound parabolic con-
centrator (CPC), which is comprised of two parabolas facing one another rotated at an
angle equal to the acceptance angle of the concentrator. The focal point, p, of each rotated
parabola is placed at the edge of the source plane (Fig. 1.13).
Figure 1.13: Diagram of the geometry of a compound parabolic concentrator.
14
The CPC design reaches the thermodynamic limit for radiative transport between surfaces[13],
otherwise know as the maximum theoretical concentration ratio, Cmax . The maximum con-
centration ratio is an inverse relationship to the sine of the acceptance angle of a concentra-
tor. This is dierent than the geometric concentration ratio which is a ratio of the aperture
to receiver area. Unlike imaging optics, nonimaging optics can reach the maximum limit of
concentration in both two and three-dimensions.
Geometric Concentration:
Aperture Area
Cg = (1.20)
Receiver Area
2D Concentration:
1
Cmax2D = (1.21)
sin θ
3D Concentration:
1
Cmax3D = (1.22)
sin2 θ
Another common nonimaging optical design is the compound elliptical concentrator
(CEC). Instead of using rotated parabolas like the CPC, the CEC uses rotate ellipses (Fig.
1.14). Aside from the shape, the key dierence between a CPC and a CEC is while a CPC
emits light to a plane innitely far away, the CEC is designed to emit light ideally to a
target plane at a nite distance. This is accomplished by placing the second focal point of
each ellipse at the edges of the target plane (Fig. 1.14). Both the CPC and the CEC can be
shaped as a reector or out of a solid dielectric media, and have been used in lighting and
solar concentration elds. Often the 2D proles are extruded as a trough, but sometimes
the proles are revolved about a central axis to create a 3D shape.
Figure 1.14: Diagram of the geometry of a compound elliptical concentrator.
15
There are other means and methods for developing a nonimaging concentrator/illuminator,
but they all are limited by the conservation of étendue, or the conservation of light momen-
tum in phase space[14, 2, 13]. Étendue is a limiting factor in the transfer of light through
an optical system, which is dened mathematically as
dG = n2 dA cos θdΩ, (1.23)
where n is the index of refraction of the media, dA is the innitesimal surface area element,
cos θ is the projection of the surface with respect to its surface normal, and dΩ is the solid
angle of the light emitted or passing through the surface. Étendue has units with respect
to area, since all quantities in eq. 1.23 are unit-less except for dA. Integrating eq. 1.23 over
the surface element and all angles, the equations for étendue can be dened in both 2D and
3D cases (Eq. 1.24 & 1.25).
2D:
G2D = 2nA sin θ (1.24)
3D:
G3D = πn2 A sin2 θ (1.25)
Étendue can be represented as a four-dimensional volume in phase space. The axes of this
phase space are X , Y , L, and M, where X and Y are the Cartesian vectors, and L and M
are the direction cosines of X and Y respectively. Étendue can be transformed though an
optical system by adjusting the angle or area through which light passes, but the quantity of
étendue must always be conserved. Nonimaging optic design focuses on engineering étendue
for tailored light transfer, either via concentration or illumination[2, 15]. With nonimaging
optics, a Lambertian source of a certain half angle of emission can be transformed into a
Lambertian source of a dierent emission half angle. This property of nonimaging optics
produces the sharp cuto angle ideal for prescribed illumination.
16
1.4 Characterization Standards and Metrics
Most modern characterization standards for lighting and illumination are based on the
photometric properties and human visual response dened by the CIE in 1931. Though
photometric properties are still the standard used within the illumination industry today,
there have been many advancements in the understanding of human vision which make the
lumen a non-ideal characterization metric. The human eye is composed of photo-receptor
cells called rods and cones. These cells, located on the inner surface of the eye, convert the
impact of photons into electrical neural signals to the brain which are interpreted as vision.
There are three types of cone cells (Erythropic, Chloropic, and Cyanopic), and one type of
rod cell (Rhodopic)[16]. The cone cells are responsible for color vision with each cell type
having a spectral response curve within a short range of wavelengths (Fig. 1.15).
Figure 1.15: Relative spectral response of human vision for the Erythropic, Chloropic, and Cyanopic cone
cells and the Rhodopic rod cells along with the Melanopic response plotted against the Photopic response
curve. Data adapted from the CIE TM 003-2015[3]
The Erythropic or L-cones respond to longer wavelengths, the Chloropic or M-cones

respond to medium wavelengths, and the Cyanopic or S-cones respond to short wavelengths
of light. The Rhodopic rod cells are responsible for dark or dim light vision. Recently a great
deal of study has been conducted into what is called the Melanopic response spectrum[17, 16],
which suppresses the production of the hormone Melatonin in humans and other mammals.
The Melanopic response and subsequent suppression of melatonin is crucial to circadian
rhythm, or the day/night, awake/sleep cycle. Based on the rods, cones, and melanopic
response spectrum in Fig. 1.15 it is plain to see that the Photopic response curve does
not encompass all the information about human vision. Therefore the lumen and other
photometric properties may not be the best means to characterize light and illumination.
Regardless, color and the perception of color are integral parts of human vision and must
be addressed when discussing illumination.
17
1.4.1 Color Metrics
Color and the concept of color rendering play an important role in illumination characteri-
zation. The color that is perceived by the eye is merely light reected o of a surface, and
is highly dependent on the surface and the light source. Color is often dened based on the
CIE 1931 x, y chromaticity diagram (Fig. 1.16), which is a 2D representation of the color
of a light source, or reected/transmitted light. The spectrum locus line in Fig. 1.16 shows
the chromaticity coordinates for monochromatic light, while a more broadband spectral light
source may fall within the boundary of the diagram. SPDs of light sources and the spectral
reectance or transmittance function of samples can be used to calculate x, y chromaticity
coordinates [4, 5].
o
Figure 1.16: The CIE 1931 2 chromaticity diagram showing the location of the spectrum locus, purple
boundary and blackbody locus. Several lines of constant CCT are shown intersecting the blackbody locus.
Figure adapted from [4]
From the x, y chromaticity diagram concepts like the correlated color temperature
(CCT), and color rendering index (CRI) can be determined for a light source. The CCT
is the blackbody temperature, in units of Kelvin [K], whose spectrum resembles the chro-
maticity of a light source. CCT is often used to categorize lighting color with lower CCTs
producing a more orange-red light, and higher CCTs producing a more white-blue light. The
CRI is a measure of how well a light source renders the color of objects in the environment
being lit. CRI is experimentally determined by measuring the average color shift of a set
of 14 standard color samples (Table. 1.4) illuminated by a light source in question when
compared to the illumination of a reference light source of the same CCT. CRI values range
from 0 to 100 with light sources that produce a large chromaticity shift in the measured
samples with respect to the reference source have a low CRI while light sources that produce
small chromaticity shifts have a high CRI.
18
Table 1.4: CIE test color samples for calculation of CRI. Adapted from [4]
Test x, y Cromaticity Color Approximate

Color # x y Description Appearance
1 0.375 0.331 Light grayish red
2 0.385 0.395 Dark grayish yellow
3 0.373 0.464 Strong yellow green
4 0.287 0.400 Moderate yellowish green
5 0.258 0.306 Light bluish green
6 0.241 0.243 Light blue
7 0.284 0.241 Light violet
8 0.325 0.262 Light reddish purple
9 0.567 0.306 Strong red
10 0.438 0.462 Strong yellow
11 0.254 0.410 Strong green
12 0.155 0.150 Strong blue
13 0.372 0.352 Light yellowish pink
14 0.353 0.432 Moderate olive green
CRI has its limitations; the biggest being the averaging of color shift in all 14 samples
and the assumption that the reference light source is the ideal. In some cases a light source
with a high CRI may render the color of 12 out of 14 standard samples fairly well, but
extremely poorly for the other two, causing those colors to appear faint or gray under that
light. The reliance on the reference source being the ideal for color rendering also causes
issue when comparing lights of dierent CCTs which will always render color dierently. A
reference source may also not produce enough perceived brightness or color saturation to
an observer, and therefore may not be an ideal source for comparison. In general, every
person perceives light and color dierently. There are some more modern metrics that try
to diversify how light and color rendering are characterized[18]. However so many color
metrics are based on the standard observer dened by the CIE photometric response, and
can only approximate the color quality of light for a general population.
19
1.4.2 Geometric Metrics
The experimental measurement of light is called photometry, and there are a variety of
dierent methods used to quantify light in terms of photometric properties. An illuminance
meter can measure the [lm/m2 ] at a target, and an integrating sphere or photometer can
be used to determine the total luminous output of a light source. However, there are
only two characterization methods which are commonly used to geometrically calculate the
performance of an optical system: view factors, and ray tracing.
Figure 1.17: Conguration parameters for calculating the view factor between two surfaces based on eq.
1.26
View factors, F, are derived from radiative heat transfer, such that the view factor is
the fraction of total radiation leaving one surface that reaches another. View factors are
purely geometric and thus unit-less. A view factor between two surfaces can be calculated
with respect to the area of both surfaces, Ai and Aj , the distance between them, S, and
cosine of the angles, θi and θj formed between each surface normal of each surface and a
vector along S as shown in Fig. 1.17. View factors are dened mathematically with values
from 0 to 1 as
Z Z
1 cos θi cos θj
FAi −Aj = dAi dAj . (1.26)
Ai Ai Aj πS 2
Equation 1.26 can be solved for a large number of geometric surface congurations to provide
unique solutions for view factors in a simpler algebraic form.
The view factor from one surface, Ai , to surface, Aj , can be used to determine the
amount of radiation from Ai that reaches Aj for a diusely emitting surface.
Φi−j = FAi −Aj Φi (1.27)
Since view factors are derived from geometric relationships, they obey a select set of algebraic
rules, such as summation (eq. 1.28) and reciprocity (eq. 1.29).
20
N
X
FAi −Aj = 1 (for a closed system) (1.28)
j=1
Ai FAi −Aj = Aj FAj −Ai (1.29)
Couping eq. 1.26-1.29 with the concepts of reectance and transmittance, solutions to the
radiative transport between multiple surfaces can be determined analytically.
Figure 1.18: Raytracing diagram depicting a bundle of rays emitted from a source. Some rays are traced
and either absorbed, transmitted, or reected by the object in their path.
The second geometric metric for analyzing optical performance, ray tracing, involves a
more numerical approach. Computer software is used to generate a ray, or bundles of rays
with some angular distribution emitting from a source object. Each ray bundle has a specic
energy associated with it as a fraction of the source energy, and multiple ray bundles of equal
energy levels are often used. The rays in each bundle are projected in straight lines out from
the source in 3D space until they impact another surface. The ray is then either reected,
transmitted, scattered, or absorbed by the surface, based on that surfaces material properties
(reectance, transmittance, etc.) shown in Fig. 1.18. A random number generator can be
used to determine which rays are either reected, transmitted, scattered, or absorbed as they
hit the surface. For example a surface that has a reectance of ρ = 0.9 and an absorptance
of α = 0.1 should reect 90% of incoming rays and absorb 10%. A random number, Γ,
ranging from 0 to 1 can be calculated for each ray such that if Γ ≤ 0.9 the ray is reected,
otherwise the ray is absorbed and ray tracing ends. Random numbers can also be used in
this way to determine the angle of reection or refraction for diuse and scattering surfaces
or media. New random numbers are generated for each ray every time they impact a new
surface. Rays can also be traced as a function of wavelength, providing dierent solutions
based on the spectral dependence of properties of the surfaces each ray interacts with. Each
ray can be followed in this manner from surface to surface using reection (Fig. 1.6) and
Snell's relationships (eq. 1.11) until all rays have been traced through an optical system
until they are absorbed or they reach their nal target.
21
The number of rays being traced dictates the statistical accuracy of the simulation being
conducted. The results of tracing N ray bundles produces a solution of T (N ), with the exact
solution to the simulation occurring as N → ∞. To estimate with what degree of condence
the solution T (N ) falls within the exact solution T (∞), the number of ray bundles, N , being
simulated can be broken up into k sub-samples such that
k
X
N= Ni (1.30)
i=1
with each sub-sample containing equal amounts of ray bundles. The total solution to the
ray tracing simulation takes the form
k
1 X
T (N ) = NI T (Ni ) (1.31)
N
i=1
The solutions of k sub-samples can then be treated as independent experimental measure-

ments of equal quantity. The variance, σm , of these measurements can be used to determine
our condence for the solution.
k
2 1 X
σm = [T (Ni ) − T (N )]2 (1.32)
k(k − 1)
i=1
From the variance the solution has 68.3% condence that the correct answer T (N ) lies
within the limits of T (N ) ± σm , 95.5% condence within T (N ) ± 2σm , or 99% condence
within T (N ) ± 2.58σm . More analysis on view factors and ray tracing can be found in Ref.
[1, 4, 2, 19].
22
1.5 Structure of this Dissertation
Within this chapter an overview of the science, mathematics, and characterization methods
for illumination optics has been shown. Though not entirely comprehensive, this background
material provides insight into how illumination optics have been developed historically as
well as some more current innovations. The lighting and illumination community uses many
of the methods outlined here, as well as others, to produce new lighting xtures, but there
is still not a clear understanding of what the ideal limits are to illumination and how to
quantify them.
This dissertation will be broken up into a total of ve chapters including this introduction.
The chapter following this one will go into the thermodynamics of illumination and what
makes an optical system an ideal illuminator. From this thermodynamic analysis a new
gure of merit for characterizing illumination will be proposed and evaluated based on
some common illumination designs. The third chapter will explore theatrical spotlighting
a eld of illumination where the optics have not signicantly changed for almost a century.
A new design of spotlight will be presented that couples imaging and nonimaging optics to
produce a highly ecient optical system that outperforms existing spotlighting technologies.
Chapter four will present a novel addition to the nonimaging optics eld in the form of a 3D
asymmetrical concentrator. Most current asymmetrical designs only exist as two dimensional
proles extruded as a linear xture. In chapter four, a method for the extrapolation of an
asymmetric prole into 3D will be presented and the optical performance will be evaluated.
The last chapter will present the overall conclusions from this work. The two innovative
designs presented in chapters three and four will be evaluated with the novel gure of merit
presented in chapter 2. It will be shown that using the principles of nonimaging optics and
my new gure of merit can help inform the designs of ideal illuminators. Finally, potential
future work will be proposed for each project, the gure of merit, the nonimaging spotlight,
and the 3D asymmetric concentrator, to express how these advancements might be expanded
on and adopted.
23
Chapter 2
The Thermodynamics of Ideal Illumination
2.1 Abstract
The advancement of modern lighting technologies has led to many revolutions in lighting
eciency and presentation. The progression from lament bulbs, to compact uorescent
(CFL), and now LED technologies have produced a bounty of energy ecient lighting op-
tions for design engineers and consumers. As the light producing elements of a lighting
xture improve, the limiting factor in ecient illumination is no longer the light source,
but the optical system itself. There are many characterization methods and standards for
dening light for illumination in terms of color and human response. With concerns of how
things like light pollution and energy requirements impact our society and the world around
us, it is critical to understand how well a lighting xture can illuminate a desired area while
minimizing light lost to the environment and maximizing the total radiative intensity (ra-
diance) of a space. This work presents two gures of merit, one for over illumination and
another for under illumination, to characterize the optics of a lighting system based on a ray
tracing methodology. Five common simplied optical design, with four varying beam angles,
were simulated to present these new gures of merit. Results showed that common imaging
optical systems such as parabolic and ellipse reectors struggled to produce a well-lit area
without over illumination, while nonimaging alternatives like the compound parabolic and
compound elliptical reectors were able to approach the thermodynamic ideal of a fully
illuminated area without light lost to the environment. This work hopes to inform illumi-
nation engineers and lighting designers to help improve their optical design to maximize
performance and minimize waste.
2.2 Introduction
There have been many techniques and standards developed over decades and centuries to
characterize the quality of light. The SI units of measuring light such as the lumen and the
candela themselves have been a metric to compare any light xture to that of a simple candle.
Over time more complex means to characterize lighting quality have been developed, such
as the standards set by the International Commission on Illumination (CIE). Many of these
standards like the color rendering index CIE Ra value, corresponding color temperature,
and the CIE 1931 color space [4] rely primarily on spectral distribution of light and how the
human eye perceives color. Other human factor approaches such as brightness requirements
for certain working environments [20, 21, 22, 23] and studies on the spectral adaptation
of human circadian rhythm [24, 25, 26] have seen signicant study. Some work has been
done to develop optical systems that produce uniform illumination across a target in terms
of color [27, 28, 29, 30, 31] and total radiative intensity (radiance) in units of W m−2 sr−1
[32, 33, 34, 35, 36, 37, 38, 39] A signicant amount of work has been done on the concept
of Etendue, or conservation of four-dimensional phase space volume (X, Y, direction cosine
of X, direction cosine of Y). The concept of etendue dilution [40] has been used to dene
when the etendue between two surfaces is not fully utilized or fully lled for the purpose of
24
energy transfer. Etendue dilution gets closer to the idea of what ideal illumination should
be, but it can be non-intuitive and dicult to calculate. Industry standards such as beam
lumens or eld lumens are sometimes used for general illumination applications. Though
useful and easy to understand, the beam and eld lumens concepts don't necessary dene
what is ideal, or the upper limit of what is achievable in a design.
Each of these characterizations of light are important, but many of them are subjective
and based on the variability of the environments being illuminated, and the individual
perception of a human viewer. Though beam lumens, eld lumens, and etendue dilution
come close to the answer, none of the other approaches seem to address major issues facing
our society when it comes to illumination, like light pollution and energy waste [41, 42, 43,
44, 45]. A new approach is necessary to dene the thermodynamic limits of illumination
for any design as to minimize wasted light while maximizing the illumination of a dened
space.
2.3 Methods
2.3.1 The Thermodynamics of Illumination
To establish a gure of merit that categorizes ideal geometrical illumination, the concepts of
under illumination and over illumination need to be quantied. An ideal illuminator should
not miss any part of the desired illumination area, nor lose any light to space outside the
desired illumination area (Figure 2.1). These two concepts can be determined independently
and then coupled together to determine the performance of a lighting system. This section
is intended to outline the thermodynamics of both over illumination and under illumination
based on the concepts of radiative transfer between surfaces.
Figure 2.1: Diagram of (a) the concepts of ideal, over, and under illumination for a desired illumination
area, and (b) an illumination setup with (1) the target illumination area (red), (2) the aperture of a lighting
xture (green), and (3) the light source plane (blue).
25
First, we examine the concept of over illumination by considering the simplied lighting
problem outlined in Figure 2.1(b). An ecient lighting system should ensure that none of
the light misses the target. In other words, all the light emitted from the light emission plane
3 reaches the target illumination area 1. The total luminous power, Φ31 , being emitted by
3 that reaches 1 can be dened as
Φ31 = A3 E3 P31 , (2.1)
where A3 is the total emitting area of the 3 [units of m2 ], E3 is the total emissive power of 3
2
[units of lumens/m or lux], and P31 is the probability of radiation leaving 3 that reaches 1.
P31 , though similar to view factors, F31 , used in radiative heat transfer, is dierent in that
radiative transfer probabilities include any radiation that is either reected, refracted, or
transmitted through an optical system [46]. By examining equation 2.1 it becomes readily
apparent that for all the radiation emitted from 3 to reach 1 with none of light missing the
target, P31 = 1. Assuming that all the radiation emitted from 3 passes through the aperture
at 2, P32 = 1, then if P31 < 1 only a fraction of the total radiation emitted from 3 reaches 1.
This implies that 1 is either over illuminated or the light emitted by 3 is missing 1 entirely.
Any lighting system illuminating a target area will have a radiative probability dening over
illumination of the form,
P31 ≤ 1 (2.2)
with P31 = 1 meaning no over illumination. Any value less than one in equation 2.2 signies
that light emitted from the source is missing the target entirely.
Under illumination is qualied by the light ux emitted from a source not being dis-
tributed across the total target area. To do so in a way that maximizes the full illumination
of the target area, the following thought experiment is proposed. Consider a light source
with a xed emissive power, Es , placed in two congurations:
A The light source placed at the aperture plane:
Φ21 = A2 Es P21 (2.3)
B The light source placed at the light emission plane:
Φ31 = A3 Es P31 (2.4)
The maximum light ux at the target illumination area is achieved when the light source
is placed at the aperture, as in conguration A. However, conguration A often loses a
signicant amount of light to the environment since P21 6= 1. Applying the theory of
reciprocity used in radiative heat transfer calculations (A2 P21 = A1 P12 , and A1 P31 = A1 P13 )
to congurations A and B, equations 2.3 and 2.4 respectively, we nd that
Φ21 = A1 Es P12 (2.5)
Φ31 = A1 Es P13 (2.6)
26
In order for conguration B to achieve the maximum light ux on the target illumination
area, equations 2.5 and 2.6 must be equal to each other, requiring that P13 = P12 . To
understand how P12 and P13 relate to under illumination, their physical meaning can provide
some insight. P12 and P13 are the probability that radiation emitted from 1 will reach either
2 or 3 respectively, so P13 < P12 signies that some of the light from 1 that passes through
2 does not reach 3. This scenario can be thought of as rays emitted from the edge of the
target illumination area that pass through the aperture and miss the light emission plane
by either being reected or refracted back out of the optical system (Figure 2.2). By taking
the ratio of P13 to P12 the amount of under illumination can be quantied as any value less
than one (equation 2.7).
P13
≤1 (2.7)
P12
Figure 2.2: Diagram of the physical signicance of the radiative probabilities P12 (red dashed line) and P13
(blue dashed line) and how they relate to under illumination.
Both equations 2.2 and 2.7 can be used together to quantify how well a lighting system
illuminates a desired area, via over illumination and under illumination respectively. Though
a blackbody assumption was used in the development of both equations 2.2 and 2.7 the
theory of radiative probabilities does not require the light source to be an ideal blackbody. If
a source has a specic angular distribution of radiative intensity (sometimes called radiance),
optics should be designed to t such a source. The radiative transfer probabilities will
still apply since they are based purely on geometry and material properties. Any lighting
conguration with a dened size and location for the emission source, aperture, and target
illumination area can be compared using both P31 and P13 /P12 by plotting them on a graph
such as shown in Figure 2.3. An ideal lighting system will have values for both equations
equal to one. It is interesting that the ideal optics from a thermodynamic point of view will
achieve both at the same time. One can easily design a device that satises equation 2.2
but not 2.7, or the other way around. Only the ideal optics can avoid both over illumination
and under illumination.
27
Figure 2.3: P31 vs P13 /P12 showing the bounding limits of illumination with the ideal located at P31 = 1
and P13 /P12 = 1.
2.3.2 Simulation and Validation

The concepts of both P31 and P13 /P12 are ideally suited to validation through ray tracing.
The denition of radiative transfer probability as dened in section 2.3.1 is the amount of
light emitted by one surface that reaches another, through any optical system including
reected, refracted, and transmitted light. Any ray tracing technique used in most modern
ray tracing software can easily determine these radiative transfer probabilities by emitting
a known number of rays from one surface and counting the number of rays that impact the
other surface. The ratio of these two numbers of rays gives the radiative transfer probability
as shown in equation 2.8. Equation 2.8 deliberately discounts non-ideal reectance and
transmittance through an optical system. If one considers any optics problem from a ray
tracing point of view, each ray in a raytracing simulation is a small volume of etendue in
the four-dimensional phase space (X, Y, direction cosine X, direction cosine Y). By having
less than 100% optical properties one eectively reduces this etendue/phase space volume
connecting the source and the target. The goal of equation 2.8 is to dene the upper bounds
of what is theoretically possible for an illumination scenario. Any optical ineciencies can
be considered within a raytracing simulation to further dene what a realistic performance
might be.
Total number of rays incident on surface j
Pij = (2.8)
Total number of rays emitted by surface i
As a benchmark for testing the under illumination and over illumination gures of merit,
raytracing simulations were done using LightTools software for ve simple reector based
optical designs: tapered/conical mirrors, an elliptical mirror, a parabolic mirror, a compound
parabolic concentrator (CPC) mirror, and a compound elliptical concentrator (CEC) mirror
(Figure 2.4). Each design consisted of a constant source size/area and four dierent beam
half-angles: 15, 30, 45, and 60 degrees. All designs were simulated using the reector proles
shown in Figure 2.4 in both 2Dand 3Dcases, keeping a constant source size/area of 0.01 m
2
and 7.85e-5 m respectively.
28
Figure 2.4: Diagrams of the ve simple reector proles used in benchmark raytracing simulations testing
the P31 and P13 /P12 gures of merit. (a) tapered/conical, (b) elliptical, (c) parabolic, (d) CPC, and (e)
CEC reectors. Blue depicts the light source area, green depicts the aperture area, gray depicts the reector
prole, red depicts the ray path, and the dotted arrow shows the beam angle for each design. Two-dimensional
designs were used as shown in the gure, and three-dimensional designs were developed by revolving the
proles about a central axis.
2.4 Results
Results are compiled from three ray tracing simulations per each reector prole and beam
angle combinations. The three simulations per design were conducted to calculate the pa-
rameters in Eqs. 2.2 and 2.7. Each simulation varied the distance between the target plane
and the aperture plane from 0 to 2 m in 0.025 m increments. The goal of varying the target
distance was to validate the usefulness of using P31 and P13 /P12 as gures of merit for char-
acterizing under and over illumination. All simulations where run in LightTools raytracing
software using ve million rays per simulation. Fig. 2.5 displays both P31 and P13 /P12 as a
function of target distance from the aperture plane, and P31 vs P13 /P12 for each of the ve
reector designs, with Fig. 2.5(a)-(e) correlating to the designs designated in Fig. 2.4(a)-(e)
respectively. Each design beam half-angle is shown as a dierent color in Fig. 2.5 with blue,
green, red, and purple representing 15 degrees, 30 degrees, 45 degrees, and 60 degrees. As
a more direct comparison between each reector design, Fig. 2.6 shows a plot of P31 and
P13 /P12 values at the desired target distance of 1 m.
29
Figure 2.5: Results of the raytracing simulations for the 2D reector proles calculating P31 and P13 /P12 by
varying the target distance L from the aperture. The rst column shows P31 in dashed lines and P13 /P12 in
solid lines with the blue, green, red, and purple representing the 15, 30, 45, and 60 degree beam half-angles
respectively. Figures a, b, c, d, and e correspond to the reector designs depicted in Fig. 2.4(a)-(e). The
second column shows P31 vs P13 /P12 for each design.
30
Figure 2.6: Results of P31 vs P13 /P12 for all (a) 2D and (b) 3D reector designs. The points plotted are
the values calculated from raytracing simulations for a target distance of 1 m from the aperture of each
luminary. Colors distinguish the reector prole and the symbols distinguish each beam half-angle.
31
2.5 Discussion
The most interesting trend in the simulation data was the noticeable transition seen between
under illumination and over illumination. For an ideal design, this transition should occur
at the specied target distance (Fig. 2.7). Except for the tapered and parabolic reectors,
many of the simplied 2D designs studied saw this transition occurring at or near the target
distance of 1 m (Fig. 2.5). In the 2D designs presented in Fig. 2.5, where the luminary
is faced directly at the target illumination plane, under illumination was found to be a
minimum of the ratio of source area to aperture area, seen as the dashed line in Fig. 2.7.
However, this was not the case for the 3D parabolic and elliptical designs. Other lighting
scenarios may produce less illumination on a target area than the source to aperture area
ratio as well. Generally, this limiting ratio makes sense when considering how P13 /P12 is
determined, with the amount of light that can reach the source from the target being limited
by the amount of light that can pass through the aperture area.
Figure 2.7: Trends seen in the P31 vs P13 /P12 data as a function of distance from the aperture to target
illumination area for an ideal illumination lighting xture.
The direct comparison of each design (Fig. 2.6(a) and (b)) is perhaps more signicant in
that it demonstrates the eectiveness of the studied luminaries at illuminating a target at a
dened distance. Almost all the designs saw little to no over illumination (P31 = 1), but only
the elliptical, CPC, and CEC designs in the 2D case and only the nonimaging CPC and CEC
designs in the 3D case approached the ideal of no under illumination (P13 /P12 = 1). Ideal
illumination with little to no loss of light or under illuminated area was only produced by the
CEC reector for all beam half-angles simulated in both the 2D and 3D designs. The CPC
did perform ideally for the 60 and 45 degree beam half-angles, but smaller beam angles fell
short of P13 /P12 = 1. This is primarily because a CEC reectors shape has design criteria
that include a target area at a dened distance, while a CPC denes a target innitely far
away. The most striking dierence between the 2D and 3D results shown in Fig. 2.6 is the
drastic shift in performance of the parabolic and elliptical designs. This is primarily due to
the source shape being cylindrical instead of a disk like the tapered/conical, CPC, and CEC
sources.
32
2.6 Conclusions
The theory and simulations presented here demonstrated the thermodynamics of illumina-
tion, and how to quantify what makes an ideal optical system. Two proposed gures of merit,
one for over illumination and another for under illumination, and the means by which to cal-
culate them via ray tracing were shown. Various luminary reector designs were presented
and compared with these gures of merit to categorize which designs approach or reach the
ideal thermodynamic limit to illumination. Nonimaging optics was the only design shown to
meet the thermodynamic ideal for both gures of merit. Other, perhaps more complex de-
signs not tested here may approach the ideal of P31 and P13 /P12 , but only nonimaging optics
can reduce light loss and maximize light transfer using a simple luminary design. Though
not entirely representative of all possible real-world lighting and illumination designs, the
simulations presented were used as a selective demonstration of how the gure of merit for
under and over illumination can be quantied for a variety of lighting scenarios. The goal of
the authors has been to provide a means by which lighting and illumination engineers can
characterize their designs. Future work hopes to develop a methodology for experimental
validation of the thermodynamic limits to ideal illumination. More work also needs to be
done to explain the appropriate ray distributions used in simulation when calculating P31
and P13 /P12 for a desired distribution of light other than Lambertian. It is hoped that the
theory of illumination presented here can help to inform and develop lighting optics which
can approach the thermodynamic limit.
33
Chapter 3
Coupling of Nonimaging and Imaging Optics for Uniform Illumination in
Theatrical Spotlighting
3.1 Abstract
It has been well established that nonimaging optics is the ideal way to concentrate or funnel
light for applications ranging from solar energy to illumination. One industry where lighting
is critical are theaters. Stage performers in theaters must be illuminated eectively and
consistently for a performance to be observed. Historically theater lighting has been done
using lament bulbs, which produce a lot of heat and provide discomfort to the actors and
patrons. With the advent of LED lighting, many theaters are starting to replace the old
lament bulbs with LED xtures to provide the same lighting eect at reduced energy need.
However, the last holdout for transition to LEDs has been the spotlight, because of the high
standards of theater experts. Typical spotlights use a halogen-tungsten bulb coupled with
an ellipsoid reector to provide illumination of about 10,000 lumens with a power input of
750 W. In this chapter work is presented on the coupling of a nonimaging optics reector
coupled with a 70 W LED xture and an imaging lens column to produce a new type of
spotlight. This new spotlight will be shown to match the 10,000 lumens required by theaters
for spotlighting, at a fraction of the power requirement and heat generation while producing
a uniform illumination.
3.2 Introduction
Stage performance has a history dating back thousands of years to ancient Greece. Since the
introduction of theatrical performance as an art form, the ability to eectively illuminate
the performers and scenery on stage has been a constant challenge. Early lighting systems
consisted of torches and oil lanterns scattered around the stage and held by performers[47].
At the dawn of the 20th century the electrically powered lament bulb provided new and
unique ways to use light to accentuate a performance and not just illuminate it. Perhaps
the most well-known form of theatrical lighting to come out of the early 1900's was the
spotlight, which provided an isolating beam of light to focus on a single performer or object
on stage. Worldwide, the most common spotlight design that is used in performances today,
was developed in 1933 by Century Lighting. Called the Leko light[47], this spotlight is a
form of ellipsoidal reector spotlight (ERS)(Fig. 3.1). The advantage of the ERS design
was the ability to collect and direct light within a specied beam angle, providing a more
uniformly illuminated spot than previously developed lighting systems[48].
In the 21st century, lighting systems have evolved to become more energy ecient. The
invention of the light-emitting diode (LED)has drastically shifted the lighting and illumina-
tion markets within the last couple of decades, including theatrical lighting systems, where
most modern theaters have changed out their old lament bulb lights for LED xtures to
cut electricity cost. Another added advantage of LED lighting over traditional lament x-
tures is that LEDs produce much less heat during operation. In theaters, where lighting
systems can include dozens if not hundreds of light xtures, heat becomes a critical issue
34
Figure 3.1: Diagram depicting a traditional Ellipsoidal Reector Spotlight design.
not only for the performers on stage, but for the audience. The last holdout in theatrical
lighting to adopt the use of LEDs has been the spotlight. This is primarily due to the
demand for illumination delity by lighting technicians and the inability of the ERS design
to easily adapt to the use of LED's in place of a high-power lament system. The ERS
design, though eective at creating a tight spotlight beam, is inherently inecient at the
ideal transfer of light from source to target. ERS designs generate a signicant amount of
wasted light, which never reaches the desired target, leading to excess energy usage. This
chapter will demonstrate the redesign of the theater spotlight system to use a nonimaging
optics compound parabolic collector (CPC) in place of the ellipsoidal reector, coupled with
an LED chip to produce a more uniform illumination and sharper angular cuto than the
traditional ERS, while maximizing light throughput to the target.
3.3 Methods
To create an ideal spotlighting system incorporating nonimaging with imaging optics, two
main concepts need to be understood. The rst of which is Köhler illumination, and the
second is the concept of étendue. Both concepts are necessary to ensure the maximum
throughput of light through an optical system. Using these two concepts together, a design
method for lens columns which can be coupled with a nonimaging CPC to produce uniform
illumination will be proposed.
3.3.1 Khöler Illumination

Köhler illumination[8, 9, 10, 11] is a lighting principle commonly used in microscope and
projection optics, where it is necessary to remove the image of the light source from the
slide plane, so that the slide is not obscured by the light source image. This is essential
when the light source is not uniform, such as a lament bulb. By using collector optics to
uniformly illuminate a slide plane, and condenser optics to project the slide plane to the eye,
microscopes using Köhler illumination provide a clear, uniformly lit, view of a slide. Before
Köhler illumination, a technique termed critical illumination was used in microscope optics
to provide light behind a slide plane. The dierence of the basic optical elements between
critical (or Abbe) and Köhler illumination is shown in Fig. 3.2.
35
Figure 3.2: Diagram depicting the dierence in optical elements of a basic microscope for (a) critical (Abbe)
illumination, and (b) Köhler illumination.
The benets of Köhler illumination for uniform projection have also been applied to so-
lar concentration, where uniform illumination is desired across the surface of a photovoltaic
(PV)cell[49, 50]. Imaging a uniform distribution of light across a plane is directly applicable
to nonimaging optics. Nonimaging optics began in the elds of solar and radiation concen-
tration. The illumination conguration of an optical system can be thought of as the reverse
of a solar concentration system. Since the aperture of a CPC is uniformly illuminated from
the light source placed at the source plane, Köhler illumination can be applied to create a
uniform projection system (Fig. 3.3). Coupling nonimaging optics with Köhler illumina-
tion ensures the maximum throughput of light from source to projection plane with sharp
angular cuto and uniform illumination distribution.
Figure 3.3: A CPC luminary can be a replacement for Köhler illumination in microscope and projection
optics systems to provide a uniform illumination at the slide plane.
36
3.3.2 Étendu Matching from Light Source to Target
Historically, lens design has been accomplished using lens makers equations (outlined in
Chapter 1) derived from Snell's and Fresnel's laws to determine lens curvature and thick-
ness parameters. Modern computational software allows for optimization of lens design by
dening desired light paths. One concept that is poorly understood in the eld of tradi-
tional imaging optics is étendue, or the conservation of light momentum in phase space.
Étendue is a limiting factor in the transfer of light through an optical system, which is
dened mathematically as
dG = n2 dA cos θdΩ, (3.1)
where n is the index of refraction of the media, dA is the innitesimal surface area element,
cos θ is the projection of the surface with respect to its surface normal, and dΩ is the solid
angle of the light emitted or passing through the surface. Étendue has units with respect to
area, since all quantities in eq. 3.1 are unit-less except for dA. Integrating eq. 3.1 over the
surface element and all angles, the equations for étendue can be dened in both 2D and 3D
cases.
2D:
G2D = 2nA sin θ (3.2)
3D:
G3D = πn2 A sin2 θ (3.3)
Étendue can be represented as a four-dimensional volume in phase space. The axes of this
phase space are X , Y , L, and M, where X and Y are the Cartesian vectors, and L and M
are the direction cosines of X and Y respectively. Étendue can be transformed though an
optical system by adjusting the angle or area through which light passes, but the quantity
of étendue must always be conserved.
Nonimaging optics design focuses on engineering étendue for tailored light transfer, either
via concentration or illumination[2, 15]. With nonimaging optics, a Lambertian source of
a certain half angle of emission can be transformed into a Lambertian source of a dierent
emission half angle. This quality of nonimaging optics produces the sharp cuto angle ideal
for prescribed illumination. The advantage of coupling nonimaging with imaging optics
arises when a uniform illumination within a beam angle is desired. By designing both
nonimanging and imaging optical components to match the étendue of a light source, the
total light throughput can be conserved. For example, a planar Lambertian source of 0.01
2
m in diameter will have an étendue in air of 2.47e-04 m , based on eq. 3.3. Any subsequent
reector or lens optic which the light passes through must maintain this étendue. So by
adding a lens of 0.1 m in diameter, the beam angle of the light must be constrained from
180.00
o to 11.48o for the maximum throughput of light to be achieved. Understanding how
the angle of the light must be adjusted to maintain the conservation of étendue can help
inform how the curvature of a lens or reector optic should be designed.
37
Figure 3.4: A Diagram of a two plano-convex lens system designed to project the uniformly illuminated
aperture of a CPC to a desired target.
There are many methods for using lens optics to image an object to a target plane. Only
one such method, shown in Fig. 3.4, will be used in this chapter. This is by no means the
only method to achieve étendue matching with nonimaging optics, but it is a simple method
that does produce a uniform distribution of light on the desired target plane. Two plano-
convex lenses are used to focus an image of the aperture of a CPC to a target plane a certain
distance away. The diameter, D, and emission angle, θo , of the nal lens are dened by the
étendue of the diameter, DCP C , and emission angle, θi , of the CPC aperture. From these
parameters, and using the lens maker's equations dened in Sec. 1.3.1, the focal length,
f, the distance between the CPC aperture and lens 1, dCP C−L1 , and the distance between
lenses 1 and 2, dL1−L2 , are determined such that:
The focal point of Lens 1 is placed at the convex surface of Lens 2.

The distance between the CPC aperture and lens 1 is limited to:
0.5(DL1 −DCP C )
dCP C−L1 < tan θi
Lens one forms a virtual image of the CPC aperture, which Lens 2 projects to the
target plane.
38
3.4 Simulation and Experimental Validation
To validate the methods for developing a nonimaging spotlight, simulation of an optical
design was done using LightTools raytracing software. A two plano-convex lens design
coupled with a CPC reector was implemented using the methods described in the previous
section. The CPC is designed with an aperture of 50 mm with a half angle of emission of
o 2
20 . This produces an étendue of 721.58 mm . Dening a nal desired emission half angle
o
of 13 limits the nal lens diameter to 76.02 mm. The distance between the CPC aperture
and Lens 1 is 22.31 mm, and the distance between lenses 1 and 2 is xed at 100.05 mm,
with a target plane distance xed at 2 m from the aperture of lens 2. Parameters for the two
plano-convex lenses used in this design are shown in Table 3.1 below. LightTools raytracing
simulations were done to validate the uniformity and angular cuto of the design method
and parameters proposed. The results of which are shown in Fig. 3.5.
Table 3.1: Lens parameters for the design of a nonimaging to imaging projection system.
Parameter Lens 1 Lens 2

Material PMMA PMMA
Index of Refraction (550 nm) 1.4936 1.4936
Diameter 76.02 mm 76.02 mm
Thickness 15.20 mm 15.20 mm
Front Surface Radius of Curvature ∞ ∞
Back Surface Radius of Curvature 49.39 mm 73.16 mm
Figure 3.5: Validation of Irradiance uniformity for the CPC and lens design proposed in Table 3.1
39
3.4.1 Optics Fabrication
The goal of the design system proposed here is primarily to fabricate a new type of theatrical
spotlight. This new nonimaging spotlight (NIS)will be able to produce a uniform distribution
of light with a sharp cuto angle, and minimize loss of light though the optical pathway.
Working with a corporate sponsor under the California Energy Commission's CalSEED
funding program, a proprietary imaging lens column and CPC were developed. ERS systems
come in various design formats, often referenced by the beam angle of the light leaving the
spotlight, either variable angle (or zoom), or xed angle. Variable angle spotlights have an
o o
adjustable lens column that can vary the beam angle of the light from 15 to 50 depending on
the manufacturer and design. These types of spotlights are used most often for illuminating
moving targets on a large stage. The more common xed angle spotlights have a single
beam angle that cannot be adjusted. Fixed angle spotlights commonly come in 19 , 26 ,
o o
o
36 , and 50
o designs, with some other rarer xed angles available from certain companies.
o
In this project the 26 xed angle spotlight was chosen to be the baseline for this design of
a nonimaging spotlight.
For the reector prole a revolved CPC was chosen. CPCs are ideally suited for use with
LEDs because of the at circuit board nature of many LED xtures. In this project, a disc
shaped LED chip was chosen as the light source for the spotlight because an ideal source for
a CPC prole revolved about an axis perpendicular to the disc center. CPCs revolved in this
way are referred to as CPC cones. CPC cones are traditionally created from either a solid
dielectric material or a reective surface that follows the CPC curvature. A revolved CPC
cone has surface curvature in two directions which makes the adherence of a at reective
lm impossible. For ease of manufacture the CPC cone in this work was approximated by
segmenting the shape into eight CPC facets. These facets produced a CPC cone with an
octagonal aperture and light distribution[51, 52]. A cylindrical extension was added to the
end of the octagonal CPC cone to even out the edges of the light producing a circular light
distribution (Fig 3.6).
Figure 3.6: Irradiance distribution at the aperture of dierent CPC cone designs. From top to bottom:
revolved CPC cone, octagonal CPC cone, octagonal CPC cone with cylindrical extension.
40
The construction of the octagonal CPC cone was done by creating eight individual,
identical segments. Each segment maintains a 2D prole of the CPC curvature design. The
segments were modeled using a computer aided design (CAD)software. The design of the
segments is such that the sides where adjacent segments line up have corresponding pins
and holes to ensure proper alignment of the curved surfaces. The 3D models where printed
on a Makerbot Z18 lament 3D printer with PLA plastic lament.
Each CPC surface was coated rst with a spray adhesive and then a 3M ESR highly
reective lm. Excess reective lm was then trimmed from the edges so that the octagonal
alignment would be maintained. When the segments were fully assembled to create the
octagonal CPC cone, the optics where visually inspected by placing a brightly colored sheet
of paper at the source opening where the LED chip would sit (Fig. 3.7). If the reection
in all sides of the octagonal CPC cone showed the color of the paper within the acceptance
angle, the optics are guaranteed.
Figure 3.7: 3D printed octagonal CPC cone lined with 3M ESR reective coating. Red paper is used as a
visual reference to ensure ideal optical reection.
The two plano-convex lenses for the NIS prototype were machined from cast acrylic
(PMMA) sheets. Each sheet was cut to a rough round shape then machined on a lathe
down to the correct diameter for each lens. These round "blanks" were then moved to a
computer numerical controlled (CNC)lathe where the exact curvature of for the surface could
be cut down precisely. The curved surface of each lens were then polished by hand with an
abrasive polishing compound and cloth until a transparent nish was achieved. These steps
are all depicted in Fig. 3.8.
41
Figure 3.8: Lens fabrication process. (a) machining of lens blanks, (b) CNC lathe of curved surface, (c)
rough lenses produced on CNC lathe, (d) nal polished lenses.
The nal lens assembly was coupled with the octagonal CPC and cylindrical extension
to perform experimental measurements. A framework was 3D printed to hold the lenses
and CPC/cylinder in the proper arrangement, shown in Fig. 3.9. The light source used
for optical validation was an LED ashlight with a thin tissue paper diuser place at its
aperture to approximate an LED chip of the correct diameter. Measurements were taken on
the wall with a Konica Minolta TL-1 illuminance meter, which returns values of lx [lm/m2 ].
Measurements were taken across the horizontal and vertical diameter of the wall spot in 25
mm increments.
Figure 3.9: Experimental benchtop setup of the octagonal CPC, cylindrical extension, and two lens spotlight
assembly.
42
3.5 Results
As a baseline for comparing the Nonimaging spotlight designs, lighting data was gathered
from product specication sheets for spotlights build by the ETC and Altman Lighting
companies[53, 54, 55, 56, 57]. Three lighting units were chosen from each company; one
being a traditional 750 W ERS and the other two being a 150 W and 250 W LED spotlight
o
respectively. These spotlights were chosen because they are all 26 beam angles with diering
light technologies, reecting the modern lighting xtures available to theaters today. All data
gathered from the ETC and Altman spotlights were freely provided by the manufacturers
on their websites. This data is used as a comparison for the simulation and experimental
results of the Nonimaging spotlight design.
Table 3.2: Industry spotlight choices for comparison to the Nonimaging spotlight designs with published
metrics for power consumption, luminous ux, and luminous ecacy of source
Manufacturer Model Power Field Ecacy

o
Consumption Lumens
ETC Source Four 26 Ellipsoidal 750 W 13690 lm 18.3 lpw
o
ETC Source Four LED Daylight 26 150 W 7237 lm 48.2 lpw
o
ETC Source Four LED-2 Daylight 26 250 W 11489 lm 46.0 lpw
o
Altman PHX ERS 26 Ellipsoidal 750 W NA NA
o
Altman PHX 150 3000K luminary 26 150 W 4114 lm 27.4 lpw
o
Altman PHX 250 3000K luminary 26 250 W 7379 lm 29.5 lpw
The rst performance metric for comparison of the NIS to industry spotlights was to
simulate target distances of 3, 5, 10, 15, and 20 m to determine the decrease in illuminance
as a function of distance. Illuminance falls as a power function of distance, as seen in Fig.
3.10.
Figure 3.10: Peak illuminance as a function of spotlight throw distance for various industry spotlights
compared to simulated results for the NIS spotlight design.
43
Another metric for comparison commonly used throughout the spotlight industry is an
intensity or candela plot. A candela plot is a luminance output plot of the cross section of
the light distribution within the beam angle. This is useful in determining how uniform the
light is across its beam angle and at which angle the light drops by 50% of its peak value. In
Fig. 3.11 normalized candela plots are shown for the industry example spotlights in contrast
to NIS simulations and experimental measurements.
Figure 3.11: Results of the angular distribution of normalized luminous intensity for (a) ETC Source Four
series spotlights, (b) Altman PHX LED series spotlights, (C) NIS designed spotlight, and (d) Combined
luminous intensity proles for comparison.
44
The next means of analyzing spotlight performance is to determine the total luminous
ux within the eld of light on the target. The NIS was simulated using a Cree CXB3070
lamp which operates with an output of 9739 lm at approximately 70 W operation. This
equates to a source ecacy around 140 lpw. A nal simulation and experiment were done
to test the ability of the NIS to focus a projected image on the target. A simple black
and white design of bubbles was placed at the aperture plane of the cylindrical extension
to the octagonal CPC and the lens column was adjusted to focus the image at a target 2
m away. The computer-generated bubble pattern was used for the simulation, while the
experimental version used an approximation drawn by hand on transparent tape Fig. 3.12.
Realistically any pattern could be printed on transparent cellophane and projected using
the NIS, however this option was not available at the time of testing.
Figure 3.12: Projected spot on a target 2 m away showing a bubble pattern for the NIS spotlight (a)
simulated color rendered result, and (b) experimental result.
45
3.6 Discussion
Comparing to some of the industry spotlights, the NIS performed as well or better than
most of the industry examples. The NIS design creates a practically at luminance output
for all angles within the 26° range. Other spotlight designs lose luminous output as the
angle approaches the beam angle edge. Looking at the source ecacy for industry designs,
the NIS simulated performance signicantly better by a factor of three to seven times. This
is signicant since the energy required to produce the same luminous output on a target
is drastically improved when ideal optics are used. The reduction in energy requirements
translate to direct cost savings for both the operation of a spotlight and the air conditioning
and cooling requirements for maintaining theater comfort.
The ERS xtures used in theaters around the country use between 500 and 750 W of
power per spotlight. That power usage equates to 0.5-0.75 kWh of energy for each hour of
usage. Even for an ERS equipped with LEDs running at 150-250 W becomes 0.15-0.25 kWh
of energy usage each hour. With ecient optics coupled with an ecient LED emitter, as
shown in Section II, power and energy consumption for a single spotlight can be reduced to
70 W and 0.07 kWh of energy usage. The potential power and energy reductions of an ideal
optical system equate to a factor of almost ten times reduction in energy usage and cost as
compared to lament ERS xtures, and more than three times reduction compared to the
inecient LED ERS xtures.
Figure 3.13: Energy usage and cost comparison to run a single 750 W ERS, 250 W LED ERS, or 70 W NIS
vs hours of operation.
* Community theater data from Theater Communications Group for 2018[58]
** Broadway theater data from the Broadway League for 2018-19[59]
+ High Performance Lamp (HPL) data from ETC[60]
46
These cost and energy savings are most easily seen in Fig. 3.13. Energy costs were
assumed to be charge at the national average rate of $0.12/kWh. Stage shows have an
average performance run time of 135 minutes per show[61]. Using data for the average
number of shows per year in 2018-2019 for both a small community theater of 92 shows[58]
and a large Broadway theater of 339 shows[59], average time for running a single spotlight
per year are calculated as shown as dashed lines in Fig. 3.13. In reality, theaters around the
country operate more than one spotlight at a time for more than just performing hours. A
Broadway theater running a single 750 W ERS light can consume over 0.57 MWh of energy
every year. Many theaters use dozens or even over a hundred of these spotlights to properly
light a stage creating multiple MWh of energy demand every year. In contrast, existing 250
W LED ERS lights reduce this energy consumption and operation costs by a factor of three,
which demonstrates the added benet of LED technology. The full benet of using LEDs is
only prevalent however when ideal optics are used. The incorporation of nonimaging optics
to spotlight design has been shown here to reduce energy usage and costs to run a single
spotlight for an average Broadway theater from 572 kWh/yr and electrical cost of $69/yr
with the 750 W ERS to 53 kWh/yr and electrical cost of $6/yr with the ideal nonimaging
LED spotlight. Even though these numbers are simplied estimations, they show the drastic
improvement of the NIS over existing ERS models.
Working with the Fresno State business school and RAF electronics, a market impact
study was conducted to determine the benets of replacing ERS xtures with the new NIS
design. Looking at data pertaining to California, over 740 venues such as theaters, churches,
arenas, and amusement parks were estimated to use more than 68,000 ERS type spotlights
throughout the state. From this data estimates for the energy and cost reduction, including
reduction of greenhouse gas (GHG)emissions were calculated as shown in Table 3.3 below.
These estimates assume a 1-10% market adoption over the next few years when calculating
savings based on current utility and GHG emission costs and standards[62, 63].
Table 3.3: Estimated annual market impact for the state of California assuming annual savings for a 1%
and 10% addoption and replacement of existing ERSs with new NIS xtures.
Savings Category 1% Market 10% Market

adoption of NIS adoption of NIS
Energy Reduction 468,559 kWh 4,685,587 kWh
∗
Operational Cost Reduction $78,709 $787,088
∗∗
GHG Emissions Reduction 223 MTon CO2e 2,232 MTon CO2e
+
Avoided Economic Damage from Climate Change $9,373 $93,728
∗
= Assuming a annual operational and utility cost savings of $114 per spotlight with the new NIS over
existing ERS xtures
∗∗
= Assuming 1 kWh produces 1.05 lbs of CO2[63]
+
= Assuming an economic impact of $42/MTon CO2e[62]
47
3.7 Conclusions
The work presented hear shows the advantage of coupling nonimaging optics with LED
lighting to produce an eective theater spotlight design. The prototype creates a soft white
spot, with correlated color temperature of 3000 K, of even illumination. It has been shown
that this NIS design can meet theater industry standards while outperforming existing com-
petitors in the market. One nal advantage to the NIS design, which cannot be understated,
is its low power consumption. The heat produced by the LED xture can so easily be con-
trolled by a simple heat sink that minimal shielding is necessary to protect operators from
injury, while also minimizing heat addition to the theater. Not only is a Nonimaging spot-
light a more electrically ecient lighting system, it can also potentially reduce cooling and
air conditioning costs for theaters around the world. These costs and energy savings are
best seen in the market impact study conducted by Fresno State for the state of California
(Table 3.3). With only a 10% addoption of the new NIS, California could reduce its energy
usage by more than 4.6 million kWh and see a reduction in GHG production by 2,232 metric
tons of CO2. Savings of this magnitude truly highlight the benets of coupling nonimaging
optics with LEDs.
48
Chapter 4
The Three-Dimensional Asymmetric CPC
4.1 Abstract
Ideal three-dimensional concentration has been sought after for decades in the solar and
optical world. Nonimaging optics has been able to achieve this thermodynamic maximum
concentration limit with certain symmetrical designs, but there has yet to be a truly asym-
metrical concentrator design that reaches the thermodynamic limit to radiation transfer in
three-dimensions. This work outlines a new design approach that hopes to move towards the
development of an ideal 3D asymmetric concentrator based on the current asymmetric com-
pound parabolic concentrator (ACPC)design principles. A geometric scheme is presented to
revolve a variable ACPC curve about a central axis. The 3D ACPC is compared to a stan-
dard compound parabolic concentrator (CPC) in a tilted and angular truncated case. Each
design was characterized for performance by examining the angular acceptance/illumination
region in angular direction cosine space. Ideal performance was demonstrated for the 3D
ACPC design for o axis circular Lambertian acceptance region in direction cosine plots,
while the CPC designs failed to maintain a circular Lambertian distribution. High aperture
tilt angles demonstrate an interesting aring of the aperture which caused a decline in ideal
performance. Despite design deformations at high aperture angles, the 3D ACPC design
method presented here is a step towards ideal 3D asymmetric concentration.
4.2 Introduction
Over the fty plus years of history of nonimaging optics[12, 64], many variations of designs
have been made. Beginning with the simplied two-dimensional form of the Compound
Parabolic Concentrator (CPC) to a multitude of more complex three-dimensional designs
for varying source/absorber geometry[65]. Most designs follow simple rules concerning the
edge ray principle[66, 67, 68, 69, 33] and the law of conservation of etendue. The 2D format
allows for a great deal of designs to be produced which can reach the maximum theoretical
limit for concentration[40, 70]. These 2D designs are often extruded as a 3D trough[71]
to allow for use in real world solar and lighting applications. When moving to a truly 3D
design, many nonimaging designs simply revolve a 2D nonimaging prole about a central
axis[72, 73, 74], or create a faceted approach[52]. This approach has worked eectively
for many solar concentrators and lighting systems, even though it may not fully reach the
thermodynamic limit of concentration. Such nonimaging optical systems revolved about a
central axis produce an inherently symmetrical concentrator. Symmetry is useful when the
source and target are both perpendicular to the axis of revolution. When the source or tar-
get are oset or angled dierently from one another, simplied symmetric designs perform
less ideally. Work has been done to develop various nonimaging designs using owline and
other methods to produce more asymmetrical concentrator proles. These asymmetrical
concentrators overcome the need for perfect symmetry and allow for much more variability
in designing ideal optical systems for any scenario. The one major limitation for asymmet-
rical nonimaging concentrators is that their asymmetry makes them dicult to extrapolate
49
into three-dimensions. Some work has been done using a trough design extruded from
a 2D prole[75, 76, 77, 78, 79]. Others have done work in generating asymmetrical rev-
olutions of generally symmetrical optical systems to account for various source/absorber
geometry[80, 81]. An example of a hyperbolic asymmetric concentrator with elliptical aper-
ture and receiver has even been shown[82], but little other work has been done to advance
three-dimensional asymmetric concentration. This paper will outline a new approach to
developing a unique three-dimensional asymmetric concentrator based on the current design
principles of asymmetrical nonimaging optics[83]. The motivation is that such a concentrator
could have major benets to solar concentration, day-lighting in buildings, and illumination
optics applications where a source/target plane is not parallel to the aperture of the optics.
4.3 Methods
4.3.1 The Asymmetrical CPC in Two-Dimensions
The design of the asymmetrical compound parabolic concentrator (ACPC) was developed
in a similar manner to the design of the traditional CPC. The major dierence being that
while two-dimensional design of the CPC has a singular acceptance angle, the ACPC has
two, one for each parabolic prole. Other than that dierence, the design procedure is
the same between the CPC and the ACPC; a parabola is generated based on the design
constraint of the source/absorber size/shape, then that parabola is rotated about its focal
axis by the desired acceptance angle. This parabolic prole is mirrored for the CPC, but the
procedure is repeated for the ACPC using a secondary rotation angle for the second prole.
Figure 4.1 shows the design parameters and proles for both the CPC and the ACPC in
two-dimensions.
(a) (b)
Figure 4.1: Design parameters and two-dimensional proles for both the (a) CPC and (b) ACPC nonimaging
concentrators.
50
While the rotation angle for the CPC proles corresponds directly to the acceptance
angle, recent work has shown that this is not the case for the ACPC[83]. The rotation or
design angles for the ACPC create a tilted aperture which creates new acceptance angles
for both sides of the ACPC. The relationship between the design angles, the aperture tilt
angle, and the acceptance angle of the ACPC as dened by Ricketts et. all[83] are shown in
Eqs. 4.1 and 4.2 below.
θL = ϕL − α t (4.1)
θR = ϕR + αt (4.2)
4.3.2 Extrapolation to Three-Dimensions

The asymmetric nature of the ACPC and the angular dierence between the source/absorber
plane and the aperture plane create the major diculties in developing a three-dimensional
asymmetric concentrator. One method to achieve 3D concentration using an ACPC design is
to start with the two parabolic proles of an ACPC in 2D and use them to generate a circular
0 − P 0 and P
aperture and receiver with diameters of PLo Ro Lo − PRo respectively (Fig. 4.2(a)).
0
New points PLi ,
0
PLi , PRi , PRi are generated along these circular proles. These points are
radially symmetric about an axis through the origin and perpendicular to the XY plane.
Points are dened on a plane at an angle β from the XZ plane (Fig. 4.2(b)). These new
points are then used to develop the parameters to generate new ACPC parabolas on this β
plane (Fig. 4.2(c)). The 3D ACPC surface (Fig. 4.2(d)) is generated by stitching together
the parabolic proles created by sweeping β from 0 to π. Ideally this "stitching" process
would create a smooth surface. However, because of design and modeling limitations, the
surfaces used in this work are faceted by interpolating at segments between swept proles.
(a) (b) (c) (d)
Figure 4.2: (a) XZ plane showing original 2D design parabolas (solid blue and red) for an ACPC, (b) XY
plane view of the circular aperture and receiver (magenta and green respectively) of the 3D ACPC revolution,
(c) the β plane showing the new ACPC parabolas (dashed blue and red) rotated by an angle β from the XZ
plane, and (d) Isomorphic view of a 3D ACPC surface generated by stitching together parabolic proles
swept around a center axis.
51
4.3.3 Simulation
Any ray of light can be dened in terms of its direction cosines in three-dimensional space.
The direction cosines of a ray can be dened as its projections along the X, Y, and Z axes
using spherical coordinate variable θ and ϕ, as shown in Fig. 4.3. The direction cosines are
dened mathematically in Eqs. 4.3-4.5 where L, M, and N are the direction cosines in X,
Y, and Z respectively.
L = sin θ cos φ, (4.3)
M = sin θ sin φ, (4.4)
N = cos θ. (4.5)
Figure 4.3: A ray (red arrow) depicted in Cartesian co-

ordinate system, dened by spherical coordinate sys-
tem variables θ and φ.
The goal of a three-dimensional ACPC is to shift the acceptance/illumination region of
light in direction cosine or LM space of the concentrator with respect to the aperture. The
key is to preserve the circular distribution in LM space of a revolved CPC while shifting
that distribution along either the L or M direction cosine axis, seen in Fig. 4.4. This shift
can be dened by using the relationship between the acceptance angle and the LM space
distribution of a revolved CPC (Fig. 4.4(a)). If the diameter of the circular distribution in
LM space is 2 sin θ, then the shifted LM distribution should maintain this diameter. The
acceptance angles for a preliminary ACPC curve can now be dened in the form such that:
sin θR + sin θL = 2 sin θ. (4.6)
Using the relationship in Eq. 4.6, a set of acceptance angles can be calculated for any
3D ACPC that maintains a desired circular distribution shifted in LM space. However,
the acceptance angles are not the parameters needed to design the parabolic curves of an
ACPC. An algorithm was developed to generate all variations of ACPC parabolic curvature
to produce a database of the relationship between ϕL , ϕR , θL , θR , and αt for all possible
ACPC designs. From this database any desired acceptance angles (Fig. 4.4(b)) could be
used as an input parameter to dene the design angles of an ACPC based on the relationships
of Eqs. 4.1 and 4.2.
52
(a) (b)
Figure 4.4: LM space direction cosine plots of the acceptance/illumination region for (a) a standard revolved
CPC, and (b) the desired shifted space for an 3D ACPC. The black circle represents the unit circle of 2π
steradians and the red circle represents the direction cosine space for the concentrator.
To characterize the performance the 3D ACPC design approach presented, comparison

simulations were made using a traditional revolved CPC of similar design angle. To create a
meaningful comparison between ACPC and CPC performance, four scenarios were presented
as outlined in Fig. 4.5 below.
(a) (b) (c) (d)
Figure 4.5: Diagrams of the reector proles for: (a) CPC with bent tube, (b) ACPC, (c) tilted CPC, and
(d) sliced CPC. Each reector design has an aperture tilted at the same angle, αt .
53
A standard revolved CPC of 30
◦ acceptance half-angle with a bent tube extension was used
as a baseline (Fig. 4.5(a)). The two other designs used for comparison to the 3D ACPC
where a standard revolved CPC, tilted and sliced at the same angle as the αt angle of the
ACPC aperture (Fig. 4.5(c) and (d) respectively). Ten trials, not including the standard
30
◦ CPC design, were conducted. The goal was to present various designs which could shift
the circular distribution of light in LM space from the center of the unit circle to the upper
edge. Angular parameters for the ten 3D ACPC designs simulated are shown in Table 4.1
below. All designs were simulated using LightTools raytracing software with ve million rays
emitted from the source/absorber plane and a spherical far-eld receiver placed to capture
the distribution of rays. The spherical far-eld receiver was oriented in a way such that
its Z-axis was perpendicular the the aperture plane of each design. MATLAB was used to
process the data from the far-eld receiver and generate LM space plots for each simulated
design.
Table 4.1: Angular parameters for the 3D ACPC designs simulated. All parameters are dened in degrees.
Trial # θL θR αt ϕL ϕR
0 30.00 30.00 0.00 30.00 30.00
1 36.53 23.87 9.53 27.00 33.40
2 39.40 21.40 13.60 25.80 35.00
3 42.27 19.13 17.57 24.70 36.70
4 45.46 16.74 21.96 23.50 38.70
5 49.42 13.78 27.52 21.90 41.30
6 52.38 12.02 31.48 20.90 43.50
7 56.59 9.51 37.29 19.30 46.80
8 60.73 7.37 43.03 17.70 50.40
9 65.87 4.83 50.47 15.40 55.30
10 72.05 2.65 59.55 12.50 62.20
54
4.4 Results
Each simulation produced results on the spherical far-eld receiver on a grid of spherical co-
ordinates, θ and φ, in mesh units of Watts/steradian. To convert these values appropriately
to LM direction cosine space, the spherical coordinates can easily be converted to L and M
coordinates according to Eqs 4.3 and 4.4. To convert the mesh units from W/steradian into
direction cosine space we can think of the steradian as being dierential element of a solid
angle, dΩ. This dierential element of a solid angle is dened by direction cosines as
dLdM
dΩ = . (4.7)
dN
The mesh grid of the spherical far-eld receiver can then be converted to the form W/dLdM
by dividing the W/steradian value by dN . Figure 4.6 shows results of a standard 30
◦ CPC
for comparison, and Fig. 4.7 show some examples of the results for the ten trials conducted
for the 3D ACPC designs, including results for the tilted and sliced CPCs. The red circle
in these gures represents the desired acceptance/illumination region, and the results are
shown as a distribution of light represented by a color gradient.
◦
Figure 4.6: LM space direction cosine simulation for the distribution of light from a standard 30 CPC.
The red circle show the desired acceptance/illumination region.
55
◦ ◦ ◦ ◦
Figure 4.7: LM space direction cosine plots of four αt angles (9.53 , 21.96 , 37.29 , and 59.55 ) for each
of the concentrator designs outlined in Fig. 4.5. The red circles show the desired acceptance/illumination
region for each αt angle.
56
The method outlined in sec. 4.3.2 for generating new parabolic curves is inherently over
constrained. The eects of this over constrained problem are apparent when high aperture
tilt angles are involved. Figure 4.8(a) shows a distinct aring eect which is produced by
the original 3D ACPC design approach.
(a) (b)
Figure 4.8: Results of extreme aperture tilt angle, αt , on the design and performance of the 3D ACPC. (a)
isomorphic rendering of the aring eect in the initial 3D ACPC design, (b) LM space distribution for the
3D ACPC design. The jagged artifacts at the edge of the aperture in (a) are due to the faceted structure of
the design.
The parabolic curves of an ACPC are dened by two parameters; the focal length, f, and
design angle, ϕ. The focal length of both the left and right parabolas can be dened as
a function of the distance between PLi and PRi , shown in Fig. 4.2(c). The design angles
←−−−→0
←−−−→0
ϕLi and ϕRi are dened by the angles formed between the lines PLi PRi and PRi PLi and
a vector normal to the XY plane respectively (Fig. 4.1(b)). A parabola fully dened by
these parameters can only be constrained in space by a single point. The point chosen to
develop the faceted stitched 3D ACPC in the original approach constrained the left and
right parabolas at PLi PRi respectively, then truncated each curve once they reached
and
the design angles ϕLi
and ϕRi . This approach maintained the circular geometry of the
source/absorber plane but created a non-circular, non-planar aperture at high αt angles (see
Fig. 4.8(a)). This non-planar aperture drastically increased the height of the concentrator
design. For this reason, a second truncation step was taken to limit the height of the
parabolic curves to the design aperture circle plane dened in sec. 4.3.2 above. The nal
design rendering and LM space results for this aperture height truncation are shown in Fig.
4.9 below.
57
Figure 4.9: Sample of results of aperture tilt angle, αt , on the design and performance of the truncated 3D
ACPC. The rst row denes the αt angle, the second row depicts the LM space distribution, and the nal
row shows the isomorphic view of each truncated 3D ACPC design.
4.5 Discussion
The most dening results seen in the 3D ACPC comparison to the other traditional CPC
designs was that the 3D ACPC maintained a circular distribution shifted in LM space for
most αt angles, while the CPC designs became more elliptical in shape as the tilt/slice angle
increased, shown in Fig. 4.6. An interesting note is that the tilted CPC no longer maintained
a Lambertian distribution in LM space, but the CPC + bent tube and sliced CPC did (Fig.
4.7), even though they had the same elliptical shaped boundary. The 3D ACPC kept a
Lambertian distribution within the desired circular boundary except at high aperture αt
angles as shown in the ared version of the ACPC in Fig. 4.8(b). The aring eect
starts to occur when the αt angle becomes greater than or equal to that of the minimum
design angle, either ϕL or ϕR . There are two typed of aring that were observed; an
outer-aring which pushes the parabolas near the side with the larger of the two design
angles outside the boundary of the aperture circle, and an inner-aring which pulls the
parabolas near the opposite side inside the aperture circle boundary (Fig. 4.10).
58
Figure 4.10: Top-down view of a 3D ACPC rendering depicting the outer and inner aring eects seen at
extreme αt angles. The jagged artifacts at the edge of the aperture are due to the faceted structure of the
design.
A secondary design step to truncate the parabolic curves at the aperture circle plane was
considered. This truncation produced a concentrator that was drastically reduced in height
compared to the original untruncated design for large αt angles. The results of the distribu-
tion of light for the truncated 3D ACPC, though still not Lambertian at extreme αt angles,
were much more uniform in LM space than the untruncated design. However even with this
truncation the geometric concentration ratio deviated signicantly from the ideal 3D con-
centration ratio for a 30
◦ half angle cocentrator (Fig. 4.11). As the aperture tilt increased,
the 'aring' eect with truncation caused the geometric concentration ratio to deviate from
the ideal by a factor of almost 1.5 times for αt = 59.55◦ . The larger aperture area led to the
collection/emission of potentially more light than would otherwise pass through the ideal
circular aperture. Aperture area and concentration ratio for untruncated designs was not
calculated, because the 'aring' produced a non-planar aperture.
59
Figure 4.11: Deviation of the geometric concentration ratio for the truncated 3D ACPC from the ideal for
o
a 30 half angle concentrator. Primary reason for the large deviation is due to 'aring' eect at increased
αt angles.
The biggest performance advantage of the 3D ACPC is that 96% to 97% of the light
remained in the desired distribution region for all aperture tilt angles, while the light within
that same region for the CPC designs fell o to 60% for the tilted CPC and 74% for the
sliced CPC (Fig. 4.12). Truncated designs saw little performance degradation when com-
pared to the standard 3D ACPC. At high αt angles, there is a drop from 96% to 91% of
light transferred within the desired acceptance/illumination region in LM space between the
untruncated and truncated 3D ACPC respectively.
Figure 4.12: Percentage of light that falls within the target region of LM space as a function of αt angle for
each of the simulated designs.
60
4.6 Conclusions
An ideal three-dimensional ACPC should be able to concentrate/emit light from within a
circular region of Lambertian distribution shifted o axis from the receiver/source plane of
the concentrator. The new design method described in this paper moves closer towards the
development of a truly three-dimensional ACPC. The 3D ACPC is shown to collect/emit
light within a desired circular region in direction cosine space, collecting/emitting greater
than 96% of the light within that region. This shows the 3D ACPC to be an eective design
for o axis concentration/illumination. At low aperture tilt angles, the 3D ACPC design
produces an ideal asymmetric concentrator, but the design breaks down as αt increases due
to over constrained geometry. This break down in the design produces a "aring" eect
which increases the concentrator height, and causes the aperture area to increase by up
to 1.5 times the thermodynamic ideal at αt = 59.55◦ . Despite the "aring" eect of high
aperture tilt, a truncated design presented in this work still collects/emits greater than 90%
of the light within the desired angular region. This perfomance drastically exceeded that of
a tilted or sliced traditional revolved CPC, which collected/emitted only 60% and 74% of
the light within the desired angular region respectively at large aperture angles.
An 3D ACPC has numerous benecial applications to solar concentration, day-lighting
in buildings, and illumination optics where a source/target plane is not parallel to the aper-
ture plane. Future work hopes to correct the "aring" eect present with large aperture
angles by incorporating ow-line variants of the ACPC proles. Other work hopes to ex-
pand the methodology by applying the same design principles outlined here towards an
asymmetric compound elliptical concentrator. Finally moving towards a fabrication method
and experimental validation of the 3D ACPC.
61
Chapter 5
Conclusions and Future Directions
5.1 Introduction
This manuscript has shown my development of a novel gure of merit for characterizing
ideal geometric illumination. This gure of merit can be highly useful in maximizing the
illumination of a given space while minimizing wasted energy and light pollution. Using
the understanding of ideal geometric illumination that my gure of merit demonstrates, two
new illumination optical designs were presented in subsequent chapters. The rst was a
nonimaging spotlight designed to replace existing theatrical spotlights, and the second was
a 3D asymmetrical concentrator design for illumination and solar applications. This chapter
will present some of the major conclusions from the gure of merit and the new illumination
designs, and the potential future directions that each of these projects should take to be
further developed.
5.2 Insights into illumination design based on the new gure of merit
The major theoretical contribution to illumination science presented in this dissertation is
the thermodynamic gure of merit for ideal illumination. The gure of merit is a relatively
simple to understand optical characterization method outlining the concepts of under and
over illumination. A major benet of the theoretical understanding gleaned by the P31 vs
P13 /P12 gure of merit, outlined in chapter 2, is its ability to be easily simulated by modern
computational ray tracing software. This is useful when demonstrating the benet of the
new optical designs presented in chapters three and four, and I will explore how the NIS
and 3D ACPC compare to their existing illumination counterparts.
To characterize and compare the performance of the NIS to existing ERS theatrical
lighting systems, a set of xed design elements were chosen to emphasize the benets of
switching from an imaging to a nonimaging reector prole. Figure 5.1 shows a diagram of
both the ERS and NIS optical systems being simulated. In each optical design the source
area, imaging plane area, the two lens column, the target distance, and the target area were
all the same. The only dierence between ERS and NIS designs were the source placement
and reector optics used.
62
Figure 5.1: Diagram of the simulation setup for comparing an ERS to the new NIS spotlighting designs.
(a) The ERS design setup, and (b) the NIS design setup. Each design uses the same lens column, target
area, and target distance for a baseline comparison.
The irradiance distribution on a circular target placed 2 m from the aperture of each
spotlight is shown in Fig. 5.2, where it is apparent that the NIS produces a much more
uniform distribution of light on the target than the ERS. Using my novel gure of merit
comparing to the ideal of P31 = 1 and P13 /P12 = 1, the ERS has values of P31 = 0.542 and
P13 /P12 = 0.805, while the NIS resulted in P31 = 0.993 and P13 /P12 = 0.715. There is a
drastic decrease in over illumination (dened by P31 → 1) for the NIS in comparison to the
ERS simulated. This shows that a majority of the light in the optical system is within the
target area. For under illumination (dened by P13 /P12 → 1) it seems that the NIS performs
worse than the ERS, even though the distribution of light across the target is a more desired
uniform illumination (Fig. 5.2). This discrepancy is most likely because the rays simulated
from the target plane when calculating P13 /P12 were of a Lambertian distribution. This
shows that there may still be some renement necessary to improve my gure of merit as
will be discussed in the future work section.
Figure 5.2: Target irradiance distribution for (a) The ERS design setup, and (b) the NIS design setup. The
desired target area is shown as a black circle and the color gradient represents the irradiance inW/m2 .
63
To characterize the performance of the 3D ACPC using the P31 vs P13 /P12 gure of
merit, a standard CPC, tilted CPC, and sliced CPC were used as comparison as outlined in
chapter four. Since the 3D ACPC has some issues with aring at large aperture tilt angles,
an ACPC design with a aperture angle which preserved the circular aperture shape was
used. An ACPC of design angles ϕL = 21.9o and ϕR = 41.3o with an aperture tilt angle
αt = 27.52 was characterized using the gure of merit. A standard CPC of 30o acceptance
o
o
angle was used for comparison, either sliced or tilted at the αt = 27.52 angle. The target
was placed 3 m away from the source in a conguration as shown in Fig. 5.3 for both the
CPC and ACPC designs. The source plane had a circular radius of 5 mm and the target
encompassed a circular area of radius 1.75 m.
Figure 5.3: Diagram of the illumination target conguration for (a) a standard CPC, and (b) an ACPC
design. The target illumination area is the same, but shifted to the right for the ACPC.
Results for the target irradiance distribution for the four designs simulated are shown in
Fig. 5.4, with a black circle representing the desired target area. An interesting result seen
between Fig. 5.4(a) and (b) is that the sliced CPC shifts the peak point of the irradiance
distribution, but maintains the same circular shape as the standard CPC. The 3D ACPC
design is the only concentrator design that fully encompasses the target area while also
shifting the peak irradiance to the center of the target (Fig. 5.4(d)). Using the P31 vs
P13 /P12 gure of merit the 3D ACPC resulted in values near the ideal with P31 = 0.93 and
P13 /P12 = 0.97.
64
o o o
Figure 5.4: Target irradiance distributions for (a) a standard 30 CPC, (b) a 30 CPC sliced at 27.52 , (c) a
o o o o
30 CPC tilted by 27.52 , and (d) 30 ACPC with aperture tilt αt = 27.52 . The target for each simulation
was placed at 3 m from source with the ideal target area shown as a black circle in each image. The color
2
gradient represents the irradiance in W/m .
Performance improvement for the new illumination designs presented in this manuscript
becomes apparent when plotting all of the results gathered on a P31 vs P13 /P12 graph (Fig.
5.5). Examining the contrast between the ERS and NIS designs it approaches the ideal
for P31 = 1 of no over illumination. The ERS by contrast may be a functional spotlight
design, but performs poorly according to the ideal geometric illumination of P31 = 1 and
P13 /P12 = 1. As discussed above, the NIS performes worse than the ERS with respect to
P13 /P12 , and this can be explained by the necessity to rene my gure of merit to express
performance based on a desired distribution of light on the target plane.
Looking at the 3D ACPC, there is little dierence between its performance and that
of the ideal P31 = 1 vs P13 /P12 = 1, whereas the other CPC designs showed signicant
under illumination with the tilted CPC performing the worst. From these results it is clear
that even though the new designs presented in chapters three and four of this dissertation
are not perfect geometric illuminators, they approach the ideal more closely than their
existing counterparts. The next section will discuss more on how these designs along with
the gure of merit could potentially be improved and the future work needed to make those
improvements.
65
Figure 5.5: P31 vs P13 /P12 plot comparing each new design presented in this work in contrast to their
existing counterparts in the literature and previous chapters.
5.3 Future Directions

The illumination gure of merit
My novel P31 vs P13 /P12 illumination gure of merit has been shown to be an eective
way at characterizing the geometric illumination performance of an optical system. There
are however some current issues that could be resolved with some minor adjustments. An
example that comes to mind is seen in Fig. 5.2 where the almost uniform illumination of the
target for the NIS is actually calculated by the underillumination metric P13 /P12 = 0.715. In
the instance of theatrical spotlighting and potentially other illumination scenarios a uniform
illumination may be desired and P13 /P12 should approach 1 to reect that. This could
potentially be solved during ray tracing by making the distribution of light rays emitted
from the target plane be structured in a way to represent the uniformity desired when
calculating P13 and P12 . Other future work should also be done to develop a methodology
for experimental P31 vs P13 /P12 validation. An experimental evaluation of P31 vs P13 /P12
would be benecial to characterize existing illumination xture designs already in use.
The Nonimaging Spotlight

The NIS is currently being developed alongside RAF electronics under a California energy
commission CalSEED grant. Next steps include working with a local engineering rm, Blue
Dolphin Engineering, in the central valley to produce a housing for all the optical and
electronic components. This housing will be structured to look and perform as a production
ready commercial prototype similar to existing spotlight models used in theaters. This
NIS commercial prototype will be tested and evaluated quantitatively by Blue Dolphin
and qualitatively by Fresno State theater department on a working stage alongside existing
market ERS spotlight models.
66
The 3D ACPC
The 3D ACPC performed admirably, and is a major step in the development of 3D concen-
tration. There was signicant design and performance defects when large aperture tilt angles
were seen which produced a "aring" eect. Future work hopes to correct this "aring" ef-
fect by incorporating ow-line variants of the ACPC proles. The same design principles
could also be applied towards an asymmetric compound elliptical concentrator, which may
produce even better illumination performance for targets at a nite distance. The ACPC
has multiple benecial applications within the illumination and solar industries. For illumi-
nation, the 3D ACPC could be used for exterior lighting of buildings or parking lots. The
solar industry could benet by incorporating the 3D ACPC into exterior wall mounted solar
collectors on buildings.
Other work should be done to develop an experimental fabrication and testing method to
characterize the 3D ACPC performance. Early steps have already been taken to develop a
3D ACPC cast out of polyester resin (Fig. 5.6). From this dielectric version of the 3D ACPC
a visual inspection, with blue paint at the source plane, shows the concentrator reecting a
majority of the blue color out of the aperture within the design angle (relative to the index
of refraction). This rough fabrication may not provide the best performance, but it gives a
glimpse into what an ideal 3D asymmetrical concentrator could look like.
Figure 5.6: A 3D ACPC cast out of polyester resin. The source plane is painted blue for contrast to visually
inspect optical performance.
67
5.4 Overall Conclusions
The elds of lighting and illumination have advanced considerably over the last few decades.
There has been a rapid shift towards the development and adoption of more energy ecient
lighting systems, such as LEDs, which illuminate the environments we use on a daily basis.
Many of the disciplines of lighting are undergoing constant research to improve our under-
standing of the human perception of color, the impact light has on our circadian rhythms,
and the detrimental eects of light pollution on the environment. Most of this research
provides pieces to the overall solution to what an 'ideal' illumination system should be,
but not the entire picture. In this manuscript I have proposed a novel gure of merit that
provides a unifying approach to characterizing geometric illumination. This metric can be
applied across all disciplines of illumination; from automobile headlights and street lights,
to a simple desk lamp or ashlight, and many more. With the knowledge gained from this
novel gure of merit, I proposed two new optical designs using the principles of nonimaging
optics; a theatrical spotlight and an asymmetric illuminator. These designs provide a sig-
nicant improvement over existing lighting xtures, in both energy usage and in how they
illuminate a desired area. My hope is that the new gure of merit and the principles behind
the designs shown will help to increase the optical performance of all types of illumination
systems while minimizing wasted energy.
68
References
[1] M. F. Modest, Radiative heat transfer. Elsevier Academic Press, third ed., 2013.
[2] R. Winston, J. C. Minano, and P. Benitez, Nonimaging Optics. San Diego: Elsevier
Academic Press, 2005.
[3] CIE, TN 003-2015, tech. rep., 2015.
[4] S. G. DiLaura DL, Houser KW, Mistrick RG, The lighting handbook reference and
application. 2011.
[5] M. Royer, Spectral Power Distribution The Building Block of Applied Lighting, in
DOE Technology Development Workshop, 2016.
[6] J. Roby and M. Aube, LSPDD: Lamp Spectral Power Distribution Database:
http://galileo.graphycs.cegepsherbrooke.qc.ca/app/fr/home, 2020.
[7] J. Needham, W. Ling, and K. G. Robinson, SCIENCE AND CIVILISATION IN

CHINA, vol. 4. Cambridge: Cambridge Univeristy Press, 1962.
[8] J. Bentley and C. Olson, Field Guide to Lens Design. SPIE - The International Socienty
for Optical Engineering, 2012.
[9] W. J. Smith, Modern Optical Engineering. McGraw-Hill, 3rd ed., 2000.
[10] M. Bass, J. M. Enoch, W. L. Wolfe, E. W. Van Stryland, D. R. Williams, and W. L.

Wolfe, Handbook of optics, vol. 3. 2000.
[11] J. E. Greivenkamp, Field Guide to Geometrical optics. SPIE - The International So-
cienty for Optical Engineering, 2004.
[12] H. Hinterberger and R. Winston, Ecient light coupler for threshold erenkov coun-
ters, Review of Scientic Instruments, vol. 37, no. 8, pp. 10941095, 1966.
[13] O. Gallagher, J. O. Gallagher, and R. Winston, Maximum Theoretical Concentration
Ratio, vol. 4, no. 1, pp. 6668, 1987.
[14] J. Chaves, Introduction to Nonimaging Optics. Boca Raton, FL: CRC Press Taylor &
Francis Group, 2nd ed., 2017.
[15] W. Welford and R. Winston, Shape Tolerances and Manufacturing Methods for Non-
imaging Optical Components, in High Collection Nonimaging Optics, ch. 9, pp. 147
163, Academi Press, Inc., 1989.
[16] R. J. Lucas, S. N. Peirson, D. M. Berson, T. M. Brown, H. M. Cooper, C. A. Czeisler,

M. G. Figueiro, P. D. Gamlin, S. W. Lockley, J. B. O'Hagan, L. L. Price, I. Provencio,
D. J. Skene, and G. C. Brainard, Measuring and using light in the melanopsin age,
Trends in Neurosciences, vol. 37, no. 1, pp. 19, 2014.
69
[17] T. M. Brown, S. I. Tsujimura, A. E. Allen, J. Wynne, R. Bedford, G. Vickery, A. Vugler,
and R. J. Lucas, Melanopsin-based brightness discrimination in mice and humans,
Current Biology, vol. 22, no. 12, pp. 11341141, 2012.
[18] Illuminating Engineering Society, TM-30-18, tech. rep., New York, New York, 2018.
[19] R. K. John, Illumination Engineering: Design with Nonimaging Optics. 2013.
[20] A. R. Bean and R. I. Bell, The CSP index: A practical measure of oce lighting
quality as perceived by the oce worker, Lighting Research & Technology, vol. 24,
no. 4, pp. 215225, 1992.
[21] I. Moreno, Illumination uniformity assessment based on human vision, Optics Letters,
vol. 35, no. 23, p. 4030, 2010.
[22] K. M. Rebec and M. K. Gunde, High-performance lighting evaluated by photobiological

parameters, Applied Optics, vol. 53, no. 23, p. 5147, 2014.
[23] H. W. Bodmann, Quality of Interior Lighting Based on Luminance, Lighting Research
and Technology, vol. 32, no. 1 IEStrans, pp. 2240, 2015.
[24] J. Hye Oh, S. Ji Yang, and Y. Rag Do, Healthy, natural, ecient and tunable lighting:
Four-package white LEDs for optimizing the circadian eect, color quality and vision
performance, Light: Science and Applications, vol. 3, no. 2, p. e141, 2014.
[25] G. E. Fernandes, J. Bohar, and J. Xu, Spectral-Temporal LED Lighting Modules for
Reproducing Daily and Seasonal Solar Circadian Rhythmicities, 2017 IEEE Interna-
tional Conference on Smart Computing, SMARTCOMP 2017, no. 1430007, pp. 16,
2017.
[26] Q. Dai, W. Cai, L. Hao, W. Shi, and Z. Wang, Spectral optimisation and a novel
lighting-design space based on circadian stimulus, Lighting Research and Technology,
vol. 50, no. 8, pp. 11981211, 2018.
[27] D. K. Son, E. B. Cho, I. Moon, Y. Park, and C. G. Lee, Development of an illumination

measurement device for color distribution based on a CIE 1931 XYZ sensor, Journal
of the Optical Society of Korea, vol. 15, no. 1, pp. 4451, 2011.
[28] Z. M. Zhu, X. H. Qu, G. X. Jia, and J. F. Ouyang, Uniform illumination design by
conguration of LED array and diuse reection surface for color vision application,
IEEE/OSA Journal of Display Technology, vol. 7, no. 2, pp. 8489, 2011.
[29] A. Teupner, K. Bergenek, R. Wirth, J. C. Miñano, and P. Benítez, Merit function for
the evaluation of color uniformity in the far eld of LED spot lights,Light-Emitting
Diodes: Materials, Devices, and Applications for Solid State Lighting XVIII, vol. 9003,
no. February 2014, p. 900303, 2014.
[30] A. Teupner, K. Bergenek, R. Wirth, J. C. Miñano, and P. Benítez, Optimization of

optical systems for LED spot lights concerning the color uniformity,Thirteenth Inter-
national Conference on Solid State Lighting, vol. 9190, no. September 2014, p. 91900J,
2014.
70
[31] A. Teupner, K. Bergenek, R. Wirth, J. C. Miñano, and P. Benítez, Optimization of
a merit function for the visual perception of color uniformity in spot lights, Color
Research and Application, vol. 40, no. 3, pp. 287296, 2015.
[32] J. M. Gordon, P. Kashin, and A. Rabl, Nonimaging reectors for ecient uniform
illumination: errata, Applied Optics, vol. 32, no. 13, p. 2303, 1993.
[33] J. M. Gordon, Tailored edge-ray designs for uniform illumination of distant targets,
Optical Engineering, vol. 34, no. 6, p. 1726, 2006.
[34] J. Muñoz, Uniform illumination of distant targets using a spherical light-emitting diode
array, Optical Engineering, vol. 46, no. 3, p. 033001, 2007.
[35] Y. Ding, X. Liu, Z.-r. Zheng, and P.-f. Gu, Freeform LED lens for uniform illumina-
tion, Optics Express, vol. 16, no. 17, p. 12958, 2008.
[36] Z. Zhenrong, H. Xiang, and L. Xu, Freeform surface lens for LED uniform illumina-
tion, Applied Optics, vol. 48, no. 35, p. 6627, 2009.
[37] R. Hu, X. Luo, H. Zheng, Z. Qin, Z. Gan, B. Wu, and S. Liu, Design of a novel freeform
lens for LED uniform illumination and conformal phosphor coating, Optics Express,
vol. 20, no. 13, p. 13727, 2012.
[38] R. Hu, Z. Gan, X. Luo, H. Zheng, and S. Liu, Design of double freeform-surface lens
for LED uniform illumination with minimum Fresnel losses, Optik, vol. 124, no. 19,
pp. 38953897, 2013.
[39] Z. Su, D. Xue, and Z. Ji, Designing LED array for uniform illumination distribution
by simulated annealing algorithm, Optics Express, vol. 20, no. S6, p. A843, 2012.
[40] N. Shatz, J. Bortz, and R. Winston, Thermodynamic eciency of nonimaging concen-
trators, Nonimaging Optics: Ecient Design for Illumination and Solar Concentration
VI, vol. 7423, no. August 2009, p. 742308, 2009.
[41] T. Gallaway, R. N. Olsen, and D. M. Mitchell, The economics of global light pollution,
Ecological Economics, vol. 69, no. 3, pp. 658665, 2010.
[42] F. Hölker, T. Moss, B. Griefahn, W. Kloas, C. C. Voigt, D. Henckel, A. Hänel, P. M.
Kappeler, S. Völker, A. Schwope, S. Franke, D. Uhrlandt, J. Fischer, R. Klenke,
C. Wolter, and K. Tockner, The dark side of light: A transdisciplinary research agenda
for light pollution policy, Ecology and Society, vol. 15, no. 4, 2010.
[43] F. Falchi, P. Cinzano, C. D. Elvidge, D. M. Keith, and A. Haim, Limiting the im-
pact of light pollution on human health, environment and stellar visibility, Journal of
Environmental Management, vol. 92, no. 10, pp. 27142722, 2011.
[44] F. Falchi, R. Furgoni, T. Gallaway, N. Rybnikova, B. Portnov, K. Baugh, P. Cinzano,
and C. Elvidge, Light pollution in USA and Europe: The good, the bad and the ugly,
Journal of Environmental Management, vol. 248, no. February, p. 109227, 2019.
71
[45] C. D. Galatanu, M. Husch, L. Canale, and D. Lucache, Targeting the Light Pollution:
A Study Case, 2019 IEEE International Conference on Environment and Electrical
Engineering and 2019 IEEE Industrial and Commercial Power Systems Europe (EEEIC
/ I&CPS Europe), pp. 16, 2019.
[46] B. Widyolar, L. Jiang, and R. Winston, Thermodynamics and the segmented com-
pound parabolic concentrator, Journal of Photonics for Energy, vol. 7, no. 2, p. 028002,
2017.
[47] L. Wild, A brief outline of the history of stage lighting, 2015.
[48] K. Shirek Doughtie and S. Nelson, Fundamentals of Lighting WORKHORSE OF THE

GRID Getting to know the ellipsoidal reector spotlight, 2015.
[49] M. Hernández, A. Cvetkovic, P. Benítez, and J. C. Miñano, High-performance Kohler

concentrators with uniform irradiance on solar cell, Nonimaging Optics and Ecient
Illumination Systems V, vol. 7059, no. September 2008, p. 705908, 2008.
[50] R. Winston and W. Zhang, Simple Köhler homogenizers for image-forming solar con-
centrators, AIP Conference Proceedings, vol. 1407, pp. 105108, 2011.
[51] a. Timinger, a. Kribus, P. Doron, and H. Ries, Faceted concentrators optimized for
homogeneous radiation., Applied optics, vol. 39, no. 7, pp. 11521158, 2000.
[52] T. Cooper, F. Dähler, G. Ambrosetti, A. Pedretti, and A. Steinfeld, Performance of
compound parabolic concentrators with polygonal apertures, Solar Energy, vol. 95,
pp. 308318, 2013.
[53] Altman, ELLIPSOIDAL LIGHTING PHX 250W LED PROFILE SPOT 250 Watt
LED, tech. rep.
[54] Altman, ELLIPSOIDAL LIGHTING PHX 150W LED PROFILE SPOT 150 Watt
LED, tech. rep.
[55] Altman, ELLIPSOIDAL LIGHTING PHX ERS ELLIPSOIDAL HPL or GLA Lamps,
tech. rep.
[56] Electronic Theater Controls, Source Four LED Photometry Guide Source Four LED
Series 2 Daylight HD Source Four LED Lustr + Source Four LED Series 2 Tungsten
HD Source Four LED Photometry Guide, tech. rep.
[57] Electronic Theater Controls, ETC SourceFour 26° Datasheet, tech. rep.
[58] Theatre Communications Group, Theatre Facts.
[59] The Broadway League, Research & Statistics STATISTICS - BROADWAY IN NYC,
2020.
[60] Electronic Theater Controls (ETC), Source Four Accessories, 2020.
[61] Y. Bressler, Opening the Stage Door for Big Data in Broadway Building Databases
from unstructured text using Machine Learning., www.towardsdatascience.com, sep
2018.
72
[62] US Environmental Protection Agency, The Social Cost of Carbon - Estimating the
Benets of Reducing Greenhous Gas Emissions, 2017.
[63] BlueSkyModel, BlueSkyModel - 1 kilowatt-hour, 2019.
[64] R. Winston, Light Collection within the Framework of Geometrical Optics*, Journal
of the Optical Society of America, vol. 60, no. 2, pp. 245247, 1970.
[65] M. Tian, Y. Su, H. Zheng, G. Pei, G. Li, and S. Riat, A review on the recent research
progress in the compound parabolic concentrator (CPC) for solar energy applications,
Renewable and Sustainable Energy Reviews, vol. 82, no. September 2017, pp. 12721296,
2018.
[66] H. R. Ries and R. Winston, Tailored edge-ray reectors for illumination, Journal of
the Optical Society of America A, vol. 11, no. 4, p. 1260, 1994.
[67] H. Ries and A. Rabl, Edge-ray principle of nonimaging optics, Journal of the Optical
Society of America A, vol. 11, no. 10, p. 2627, 1994.
[68] P. T. Ong, J. M. Gordon, and A. Rabl, Tailored edge-ray designs for illumination with
tubular sources, Applied Optics, vol. 35, no. 22, p. 4361, 1996.
[69] D. Jenkins and R. Winston, Tailored reectors for illumination, Applied Optics,
vol. 35, no. 10, p. 1669, 1996.
[70] R. Winston and W. Zhang, Pushing concentration of stationary solar concentrators to

the limit, Optics express, vol. 18, no. 9, pp. A6472, 2010.
[71] B. Norton, P. C. Eames, and Y. P. Yadav, Symmetric and asymmetric linear com-
pound parabolic concentrating solar energy collectors: The state-of-the-art in optical
and thermo-physical analysis, International Journal of Ambient Energy, vol. 12, no. 4,
pp. 171190, 1991.
[72] V. J. Sevcik and R. Winston, Dielectric Compound Parabolic Concentrators., Applied

Optics, vol. 15, no. 2, pp. 143149, 1975.
[73] X. Ning, Three-dimensional ideal θ_1/θ_2 angular transformer and its uses in ber
optics, Applied Optics, vol. 27, no. 19, p. 4126, 1988.
[74] S. Kudaev and P. Schreiber, Optimization of symmetrical free-shape non-imaging con-
centrators for LED light source applications, Nonimaging Optics and Ecient Illumi-
nation Systems II, vol. 5942, no. August 2005, p. 594209, 2005.
[75] T. K. Mallick, P. C. Eames, T. J. Hyde, and B. Norton, The design and experimental
characterisation of an asymmetric compound parabolic photovoltaic concentrator for
building façade integration in the UK, Solar Energy, vol. 77, no. 3, pp. 319327, 2004.
[76] T. K. Mallick, P. C. Eames, and B. Norton, Non-concentrating and asymmetric com-
pound parabolic concentrating building façade integrated photovoltaics: An experimen-
tal comparison, Solar Energy, vol. 80, no. 7, pp. 834849, 2006.
73
[77] T. K. Mallick, P. C. Eames, and B. Norton, Using air ow to alleviate temperature
elevation in solar cells within asymmetric compound parabolic concentrators, Solar
Energy, vol. 81, no. 2, pp. 173184, 2007.
[78] A. Harmim, M. Merzouk, M. Boukar, and M. Amar, Performance study of a box-
type solar cooker employing an asymmetric compound parabolic concentrator, Energy,
vol. 47, no. 1, pp. 471480, 2012.
[79] W. Lu, Y. Wu, and P. Eames, Design and development of a Building Façade Integrated
Asymmetric Compound Parabolic Photovoltaic concentrator (BFI-ACP-PV), Applied
Energy, vol. 220, no. December 2017, pp. 325336, 2018.
[80] S. H. Abu-Bakar, F. Muhammad-Sukki, R. Ramirez-Iniguez, T. K. Mallick, A. B. Mu-
nir, S. H. Mohd Yasin, and R. Abdul Rahim, Rotationally asymmetrical compound
parabolic concentrator for concentrating photovoltaic applications, Applied Energy,
vol. 136, pp. 363372, 2014.
[81] S. H. Abu-Bakar, F. Muhammad-Sukki, D. Freier, R. Ramirez-Iniguez, T. K. Mallick,

A. B. Munir, S. H. Mohd Yasin, A. Abubakar Mas'ud, and N. Md Yunus, Performance
analysis of a novel rotationally asymmetrical compound parabolic concentrator, Ap-
plied Energy, vol. 154, no. April, pp. 221231, 2015.
[82] A. Garcia-Botella, A. A. Fernandez-Balbuena, D. Vázquez, and E. Bernabeu, Ideal 3D
asymmetric concentrator, Solar Energy, vol. 83, no. 1, pp. 113117, 2009.
[83] M. N. Ricketts, R. Winston, and J. Ferry, Étendue and Angular Acceptance of the
Asymmetric Compound Parabolic Concentrator, Journal of Photonics for Energy,
vol. 7, no. 2, p. 028003, 2017.
74

Non Imaging Optics Dizertation

Uploaded by

Copyright:

Available Formats

Non Imaging Optics Dizertation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Non Imaging Optics Dizertation

Uploaded by

Copyright:

Available Formats

UNIVERSITY OF CALIFORNIA, MERCED

Nonimaging Optics and the Radiative Transport Properties of Ideal Illumination

A dissertation submitted in partial satisfaction of

Jonathan James Ferry

Professor Gerardo Diaz, Chair of Advisory Committee

Professor Arnold Kim

Professor Sarah Kurtz

Professor Roland Winston, Supervisor

Jonathan James Ferry, 2020

All rights reserved.

Professor Arnold Kim Date

Professor Sarah Kurtz Date

University of California, Merced

To Kathryn Fischer, my best friend and the love of my life.

1 Illumination Optics and the Characterization of Light 1

2 The Thermodynamics of Ideal Illumination 24

3 Coupling of Nonimaging and Imaging Optics for Uniform Illumination

4 The Three-Dimensional Asymmetric CPC 49

5 Conclusions and Future Directions 62

Candela [cd = lm/sr]. Lux [lx = lm/m2 ]. Steradians [sr].

α Absorption [−] Ee Irradiance [W/m2 ]

Cmax Maximum concentration ratio [−] Qe Radiant energy [J]

ACPC Asymmetric compound parabolic concentrator.

CAD Computer aided design.

CCT Correlated color temperature.

CEC Compound elliptical concentrator.

CFL Compact uorescent.

CIE International Commission on Illumination.

CNC Computer numerical control.

CPC Compound parabolic concentrator.

CRI Color rendering index.

ERS Ellipsoidal reector spotlight.

GHG Greenhouse Gas.

LED Light emitting diode.

NIS Nonimaging Spotlight.

SPD Spectral power distribution.

3.1 Diagram depicting a traditional Ellipsoidal Reector Spotlight design. . . . . 35

for 2018-19[59] + High Performance Lamp (HPL) data from ETC[60] . . . . . . . . . . . 46

deviation is due to 'aring' eect at increased αt angles. . . . . . . . . . . . . 60

University of California, Merced

1. J. Ferry, B. Widyolar, L. Jiang, and R. Winston, Solar thermal wastewater evapo-

2. J. Ferry, L. Jiang, and R. Winston, "The thermodynamics of ideal illumination: a

3. B. K. Widyolar, L. Jiang, J. Ferry, J. Brinkley, Y. Bhusal, and R. Winston, Nonimag-

4. Y. Bhusal, B. Widyolar, L. Jiang, J. Ferry, J. Brinkley, and R. Winston, High

5. B. Widyolar, L. Jiang, J. Ferry, R. Winston, A. Kirk, M. Osowski, D. Cygan, and

6. B. Widyolar, L. Jiang, J. Ferry, R. Winston, D. Cygan, and H. Abbasi, Experimental

7. B. Widyolar, L. Jiang, J. Ferry, and R. Winston, "Non-tracking East-West XCPC

8. J. Ferry, R. Winston, and R. A. Flasck, Nonimaging optics applications in theater

13. M. Ricketts, J. Ferry, L. Jiang, and R. Winston, "Nonimaging optics in lighting

14. J. J. Ferry, F. S. Alleyne, R. R. Milczarek, R. Winston, and D. A. Olson, "Eciency

15. R. Winston, L. Jiang, M. Abdelhamid, B. K. Widyolar, J. Ferry, D. Cygan, H. Abbasi,

 Mechanical Engineering Bobcat Fellowship, University of California, Merced . . . 2018

1.2 The Basics of Light Transfer

1.2.1 Quantication of Light

L = sin θ cos φ, (1.3)

M = sin θ sin φ, (1.4)

Table 1.1: Radiometric properties of light.

Term Symbol Denition Units

Radiant Energy Qe Energy emitted by a light source Joules [J]

CFL Compact uorescent.

ERS Ellipsoidal reector spotlight.

3.1 Diagram depicting a traditional Ellipsoidal Reector Spotlight design. . . . . 35

deviation is due to 'aring' eect at increased αt angles. . . . . . . . . . . . . 60

1. J. Ferry, B. Widyolar, L. Jiang, and R. Winston, Solar thermal wastewater evapo-

3. B. K. Widyolar, L. Jiang, J. Ferry, J. Brinkley, Y. Bhusal, and R. Winston, Nonimag-

4. Y. Bhusal, B. Widyolar, L. Jiang, J. Ferry, J. Brinkley, and R. Winston, High

6. B. Widyolar, L. Jiang, J. Ferry, R. Winston, D. Cygan, and H. Abbasi, Experimental

8. J. Ferry, R. Winston, and R. A. Flasck, Nonimaging optics applications in theater

14. J. J. Ferry, F. S. Alleyne, R. R. Milczarek, R. Winston, and D. A. Olson, "Eciency

Mechanical Engineering Bobcat Fellowship, University of California, Merced . . . 2018

1.2.1 Quantication of Light

Term Symbol Denition Units

Term Symbol Denition Units

The reectance, ρ, of a surface is dened as

reected part of incoming radiation

Similarly to reectance, the amount of light transmitted by a surface can be expressed as

where the superscripts s and d represent specular and diuse respectively.

Table 1.3: Eect of κ values on an aspherical lens shape.

The focal point of Lens 1 is placed at the convex surface of Lens 2.

Manufacturer Model Power Field Ecacy