Solar Energy Conversion and PEM Technology
for a Residential Backup Power System
by
Aaron Biddings
A thesis submitted to the Graduate Faculty of
Auburn University
in partial fulfillment of the
requirements for the Degree of
Master of Science
Auburn, Alabama
May 4, 2013
Keywords: photovoltaic, electrolysis, proton exchange membrane
Approved by
Steven M. Halpin, Chair, Professor, Electrical and Computer Engineering Department
Charles A. Gross, Professor Emeritus, Electrical and Computer Engineering Department
Mark Nelms, Professor and Chair, Electrical and Computer Engineering Department
ii
ABSTRACT
Shown in this work are the characterization, simulation, and implementation of a small
scale backup power system powered through solar energy conversion. The technical approach
utilizes solar arrays with a proton exchange membrane (PEM) electrolyzer and fuel cell. This
facilitates the production and storage of electrical energy using hydrogen as a carrier. Proton
exchange membrane (PEM) technology was chosen due to its relatively high power density (in
comparison with other fuel cell types), its low operating temperature range, and lack of noxious
emissions. The setup employs a solar array designed to provide sufficient power to the
electrolyzer under varying temperature and lighting conditions. The electrical power is derived
directly from the electrochemical energy stored in the hydrogen supply. In this way, the inherent
instability and fluctuations in power due to the intermittent nature of solar energy are diminished.
The focus of this study is strictly technical. Monetary cost considerations are beyond the
scope of this document.
iii
TABLE OF CONTENTS
Abstract??????????????????????????????? ..? .?. ...ii
List of Tables?? ???????????????????????????? ..?? iv
List of Illustrations???????????????????????????? ..?? .v
List of Abbreviations??????????????????????????? ..? ? vi
Chapter 1: Introduction????????.. ??????????? ????? ??. ..?. .1
Chapter 2: Proton Exchange Membrane Fuel Cell?. ????????????????. ....3
Chapter 3: Proton Exchange Membrane Electrolyzer..????????????????. 10
Chapter 4: Solar Energy Conversion?????. ?????????????????? 16
Chapter 5: Conclusion????????????????????????????... 33
Chapter 6: Further Research??????????????????????????.. 35
References?????????????????????????????... ?? ? ..36
Appendix A: Determination of Irradiance????????????????????? .38
Appendix B: Matlab Code For Simulations???????????????????? ...43
iv
LIST OF TABLES
Table A1. Light Energy, Power, and Irradiance??????????????????... 40
Table A2. Error Calculations?????????????????????????? 41
Table A3. Adjusted Light Energy, Power, Irradiance, and Percent Error????????? 42
v
LIST OF ILLUSTRATIONS
Figure 21. A) Membrane Electrode Assembly, B) Gas Diffusion Layers, C) Bipolar Plates
D) Current Collectors, E) Compression Plates? ???? ?. ??????? .?... 3
Figure 22. Activation Loss: 0A0.2A; Resistive Loss: 0.2A0.6A;
Mass Transport Loss: 0.6A0.99A???????????????????? 6
Figure 31. A) Membrane Electrode Assembly, B) Gas Diffusion Layers, C) Bipolar Plates
D) Current Collectors, E) Compression Plates??????????????... 10
Figure 32. Input Voltage vs. Input Current?.. ??. ????????????????... 12
Figure 33. Hydrogen Production Rate vs. Input Power.???????????????... 13
Figure 41. Silicon Solar Cell Crosssection????????????????????.. 17
Figure 42. Simple Model of Silicon Solar Cell?. ?????????????????... 18
Figure 43. Typical Solar Cell IV Curve????????????... ????????... 19
Figure 44. Effect of Decreasing Shunt Resistance?????????????? .???.. 21
Figure 45. Effect of Increasing Series Resistance????? .??????????? ?... 22
Figure 46. Effect of Increasing Current????????????? ..????????.. 24
Figure 47. Measured Output Current/Voltage Characteristic?????????????... 25
Figure 48. Impedance Plot (0.001 Hz ? f ? 1MHz)?????????????????. 27
Figure 49. Nyquist Plot Representative Circuit??????????????????? 27
Figure 410. Effect of Light Intensity on Solar Cell Performance??????? .???... ? 29
Figure 411. VOC and ISC as Functions of Temperature (?C)??????????????. 31
vi
LIST OF ABBREVIATIONS
EIA Energy Information Administration
H2O Water (Dihydrogen Monoxide)
M Molar Solution (mol/L)
MEA Membrane Electrode Assembly
PEM Proton Exchange Membrane
1
CHAPTER 1: INTRODUCTION
Since the late nineteenth century, when Thomas A. Edison premiered his Pearl Street
Station in New York City on September 4, 1882, the generation, transmission, and utilization of
electric power has become an indispensible part of everyday living. Much advancement in both
equipment and operating practices has been put into effect since those early days. In the United
States, the decade of the 1960?s was a time of high load growth due to both industrial and
commercial expansion. Initially, regulated utilities generated and delivered electric power within
localized service areas. Today, the transmission system delivers across states or even regions
[1].
In the United States, due to the ubiquity and reliability of the grid, the availability of
electric power is generally taken for granted by the public at large. That is, until problems arise
and the system fails to deliver. One extreme example of such an outage occurred in 2003 when a
widearea power failure affected millions of Americans across several northeastern states. While
outages of this magnitude aren?t the norm in the United States, smaller failures do happen
occasionally due to various causes ranging from bad weather to aging equipment. In the absence
of the main power source, alternate sources may be desired for residential application. Solar
arrays used in conjunction with PEM technology may make this a possibility.
Proton exchange membrane (PEM) fuel cells run on hydrogen, which is reported to be
the most abundant element on earth. However, it is not usually found in its pure form and must
be extracted from a compound. One such wellknown compound is none other than H?O
2
(water). In the United States, water is supplied directly to most homes. Furthermore, it is
unlikely that both the electric power and water supply would fail simultaneously. Therefore, a
backup power system utilizing solar arrays, a PEM electrolyzer, and a PEM fuel cell may be a
viable source of intermittent power. The U.S. Energy Information Administration (EIA)
reported the average annual electricity consumption for a U.S residential utility customer to be
11,496 kWh in 2010, an average of 958 kWh per month [2]. Thus, the backup system would
need to generate and store enough hydrogen to supply about 32 kWh for daily use.
3
CHAPTER 2: PROTON EXCHANGE MEMBRANE FUEL CELL
The proton exchange membrane (PEM) fuel cell is an electrochemical cell which
functions to combine gaseous hydrogen with oxygen. In the process, only electricity, water, and
heat are generated. The basic structure of the PEM fuel cell consists of a membrane electrode
assembly (MEA), two gas diffusion layers, two bipolar plates, two current collectors, and two
compression plates. See Figure 21 for the diagram.
Figure 21. A) Membrane Electrode Assembly, B) Gas Diffusion Layers, C) Bipolar Plates
D) Current Collectors, E) Compression Plates
The electrochemical reaction that occurs within the fuel cell is described by the equations 21
through 23 [3].
Anode (oxidation):
Cathode (reduction):
Overall reaction:
The relationship between the fuel cell potential and the concentrations of the products and
reactants involved in the electrochemical reaction is described by equations 24 and 25 [4].
(21)
(22)
(23)
4
? ?
Equation 25 is known as the Nernst equation, where the following definitions apply:
( )
( )
Note that in these definitions, standard conditions are P = 1 atm, T = 298.15 K (77 ?F), and 1.0
M concentrations.
In order to calculate the fuel cell potential from the redox reaction, it is necessary to
obtain the standard potential values for each halfcell reaction within the cell. These may be
obtained from the list of standard reduction potentials and are exemplified by equations 26 and
27 [5].
(0.0 V)
(1.23 V)
(25)
(26)
(27)
(24)
5
Each standard reduction potential equation represents a reduction reaction. The cell potential
(for nonstandard conditions) is found using the concentration values for pure water. Note that
nonstandard conditions are P = 1 atm, T = 300.4 K (81 ?F), [H+] = 107 M (for pure water) [6].
So, according to the Nernst equation, the fuel cell output voltage is found as shown in equations
28 to 212 [4].
?
The positive result of equation 212 implies that the reaction is spontaneous. Because this value
represents the output,   is the theoretical magnitude of the opencircuit voltage (difference
between the cathode and anode potentials) of the fuel cell. The magnitude of the measured open
circuit voltage of the fuel cell tested in this work was less than this calculated value.
Three main kinds of losses (also known as overpotentials) affect the operation of a fuel
cell: activation loss, resistive loss, and mass transport loss. Each loss causes a reduction in the
voltage at the output of the fuel cell. Activation loss is dominant at low current levels. The
cause is sluggishness of reactions occurring at the active sites within the fuel cell. Increasing the
area of the catalyst may reduce this type of loss. Resistive loss is caused by current flow through
(28)
(29)
(210)
(211)
(212)
6
resistive components of the fuel cell such as the membrane, electrodes, contacts, etc. This kind
of loss is dominant for moderate current levels. Increasing the hydration of the membrane may
reduce this type of loss. Mass transport loss is dominant when current levels are high. The cause
of this type of loss is the depletion of reactive species at the electrode surface during heavy
loading. More efficient evacuation of reaction product, i.e. water, at the anode may reduce this
type of loss. Using a 50 ? variable resistor as load on the fuel cell under test in this work, a plot
of the output voltage versus output current was generated to exemplify these losses. See Figure
22.
Figure 22. Activation Loss: 0A0.2A; Resistive Loss: 0.2A0.6A;
Mass Transport Loss: 0.6A0.99A
The ratings of the fuel cell used to produce the polarization curve are: V = 0.6 V, I = 0.4 A, and
P = 0.24 W. The membrane electrode assembly of the cell has a measured area of 3.61 cm2. As
observed, a maximum power of 227.9 mW was produced for a voltage of 0.518 V at a current of
0.0000
0.0500
0.1000
0.1500
0.2000
0.2500
0.000
0.100
0.200
0.300
0.400
0.500
0.600
0.700
0.800
0.900
1.000
0.000 0.200 0.400 0.600 0.800 1.000 1.200
Power
(W)
Voltage
(V)
Current (A)
Fuel Cell Polarization Curve
Voltage
Power
T = 27.2 ?C
7
0.44 A. At maximum power, the hydrogen fuel consumption rate may be calculated as shown in
equations 213 to 215.
The following definitions apply in equations 213 through 215:
( )
( )
The efficiency of this fuel cell at maximum power is calculated in equation 216.
Maintaining the assumption that the average American household uses about 32 kWh per
day when running at full capacity, the fuel cell will need to be scaled up to meet this demand.
The rate of daily energy usage, however, will not be constant throughout the day. In order to
(213)
(214)
(215)
(216)
8
simplify the problem, the typical day is broken into 3 intervals: 12am8am (morning), 8am4pm
(noon), and 4pm12am (evening). It is likely that most U.S. citizens will sleep during the
morning interval, work or attend school during the noon interval, and arrive home to actively use
electricity during the evening interval. For each interval, the following assumptions are made:
1) morning energy consumption: 33%,
2) noon energy consumption: 27%, and
3) evening energy consumption: 40%.
The fuel cell?s capacity must, therefore, match the greatest load requirement of 40% of daily
usage during an 8hour interval.
The following definitions apply in equations 217 and 218:
Nafion membranes can have a thickness in the range of 127 to 254 microns [7]. The thickness of
the fuel cell used in this analysis was measured to be 0.2 cm. Thus, a 13cell stack with an MEA
area of per cell would have minimal space requirements. The average rates of hydrogen
consumption by the stack are calculated in equations 219 through 221.
(217)
(218)
9
Moreover, the rate of water consumption necessary to sustain the maximum average demand was
found according to equations 222 and 223.
The total water consumed would amount to less than . An electrolyzer would need to
replenish the supply of hydrogen lost to the fuel cell.
(219)
(220)
(221)
(222)
(223)
10
CHAPTER 3: PROTON EXCHANGE MEMBRANE ELECTROLYZER
The proton exchange membrane (PEM) electrolyzer is an electrochemical cell which
functions to separate water into its constituent elements using electrolysis. The basic structure is
the same as that of a PEM fuel cell: a membrane electrode assembly (MEA), two gas diffusion
layers, two bipolar plates, two current collectors, and two compression plates. See Figure 31 for
the diagram.
Figure 31. A) Membrane Electrode Assembly, B) Gas Diffusion Layers, C) Bipolar Plates
D) Current Collectors, E) Compression Plates
Equations 31 to 33 govern the reactions within the electrolyzer cell [3].
Cathode (reduction):
Anode (oxidation):
Overall reaction:
(31)
(32)
(33)
11
The electrolyzer cell potential is calculated in a way similar to that of the fuel cell. From the list
of standard reduction potentials, the standard potentials for each half cell reaction are again
shown in equations 34 and 35 [5].
(0.0 V)
(1.23 V)
The cell potential for the electrolyzer cell is found using the Nernst equation [4].
?
The negative sign of the result implies that the reaction is not spontaneous. Because this result is
the value of the input voltage to the electrolyzer cell,   represents the theoretical magnitude
of the minimum applied voltage (difference between cathode and anode potentials) for which the
electrolyzer will begin to separate hydrogen from the water molecules for nonstandard
conditions, i.e P = 1 atm, T = 296.5 K (74 ?F), and [H+] = 107 M.
Whereas the MEA was measured to be 6.25 cm2, the specifications provided by the
manufacturer for the electrolyzer used in this analysis are the following [8]:
(34)
(35)
(36)
(37)
(38)
(39)
12
1) 1.7 V to 3.0 V (input), and
2) 1 A current @ 2.0 V (input).
Using the Tenma? Laboratory DC power supply 722080, an input voltage/current plot was
produced as shown in Figure 32 for 100 mA increments in the range from 0.1 A to 1.2 A.
Figure 32. Input Voltage vs. Input Current
Again using the DC power supply, along with a graduated cylinder and a timer, the
hydrogen production rate was plotted as a function of the input power. The current served as the
controlled variable in increments of 100 mA from 0.1 A to 1.2 A. See Figure 33 for the plot.
1.527
1.604
1.598
1.679
1.789
1.86
1.813
2.02
2.24
2.63
3.25
4.28
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Voltage
(V)
Current (A)
Electrolyzer Input Characteristic
T = 23.3 ?C
13
Figure 33. Hydrogen Production Rate vs. Input Power
As is visible from Figure 33, the hydrogen production rate increased at a diminishing pace as
the input power increased. This phenomenon is most likely due to mass transport limitations of
the reactive species that become more dominant as the current increases. Using Excel curve
fitting functionality, the input to output relationship is shown to closely approximate a
logarithmic function. (Current levels above 1.2 A were avoided due to an issue with overheating
and destruction of the MEA gasket.)
According to calculations in Chapter 2, the postulated PEM fuel cell stack would
consume hydrogen at a maximum average rate of during the evening interval. The
electrolyzer cell under test in this work (MEA of 6.25 cm2) produced hydrogen at a rate of
at 5.136 W for a current of 1.2 A. By the ideal gas law [9], the hydrogen production
rate by mass at 23.3 ?C (74 ?F) was approximated in equations 310 and 311.
0.614
1.357 1.925
2.382
3.219 3.838
4.329
4.980
5.771
6.497
7.695 8.188
y = 2.3667ln(x) + 4.007
1.000
0.000
1.000
2.000
3.000
4.000
5.000
6.000
7.000
8.000
9.000
0 2 4 6
Ra
te
(mL/
min)
Power (W)
H2 Rate vs. Input Power
Electrolyzer
Log. (Electrolyzer)
T = 23.3 ?C
14
The overall efficiency of the electrolyzer cell was calculated in equations 312 and 313.



The electrolyzer would be required to replenish the hydrogen supply used by the fuel cell.
Therefore, the amount of hydrogen lost is needed to calculate the size of the electrolyzer stack.
Because the system will derive its power from solar energy conversion, the noon interval
would be the optimal time for hydrogen production. Starting with a full tank of hydrogen, the
worst case scenario would be a total main power failure beginning a 4pm and lasting for 24
hours. The amount of hydrogen consumed by the fuel cell during the evening and morning
intervals (while the electrolyzer is powered down) is calculated in equations 314 and 315.
( )
( ) ( )
The electrolyzer will need to replenish this amount in addition to keeping up with the noon
energy demand. The necessary rate of hydrogen production was found according to equation 3
16.
(310)
(311)
(314)
(315)
(316)
(312)
(313)
15
Scaling up the electrolyzer to match this hydrogen production rate gives a membrane electrode
assembly with an area as found by equation 317.
The following definitions apply in equation 317:
Similar to the fuel cell, the thickness of the electrolyzer?s membrane is small. The thickness of
the electrolyzer cell used in this analysis was measured to be 0.3 cm. So, a ninecell electrolyzer
stack with an MEA area of per cell is sufficient to meet the specified mass flow rate.
Also, the space requirements are not significant.
Thus, the daily energy demand for the electrolyzer stack is calculated in equations 318
and 319.
If a solar array is incorporated into the system design, it must match the energy need of this
electrolyzer stack.
(317)
(318)
(319)
16
CHAPTER 4: SOLAR ENERGY CONVERSION
A solar array is a photovoltaic device which consists of a collection of solar cells
arranged in a combination of rows and columns. The solar cell is the basic unit of photovoltaic
energy conversion and is manufactured from semiconductor materials. There are three general
categories of photovoltaic cells currently on the market. They are the monocrystalline,
polycrystalline, and thin film (or amorphous) cells [10]. The monocrystalline solar cell is cut
from a continuous crystal of silicon. While is has the greatest efficiency, it is also the most
expensive type of solar cell. The polycrystalline solar cell is cut from a block of silicon which
had been previously melted and poured into a mold. As the silicon cools, many crystals are
formed throughout its bulk. Polycrystalline solar cells are slightly less efficient than the
monocrystalline variety. Finally, the thin film (or amorphous) solar cell contains no crystals and
is made by depositing a microscopically thin layer of silicon on a backing material such as some
other metal or glass. For the purpose of this analysis a polycrystalline silicon solar cell was
utilized.
The structure of a solar cell is such that it consists of several layers: the front contacts, the
antireflection coating, the nsubstrate material, the psubstrate material, and the rear contacts.
See Figure 41 for the solar cell diagram.
17
Figure 41. Silicon Solar Cell Crosssection (Not to scale)
When a photon of sufficient energy strikes the cell it gets absorbed. In the process, the photon
transfers its energy to the silicon lattice creating an electron/hole pair. The electron is
simultaneously raised from the valence band to the conduction band where it is capable of doing
electrical work, e.g. flowing though an external circuit.
The simplest model of a solar cell takes into account the effects of the pn junction, light
generated current, series resistance, and internal shunt resistance. The simple solar cell model
circuit is shown in Figure 42.
18
Figure 42. Simple Model of Silicon Solar Cell [11]
In the model, the light generated current is represented by the independent current source; the
diode represents the pn junction; and the series and shunt resistances are represented by the
resistors RS and RSH, respectively.
Analysis of a solar cell is usually done in terms of its output current/voltage relationship.
Three methods used differ in their results due to the internal resistance of the solar cell. These
three characteristics are the following: photovoltaic output characteristic, diode forward
characteristic, and the pn junction characteristic [12]. The first method employs a fixed
illumination (usually) of known intensity while varying a load resistance between shortcircuit
and opencircuit. Voltage and current measurements are recorded for each resistor setting. The
second method uses no illumination, but instead applies a variable source to the output terminals
of the cell. The current is, therefore, measured and plotted against the voltage as it is varied over
a certain range. The third method varies the light intensity while measurements of the open
circuit voltage and shortcircuit current are taken for each light setting. Because the conditions
0
Ig
D (p n junc t ion)
Rsh
Rs
19
of the photovoltaic analysis are similar to those of normal operation, this type of analysis is the
focal point of this study.
The pn junction of a solar cell determines the output current/voltage relationship.
However, the internal resistance can cause significant effects. A good quality cell has a large
shunt resistance and a small series resistance and produces an IV curve similar to that shown in
Figure 43. This is an example of a typical photovoltaic output characteristic curve.
Figure 43. Typical Solar Cell IV Curve (Source: Everbright Solar, Inc.)
In order to understand how the series and shunt resistances affect the output characteristic,
MATLAB? simulations were performed in which one quantity was held constant while the other
was varied. To facilitate the analysis, the circuit equation 41 was derived from Figure 42
(incorporating a load resistance).
20
(
)
The following definitions apply in equation 41:
( )
In the first simulation, the series resistance was held constant (RS = 0.167 ?). (This value
represents the actual series resistance of the cell under test in this work.) Four separate plots
were generated for distinct values of shunt resistance while the load resistance was varied from
RL = 0 ? to 50 ? in each case. The values chosen for the shunt resistance represent the
measured resistance of the solar cell under test in this work and three submultiples of the same.
See Figure 44 for the plots of load current versus the load voltage.
(41)
21
Figure 44. Effect of Decreasing Shunt Resistance
The initial effect of reducing the shunt resistance is that both the shortcircuit current and open
circuit voltage are slightly diminished. However, upon further decrease, the shunt resistance
begins to dominate the pn junction; the exponential curve eventually become linear and appears
as purely resistive.
In the second simulation, the shunt resistance was held constant (RSH = 211 ?). (This
value represents the actual shunt resistance of the cell under test in this work.) Four separate
plots were generated for distinct values of series resistance while the load resistance was varied
from RL = 0 ? to 50 ? in each case. The values chosen for the series resistance represent the
0 0 . 1 0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7
0
0 . 2
0 . 4
0 . 6
0 . 8
1
1 . 2
1 . 4
1 . 6
1 . 8
2
L o a d V o l t a g e ( V )
L
o
a
d
C
u
r
r
e
n
t
(
A
)
V a r y i n g S h u n t R e s i s t a n c e
R s h = 2 1 1 o h m s
R s h = 2 1 . 1 o h m s
R s h = 2 . 1 1 o h m s
R s h = 0 . 2 1 1 o h m s
22
measured resistance of the solar cell under test in this work and three integer multiples of the
same. See Figure 45 for the plots of load current versus the load voltage.
Figure 45. Effect of Increasing Series Resistance
Increasing the series resistance produces a similar but more dramatic reduction in the short
circuit current as compared to the incremental reduction in the shunt resistance. Also, the series
resistance begins to dominate the output characteristic producing a straight line. However,
unlike the case with shunt resistance, the opencircuit voltage remains unaffected. Thus, the
combination of a large series resistance and a small shunt resistance works to lower both the
0 0 . 1 0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7
0
0 . 2
0 . 4
0 . 6
0 . 8
1
1 . 2
1 . 4
1 . 6
1 . 8
2
L o a d V o l t a g e ( V )
L
o
a
d
C
u
r
r
e
n
t
(
A
)
V a r y i n g S e r i e s R e s i s t a n c e
R s = 0 . 1 6 7 o h m s
R s = 0 . 3 3 4 o h m s
R s = 0 . 6 6 8 o h m s
R s = 1 . 3 3 6 o h m s
23
opencircuit voltage and shortcircuit current from their ideal. In solar cells of lower quality, the
output characteristic begins to look more resistive.
One additional note on the output current/voltage relationship of a solar cell is that for
given shunt and series resistances, large lightgenerated current may also produce linearization of
the output characteristic due to the fact that much of this current effectively ?bypasses? the pn
junction but passes through the resistances. Holding the series resistance constant (RS = 0.167
?) and the shunt resistance constant (RSH = 211 ?), six plots were produced for distinct values of
lightgenerated current while the load resistance was varied from RL = 0 ? to 50 ? in each case.
The values chosen for the light generated current represent the rated current of the solar cell
under test in this work and submultiples of the same. In Figure 46 is shown plots of load current
versus load voltage.
24
Figure 46. Effect of Increasing Current
Now that the ground work has been established for properly analyzing the photovoltaic
output characteristic of a solar cell, the actual measurements may be interjected. The ratings of
the cell under test were 1.8 W, 0.5 V, and 3.6 A. Using a 50 ? variable resistor as load and a
1500 lumens/100 W bulb (at a distance of 30.48 cm) for constant illumination, the solar cell
output characteristic was plotted. The outcome of the analysis is shown in Figure 47.
0 0 . 1 0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7
0
0 . 5
1
1 . 5
2
2 . 5
3
3 . 5
L o a d V o l t a g e ( V )
L
o
a
d
C
u
r
r
e
n
t
(
A
)
V a r y i n g L i g h t G e n e r a t e d C u r r e n t
I
G
= 0 . 6 A
I
G
= 1 . 2 A
I
G
= 1 . 8 A
I
G
= 2 . 4 A
I
G
= 3 A
I
G
= 3 . 6 A
25
Figure 47. Measured Output Current/Voltage Characteristic
The good curvature of the current/voltage plot implies that neither the series nor shunt
resistances dominate the pn junction. It should be noted that the nonzero resistance of the
tabbing wire used to short the solar cell terminals prevented the output voltage from reaching
zero.
The irradiance on the cell for the light source used was found to be 0.07 W/cm2. (See
Appendix A for an explanation of how to determine irradiance.) With a solar cell surface area of
116.13 cm2, the incident power was calculated to be 8.89 W. The maximum cell output power of
0.893 W yielded an efficiency shown by equation 42.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.5
1
1.5
2
2.5
3
0 0.1 0.2 0.3 0.4 0.5 0.6
Power
(W)
Curr
ent (A
)
Voltage (V)
Photovoltaic Curve
Voltage
Power
Irradiance = 0.07 W/cm?
Cell Temperature = 36.1 ?C
(42)
26
The series resistance can usually be estimated from the slope of the curve at the open
circuit point; the shunt resistance can be estimated from the slope at the short circuit point.
However, a more accurate method for acquiring both resistances entailed using AC analysis.
This analysis involved application of a sinusoidal voltage source of constant magnitude to the
solar cell. The frequency of the source was varied over a range of values; the resistance and
reactance of the cell were simultaneously measured for each frequency. Taken together, the
resistance and reactance represent the output impedance of the solar cell. (Given a zero
resistance value, a positive reactance value represents pure inductance; a negative reactance
value represents pure capacitance. A zero reactance value represents pure resistance. All other
numerical combinations signify some interconnection of resistance, capacitance and/or
inductance.) From the measured resistance and reactance values a Nyquist plot of the solar cell?s
output impedance was subsequently generated. This plot is reproduced here in Figure 48.
27
Figure 48. Impedance Plot (0.001 Hz ? f ? 1MHz)
The Nyquist plot of the solar cell impedance was created for a frequency range of 0.001 Hz to 1
MHz. Observance of the plot shows that there were two frequencies for which the reactance was
negligible. At each of these frequencies, the solar cell output impedance appeared purely
resistive to a good approximation. (Resistance and reactance measurements were simultaneously
captured using the Solartron Instruments SI 1260 Impedance/GainPhase Analyzer with the SI
12601 component test module.) Figure 48 is representative of an RR/C combination. See
Figure 49 for the circuit.
Figure 49. Nyquist Plot Representative Circuit
120.000
100.000
80.000
60.000
40.000
20.000
0.000
20.000
0 50 100 150 200 250
Rea
ctance (
?)
Resistance (?)
Solar Cell Impedance
Impedance
R s hC
Rs
28
To find the resistances, the circuit equation was derived for Figure 49. See equations 43 to 47
for the derivation.
? ( )
?


At a frequency of 96,813.02 Hz the measured reactance was at its smallest value and the
corresponding impedance was measured to be 0.167 ?. At the opposite end of the spectrum, for
a frequency of 7.71 mHz, the measured reactance was again at its smallest value; the
corresponding impedance was measured to be 211.74 ?. Thus, the series and shunt resistances
were found in equations 48 and 49.


The daily light level varies between before sunrise/after sunset and
(average value) at noon. Luminous intensity has a considerable effect on the performance of a
solar cell. The shortcircuit current is directly proportional to the luminous intensity. From
equation 41, for negligible series (RS = 0) and shunt (RSH = ?) resistances, the opencircuit
(43)
(44)
(45)
(46)
(47)
(48)
(49)
29
voltage has a logarithmic dependence on light intensity (through the short circuit current)
according to equation 410.

In Figure 410 is shown the effect of light level on both the opencircuit voltage and the short
circuit current for the solar cell under test.
Figure 410. Effect of Light Intensity on Solar Cell Performance
A cursory evaluation of Figure 410 shows that the shortcircuit current increases linearly as
luminous intensity rises. Additionally, the opencircuit voltage continues to increase for higher
light intensity levels. (When loaded, the shunt resistance effect becomes dominant at low current
levels; the series resistance effect becomes dominant at high current levels.)
0.000
1.000
2.000
3.000
4.000
5.000
6.000
0.530
0.540
0.550
0.560
0.570
0.580
0.590
0.600
0.610
0.0000 0.0500 0.1000 0.1500
Curr
ent (A
)
Voltage
(V)
Irradiance (W/cm?)
Voc and Isc vs. Irradiance
Voc
Isc
26.7 ?C
(410)
30
Changes in temperature will adversely affect the output power of a solar cell. Increasing
temperature reduces the band gap of a semiconductor which effectively increases the energy of
electrons within the material. Therefore, less energy is necessary to break the bonds and raise
those electrons from the valence band to the conduction band. Theoretically, this means that as
the temperature rises, the shortcircuit current increases; but the opencircuit voltage is reduced.
In practice, changes in the internal resistance that occur with temperature may also produce
visible effects in the output. According to the simplified pn junction (diode) equation [13]:
The opencircuit voltage is directly proportional to the temperature. However, it is also
dependent upon the saturation current IS which approximately doubles for every 10 ?C rise in
temperature [14]. In Figure 411 is shown the measurement of both the opencircuit voltage and
shortcircuit current as functions of temperature.
31
Figure 411. VOC and ISC as Functions of Temperature (?C)
The temperature range over which the solar cell was tested is 26.751.7 ?C (80125 ?F), with
incremental changes of 2.8 ?C (5 ?F). (This test was conducted by direct application of radiant
heat energy to the solar cell.) As expected, the opencircuit voltage decreases and the short
circuit current increases as the temperature rises.
The area of the solar array that is necessary to produce enough power to run the
electrolyzer stack postulated in Chapter 3 using the solar cell under test in this work is calculated
in equations 414 to 416.
2.400
2.420
2.440
2.460
2.480
2.500
2.520
0.510
0.520
0.530
0.540
0.550
0.560
0.570
0.580
0.590
25.0 30.0 35.0 40.0 45.0 50.0 55.0
Curr
ent (A
)
Voltage
(V)
Temperature (?C)
Voc and Isc vs. Temperature
Voc
Isc
(414)
(415)
(416)
32
According to the National Association of Home Builders, a 2009 survey of 54 home builders
states that the average home size was slightly greater than 2700 square feet [15]. Consequently,
this solar array would fit comfortably on the roof of the average U.S. home represented in the
survey (assuming a footprint greater than 600 square feet).
33
CHAPTER 5: CONCLUSION
The scaled up versions of the electrolyzer, fuel cell and the solar array (based on the solar
cell utilized in this study) presented no technical problems with regard to power or spacing
constraints. Notwithstanding, the system would work best in cool, sunny environments. The
amount of hydrogen equivalent to 32 kWh of fuel is calculated in equation 51.
The volume of this amount of hydrogen gas is approximated using the ideal gas law [9] as shown
in equations 52 and 53.
(
)
This volume represents a relatively large tank at atmospheric pressure. Since it is more likely for
the hydrogen to be stored at pressures greater than one atmosphere, the size of the storage tank
may be reduced significantly. However, the added need for a compressor would increase the
total energy consumption of the system. Furthermore, because the backup power system
generates dc output, additional power conditioning equipment such as an inverter may be
necessary if the end use requires ac power.
Assuming a tank volume of 40 gallons, the amount of energy required to pressurize the
hydrogen at an ambient temperature of 23.9 ?C (75 ?F) is calculated in equations 54 and 55.
(51)
(52)
(53)
34
(
)
Utilizing a compressor of 70% efficiency (typical of the rotary type) would increase the overall
system energy requirement by approximately 0.40 kWh. This amount of energy is equivalent to
an additional 10.2 g H2 that must be generated.
(54)
(55)
35
CHAPTER 6: FURTHER RESEARCH
Further investigation will be necessary to gain more insight into the feasibility of a
hydrogen based residential backup power system. The hope is that through the analysis of more
and different data, greater power densities and higher operating efficiencies may be achieved.
The modular construction of the system lends itself well to the optimization of individual
components. For the solar array, additional research could include solar energy concentration,
sun tracking hardware, and the use of novel substrate materials. Fuel cell research could
incorporate operation at higher pressures (> 1 atm), use of pure oxygen (as opposed to ambient
air), and more efficient water management. Finally, electrolyzer research could integrate new
data regarding higher operating pressures (>1 atm) and the use of better materials for the catalyst
and membrane electrode assembly.
36
REFERENCES
[1] J. Duncan Glover, Mulukutla S. Sarma, and Thomas J. Overbye. Power System Analysis and
Design, 4th edition: 2. Stamford, CT: Cengage Learning, 2008. Print.
[2] ?How much electricity does an American Home Use?? EIA, 2011. Web. 27 Jan. 2013.
< http://www.eia.gov/tools/faqs/faq.cfm?id=97&t=3>
[3] XiaoZi Yuan, Chaojie Song, Haijiang Wang, and Juijun Zhang. Electrochemical Impedance
Spectroscopy in PEM Fuel Cells: Fundamental and Applications, 1st edition: 2. London:
Springer, 2010. Print.
[4] XiaoZi Yuan, Chaojie Song, Haijiang Wang, and Juijun Zhang. Electrochemical Impedance
Spectroscopy in PEM Fuel Cells: Fundamental and Applications, 1st edition: 13. London:
Springer, 2010. Print.
[5] "Standard Reduction Potentials." Jesuit New Orleans AMDG, 2011. Web. 3 Mar. 2013.
[6] David Sadava, H. Craig Heller, Gordon H. Orians, et al. Life: The Science of Biology, 8th
edition: 34. Sunderland, MA: Sinauer Associates Inc., 2008. Print.
[7] ?Dupont Fuel Cells: DuPont Nafion PFSA Membranes.? DuPont, 2009. Web. 4 Oct. 2012.
[8] ?PEM Electrolyzers.? Horizon Fuel Cell Technologies, 2007. Web. 4 Oct. 2012.
[9] William L. Masterton, Cecile N. Hurley. Chemistry: Principles and Reactions, 6th edition,
107. Belmont, CA: Brooks/Cole Cengage Learning, 2009. Print.
[10] ?Three Photovoltaic Technologies: Monocrystalline, Polycrystalline and Thin Film.?
Wholesale Solar, 2011. Web. 4 Oct. 2012.
[11] (2012). ?Effect of Parasitic Resistance.? Photovoltaic Education Network, n.d. Web. 4 Oct.
2012.
37
[12] Martin Wolf and Hans Rauschenbach. ?Series Resistance Effects on Solar Cell
Measurements.? Advanced Energy Conversion, vol. 3 1963: 455479. Pergamon Press
[13] Theodore F. Bogart Jr., Jeffrey S. Beasley, and Guillermo Rico. Electronic Devices and
Circuits, 6th edition: 1213. Upper Saddle River: Pearson, 2004. Print.
[14] Theodore F. Bogart Jr., Jeffrey S. Beasley, and Guillermo Rico. Electronic Devices and
Circuits, 6th edition: 18. Upper Saddle River: Pearson, 2004. Print.
[15] ?Breaking Down House Price and Construction Costs.? National Association of Home
Builders, 2010. Web. 27, Jan. 2013.
< http://www.nahb.org/fileUpload_details.aspx?contentID=85253&fromGSA=1>
38
APPENDIX A: DETERMINATION OF IRRADIANCE
The light source used to illuminate the solar cell under test in this work was a 1500
lumens/100 W halogen floodlight. In order to calculate the irradiance of the incident light
striking the surface of the solar cell, it was first necessary to determine the bulb?s output power.
The equipment used to accomplish this task was the following: a halogen flood light, a flat
circular aluminum pan (painted black) filled with water, a thermometer, and a stop watch. To
begin the process, the water volume was measured. The mass to volume ratio of water is close to
one; therefore, the volume in milliliters is numerically equal to the mass in grams. After
measuring the diameter of the aluminum container, the surface area of the water was calculated
in equation A1.
( ) ( )
The following definitions apply in equation A1:
Subsequent to allowing the water temperature to reach equilibrium with that of the ambient air,
the light source was placed a distance of 10.16 cm (4 in) away from it. This initial temperature
was measured and recorded. Next, the light source was switched on while the timer was
simultaneously started. After permitting the water temperature to rise for a total of 15 minutes,
the timer was stopped and the light source switched off. The new water temperature was
immediately measured and recorded. This process was repeated six more times for incremental
(A1)
39
increases in distance of 10.16 cm (4 in). Thus, the light energy, power, and irradiance over the
water surface at each distance were determined according to equations A2 through A4.
The following definitions apply in equations A2 to A4:
( )
The results are show in Table A1.
(A2)
(A3)
(A4)
40
Distance (cm) Ti (?C) Tf (?C) ?t E_H?O (J) P_H?O (W) Irr_H?O (W / cm?)
10.16 24.4 37.2 12.8 25315.2 28.13 0.1174
20.32 24.4 35.6 11.1 22013.2 24.46 0.1021
30.48 24.4 32.8 8.3 16509.9 18.34 0.0766
40.64 24.4 30.6 6.1 12107.3 13.45 0.0562
50.8 23.3 28.3 5.0 9905.9 11.01 0.0460
60.96 24.4 28.3 3.9 7704.6 8.56 0.0357
71.12 24.4 26.7 2.2 4402.6 4.89 0.0204
Table A1. Light Energy, Power, and Irradiance
In order to evaluate the percent error, the thermometer was used to measure both the
freezing point and boiling point of water at an atmospheric pressure of 30.11 inHg. The freezing
point of water was measured as 1.1 ?C (34 ?F); the boiling point was measured as 99.4 ?C (211
?F). Assuming the error to be linear with respect to the temperature gives a slope as found in
equation A5.
The following definitions apply in equation A5:
Using the calculated error slope, the correction factors for the initial and final measured water
temperatures may be found according to equations A6 and A7.
(A5)
(A6)
41
The results for equations A6 and A7 are presented in Table A2.
Ti (?C) Tf (?C) e_b (?C) e_z (?C) e_i (?C) e_f (?C)
24.4 37.2 0.6 1.1 0.7 0.5
24.4 35.6 0.6 1.1 0.7 0.5
24.4 32.8 0.6 1.1 0.7 0.6
24.4 30.6 0.6 1.1 0.7 0.6
23.3 28.3 0.6 1.1 0.7 0.6
24.4 28.3 0.6 1.1 0.7 0.6
24.4 26.7 0.6 1.1 0.7 0.7
Table A2. Error Calculations
Now, the new light energy, from which power and irradiance are found, can be calculated
according to equation A8.
{ }
Of course, the power and irradiance may be found using equations A3 and A4. Finally, the
percent error can be calculated in accordance with equation A9.
The results for the newly calculated light energy, power, irradiance, and percent error are all
shown in Table A3.
(A7)
(A8)
(A9)
42
Distance (cm) Ti (?C) Tf (?C) E_H?O (J) P_H?O (W) Irr_H?O (W / cm?) % error
10.16 24.4 37.2 25744.3 28.60 0.1194 1.7
20.32 24.4 35.6 22386.3 24.87 0.1039 1.7
30.48 24.4 32.8 16789.7 18.66 0.0779 1.7
40.64 24.4 30.6 12312.5 13.68 0.0571 1.7
50.8 23.3 28.3 10073.8 11.19 0.0467 1.7
60.96 24.4 28.3 7835.2 8.71 0.0364 1.7
71.12 24.4 26.7 4477.3 4.97 0.0208 1.7
Table A3. Adjusted Light Energy, Power, Irradiance, and Percent Error
These small errors are negligible.
43
APPENDIX B: MATLAB CODE FOR SIMULATIONS
% Solar cell simulation plots from Chapter 4
clc
clear all
close all
Is = 1e6; % Diode Reverse Current
q = 1.60217733e19; % Electron Charge
eta = 1.583722347911; % Diode Emission Coefficient
k = 1.380658e23; % Boltzmann Constant
Tf = 85; % Fahrenheit Temperature
T = (5/9) * (Tf + 459.67); % Kelvin Temperature
Vt = (k * T) / q; % Thermal Voltage in Volts
Ig = 0.6; % LightGenerated Current in Amps
VL = zeros(1,1001); % Load Voltage Vector (Volts)
IL = zeros(1,1001); % Load Current Vector (Amps)
% Vary LightGenerated Current (Figure 46)
Rs = 0.167;
Rsh = 211.573;
color = ['b', 'g', 'r', 'c','m','y','k'];
hold on
for n = 1:6
num = 1;
for RL = 0:0.05:50
VD = solve('Ig = Is*(exp(VD/(eta*Vt))1) + VD/Rsh + VD/(Rs+RL)',
'VD');
IL(num) = subs(VD) / (Rs + RL);
VL(num) = subs(VD)  IL(num) * Rs;
num = num + 1;
end
plot(VL,IL,color(n), 'LineWidth', 2)
Ig = Ig + 0.6;
end
legend('I_G = 0.6 A', 'I_G = 1.2 A', 'I_G = 1.8 A', 'I_G = 2.4 A', 'I_G = 3
A', 'I_G = 3.6 A')
xlabel('Load Voltage (V)')
ylabel('Load Current (A)')
title('Varying Light Generated Current')
hold off
Ig = 2;
% Vary Shunt Resistance (Figure 44)
figure
Rs = 0.167;
Rsh = 211.573;
color = ['b', 'g', 'r', 'm'];
44
hold on
for n = 1:4
num = 1;
for RL = 0:0.05:50
VD = solve('Ig = Is*(exp(VD/(eta*Vt))1) + VD/Rsh + VD/(Rs+RL)',
'VD');
IL(num) = subs(VD) / (Rs + RL);
VL(num) = subs(VD)  IL(num) * Rs;
num = num + 1;
end
plot(VL,IL,color(n), 'LineWidth', 2)
Rsh = Rsh / 10;
end
legend('Rsh = 211 ohms', 'Rsh = 21.1 ohms', 'Rsh = 2.11 ohms', 'Rsh = 0.211
ohms')
xlabel('Load Voltage (V)')
ylabel('Load Current (A)')
title('Varying Shunt Resistance')
hold off
% Vary Series Resistance (Figure 45)
figure
Rsh = 211.573;
hold on
for n = 1:4
num = 1;
for RL = 0:0.05:50
VD = solve('Ig = Is*(exp(VD/(eta*Vt))1) + VD/Rsh + VD/(Rs+RL)',
'VD');
IL(num) = subs(VD) / (Rs + RL);
VL(num) = subs(VD)  IL(num) * Rs;
num = num + 1;
end
plot(VL,IL,color(n), 'LineWidth', 2)
Rs = Rs * 2;
end
legend('Rs = 0.167 ohms', 'Rs = 0.334 ohms', 'Rs = 0.668 ohms', 'Rs = 1.336
ohms')
xlabel('Load Voltage (V)')
ylabel('Load Current (A)')
title('Varying Series Resistance')
hold off