abstract
Air pollution is of serious concern in many Asian countries, especially in densely-populated cities
with many highly-polluting two-stroke engine vehicles.The present value of health effects have
been estimated at hundreds of dollars or more, over each vehicle’s lifetime, for a reasonably
wealthy country like Taiwan. Four-stroke engines and electric battery-powered scooters are often
proposed as alternatives, but a fuel cell scooter would be superior to both by offering both zero
tailpipe emissions and combustion-scooter class range (200 km).
Unlike 50 kW automobile-sized fuel cell stacks, the vehicular 5 kW fuel cell needed here has not
received much attention. This niche is examined here with a conceptual design and consideration of
the issues of water, heat, and gas management. The application is extremely sensitive to size,
weight, and cost, so a proton exchange membrane fuel cell using hydrogen stored in a metal
hydride is best. Hydrides also act as sinks for waste heat due to the endothermic hydrogen
desorption process. Pressurized operation is found to be ineffective due to high parasitic power
demands and low efficiencies at the low powers involved.
A computer simulation is developed to examine overall vehicle design. Vehicle characteristics
(weight, drag, rolling resistance), fuel cell polarization curves, and a Taiwanese urban driving
cycle are specified as inputs. Transient power requirements reach 5.9 kW due to the rapid
accelerations, suggesting a large fuel cell. However, average power is only 600 W: a hybrid vehicle
with a small fuel cell and peaking batteries could also handle the load. Results show that hybrid
vehicles do not significantly improve mileage, but are certain to precede pure fuel cell scooters
while fuel cells are still more expensive than peaking batteries.
i
System size is approximately the same as current electric scooters, at 43 L and 61 kg for the fuel
cell, hydrogen storage, and electric motor / controller. Manufacturing costs of fuel cell scooters are
expected to decrease to under $1,300 in the long term, with per-km fuel costs half of those for
gasoline scooters. Hybrid zinc-air scooters offer similar performance at slightly lower vehicle
price, but the fuel infrastructure costs may be prohibitive.
ii
acknowledgments
With periods of hard acceleration, rapid decelerations, and occasional stalls in the course of writing
this thesis, sometimes I felt that I was on the Taipei Motorcycle Driving Cycle myself. Thanks to
everyone who had a part in this effort.
Thanks to my advisors Robert Socolow, Bob Williams, and Joan Ogden, and my thesis reader
Enoch Durbin.
Thanks to the many people from various research groups, companies, and academic institutions
who helped with guidance, hard data, and advice.
Thanks also to my family and friends and colleagues who supported me in the past twelve months,
and for many, much longer than that.
Support for this research came from the Center for Energy and Environmental Studies, the
Mechanical and Aerospace Engineering Department (including a Daniel and Florence Guggenheim
Fellowship and a Sayre Prize), the United States Department of Energy, and the Energy
Foundation.
This thesis carries 3055-T in the records of the Department of Mechanical and Aerospace
Engineering.
Motorcycle populations in selected countries, 1993
6
1.2
VMT data for Taipei, 1987
11
1.3
A comparison of vehicle emissions standards
24
1.4
Data on motorcycle emissions: four-strokes and catalysts
26
1.5
Simulated emissions from more realistic driving cycle
26
1.6
PSI subindex pollutants in Taiwan
29
1.7
Cleanup technology, effects and prices
35
1.8
ITRI prediction of effects of scooter replacement on pollution
37
1.9
Estimate of externality damages from air pollutants
38
1.10
Electric Motorcycle Development Action Plan
43
xi
Chapter 2
2.1
Comparison of power systems
54
2.2
Motor specifications: UQM brushless and NGM hub motors
61
2.3
ZES-2000 electric scooter performance
66
2.4
Battery goals for various time frames
67
2.5
Peaking power battery characteristics
76
Chapter 3
3.1
Stack size, weight, cost summary
110
3.2
Fuel gravimetric and volumetric energy densities, lower heating value basis
121
3.3
Steam reforming versus partial oxidation
122
3.4
Hydrogen output from reformed hydrocarbon fuels
124
3.5
Reformer performance
126
3.6
Chemical hydride comparison
129
3.7
Theoretical performance of various metal hydrides
138
3.8
Metal hydride systems comparison
141
3.9
Compressed gas options
145
3.10
Storage technology comparison
148
Chapter 4
4.1
Performance of various vehicles of about 5 kW power
157
4.2
Fuel cell scooter performance requirements
159
4.3
Typical modeling parameters
164
4.4
Validation of physical model
168
4.5
Driving cycle comparison
174
4.6
Effects of “jitter”
175
xii
4.7
Results of different algorithms applied to TMDC; comparison to FTP
178
4.8
Taiwan battery-powered scooter
185
4.9
Various battery-powered designs for Taiwan scooter
185
4.10
Fuel cell design parameters at maximum power
190
4.11
Flow rate parameters at maximum power
191
4.12
Stack temperature model parameters
200
4.13
System performance under TMDC and at cruising speed
208
4.14
Subcomponent summary
208
4.15
Size of various storage designs
209
4.16
Hybrid 1.1 kW scooter inadequacies
219
4.17
Hybrid fuel cell stack designs
221
4.18
Peaking power battery characteristics
221
4.19
Hybrid performance at 30 km/h
223
4.20
Hybrid performance under TMDC
223
4.21
Hybrid system design
229
4.22
Component breakdown for 3.2 kW scooter
230
4.23
Component breakdown for 1.1 kW scooter
232
4.24
Hybrid battery configuration for Taiwan scooter model
233
4.25
Hybrid power system summary
235
4.26
Performance metrics
236
4.27
Near term 1 kW fuel cell hybrid designs
238
Chapter 5
5.1
Internal combustion engine scooter parts
244
5.2
Battery-powered electric scooter parts
245
5.3
Metal hydride storage costs
245
5.4
Long-term scooter cost to manufacture
247
xiii
5.5
Summary of cost estimates
248
5.6
Short term bridging to the future
249
5.7
Taiwan vs. USA energy prices, 1997 USD
252
5.8
Fuel costs of Taiwan in $/GJ LHV
252
5.9
Comparison of assumptions for zinc-air electrowinning costs
253
5.10
Fuel cost summary
258
5.11
Fuel cell scooter performance requirements
261
5.12
System design results
263
5.13
Long-term cost of hybrid fuel cell scooters
264
5.14
Fuel cost summary
264
list of figures
Chapter 1
1.1
A scooter
3
1.2
Taiwan vehicle mix 1991-1998
9
1.3
Scooter distribution in Taiwan 1991-1998
9
1.4
Four-stroke cycle
15
1.5
Two-stroke cycle
18
1.6
Carbon monoxide emissions by source
28
1.7
Hydrocarbon emissions by source
28
1.8
PSI in Taiwan, 1994-1996
50
Chapter 2
2.1
Axial-gap pancake motor
59
2.2
Typical torque vs. rpm curve for DC motor
63
xiv
2.3
Voltage and internal resistance of Bolder peaking battery
78
Chapter 3
3.1
Fuel cell schematic
87
3.2
Tafel plot
92
3.3
Effects of pressurization on polarization curves
94
3.4
Nafion chemical structure
98
3.5
Stack diagram
100
3.6
Active cell
109
3.7
Ignition energy of hydrogen
132
3.8
Metal hydride adsorption curve
136
Chapter 4
4.1
Free body diagram of scooter
161
4.2
Cruising power required at various speeds
166
4.3
Power required to climb various slopes at 15 km/h
166
4.4
Power required for various accelerations from 30 km/h
167
4.5
Validation of physical model
169
4.6
mFTP: modified Federal Test Procedure
171
4.7
ECE-40
172
4.8
Taipei Motorcycle Driving Cycle (TMDC)
173
4.9
Smoothed TMDC
180
4.10
Torque vs. rpm during TMDC
182
4.11
Power required in TMDC
183
4.12
Polarization curve
187
4.13
Metal hydride cooling vs. power
195
4.14
Heat generation as a function of time in TMDC
198
xv
4.15
Stack temperature as a function of time in TMDC
201
4.16
Parasitics as a function of power
204
4.17
Parasitics as a percentage of power
205
4.18
Effect of parasitics on efficiency
206
4.19
Weights of subsystems
212
4.20
Volumes of subsystems
212
4.21
Atmospheric power versus 3 atm power
214
4.22
Division of power between fuel cell and battery during TMDC, 3.2 kW stack
225
4.23
State of charge of battery over TMDC, 3.2 kW stack
226
4.24
Division of power between fuel cell and battery during TMDC, 1.1 kW stack
227
4.25
State of charge of battery over TMDC, 1.1 kW stack
228
xvi
~
Chapter One
Introduction
~
1
The hydrogen fuel cell has received extensive attention in the scientific community and the public
at large since about 1990. The first experimental fuel cell was developed in the nineteenth century,
and a 6 kW alkaline fuel cell in conjunction with a battery bank was used to power a small car as
early as 1966, but it would not be until major improvements in power density were made in the
1990's that major car companies took serious interest in fuel cells.1 Although the technology is
currently quite expensive, fuel cells offer significant benefits including high overall efficiency, quiet
operation due to few moving parts, and good efficiency over a wide range of operating points.
Predicted cost reductions mean that in the near future, fuel cells could power everything from
homes to vehicles to cell phones.
Although extensive research has been done into fuel cells for stationary power and for automobiles,
and some research has been done for portable power applications like soldier power and devices
like telephones and computers, virtually no work has been done in the field of small vehicles
requiring under 10 kW of power.2 This is an interesting option for small vehicles because the
market – and governments – are beginning to put a high value on options offering low or zero
emissions. Moreover, the challenge of putting fuel cells in scooters is an interesting technical
problem because, due to weight and cost restrictions, power systems in these vehicles cannot be as
complex as those found in cars. Yet, there is a high value on clean power. Subsystems like air
compressors, reformers, and hydrogen storage tanks are all reduced in size and complexity, so
production is made easier. On the other hand, efficiencies do not remain constant at small size so
performance in this type of application will be poorer than in automobile fuel cell power systems.
The purpose of this study is to examine a particular application of fuel cell technology: the electric
scooter. Scooters are small two-wheeled vehicles that can carry one or two people. They are unlike
motorcycles in that they are ridden in a seated position with feet forward on a platform. Although
2
in North America they are most associated with 1950's Vespas and the mod scene of later decades,
these small and cheap vehicles remain a major mode of transportation in Asia and Europe today.
(Note that the distinction between “scooters” and “motorcycles” is not always made in the
literature, especially by Asian researchers. Here it is assumed that “motorcycles” refers to scooters;
this assumption is almost certain when it comes to vehicles less than 50 cc in displacement.)
Figure 1.1 A Scooter
Honda CUV-ES electric scooter3
Due to their small size and low price point, scooters have traditionally been powered by high power
density two-stroke internal combustion engines, (although some of the larger models use fourstroke engines). Two-stroke engines produce a great deal of pollution and are an object of concern
in many Asian countries.
3
Severe pollution from two-stroke engines is a significant driver for cleaner technology. Thus, the
target market for this study is the Asian urban commuter, since scooter use is so heavy in many
Asian cities, and air pollution is a major problem in the crowded cities of the Far East.
Specifically, Taiwan (i.e. the Republic of China) is a prime example, with twenty million people
sharing an area the size of Vancouver Island with ten million scooters. Compared to the batterypowered scooters currently being promoted by the Taiwan government, fuel cell engines offer the
advantages of extended range and quick refueling.
Some countries in Europe, like Italy, also have extensive scooter populations and might also be
able to afford expensive new technology more readily. Poorer countries like China and India are
facing dramatic growth rates in two-stroke vehicle population as rickshaws and bicycles are being
replaced, and low-powered but clean scooters would be a major step in providing mobility without
compromising urban air quality.
Five chapters comprise the thesis.
The first outlines the pollution situation, includes a description of the two-stroke engine’s pollution
characteristics, and outlines Taiwan air pollution policy. A possible method for valuing reductions
in air pollution is presented.
The second chapter discusses electric scooters and battery power for them. Hybrid vehicles and
peaking power batteries are explained. The new zinc-air batteries, with their excellent energy
storage densities, are examined as some scooter researchers and manufacturers are carefully
looking at them for second generation zero emission scooters.
4
The third chapter describes in detail the engineering issues and science behind fuel cell technology
and hydrogen storage. Both advantages and disadvantages of this type of power are examined.
Hydrogen storage in the form of metal hydrides, and a proton exchange membrane fuel cell running
at low temperatures, are chosen for the reasons of ease of manufacture and operation, low cost,
and minimal volume.
The fourth chapter is the simulation and conceptual design core of the thesis. It explains the
physical vehicle simulation used to simulate vehicle power requirements during typical urban
driving. Using the specifications produced by the driving simulation, a fuel cell power system is
designed. The fuel cell components are selected along with the hydrogen storage subsystem. The
possibility of “hybridizing” the system by using a battery energy storage system is treated; this idea
offers possible energy savings from regenerative braking and reduces the maximum size of the fuel
cell, reducing cost. The performance of such a vehicle is examined in terms of technical
performance metrics: total weight, fuel economy. (Note that this thesis did not involve construction
of a physical prototype construction; the interested reader is referred to Appendix G for more
information on that topic.)
The final chapter describes how these scooters might be brought to market. How much would a
prototype cost? Could a fully-developed scooter be competitive with electric or two-stroke
scooters? How would fuel costs compare to battery-powered scooters and gasoline-powered
scooters? Infrastructure issues are briefly discussed. With the cost information finishing off the
body the study, a final summary is presented that recapitulates the findings.
5
1.1 Transportation Background
1.1.1 Why Taiwan?
There are approximately 100 million motorcycles in the world. The greatest numbers are
concentrated in Asia, and it is here that alternative scooters could have a major impact. Some
illustrative countries are listed below:
Table 1.1. Motorcycle populations in selected countries, 1993
Country
Motorcycles
% of
total vehicles
Country
Motorcycles
% of total
vehicles
Argentina
882,000
15.5%
Switzerland
834,900
20.7%
Brazil
1,371,800
9.6%
Spain
2,655,900
17.1%
Canada
434,200
7.0%
UK
913,600
3.6%
Chile
37,120
3.9%
Mexico
661,230
7.8%
Bangladesh
119,790
50.0%
Peru
86,940
12.4%
China
3,047,520
41.2%
USA (1991)
6,830,000
3.7%
Hong Kong
17,100
5.0%
Venezuela
580,920
25.3%
India
7,666,640
69.6%
Indonesia
5,890,760
74.6%
Austria
601,160
14.9%
Japan
18,451,300
26.0%
Belgium
131,670
3.2%
Korea
1,066,800
34.4%
France
3,661,450
12.6%
Malaysia
2,460,640
59.0%
Germany
2,427,480
7.3%
Pakistan
627,170
48.8%
Italy
7,938,420
23.8%
Philippines
281,530
27.2%
Norway
202,860
9.5%
Taiwan (1991)
9,232,889
73.4%
Portugal
51,500
2.9%
Thailand
6,343,558
66.1%
Data from Weaver and Chan 4
6
Numbers of scooters in use are high in Asia, and growth rates are also high. The People’s Republic
of China, for instance, had 500,000 motorcycles in 1980, and 10 million in 1994 - an annualized
growth rate of 24%, faster than the 15-20% of Chinese urban vehicles in general.5 India had an
average annual growth rate of 16% for two-wheeled vehicles from 1981 to 1998.6
Worldwide scooter production is estimated at 17 million per year.7 In 1994, Taiwan’s motorcycle
industry included 418 assemblers and manufacturers of parts and 16,000 employees. Revenues
totaled $2.4 billion that year while total domestic production reached $3.2 billion (all figures US
dollars unless otherwise noted.) 8
As one of the “Five Tigers”, Taiwan experienced rapid growth in the latter half of this century and
became a manufacturing power; its vast foreign reserves helped it weather the Asian economic
problems of the summer of 1998. Average household income in 1995 was $36,470 for an average
household size (1996) of 3.6; transportation costs were estimated at $4,000 per year, behind
household expenditures for food; rent, fuel, and power; and education.9 Household income is fairly
large compared to Taiwan’s poorer neighbours, so adoption here is (i) easier than elsewhere and
(ii) may ease development of advanced scooters elsewhere.
(In 1998, the U.S. dollar was equal to approximately 30 New Taiwan Dollars).
Air pollution is a major problem on this 400-km long island with an area of 35,873 km2. Industry,
diesel-powered vehicles, and the omnipresent two-wheeled, two-stroke scooters all contribute to the
extremely dirty air. In 1997, the overall population was 21.7 million and the population density
was 601 persons per square kilometer. In the same year, the city of Taipei’s population density was
9560 persons per square kilometer while the second largest city, Kaohsiung, had a population
7
density of 9350 persons per square kilometer. Urban centres with population over 1 million
contained 67.8% of Taiwan’s population.10
Taiwan is focused on here, because of the high fraction of scooters in its vehicle fleet, its poor air
quality, and because it is one of the top six producers of scooters in the world. Being wealthier than
many of the other countries with high scooter densities, Taiwan can afford to spend money on
novel vehicle designs; on the other hand, it should be noted that any improved scooters that were
developed would be of great benefit in reducing high air pollution levels in other developing
countries.
1.1.2 Taiwan vehicle fleet
Taiwan’s transportation split is interesting. Historically, the lack of an automotive industry in the
critical growth period meant that people rapidly adopted scooters and then did not switch to
automobiles as they became more wealthy. The crowded cities, warm weather, dense population
and limited land continue to make scooters popular. Car use is increasing, but scooters have the
advantage of being able to swarm through the congested car traffic in cities. This explains the over
ten million scooters currently in Taiwan, of which approximately 60% are low-power scooters
under 50 cc (cubic centimeters) in cylinder displacement.11 The largest cylinder size (i.e. most
powerful engine) allowed in Taiwanese scooters is 150 cc. Especially high-polluting two-stroke
vehicles made up 40% of all vehicles in Taipei in 1996.12
8
Figure 1.2. Taiwan vehicle mix 1991-1998
20
vehicles (millions)
18
16
14
12
10
8
6
4
2
0
19911992199319941995199619971998
passenger car
motorcycle 50cc
other
Data from the Monthly Bulletin of Statistics of the Republic of China,
February 1999.13
Figure 1.3. Scooter distribution in Taiwan 1991-1998
50%
fraction of motorcycles 100 (data for years since 1993), but these were the extreme
cases. For 5,690 county-years between 1993 to 1998 for counties across the United States, fully
99.2% had less than 6% exceedances per year.46 In fact, 83.5% of the county-years had less than
1% PSI exceedances. In other words, Taiwan’s overall pollution rate of 6% exceedances per year
was worse than all but 0.8% of American individual county readings.
In addition, the average level of pollution has been decreasing over time, as indicated by various
subsets of the seventy-one monitoring stations scattered across Taiwan:
Figure 1.8 PSI in Taiwan, 1994-1996
100
PSI (annual average)
90
80
industrial area
70
national park
60
50
"background"
40
traffic area
30
20
overall
10
0
1994
1995
1996
Data is from the Taiwan Environmental Protection Agency.47
30
1.2.5 Cleaner combustion technology
There are numerous options for reducing vehicle emissions. Over the years, four-stroke engines
have received much more research attention than two-stroke engines, due to the overwhelming
number of automobiles in the world, and this is part of the reason automobile technologies like
catalysts and fuel injection have not been adapted for the two-stroke market. The most important
reason, however, is that most of the pollution cleanup technologies add weight and cost, eroding the
original benefits of two-stroke engines.
1.2.5.1 Exhaust gas recirculation
EGR is mainly used in automobiles to reduce NOx production, but it is discussed here mainly to
explain why two-stroke scooters produce low levels of that particular pollutant. Exhaust gas
recirculation causes some of the burned gases to combine with the incoming air/fuel. This lowers
the engine temperature because the relatively large heat capacity of triatomic species like CO2 and
H2O in the recirculated exhaust gas dilute the contents and steal heat from the combustion process,
and this means less NOx production. In fact, due to the nature of two-stroke engines (with the
incoming charge partially mixing with the outgoing gases), some EGR occurs automatically. EGR
has the disadvantage of slowing combustion rate and thus making stable combustion more difficult;
there is increased possibility of unburnt hydrocarbon emissions.48
1.2.5.2 Superchargers
Superchargers allow precompression of the air/fuel without requiring crankcase compression, and
avoid mixing lubricating oil with the fuel, but add expense. Essentially, a supercharger is a
31
compressor or blower that increases the pressure of the intake air. It may be powered off the
crankshaft (thus parasitically consuming some of the developed power), electrically, or by a turbine
driven by the exhaust gas flow in which case it is called a turbocharger.
1.2.5.3 Fuel injection
As alluded to earlier, with fuel injection systems only air is compressed in the crankcase, not an
air/fuel mixture. The fuel spray is then injected at high pressure into the compressed air stream just
before intake or directly into the combustion chamber; such a system allows more precise,
electronic control of the air-to-fuel ratio than a carburetor. Orbital Engineering in Australia is one
company trying to combine the high power density of two-stroke engines with the efficiency of
direct fuel injection; the company also modifies the combustion chamber to improve emissions.
Fuel injection is estimated to reduce fuel consumption by 25-35% due to the more complete
combustion, and to reduce unburned hydrocarbon emissions by 75-85% and carbon monoxide by
50% for the same reason.49
However, the pumping system required to maintain injection pressure reduces the power density
advantage of two-stroke engines.50 A Piaggio study estimates the pump power at 300 W at
maximum speed of 8000 rpm, for a test 4 kW engine; this amounts to a 7.5% parasitic power
loss.51
32
1.2.5.4 Catalysis of exhaust gases
Three-way catalysts like those found in typical automobile engines are composed of alloys of
expensive metals like platinum and/or palladium with rhodium. Three-way catalysts are so named
because they simultaneously oxidize hydrocarbons and CO, and reduce NO to nitrogen. A rich
air/fuel ratio is needed for the NOx reduction; this richer condition increases exhaust pollutants and
partially offsets the benefit of catalysis. These systems require fairly precise stoichiometry and
typically electronic control using oxygen sensors in the exhaust pipe is needed to maintain this
ratio.
Oxidation catalysts, on the other hand, use metals like platinum and/or palladium to increase the
rate of oxidation of exhaust molecules like CO and hydrocarbons; essentially, this is catalytic
combustion. Catalysts are “poisoned” by lead in the fuel, and sulfur or phosphorous compounds
that may be found in the lubricating oil; active sites are taken up by these compounds and the
catalysts must be thermally or chemically treated to restore their function.
The high proportion of scavenged unburnt air/fuel in the exhaust gas is problematic for catalytic
converters. On the one hand, the heat capacity of the hydrocarbons reduce the temperature of the
exhaust, delaying catalyst activation. On the other hand, catalyst oxidation of the unburned A/F
may increase temperature to too high a level, causing catalyst durability problems. The solution is
sometimes to use two-stage catalysts, with the rate of catalysis controlled by admitting “secondary
air”52
In the National Taiwan University experiment described in Table 1.4, an oxidation catalyst
attached to a two-stroke motorcycle was found to reduce 45.4% of total hydrocarbons, and 61.2%
33
of CO; emissions for NOx were already extremely low. (In comparison, the same study found that
car three-way catalysts achieved reductions of 90.5% of total hydrocarbons, 88.0% of CO, and
94.2% of NOx of automobile exhaust).
1.2.5.5 Replacement by four-stroke engines
The more and more stringent emissions standards have made switching to four-strokes an
increasingly attractive option. In fact, a researcher at ITRI wrote that the announced year 2003
fourth-stage standards would be “too tough for 2 stroke [engines] to survive. This is an
understanding between Taiwan EPA and motorcycle makers to phase out 2-strokes by that time.”53
This is an easy solution because it leverages well-understood existing technology. Drawbacks
include greater vehicle weight and larger engine sizes, and of course more expensive engines.
Advanced four-strokes would follow the advances made on the automobile side, with three-way
catalysts, engine timing optimization for reduced emissions rather than specific power, etc.
1.2.5.6 Relative costs and benefits of various technologies
A Piaggio study also estimated the costs for various clean two-stroke technologies that they
considered for new high-efficiency and low-emissions two stroke engines. The relative costs are
reproduced below, along with estimates of air pollution reduction. Note that the base cost to
manufacture a two-stroke scooter engine is approximately $150.54
34
Table 1.7 Cleanup technology, effects and prices
relative
cost
THC
reduction
CO
reduction
1.0
baseline
baseline
Fuel injection with external scavenge
(separate blower to scavenge cylinder)
1.5-1.7
75% - 85%
50%
FAST (“Fully Atomized Stratified
Turbulence”) - mechanical control
1.2-1.4
68% - 76%
65% - 80%
FAST - electronic control
1.4-1.6
68% - 76%
65% - 80%
1.7
45% - 80%
61 - 95%
1.5-1.7
83%
47%
Type of engine
Two-stroke standard
catalytic converter
Equivalent four-stroke
Sources: for catalyst cost, Felton.55 For other costs, Piaggio study.56 For
reductions in pollution, National Taiwan University study57, Piaggio
study for direct injection with electronic-control pollution reductions
assumed to be the same as mechanical-control reductions.58
Improvements that may seem simple, technologically speaking, actually add about 50% to the
engine cost. On the other hand, the standard engines cost only about $150 to manufacture, so the
difference in dollars is not great - perhaps $150 once manufacturing and markups are included.
So significant reductions are possible using relatively inexpensive improved combustion
techniques, the easiest of which is a transition to only four-stroke vehicles. Are electric vehicles
necessary, then? Or in other words, is it worth spending additional money on “zeroemission”vehicles to reduce emissions further?
1.2.6. Assessing the damage
The process of establishing - and quantifying - a causal link between scooter tailpipe emissions and
health and environmental damages is a long one with many steps. In general, researchers have
35
proceeded through the following stages:
1. Measurement of pollutant emissions by collecting tailpipe exhaust under various simulated
driving cycles, as tested on a dynamometer.
2. Dispersion modeling, based on local wind patterns and atmospheric models, to proceed from
pollution emitted per kilometer on the streets to ambient concentrations in the local environment.
3. Estimation of individual exposure to various pollutants by studying population distributions
4. Dose-response modeling of health effects resulting from exposure. Epidemiological studies are
generally used to try to correlate incidences of high pollution with acute and chronic negative
health effects, which are measured in terms of deaths (mortality) and loss of useful function
(morbidity).
5. Estimates of the cost of health damages, either by calculating the value of lost work-days or by
contingent-valuation surveys that aim to capture the value of health externalities.
Similar processes are applied to damage caused to buildings and other material objects. The
literature contains little data quantifying specific Taiwan conditions, although studies have been
done to estimate valuation of health episodes (i.e. the fifth step).
1.2.6.1 Reduction estimate
Also, a previous ITRI study estimated reductions in CO and THC levels in the air assuming
36
current emission rates and increasing vehicle populations after 1991 (steps 1 and 2 of the costbenefit analysis process enumerated above). The CALINE-4 line source and dispersion model was
used to estimate ambient pollution concentrations near roads. Three situations were studied and the
following results were obtained for the 1991-1996 time period:
Table 1.8 ITRI prediction of effects of scooter replacement on pollution
Scenario
CO
HC
Baseline: no change in
scooter pollution levels
0%
0%
All motorcycles after 1991
meet second-stage standards
17.4%
-5.8%
As above but with 20% of
scooters replaced by electric
24.8%
12.2%
As above, but with 50% of
scooters replaced by electric
35.1%
17.2%
Data from ITRI study 59
These reductions are almost equal to the fractions of carbon monoxide and hydrocarbons emitted
by scooters overall, but it should be kept in mind that these measurements are for roadside ambient
concentrations, not overall emissions.
It is not clear whether it is old, highly polluting scooters or a random sample of scooters that are
being replaced with battery-powered scooters. However, the results clearly demonstrate how
important scooter pollution reductions are in improving localized air quality; scooters clearly were
predicted to produce at least 35% of roadside CO and 17% of roadside HC.
37
1.2.6.2 Externality damage estimate
A systematic study of the health and environmental benefits of reduced air pollution is not within
the scope of this study. However, as a rough estimate of the benefits of cleaner air, the particulate
(PM2.5) pollution emitted by four-stroke scooters was calculated as if it were an automobile, but
factored by the greater fuel economy of the scooter. Next, the externality cost of air pollutants was
obtained from a recent study by Spadaro and Rabl.60 They calculated the following externality
costs of various vehicle air pollutants:
Table 1.9 Estimate of externality damages from air pollutants
Pollutant
Euros / tonne
$ / gram
Urban nitrate
1.6 x 104
$0.0200
Ozone from NOx
1.45 x 103
$0.0018
Fine particles
2.2 x 106
$2.75
(PM2.5)
Note: at the time the study was done (October 1998), 1 ecu was equal
to 1.25 US dollars. The ecu has since been replaced by the euro.
The authors used a fine particulate (PM2.5) emission rate of 0.75 grams for a 43.3 km Paris trip in
a three-way-catalyst-equipped automobile, or 0.017 g/km. This is based on Heywood, which
specifically quotes a figure of 0.020 g/km for particulates for a car running on unleaded fuel with
no catalyst.61 Using the authors’ figure and their implied fuel economy of 9.14 km/L (21.5 mpg)
gives an emission rate of 0.155 grams of fine particulates per liter of fuel. An equal emission rate
was assumed for scooters running on the same four-stroke cycle as the automobiles. With a 100
mpg fuel economy (4.6 times better than the car), the per-kilometer emission rate is only 0.0037
g/km. Annual health costs for 12,000 km/y driving means annual health externalities of $120 per
38
year per four-stroke scooter.
The figure of 0.51 g/km of NOx from Table 1.5 for simulated urban scooter driving, when taken
through similar calculations, yields externalities of $133 per year, for a total pollutant damage of
$253 per year. Damages from other pollutants are not included.
The health cost is scaled by a factor proportional to the ratio of Taiwan GNP per capita to French
GNP per capita, taken to the power of the elasticity of willingness to pay for health with respect to
GNP:
ela stic ity
damage in Taiwan = damage in France x
T a iw a n G N P pe r ca pita
F ran ce G N P pe r ca pita
In 1998, French GNP per capita was $23,789 in 1997 US dollars, and the Taiwan GNP per capita
was $13,819 in 1997 dollars.62,63 The elasticity of willingness to pay with respect to income was
estimated at 0.4 for Taiwan by an Alberini, Cropper, et al. study.64 Note, however, that the original
damages were obtained for a uniform distribution of population around emission sources, and a
population density of 7500 persons per km2 was used. Average density in Taipei is 1.27 times
greater, for a final ratio of 1.022 for Taipei damages to Paris damages in terms of g/km.
Assuming a ten year vehicle lifetime and 10% discount rate means a final present value cost of
emissions of $1,590. Spadaro and Rabl quote a very broad uncertainty in terms of a geometric
mean standard deviation of 4.0, so that a 68% confidence level corresponds to $400 to $6,355. If
elasticity is 1.0 rather than 0.4 (i.e. damage scales linearly with GNP), then the equivalent interval
is $290 to $4,590.
This is a large amount and suggests that improvements in air quality would produce a significant
39
benefit; a complete elimination of tailpipe emissions from the use of electric scooters could be a
significant benefit over even four-stroke engines, the expected replacement for two-stroke engines.
Of course, as the large geometric standard deviation suggests, this is only an attempt to broadly
quantify the problem and is subject to the large uncertainties involved in any cost-benefit analysis.
1.2.7 Government policy approaches
Due to the problem of air pollution, several Asian governments have implemented measures to
control two-stroke vehicles. For example, Thailand motorcycles have been restricted to 150 cc,
“presumably ... to limit the maximum engine power, and thus the acceleration rates, top speed, and
fuel consumption”65. In practice, the policy encourages use of more polluting two-stroke engines
for their higher power at the same 150 cc displacement. Similarly, the city of Shanghai currently
tightly restricts the supply of motorcycle licenses, although as previously discussed, all predictions
point to vastly increasing Chinese vehicle usage. Taiwan’s policy towards scooters, which is more
advanced than virtually all of its neighbours, is discussed below.
Taiwan’s 1991 (“second stage”) standards were the strictest in the world, and essentially forced
two-stroke motorcycles to be equipped with oxidation catalytic converters to meet these
requirements. These catalytic converters cost approximately $80 for the manufacturer. To meet the
1991 standards, four-stroke motorcycles required modification to allow exhaust air injection
(estimated cost $40-$60).66
40
Most Taiwanese motorcycle manufacturers depend on foreign (Japanese) parent companies for
advanced technology, and catalytic converters are all imported. Naturally, Taiwanese
manufacturers would like to produce them domestically.67
1.2.7.2 Later years: inspection and maintenance
In addition to tightening standards for new motorcycles and scooters, the Taiwanese government
has acted to control emissions from highly polluting existing motorcycles. This is an effective
method of cleaning up the worst offenders, which can account for the majority of the pollution.
Originally, this consisted solely of roadside testing of emissions from randomly selected vehicles.
Annual stationary emissions testing was begun in several counties. Testing was voluntary and there
were no fines; incentive was provided by a system where drivers who brought in their vehicles for
testing were given ballots to enter in a cash lottery.68
Later, a sticker system was implemented and as of 1998, motorcycle and scooter licenses can be
revoked if the owner fails to bring in the vehicle for annual emissions testing. Vehicles that fail the
test must be tuned up and brought in a month later for retesting; a second failure also means license
revocation. By tackling the most errant vehicles, significant gains can be made; statistics report an
average 48% reduction in CO emissions and 35% reduction in total hydrocarbons in offending
vehicles after rechecking69. Currently, there are 456 inspection and maintenance (I&M) stations
across the island, and approximately 400,000 scooters were inspected in 1996. Remote sensing is
employed to measure CO and hydrocarbon emissions. 70
41
1.2.7.3 Future direction: zero-emission vehicles
In 1991, ITRI researchers estimated that by 1996, up to 50% of scooters would have to be
replaced by electric scooters to prevent continued degradation of air quality. This was in addition
to the adoption of second stage emission standards and gradual replacement of the existing fleet by
more advanced vehicles. No 50% replacement occurred, but a government policy was defined that
required 2% of the scooter fleet to be zero-emission scooters by the year 2000.71 This is almost as
aggressive as California’s clean air policy requiring zero emission automobiles, but whether either
will succeed is uncertain.
The objective appears to be to convert small engine (50 cc) scooters to electric power, while
keeping clean four-stroke engines for the larger (100 cc and up) scooters.72 By 1997, there were
approximately 300 electric scooters in use in Taiwan.73 There is currently a $5,000 New Taiwan
Dollar (USD $150) subsidy for each electric vehicle purchased.
This “Electric Motorcycle Development Action Plan” will be funded by the government at a cost of
NTD $3.8 billion (USD $115 million) from 1999 to 2002. This money is to go to research funding
and subsidizes for electric scooter purchases. Details of the plan are listed below.
42
Table 1.10 Electric Motorcycle Development Action Plan
Year
Number of electric
vehicles to be sold
1999
10,000
Notes
Republic of China EPA to select specially designated locations for
initial promotion
The Kwang Yang Motor Co. (Kymco) plans to begin mass production
in March, 1999
5% of annual motorcycle sales by manufacturers producing more than
50,000 motorcycles per year must meet special “low emission
motorcycle” standards (see Table 1.3)
2000
40,000
Electric motorcycle sales required to comprise 2% of all motorcycle
sales
2001
80,000
Electric motorcycle operating environment [recharging infrastructure,
etc.] to be gradually put in place; sales to increase
2002
150,000
50% of two-stroke motorcycle sales anticipated to be replaced by electric
motorcycle sales; four-stroke motorcycles will absorb the other half.
2003
200,000
Electric motorcycle technology to become mature; production of nickel
[metal] hydrogen batteries to begin
Introduction of fourth stage emissions standards; improvements in
battery-powered scooter technology to reduce price below that of (not
necessarily equivalent) four-stroke motorcycles.
2006
400,000
Continued growth of electric motorcycle sales; annual sales of electric
motorcycles to reach 40% of total motorcycle sales.
The description of this plan is paraphrased from an Engine, Fuel, and Emissions Engineering, Inc.
study and the March 1998 issue of the Taiwan EPA’s Environmental Policy Monthly.74,75 The
latter source writes that
Current trends indicate that by 2010 annual sales of motorcycles will reach 9
million units. It is estimated that electric motorcycles will make up one-third of
this total, or three million units sold. If this sales rate is achieved, the EPA has
calculated that carbon monoxide (CO) emissions can be reduced by 42,000
metric tons annually . . . Hydrocarbon and nitrogen oxide (NOX) emissions can
be reduced by 23,400 tons, and carbon dioxide (CO2) can be reduced by 62,800
tons annually. As for energy savings, each year 2.2 million megawatt hours can
be saved and off-peak electricity use rates can be raised.76
43
The “three million electric scooters” target seems extremely high, consider that only 400,000 are
expected to be sold by 2006. A TTVMA (Taiwan Transportation Vehicle Manufacturers’
Association) study was also not as optimistic, estimating that by 2010 only 150,000 electric
motorcycles will be produced, for an average unit price of $909 and a sales value of $136
million.77
For comparison, current electric scooters like the SWAP (Shang Wei Air Preserver) cost
approximately $2,00078, while ordinary two-stroke scooters cost on the order of $1,000.79 Scooters
currently have very low range and recharging is inconvenient due to the times involved and the fact
that not all scooter owners have access to an electric outlet (for example, they may not have
enclosed garages or parking off the street).
1.2.7.4 Research interest in fuel cell scooters
As of July 1999, several Taiwanese scooter manufacturers have explored the possibility of fuel cell
scooters with North American research groups, hydride supplies, and fuel cell companies. Projects
begun in the past two years include the following:
&
A Department of Energy - funded contract awarded in 1998 to Energy Conversion
Devices, a Michigan metal hydride manufacturer, to study hydrogen fuel for transportation
(especially scooters) in developing countries.80
&
A 1998 feasibility study of fuel-cell powered scooters performed by Sanyang Industry
Company, the Taiwan Institute of Economic Research, the Desert Research Institute, and
Texas A&M's Engineering Experiment Station. Partial funding supplied by the W. Alton
44
Jones foundation; one of the projects was to build a prototype scooter.81 Please see
Appendix G for more information on this prototype scooter.
&
A collaborative fuel cell scooter development project spearheaded by the Taiwan Institute
for Economic Research and including other Taiwan scooter concerns, announced in July
1999, and growing out of the previous study. The power system is to be based on a 3 kW
fuel cell stack developed by fuel cell scientist John Appleby of Texas A&M, and hydrogen
is expected to be produced by the China Petroleum Corporation, the Taiwan government’s
gasoline monopoly.82
Interest in fuel cell scooters has been growing rapidly in 1998 and 1999. The first step in
understanding the technological issues, then, is the next chapter which describes electric scooters,
and how they can be powered by either batteries or fuel cells.
1.2.8 Conclusion
Historically, two-stroke engines have been used for their high power density and low cost. Two
stroke scooters have become a major cause of concern in many Asian countries due primarily to
hydrocarbons short-circuiting through the chamber and escaping, unburned, in the exhaust.
Tightening government air pollution standards are squeezing out two-stroke engines in the policy
leader, Taiwan. Four-stroke engines are a cheap and effective replacement. However, an additional
benefit of several hundred, or even a thousand dollars could be realized by switching to electric
scooters with their zero tailpipe emissions.
45
The target vehicle in this study is thus an electric replacement for the small 5 kW two-stroke
scooter. A detailed comparison will be made between fuel cell scooters and battery-powered
scooters using both conventional and advanced technologies
References for Chapter 1
1. A. John Appleby and F. R. Foulkes. Fuel Cell Handbook (Van Nostrand Reinhold: 1989) pp.193-196.
The vehicle described was an Austin A-40 developed by Karl Kordesch at Union Carbide and run off
compressed hydrogen tanks. It was driven by Kordesch on public roads from 1971-1975.
2. Although relatively little work has been done, the interest is there from several parties as of September
1999. Vehicle manufacturers from Taiwan, which has more scooters than any other country in the world,
have explored the possibility of fuel cell scooters with several North American fuel cell companies. The
Desert Research Institute in Nevada had run a program to produce a prototype fuel cell scooter, while
Energy Conversion Devices, a Michigan metal hydride and battery manufacturer, has analyzed the market
possibilities for both combustion and fuel cell scooters run on hydrogen. Yamaha of Japan has studied the
prospects for fuel cell motorcycles. So far, no results have yet been published aside from this work.
3. Yoshihiro Nakazawa, Chiaki Kumagai, Mikio Kato. “Development of an electric scooter for practical
use” JSAE Review 15 (1994) pp. 373-377
4. Christopher S. Weaver, Lit-Mian Chan.“Motorcycle Emission Standards and Emission Control
Technology” Revised Final Report, submitted to The World Bank. (Engine, Fuel, and Emissions
Engineering, Inc.: August 1994), p.3
5. Xiannuan James Lin, Karen R. Polenske. “Energy Use and Air Pollution Impacts of China’s
Transportation Growth” in Energizing China: Reconciling Environmental Protection and Economic
Growth. Editors Michael B. McElroy, Chris P. Nielsen, Peter Lydon. Harvard University Press, 1998. p.
211
6. Dr. Pradeep Monga. United Nations Development Programme. “Discussion on Better Environment &
Natural Resource Management”. http://rrmeet.undp.org.in/_disc2/00000008.htm Accessed July 11, 1999
7. Taiwan Transportation Vehicle Manufacturers’ Association web page. “Taiwan Transportation Vehicle
Manufacturers’ Association - IndustryUpdates”
http://ttvma.asiansources.com/INDUSTRY/IUPDAUTO.HTM. Accessed May 16, 1999
8. Industrial Technology Information Service, Industrial Technology Research Institute.“ROC: Republic
of Computers -- Taiwan” 1997 Taiwan Industrial Outlook http://www.itis.itri.org.tw/eng/rep9712.html
Accessed March 26, 1999. Updated January 20, 1998.
9. Encyclopædia Britannica Online. “Encyclopaedia Britannica: Taiwan”
http://www.eb.com:180/bol/topic?tmap_id=205530000&tmap_typ=gd. Accessed April 1999
46
10. Government Information Office of Taiwan web page. The Republic of China Yearbook 1997.
“Chapter 2 - People” http://www.gio.gov.tw/info/yb97/html/ch2.htm. Accessed March 9, 1999
11. Directorate-General of Budget, Accounting and Statistics, Executive Yuan, the Republic of China.
Monthly Bulletin of Statistics of the Republic of China, February 1999. “Table K-5 Number of Registered
Motor Vehicles” http://www.dgbasey.gov.tw/dgbas03/english/bulletin/k5.htm. Accessed March 25, 1999
12. Laurie Underwood. “Airborne Menace” in Free China Review September 1, 1996
13. Directorate-General of Budget, Accounting and Statistics.
14. Directorate-General of Budget, Accounting and Statistics.
15. Department of Budget, Accounting, and Statistics - Taipei City Government. Summary of Statistics
Taipei City for 1998. “Vehicle Registration”. http://www.dbas.taipei.gov.tw/stat/summeng/sume19.htm
Accessed March 25, 1999
16. Hsiung-Wen Chen, Hui-Chuan Hsaio, Sheng-Jonh Wu. Environmental Protection Agency, Republic
of China (EPA ROC). “Current Situation and Prospects of Motorcycle Pollution Control in Taiwan,
Republic of China” 1992. Society of Automotive Engineers (SAE) 922176.
17. T. J. Price and S. D. Probert. “Taiwan’s Energy and Environmental Policies: Past, Present and Future”
Applied Energy 50 (1) (Elsevier:1995), p. 56
18. United States Energy Information Administration (EIA). “Taiwan”
http://www.eia.doe.gov/emeu/cabs/taiwan.html. Accessed May 8, 1999. Page updated December 1998.
19. United States Energy Information Administration (EIA). “Taiwan”
20. United States Energy Information Administration (EIA). “Taiwan”
21. Willard W. Pulkrabek. Engineering Fundamentals of the Internal Combustion Engine. Prentice Hall
(Upper Saddle River: 1997), p. 26
22. Pulkrabek, p. 28
23. Weaver and Chan, Appendix p. 16
24. Mark L. Poulton. Alternative Engines for Road Vehicles (Computational Mechanics Publications:
1994), p.54
25. Toshiharu Sawada, Minoru Wada, Masanori Noguchi, Buhei Kobayashi. “Development of a Low
Emission Two-Stroke Cycle Engine”. Society of Automotive Engineers (SAE) 980761. 1998. p. 81
26. Todd D. Fansler, Donald T. French, Michael C. Drak. General Motors Global Research and
Development Operations. “Individual-Cycle Measurements of Exhaust-Hydrocarbon Mass from a DirectInjection Two-Stroke Engine”. SAE T980758. 1998. p. 50
27. Sawada, Wada, Noguchi, Kobayashi p. 89
28. Weaver and Chan, Appendix p. 7
47
29. Mark L. Poulton. Alternative Engines for Road Vehicles (Computational Mechanics Publications:
1994), p. 50
30. Note that methane is often excluded from the category of hydrocarbon air pollutants because it is not
reactive in forming secondary air pollutants; the category then becomes non-methane hydrocarbons
(NMHCs).
31. TVS Suzuki Limited. TVS Scooty KS specifications. http://www.tvssuzuki.com/html/sco_ks.html.
Accessed May 12, 1999
32. OECD. Motor Vehicle Pollution: Reduction strategies beyond 2010 OECD, 1995. p. 57
33. Weaver and Chan, p. 47
34. Republic of China EPA document. “Emission Standards of Air Pollutants for Transportation
Vehicles” (no author or date)
35. Weaver and Chan, p. 62
36. Chang-Chuan Chan, Chiu-Kuei Nien, Cheng-Yuan Tsai and Guor-Rong Her, “Comparison of TailPipe Emissions from Motorcycles and Passenger Cars” in J. Air & Waste Management Association 45 (2)
February 1995, pp. 116-124
37. H-W Chen, S. Lu, Y-P Yu, Y-W Lin. Environmental Protection Agency ROC. “The study of driving
cyclye [sic] and emission factors estimation of vehicle in Taipei Metropolitan ROC”. Presented at
International Symposium on Automotive Technology & Automation (ISATA) 1997, paper number
97EN026
38. Government Information Office of Taiwan web page. The Republic of China Yearbook 1997.
“Chapter 13 - Environmental Protection” http://www.gio.gov.tw/info/yb97/html/ch13.htm. Accessed
March 9, 1999
39. ROC EPA web page. “Air Pollution Control Approach in Taiwan”.
http://www.epa.gov.tw/english/Offices/taiwan.htm. Accessed May 16, 1999 .
40. P. H. Jet Shu, Wei-Li Chiang, Bing-Ming Lin, Ming-Chou Cheng. Mechanical Industry Research
Laboratories, Industrial Technology Research. “The Development of the Electric Propulsion System for
the Zero Emission Scooter in Taiwan” SAE 972107, JSAE 9734403. 1997
41. Shu, Chiang, Lin, Cheng
42. ROC EPA web page. “Air Quality Monitoring Network In Taiwan”.
http://www.epa.gov.tw/english/Offices/aqmnit.htm Accessed May 16, 1999.
43. Shu-Hwei Fang, Hsiung-Wen Chen. “Air Quality and Pollution Control in Taiwan”. Atmospheric
Environment 30 (5) pp. 735-741, Elsevier 1996
44. ROC EPA web site. “Air Pollution Control Approach in Taiwan”
45. ROC EPA web page. “Air Pollution Control Approach in Taiwan”
46. EPA Office of Air Quality Planning And Standards. “AIRSData - Monitor PSI Report”
http://www.epa.gov/airprogm/airs/data/monpsi.htm Accessed June 23, 1999
48
47. ROC EPA web page. “Air Pollution Control Approach in Taiwan”
48. John B. Heywood. Internal Combustion Engine Fundamentals. McGraw-Hill 1988. p. 837
49. Marco Nuti and Roberto Pardini. Piaggio V.E. S.p.A. “Twenty Years of Piaggio Direct Injection
Research to Mass Produced Solution for Small 2T SI Engines” SAE 980760. 1998
50. Weaver and Chan, Appendix p. 17
51. Marco Nuti, Roberto Pardini, David Caponi. Piaggio V.E. S.p.A. “FAST Injection System: PIAGGIO
Solution for ULEV 2T SI Engines” SAE 970362. 1997.
52. Mark L. Poulton, p. 53
53. Wei-Li Chiang, Mechanical Industry Research Laboratories, ITRI. Personal communication, July 16
1999
54. $150 estimate of cost from Dr. Philip G. Felton, Princeton University Department of Mechanical and
Aerospace Engineering, personal communication July 26 1999
55. Dr. Felton from Princeton University’s Mechanical and Aerospace Engineering department estimated
catalytic converter costs at approximately $100 and engine costs at $150, both manufacturing prices.
Personal communication, July 26 1999.
56. Nuti and Pardini, p. 73
57. Chan, Nien, Tsai, Her.
58. Nuti, Pardini, Caponi
59. Chien-Tung Liu, C. C. Kuo, Jyh-Sheng Pan, Bing-Ming Lin. “Development of electric motor cycle
technologies in Taiwan” J. Power Sources 48 (1994) p. 244
60. Joseph V. Spadaro and Ari Rabl. Ecole des Mines de Paris, Centre d’Energetique, Paris. “Social Costs
and Environmental Burdens of Transport: An Analysis using Two Case Studies in France”. October 1998
61. Heywood, p. 626
62. OECD web site. February 1999. “GDP per capita based on exchange rates and on PPP”,
http://www.oecd.org/std/gdpperca.htm. Accessed March 16 1999.
63. Encyclopædia Britannica Online. “Encyclopaedia Britannica: Taiwan”
64. Anna Alberini, Maureen Cropper et al. “Valuing Health Effects of Air Pollution in Developing
Countries: The Case of Taiwan” J. Environmental Economics and Management 34, 1997. p. 123
65. Weaver and Chan, p. 69
66. Weaver and Chan, p. 45
67. Chen, Hsaio, Wu.
49
68. In much the same way, a Taiwan government lottery based on numbers automatically printed on cash
register receipts is used to encourage customers to get receipts from legitimate businesses rather than deal
with the black market.
69. Republic of China EPA, “Periodic Motorcycle Inspections fully implemented” Environmental Policy
Monthly, Volume 1, Issue 4. November 1997
70. ROC EPA web page, “Air Pollution Control Approach in Taiwan”
71. ROC EPA, “Electric Motorcycles Targeted as Key Industry for Development” Environmental Policy
Monthly, Volume 1 Issue 9. March 1998. http://www.epa.gov.tw/english/Epm/issue9803.htm. Accessed
May 7, 1999
72. Chen, Hsaio, Wu
73. ROC EPA web page. “Air Pollution Control Approach in Taiwan”
74. Weaver and Chan, p. 42
75. ROC EPA, “Electric Motorcycles Targeted as Key Industry for Development”
76. ROC EPA, “Electric Motorcycles Targeted as Key Industry for Development”
77. Industrial Technology Information Service, Industrial Technology Research Institute.“ROC: Republic
of Computers -- Taiwan” 1997 Taiwan Industrial Outlook http://www.itis.itri.org.tw/eng/rep9712.html
Accessed March 26, 1999. Updated January 20, 1998.
78. T. Chi Kung, Shang Wei EV Tech Inc. Personal communication, May 3 1999.
79. Chien-Liang Lin, Energy and Resources Lab. Industrial Technology Research Institute. Personal
communication, April 29 1999.
80. Energy Conversion Devices, Inc. press release. “ECD AWARDED $3.5 MILLION IN NEW DOE
COST-SHARED CONTRACTS” http://www.ovonic.com/news/Oct8_1998.html Published October 8,
1998.
81. “Desert Research Institute Designs Prototype Fuel Cell Scooter for Taiwan Manufacturer” Hydrogen
and Fuel Cell Letter, October 1998.
82. “Taiwan Group Launches PEM Scooter Project With Help from Texas A&M, DRI” Hydrogen and
Fuel Cell Letter, July 1999. http://mhv.net/~hfcletter/letter/july99/JulyFeature.html
50
~
Chapter Two
Electric Vehicles
~
51
The purpose of this chapter is to discuss various electric drive and battery possibilities and obtain
data for these options. Electric scooters require drive systems (motor, controller, transmission) and
power sources (batteries or fuel cells). Since a fuel cell scooter is essentially a battery-powered
scooter with battery replaced by fuel cell plus hydrogen storage, the basic electric scooter is
described first.
Components are chosen for the electric scooter on the basis of technical qualifications and
economic considerations. The resulting electric battery-powered vehicle is used as a base platform
to develop the fuel cell scooter design.
The chapter also includes a section on high-power batteries that would supply transient bursts of
peaking power for hybrid scooters.
It should be noted that battery-powered scooters are currently commercially available in North
America, Asia, and Europe, although they have met with only limited success.
52
2.1 Drive systems
The speed of a vehicle is determined by the output of its engine. This is more accurately
characterized in terms of the motor’s torque and angular velocity, but often measured in terms of
the product: power.
Torque is measured in lbf-in, ft-lb, or N-m, power in hp or kW, force in lbf or N, and motor speed
in rpm or rad/s. Essentially, an internal combustion engine or electric motor generates rotation at a
specific speed and torque; this rotation drives a transmission of which changes the speed of the
rotation through gearing, and transmits the force to the slower-turning wheel axle.
A transmission decreases engine speed by the same factor it increases engine torque. Its influence is
measured by the gear ratio, the ratio of engine revolution speed to driveshaft revolution speed; the
slower the output speed, the greater the output torque. Car linear speed v is the wheel
circumference (2%r) times the engine speed 7divided by the gear ratio u:
v = 2% r 7/u
Typical passenger cars use engine speeds between 2000 and 3500 rpm. Transmissions allow the
construction of a smaller motor and allow operation closer to the motor’s optimum efficiency at all
speeds by changing the gear ratio as a function of output desired. As the car accelerates, the
maximum desirable engine speed is reached and the gear ratio is reduced to the next level to stay
within an efficient zone; engine speed is allowed to drop again, but with the reduction in gear ratio,
velocity increases continuously (at the expense of torque). Essentially, torque translates to force at
53
the wheel, and large torques are required for steep hill climbing and fast acceleration.
Scooters are single-wheel drive devices; the engine powers the rear wheel, while the unpowered
front wheel is used for steering.
2.1.1 Electric drive systems: introduction
An electric drive system replaces the internal combustion engine and assorted transmission systems
with an electric system. The various differences are listed below.
Table 2.1. Comparison of power systems
Internal combustion system
Electric battery system
Electric fuel cell system
•
•
•
•
•
•
•
•
•
•
•
•
engine including
cylinders, air intake
fuel tank, carburetor, air
filter
pump for lubricating oil,
oil tank
exhaust system
transmission and chain
starter battery
battery
electric motor(s)
motor controller
transmission and
chain
•
•
•
•
•
fuel cell stack
fuel cell subsystems including
cooling system, air intake,
hydrogen intake, humidification
system if any
hydrogen storage device
electric motor(s)
motor controller
transmission and chain
battery for startup
A small internal combustion scooter engine of 50-80 cc weighs about 32 kg and occupies under 50
L of space.1 Note that current 50 cc scooters require a 12 V starting battery on the order of 1.3 kg
and 0.7 L.2 The cost is on the order of $20 retail. In a fuel cell scooter this extra battery, or
something similar, would likely be necessary to start the intake blower, open valves, activate
electronic controls, and perhaps even warm up the fuel cell. It would not be necessary at all in a
battery-powered scooter.
54
Three different groupings are defined here for the advanced scooters presented here: “fuel cell
stack”, “power system”, and “drive system”. “Fuel cell stack” refers to the series-connected
electrochemical cells that make up the core power source of the system, and includes manifolding
and a plastic insulating housing around the stack. “Power system” includes not only the stack, but
also subsystems like the blower to supply air to the fuel cell, the radiator to cool down the stack,
and the coolant pump. The most inclusive term, “drive system,” includes the power system and
peaking power battery (if any), hydrogen fuel storage, and the electric motor and controller. For an
equivalent battery-powered scooter, “drive system” includes the storage batteries, any peaking
power batteries, and the motor and controller pair.
2.1.2 Electric motor theory
Electrical motors operate on the basic principle that a current-carrying wire in a magnetic field will
experience a force. The magnetic field can be generated by permanent magnets or by a current in
an electromagnet. The stator is stationary and produces the magnetic flux, while the rotating
armature or rotor contains the coils that carry the armature current. In general, motor speed is
controlled by increasing the armature voltage, while torque is controlled by increasing the current
flowing through the armature.
In a combustion engine, the explosions of the air/fuel mix directly produce rotation with a fixed
velocity-torque relation. More flexibility can be achieved in an electric motor, where the ratio
between torque and speed can be controlled independently and electronically within the
motor/controller. For example, in a pulse-width-modulated system the frequency of rotation of the
magnetic field governs the speed output, while the phase difference between the rotor and stator
55
fields determines torque. Transmissions are often not be necessary at all; where used, they offer
optimum efficiency (since the output mechanical power can be remapped by the transmission to the
higher efficiency portions of the electric motor output) for both driving and regenerative braking.3
In this study, no transmission was assumed - only a fixed final gearing between the motor output
and the wheel.
2.1.2.1. DC motors
DC motors employ a fixed current that causes the rotor to “want” to turn to line up with the poles
in the stator. However, the current in the stator is commutated, often by a split-ring brush system,
so that the direction of the current in the poles switches as the rotor passes by. This ensures that the
rotor stays in continual motion. Multiple sets of poles are used to smooth out the rotation. In
general, controllers are cheaper than for AC motors; on the other hand, the motors themselves tend
to be bulkier and heavier and more expensive.4
In the basic field-wound motor described above, the stator field is provided by an electromagnet.
Speed and torque are controlled by changing the current in the stator field and/or rotor windings.
In a variant, permanent magnet motors use permanent magnets rather than electromagnetic
windings in the stator. The presence of brushes means relatively high maintenance, but these
motors tend to have higher efficiencies than other DC motors due to the lack of stator field
windings.5 They have a narrow peak efficiency, so transmissions are required.
With brushless DC motors, it is the rotor that is a permanent magnet. The stator electromagnet
56
current is switched on and off at the correct frequency (instead of commutated to reverse current
direction), creating a rotating magnetic field in the stator and causing rotation in the rotor.
Changing the speed of the rotating magnetic field effects rotor speed control. Torque is controlled
here by varying the magnitude of the magnetic flux of the stator. (The flux, in turn, is controlled by
changing stator current). These motors are relatively efficient due to the absence of brushes and
can achieve average efficiencies of about 84% for both motor and controller together.6 Control,
however, is more complex.7
2.1.2.2 AC motors
Alternating current motors are inexpensive, simple, and reliable. They operate by take advantage of
the changing phase of the stator current. AC motor control is expensive, however, and implemented
by changing input frequency (“dragging” the actual rotor frequency ahead to match the input
frequency) or by changing voltage. Also, an inverter is needed to produce AC from the fuel cell or
battery’s DC output.
Induction (asynchronous) motors apply alternating current to the stator winding, creating a
rotating magnetic field. There is no current in the rotor windings; the stator induces a current in the
rotor which creates torque. Efficiencies are greater than those of DC motors, on the order of 85%91%.8
An AC synchronous motor is identical to an asynchronous motor, but with a magnet (permanent or
electromagnet) in the rotor. In other words, the permanent magnet AC synchronous motor is
identical to a brushless DC motor except that the frequency of the supplied alternating current
57
controls the rotation speed of the magnetic field, not the on-off switching of a pulsed DC current.
2.1.2.3 Hub motors
Hub motors, included as a separate section, are an interesting development which could offer
benefits for electric vehicles. These motors have stators fixed at the axle, with the permanent
magnet rotor embedded in the wheel. By directly driving the wheel, they eliminate the inefficiency
of a transmission and chain connecting the motor to the axle. Other advantages include higher
efficiencies, less space, and often easier servicing.
The more traditional “exterior rotor” design has the rotor in a hollow cylinder shape and spinning
around a stator axle. The rotor consists of permanent magnets, and this is a “radial gap” motor
because the air gap between the stator and rotor extends in the radial direction. In a slightly
different option known as the pancake or disc-type brushless motor, the rotor is not around the
stator, but rather a flat disc of permanent magnets sitting on top of another flat disc which contains
the stator coils. These are “axial gap” motors because the space between stator and rotor is in the
direction of the axis. The stator can be a Mylar plate, stack of silicon steel plates with wound coils,
or even a printed circuit for small, flat applications.
58
Figure 2.1 Axial-gap pancake motor
Source: Hendershot Jr. and Miller, p. 2-11.9
Pulse width modulated (PWM) current is used to supply current to the stator, so in essence the
system is a DC brushless motor. Hub motors must run at relatively low speed – equal to the actual
rotation of wheel if there is no final gearing stage. The benefit is about a 10% increase in efficiency
due to the lack of transmission.
2.1.3 Converters and controllers
The controller connects the power source - fuel cell or battery - to the actual motor. It controls
speed and direction, and optimizes energy conversion. While batteries produce fairly constant
voltages which decrease as they are used up, the voltage output by fuel cells varies as a function of
power. Some controllers require a DC-to-DC converter to step down this changeable voltage to the
motor’s expected constant operating voltage, but other controllers incorporate a DC-to-DC
converter and can accept a varying voltage. In either case, DC-to-DC conversion losses are
minimized if the fuel cell output voltage is near the operating voltage. Converter efficiencies are
typically greater than 90%.
59
The controller varies the speed and torque of the motor. Today voltage control is almost always
achieved by “chopping” the source current - the voltage is switched on and off, with the ratio of onto-off determining the average voltage. The number of constant-width “on” pulses per unit time
can be varied, or the width (duration) of the pulses can be varied. Chopping is performed by power
electronics circuitry - diodes and thyristors and silicon control rectifiers (SCRs)
Controllers also effect regenerative braking, which is the process of driving the motor as a
generator to recharge the batteries. In practice, about a third of total energy is discarded in ordinary
friction braking (the other two-thirds is lost to rolling resistance and drag and auxiliary power).
Due to inefficiencies in the regeneration process, only about 70% of this third can be recovered in
regenerative braking.10 (In the Taipei driving cycle studied later, approximately 20% rather than
one third of the total mechanical energy output is lost as friction in braking).
2.1.4 Choice
Standard DC electric motors run at 24 or 48 V; The Taiwan scooter industry appears to be moving
towards a de facto standard of 48 V and this was the chosen operating point.11 For the fuel cell
design developed for this thesis and described in later sections, voltage varies from 56 V at
minimum power to 34 V at maximum power (5.6 kW). Current varies from zero to a maximum of
163 A over this range.
Most electric scooter motors surveyed use DC motors, the majority brushless rather than brushed
(see Appendix A). For the reason of good DC efficiency and the lack of need for an AC inverter,
60
this is the type chosen here. Two systems are examined: a New Generation Motors (NGM) hub
motor, and a Unique Mobility (UQM) axial gap DC brushless motor.
Table 2.2 Motor specifications: UQM brushless and NGM hub motors
spec
UQM motor
NGM motor
Model
SR121/1.5 L
SC-M150-04
3.6 kW
2.5 kW
115 N•m at 2000 rpm
105 N•m at up to 300 rpm
Speed
0-800 rpm
0-1300 rpm
Controller voltage
40-60 VDC
30-68 VDC
95 A
260 A
Efficiency
up to 87%
up to 95%
Motor cost
$250 (estimated)
$7,000 (retail)
Motor diameter
20 cm
31.5 cm
Motor volume
5.0 L
unk.
Motor weight
11.4 kg
20 kg
Controller volume
4.1 L
7L
Controller weight
4.1 kg
5 kg
$300 (projected)
$3,000 (current)
Maximum power output
Peak torque (geared)
Maximum controller current
Controller cost
The Unique Mobility SR121/1.5L brushless, permanent magnet design has a maximum output of
3.56 kW. The CD 05-100A controller system includes a battery charger that can rectify AC
voltage to battery charging voltage and allows regeneration. A final gearing ratio of 7.24 increases
torque and reduces speed.12 Current prices(May 1999) were estimated by Unique Mobility
representatives at $250 for the motor plus $300 for the controller if the motor were to be mass
produced immediately.13 In comparison, a Lynch Motors brushed DC motor with 3 kW continuous
output (6 kW peak output), weighing approximately 9 kg, has a retail price of $1,000, suggesting
61
that current prices are higher than Unique Mobility’s estimates.
Hub motors offer higher efficiency, but can cost thousands of dollars due almost entirely to their
relative newness and lack of development. For example, the New Generation Motors SC-M150-04
motor described above costs almost $7,000, plus $3,000 for the controller, and is used in solar cars
- a very small market.14
The arrival of cheap electric bicycles run on hub motors and made in the People’s Republic of
China promise to reduce hub motor prices, though, and the New Generation Motors president
stated a high-volume price target of $500 in the future for scooter-sized hub motors.15 For now, the
UQM motor was chosen due to the more established nature of its technology. Hub motors will play
a greater role in the future for electric scooters. A 77% efficiency was chosen for the electric
drivetrain system, consisting of the DC motor efficiency, fuel cell DC-to-DC conversion, and gear
chain / transmission losses. This is at the low end of the motor map, and performance might in
reality be slightly greater. The figure was based on previous research in electric vehicles. 16 The
variation in efficiency over a DC motor “efficiency map” (plot of iso-efficiency contours on a
torque/speed graph) is only 3% so a single value is justified.
The following curve gives an example of how maximum torque decreases as a function of speed;
the space of possible torque/speed combinations lies under this curve. For the most part, the curve
follows a hyperbola since the product of torque and angular speed is equal to a fixed maximum
power. However, the torque is capped at low speeds by the maximum current the electric motor
can handle.
62
Figure 2.2 Typical torque vs. rpm curve for DC motor
120
torque (N-m)
100
80
60
40
20
0
0
100 200 300 400 500 600 700 800
angular speed (rpm)
The data presented is from a Unique Mobility data sheet for the
SR121/1.5 L brushless motor.17
2.2. Chemical batteries
Rechargeable chemical batteries are the traditional option for electric vehicles. They tend to be
heavy and expensive to replace over their limited lifetimes. In this section, the theory behind battery
operation is laid out with some discussion of various battery energy storage options for electric
vehicles. A final section deals with the use of specialized high-power batteries to provide surge
peaking power during moments of high energy demand, and thereby allow design with a smaller
primary power source.
63
2.2.1. Theory
A battery pairs reduction and oxidation half-reactions to generate electricity. At the anode, one
substance is oxidized. The electrons flow through the external circuit (power load) and arrive at the
cathode, where a different substance is reduced. Electrochemical equilibrium is maintained by
cations in the solution flowing across an ion bridge.
Batteries are divided into primary and secondary cells. Primary cells are those that are used once
and cannot be restored by reversing the current flow, because the half-reactions are irreversible.
Voltage decreases over time as the reactants are depleted and the concentrations decrease, and
eventually the cell becomes useless and must be disposed of, ideally after the chemicals inside are
recycled .
Secondary (sometimes termed “storage”) batteries can recover the original reactants by reversing
the current flow. For example, in the common lead-acid secondary battery, the following halfreactions take place: at the anode, lead metal is oxidized to lead sulfate (PbSO4); at the cathode,
lead oxide (PbO2) is reduced to PbSO4. The electrolyte is a sulfuric acid (H2SO4) solution.
The overall reaction is:
Pb(s) + PbO2(s) + 4 H+(aq) + 2 SO4 2- (aq) :2 PbSO4(s) + 2 H2O
64
The reversible electrochemical potential Ero in this case is approximately 2.04 V, and driving the
cell in reverse (i.e. as an electrolytic reaction) regenerates the lead metal and lead oxide. However,
there is a ceiling to the number of times any battery can be “cycled” in this way: when recharged
the metals tend to precipitate in low-energy configurations like metallic needles or dendrites that
eventually grow close to each other, and internal short circuits make the battery useless. After this
point, reprocessing is needed if the chemicals inside are to be reused. This accounts for the limited
lifetime of rechargeable batteries.
2.2.2. Technology
2.2.2.1 Existing scooter battery systems
As an example of current electric vehicle battery technology, the prototype Taiwan ZES-2000
battery-powered electric scooter developed by ITRI runs on a 24 VDC power system with two
configurations: sealed lead-acid batteries, or the more advanced nickel metal-hydride (NiMH)
batteries. Each configuration is designed to store the same amount of energy, but the more
expensive NiMH batteries do so in about 60% of the lead-acid batteries’ weight.
65
Table 2.3 ZES-2000 electric scooter performance
sealed
lead-acid
nickel
metal-hydride
1.34 kWh
1.34 kWh
Total weight
44.0 kg
26.1 kg
Total volume
14.5 L
12.8 L
Specific energy density (Wh/kg)
31
51
Volumetric energy density (Wh/L)
92
105
Range at 30 km/hr
65 km
78 km
Range under ECE 40 driving
cycle
35 km
46 km
Total stored energy (output)
Data from the ITRI ZES 2000 project18
The NiMH batteries a much smaller package but, as discussed in more detail later, are very
expensive: $900 or more. Due to the lower weight of the NiMH vehicle, it uses less energy in
driving and obtains higher fuel economy. This accounts for the improved range for the same energy
storage.
Note that current battery specific energy is very low; to compare, gasoline has an energy density on
the order of 920 Wh/L (although conversion to propulsion energy is on the order of 15%, more
than five times worse than the round-trip efficiency of battery-powered scooters at about 80%).
2.2.2.2. Technology predictions
The United States Advanced Battery Consortium (USABC) was formed by Chrysler, Ford, and
GM in 1991 to accelerate development of electric vehicle batteries. In 1992, goals were set for
“mid-term” and “long-term” battery performance. The table of goals is reproduced below.
66
Table 2.4 Battery goals for various time frames
1992
leadacid
Mid-term
prediction
Long-term
prediction
Specific power (W/kg)
80% DoD, 30 seconds
67-138
150
400
Energy density (Wh/L)
C/3 discharge rate
50-82
135
300
Specific energy (Wh/kg)
C/3 discharge rate
18-56
80
200
2-3
5
10
Cycle life at 80% DoD
450-1000
600
1000
mass-produced target
cost ($/kWh) set by USABC
$70- $100
100 mpge
fuel efficiency
acceleration
0-30 m in less than 5 seconds
speed on 15( slope
10 km/h
speed on 12( slope
18 km/h
maximum speed
60 km/h
maximum curb weight
140 kg
The slope climbing, acceleration, and maximum speed requirements are based on a Taiwan
research lab electric scooter proposal, which in turn was based on surveys of scooter users.8 (As a
comparison point for the speed on inclines, San Francisco’s Lombard Street wends up a hill with a
40(slope, although the twisting road itself is limited to 21().
A broader list of electric scooters that are being developed now is included in Appendix A.
159
4.2 Vehicle modeling
Essentially, the purpose of vehicle modeling is to convert input parameters (performance
measurements like desired range, driving cycle that must be sustained, and types of power and
storage components) into the output parameters of curb weight, size of engine required, heating
system design, cost, and convenience. The process is iterative; for example, the size of the cooling
system is a function of the average vehicle power, but larger cooling system itself requires more
power to drive the fans and pumps, increasing the power load, which necessitates more cooling.
4.2.1 Physical model
To properly simulate the performance of a fuel cell scooter, a computer model was created based
on the physical properties of the scooter. This model calculates the instantaneous power required
from a scooter’s engine as it travels through various driving patterns, and derives various
numerical performance characteristics: fuel consumed per kilometer of travel, maximum power
during the driving cycle, average power during the driving cycle, amount of energy recovered in
regenerative braking, and overall hydrogen-to-mechanical-work conversion efficiency.
For the MATLAB program listing, please see Appendix F.
This power is calculated as the dot product of the current velocity of the vehicle and the various
forces acting upon it, divided by the motor and controller efficiency, plus the auxiliary power
demanded by various lighting and control systems, plus “parasitic” power required by the fuel cell
blowers and coolant pumps. The road is assumed to be level. There are several different physical
160
forces to consider: air resistance (drag); the rolling resistance of the wheels; the force of gravity,
which is not necessarily perpendicular to the velocity if the vehicle is traveling uphill or downhill;
and the normal force of the ground acting upon the vehicle. These forces must sum to zero if the
vehicle is to be held at a constant velocity, or to a net forward acceleration times mass if the vehicle
is to accelerate.
Figure 4.1 Free-body diagram of scooter
In the computer model used to simulate vehicle performance, the various power demands are
summed to a total mechanical power Pwheels demanded “at the wheels” by the motion of the vehicle.
Pwheels = ( ma v) +( mgvsin ) + ( mgvCRR cos ) + ( ½ 'ai rCD AF v3 )
The variables are listed below.
161
m = total mass of vehicle, passengers and cargo
= angle of slope
a = acceleration of vehicle
v = velocity of vehicle
CRR = coefficient of tire rolling resistance.
'air = density of air, approximately 1.23 kg•m-3
CD = drag coefficient
AF = frontal area
The terms are described one at a time:
Acceleration term. If the acceleration is negative - that is, the vehicle is decelerating - the first term
can be negative. If the overall expression for Pwheels is still positive even though the first term is
negative, it means that energy must still be supplied by the power source to the wheels to maintain
the desired deceleration rate, because the drag and rolling resistances are so large. If Pwheels is
negative, then the motor can be driven as a generator to regenerate some of the energy expended. A
battery capable of reabsorbing this energy is needed, and less than 70% of the kinetic energy is
recoverable. This figure is reduced if rapid deceleration is required, because the battery can only
charge up at a certain maximum rate.
Slope term. The second term is the force of gravity resolved opposite to the direction of motion.
Rolling resistance term. The coefficient of rolling resistance is a function of tire pressure and
deformation, and is the ratio of rolling resistance force to the load on the tires; it is fairly constant
for a given tire.9 A perfectly rigid wheel on a rigid, flat surface would have no rolling resistance,
162
but minor deformations in the wheel and properties of the road cause deviation from ideal geometry
and thus irreversible losses.
Aerodynamic drag term. The drag coefficient CD is a dimensionless constant that attempts to
capture, in one term, an object's resistance to flow. CD can vary from as high as 1.2 for a bicycle
with erect rider to 0.47 for a sphere to 0.20 for a very aerodynamically-styled modern
automobile.10 Although the equation used to determine the drag power is a simplification, it avoids
complex air flow simulation while preserving the general behaviour of the drag force with respect
to velocity. The frontal area used here was measured for the scooter by projecting a bright light
parallel to the front of the scooter and then measuring the area of the shadow on a wall behind.11
Typical values are listed in Table 4.3.
The inefficiencies in the system are applied afterwards to determine how much power must be put
out by the power source:
Pauxiliary = power needed by auxiliary systems - headlights, signal lights, dashboard, etc.
drivetrain = efficiency of the electric motor and controller subsystem - 77%
Pparasitics = parasitic power needed by fuel cell system - blowers, fans, etc.
The parasitic and auxiliary powers are electric power requirements so they do not go through the
77% efficiency loss. A more sophisticated model would not use a single value of drivetrain but rather
employ an efficiency map to determine electric motor efficiency as a function of wheel speed and
torque.
163
Factors not accounted for in this model include: turning, where the velocity is not parallel to the
acceleration / deceleration direction; wind blowing at an angle to the direction of motion;
resistances in other parts of the scooter. Friction in the transmission and similar losses are assumed
to be captured by the drivetrain efficiency of 77% above, discussed in section 2.1.4.
4.2.2 Modeling parameter selection
Vehicle modeling parameters for a typical scooter are listed below with data for other vehicles for
comparison. Although most two-stroke scooters weigh about 80 kg, the presence of lead-acid
batteries and/or fuel cell plus hydrogen storage brings the mass of an electric scooter up to
approximately 130 kg as in the case of the Honda CUV-ES with NiCd batteries.
Table 4.3. Typical modeling parameters
CRR
CD
AF (m2)
curb
weight (kg)
auxiliary
power (W)
Electric Scooter
0.014
0.9
0.6
130
60
Roadster Bicycle
0.008
1.2
0.5
10
0
unknown
0.6
0.8
300
unknown
Ford AIV Sable
0.0092
0.33
2.13
1291
500
PNGV Automobile
0.007
0.20
2.0
920
400
Vehicle
Motorcycle
The PNGV Automobile properties are targets set out by the
Partnership for a New Generation of Vehicles,12 except for the rolling
resistance and auxiliary power which were obtained from separate
studies.13,14 The Ford AIV Sable is a light weight “aluminum intensive
vehicle”, a modern mid-sized sedan.15 Motorcycle data was obtained
for some of the parameters.16 Finally, the bicycle data is for a“roadster”
upright model.10
The scooter coefficient of tire rolling resistance was estimated to be 0.014, based on measurements
done at the Desert Research Institute17, while the drag coefficient and frontal area were obtained
164
from researchers at the ITRI Mechanical Industry Research Laboratory.18 A slight mass
dependence (less than 6%) in the drag coefficient reported by the MIRL researchers was ignored,
and the largest measurement taken; the velocity dependence of the rolling resistance coefficient was
likewise neglected. Note also that in scooters and bicycles the product of drag coefficient and
frontal area can vary dramatically, depending on how the driver sits on the scooter. The values
chosen were assumed to be for the rider in a typical position.
The curb weight was set at 130 kg, which was 30 kg more than the ZES-2000 but equal to that of
the CUV-ES electric scooter. This choice was made to ensure that performance requirements
would be met even if the lower ZES-2000 weight could not be reached, whether due to the extra
structural weight needed to support the heavier power system, or due to the weight of the
components themselves. The driver weight is defined as 75 kg.
Auxiliary power: in a typical scooter, the head lights, tail lights, and dashboard total about 50 W;
assuming that these lights are always on, and that the 26.4 W turn lights are on 30% of the time,
yields an average load estimate of 60 W.19
4.2.3 Relative importance of various factors
Now that the parameters are defined, power requirements are calculated for (i) a scooter traveling
at constant velocity at various slopes, and (ii) a scooter traveling with constant velocity at various
speeds, and finally (iii) power required for various accelerations starting from 30 km/h. The total
power shown below is the electric output from the power source including auxiliary power, but not
subsystem parasitic loads (blowers, pumps, etc.) which are calculated later.
165
Figure 4.2. Cruising power required at various speeds.
3000
2500
power (W)
2000
1500
615 W at
30 km/h
1000
500
0
0
5 10 15 20 25 30 35 40 45 50 55 60
speed (km/h)
auxiliary power
rolling resistance
aerodynamic drag
Figure 4.3. Power required to climb various slopes at 15 km/h
4000
power (W)
3000
2000
1000
0
0
2
4
6
8
10
12
14
16
18
slope (degrees)
aerodynamic drag
rolling resistance
auxiliary power
slope effect
166
20
Figure 4.4 Power required for various accelerations from 30 km/h
6000
5000
power (W)
4000
3000
2000
1000
0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
acceleration (km/h/s)
rolling resistance
aerodynamic drag
auxiliary power
power to accelerate
According to the model, continuous hill climbing as set out in the requirements (10 km/h at 15(, 18
km/h at 12() require 2050 W and 3020 W out of the electric power source, respectively. Cruising at
30 km/h requires 615 W.
Due to the low speeds scooters are typically operated at, and the relative insignificance of the tire
rolling resistance, the power needed for acceleration dominates the maximum power need. The effect
of gravity on scooters traveling up a slope is also significant. In other words, for a scooter traveling
on a level road and accelerating and decelerating in a typical stop-and-start urban driving cycle, the
dominant term is mav, and the total power requirement is thus dominated by the mass of the scooter
and the velocity/acceleration profile. Aerodynamic drag, which is not proportional to mass, tends to
be a minor factor at the low speeds most scooters are driven at, especially in urbanized areas.
167
4.2.4 Validation
Some comparisons with others’ results were provided to check the physical model. As in Figures 4.2
to 4.4, these results are for Poutput without parasitic power, although they do include auxiliary power
load and drivetrain efficiencies.
Table 4.4 Validation of physical model
ITRI
ZES2000
DRI
Sun Com
T. C. Pong
requirements
This physical model
2800 W
0 to 30 m in five seconds
Maximum power
(sustained for 5 seconds)
3800 W
3600 W
Climbing a 15( hill at 10
km/h
2050 W
Climbing a 12( hill at 18
km/h
2160 W
2700 W
3020 W
Cruising at 50 km/h
(sustained for 30 minutes)
1870 W
1800 W
1720 W
1200 W
Cruising at 35.0 km/h
Cruising at 30 km/h
(sustained for 2 hours)
500 W
810 W
600 W
615 W
Cruising at 26.0 km/h
550 W
490 W
Cruising at 15.5 km/h
275 W
250 W
Cruising at 10.3 km/h
225 W
175 W
T. C. Pong, an electric scooter designer, listed a set of electric scooter
power requirements for a 48 V motor system20, while another set of
data was measured from road tests of a Sun Com scooter by Arne
LaVen of the Desert Research Institute.21 The ITRI ZES-2000 results
were from published papers.22,2
The physical properties of the other scooters were unknown but likely not more than 20% different
from the drag coefficient, rolling resistance, and frontal area used in this study. A 77% drivetrain
168
efficiency was assumed for the ZES-2000, since these data were based on motor output; the other
data points were from battery output measurements, and thus no drivetrain efficiency had to be
assumed. The tabulated results are presented graphically below.
Figure 4.5 Validation of physical model
2500
power (W)
2000
1500
1000
500
0
0
5
10
15
20
25
30
35
40
45
50
55
speed (km/h)
model
ITRI
DRI data
T.C. Pong
To verify the model in a different way, the cruising power of 615 W for 30 km/h was used to
calculate average thermal efficiency if the drivetrain efficiency was also 77% for a mechanical
system, and net fuel economy was 100 mpge as reported by various manufacturers. This means an
average 9.5% thermal efficiency, a reasonable estimate for small two-stroke engines (As an
example, a 34 cc engine designed for a blower, hedge trimmer, or chain saw has a peak thermal
efficiency of 13.6%, or 20.6% for a prototype advanced stratified lean-burn design.23)
If a 50% fuel cell conversion efficiency is assumed and parasitic losses are not included yet, then the
equivalent fuel economy is 560 mpge; detailed analysis later will provide a more accurate result.
169
4.3 Driving Cycle
The main purpose of driving cycles, in the past, was to provide a schedule to put cars through to
collect tailpipe emissions and, in the United States, to compute mileages for CAFE (Corporate
Average Fuel Economy). Simulated driving patterns have high-power peaks that produce more
emissions than a constant driving speed, and are more representative of actual driving behaviour.
The typical procedure is to place the vehicle on a wheeled dynamometer and then to execute the
driving cycle. Emissions are collected in a sampling bag and diluted with a predefined amount of air
to obtain vehicle emissions in terms of grams per kilometer.
Typical automobile test cycles include the American FUDS (Federal Urban Driving Schedule),
FHDS (Federal Highway Driving Schedule), and FTP (Federal Test Procedure 1975). American
motorcycles are tested using a modified version of FTP (“mFTP”), with one part scaled down for
motorcycle with engines less than 170 cc, so that the maximum speed is reduced from 91 km/h to 59
km/hr. Another method for testing motorcycles is the ECE-40 (Economic Commission for Europe)
test procedure, which employs a much simpler and more abstracted driving cycle. Low acceleration
rates and lack of transients mean that pollution is generally underestimated.
170
Figure 4.6 mFTP: modified Federal Test Procedure
171
Figure 4.7 ECE-40
In recent years, driving cycles have been paired with computer models of road load, as described
above, to predict and model vehicle performance.
4.3.1 TMDC
Driving patterns in Asian cities are significantly different from American highway driving, and even
American city driving. For example, Taipei’s congestion and frequent stops mean that average
driving speed is less than 15 km•h-1, and driving speed exceeds 40 km•h-1 only 10% of the time.24
The driving cycle used here, therefore, should be specifically targeted for the Asian driver. One
172
candidate is the Taipei Motorcycle Driving Cycle (TMDC), developed by researchers at the
Institute of Traffic and Transportation at Taiwan’s National Chiao Tong University. The TMDC is
an actual velocity trace obtained by researchers who followed target vehicles on an instrumented
“chase vehicle”. The driving cycle consists of 950 velocity measurements (one per second) and each
velocity measurement is rounded to the near km•h-1. An acceleration profile was derived by taking
finite differences in temporally-adjacent velocity measurements.
Figure 4.8 Taipei Motorcycle Driving Cycle (TMDC)
173
As a comparison of several characterizing parameters shows, the TMDC is different from FUDS
and the motorcycle-modified FTP:
Table 4.5 Driving Cycle Comparison
TMDC
modified
FTP
FUDS
950 s
1873 s
1372 s
5109 m
15537 m
7450 m
19.3 km/h
29.9 km/h
19.6 km/h
46 km/h
59 km/h
57 km/h
Maximum acceleration
13.0 km/h/s
5.4 km/h/s
3.6 km/h/s
Maximum deceleration
-15.0 km/h/s
-5.4 km/h/s
-3.3 km/h/s
Fraction of time spent accelerating
31.5%
42.7%
39.7%
Fraction of time spent decelerating
30.3%
56.3%
34.6%
Fraction of time at steady non-zero speed
18.5%
1.0%
6.6%
Fraction of time at standstill (v = 0)
19.7%
15.2%
19.0%
Total time
Total distance traveled
Average speed
Maximum speed
The TMDC exhibits especially severe accelerations and decelerations - maximums of +13.0 km/h/s
and -15.0 km/h/s, respectively. The maximum acceleration in the modified FTP cycle often used for
testing vehicle emissions is 5.4 km/h/s, but the maximum acceleration observed in Bangkok
motorcycle traffic is quoted as being 12 km/h/s.25 (It is not clear what size of motorcycle the
Bangkok number refers to). While it is true that scooters in Taiwan are driven in a more aggressive
way than cars in American cities, these accelerations and their consequent power requirement of
over 12 kW at the wheels significantly exceed the maximum performance capabilities of scooters
125 cc or less.
One important ramification of the high accelerations and decelerations is that a significant amount
174
of energy can theoretically be recovered by regenerative braking. Also, maximum power is much
larger than average power.
Note also in Figure 4.8 “jitter” in the velocity reading. Errors due to rounding of the speedometer
reading to the nearest km/h when the data was recorded create exaggerated accelerations and
decelerations that did not reflect reality. For example, a speed that varied from 20.4 to 20.6 and
back to 20.4 in two seconds would appear in the data as a one-second acceleration from 20 to 21
and back to 20. This jitter was analyzed below using the scooter model described previously. (The
jitter starts from the base speed and oscillates up and down by 1 km/h)
Table 4.6 Effects of “jitter”
initial
speed
(km/h)
average power to
accelerate 1 km/h faster in
1 second, then return
to original speed (W)
power if speed
was a constant
0.5 km/h faster
(W)
jitter increases
power by
this fraction
5
179
117
53%
10
290
177
64%
15
416
252
65%
20
564
348
62%
25
740
472
57%
30
951
632
50%
35
1204
834
44%
40
1507
1085
39%
45
1866
1393
34%
The test shows that these oscillations produce significant variations in power required by the model
that are not representative of actual driving.
175
4.3.2 Modification of TMDC
The TMDC is more representative of Taipei driving conditions than FTP or FUDS, but it has flaws
that should be compensated for. As given, to achieve the performance of the TMDC simulation, a
total power output of more than 12 kW is needed, significantly greater than the maximum power
achieved by the internal combustion engines of even 125 cc scooters.
As a first attempt at improving the TMDC, the maximum speed was clipped to 40 km/h to more
closely approximate reality. Accelerations were calculated from this less strenuous driving cycle;
fortunately, these peaks were very brief and the integral under these parts of the velocity curve was
small, so that there was very little difference in parameters of the driving cycle like total distance
and fuel consumption. However, this first cut reduced maximum power only to 9.7 kW, so this
technique was discarded.
Smoothing with a moving three-second box, as suggested by the researcher who developed the
TMDC, was not adequate in attenuating the maximum peaks.
Another method of adjusting the driving cycle to reflect a more realistic assumption was to use a
low-pass filter to get rid of the quantization jitter and to attenuate the very quick accelerations.
Changing the characteristics of the low-pass filter would modify the final modeling results, so to
choose an appropriate filter function, the results of the adjusted cycle were compared against
reported data on scooter performance levels. (For example, the ZES-2000 was specified as being
able to travel 30 m from a standstill in 4.5 seconds at maximum acceleration1, 7, and the power
required to achieve this acceleration was found to be 3.7 kW using the physical model. Similarly, a
standard gasoline-powered scooter was quoted elsewhere as having a maximum power of 5.0 kW 2.
176
A maximum net electrical output power of about 5.0 kW seemed reasonable given these examples.)
The low pass filter was created as a transfer function defined to have a DC gain of 1 to not change
the average value of the function acted upon (in this case the velocity as a function of time).
H ( s) =
1
2π s
1+
τo
-o is the characteristic time, so that larger values of -o produce greater smoothing. This function
was convolved with the velocity profile of the driving cycle to reduce high-frequency jitters and to
attenuate accelerations and decelerations. The smoothing effect can be thought of as a multiplication
(in the frequency domain) of the transfer function and the frequency spectrum of the driving cycle;
due to the hyperbolic shape of the transfer functions, high-frequency components are attenuated
while low-frequency components are increased.
The results of running the simulated scooter under the TMDC for various manipulations of the
TMDC are presented below. The scooter parameters given in Table 4.3 were used. The FTP was
compared to the smoothed driving cycles as a check to see if the results were close. The italicized
choice was the one eventually selected.
177
Table 4.7 Results of different algorithms applied to TMDC; comparison to FTP
original
clip at
40 km/h
1/(s+4)
smooth
1/(s+1.5
)
smooth
1/(s+1)
smooth
mod.
FTP
average speed over cycle
(km/h)
19.3
19.3
19.3
19.3
19.3
29.9
max net power from engine
(includes drivetrain)
12.6
kW
9.8 kW
9.0 kW
5.6 kW
4.2 kW
5.5 kW
935 W
759 W
651 W
566 W
537 W
1254 W
max acceleration (km/h)
13.0
13.0
9.8
6.4
5.1
5.4
max deceleration (km/h)
-15.0
-15.0
-9.2
-8.6
-7.7
-5.4
std. dev of accelerations
0.79
0.76
0.61
0.48
0.44
0.58
avg. acceleration power
(avg. of positive and negative)
63 W
59 W
38 W
24 W
20 W
34 W
avg. acceleration power
(positive only)
370 W
355 W
285 W
215 W
188 W
289 W
avg. rolling resistance power
151 W
151 W
151W
151 W
151 W
234 W
avg. aerodynamic drag power
118 W
117 W
117 W
117 W
116 W
425 W
accel. power
(when in motion)
1178 W
1129 W
678 W
517 W
446 W
674 W
rolling power
(when in motion)
189 W
188 W
184 W
179 W
175 W
272 W
aerodynamic drag
(when in motion)
147 W
146 W
143 W
138 W
134 W
495 W
avg power from engine (no
parasitics; includes drivetrain)
Note that the average acceleration powers are very low because acceleration can be negative; the
average acceleration power for positive results only is more indicative of how the energy is split
between the various components. It should be noted, however, that negative acceleration power can
be used to “cancel out” power demands from aerodynamic drag and rolling resistance. That is, when
the vehicle is decelerating, the drag power and rolling resistance can be allowed to slow down the
vehicle so negative accelerations are not entirely meaningless to the power calculation.
178
The mFTP results show much higher average power than the selected smoothed curve, due to rolling
resistance and aerodynamic drag, which in turn are due to the average speed being 50% higher than
the TMDC. On the other hand, the maximum power is very close to that of the smoothed curve,
suggesting that a scooter designed for the TMDC will be capable of sustaining the FTP driving
cycle which was originally designed for the more powerful motorcycles. The maximum mFTP
power of only 5.5 kW is a telling indicator that the TMDC (unsmoothed) is too severe.
Smoothing dramatically decreases maximum accelerations and decelerations, and changes the
acceleration characteristics of the driving cycle, but does not significantly change the other
components; this is because acceleration is such a large component of maximum power but not such
a strong determinant of average power.
The low-pass filter with a 3.1 second smoothing interval (italicized in Table 4.7) was chosen to
smooth out jitter and reduce the maximum power required to on the order of 6 kW - specifically, 5.6
kW including auxiliaries but not including parasitics because parasitics are dependent on later
calculations based on these results. This is comparable to modern 50 cc scooters, with maximum
power closer to those of the mFTP.
179
Figure 4.9 Smoothed TMDC
The “smoothed TMDC” driving cycle was used for all further calculations, and referred to as
simply the TMDC. Note that the energy finally dissipated in braking (i.e. where deceleration
“power” is greater than aerodynamic drag and rolling resistance powers) is 116 kJ over the 950
second cycle, or 122 W in the smoothed TMDC. This is about 20% of the 566 W engine output.
4.3.3 Torque vs. rpm requirements
Looking only at the maximum power produced by the electric motor and the maximum power
produced from the fuel cell or battery is not sufficient to ensure that the power demands of everyday
driving are met. This is because the total output is limited by the maximum power of the motor, but
180
also by other factors like the maximum current (which sets a maximum torque even when the speed
might be low).
Torques are summed about the axle of the drive wheel; the rolling resistance and acceleration
“torque” have moment arms equal to the radius of the wheel (15.8 cm), while the drag force is
assumed to be act at the center of mass of the scooter, 20 cm above the axle. In order to ensure that
the required torques could be produced, the model results were illustrated as a scatterplot of torque
versus speed. Imposed on this graph was the maximum performance curves of the chosen Unique
Mobility (UQM) brushless DC scooter motor rated at a nominal 3.6 kW.
181
Figure 4.10 Torque vs. rpm during TMDC
120
UQM motor
nominal limit
torque (N-m)
100
80
TMDC data
60
40
20
0
0
100 200 300 400 500 600 700 800
angular speed (rpm)
Note that at three one-second time intervals, torque required exceeds that available from the UQM
motor. These peaks represent the maximum power of 5.6 kW of electrical output, which translates
to 4.3 kW of mechanical power after the 77% drivetrain efficiency - greater than the 3.6 kW
maximum output of the UQM motor. The assumption was made that the three outliers were
sustainable for short periods of time.
182
4.3.4 Modeling results
The results of the TMDC driving cycle are shown below.
Figure 4.11 Power required in TMDC
183
Acceleration power demands the greatest peaks in the cycle, and also accounts for most of the
energy in the cycle (not including braking energy recovered from negative accelerations). Rolling
resistance accounts for somewhat less energy, and aerodynamic drag even less. The 60 W of
auxiliary power is at most 10% of the average power.
Interestingly, the average power is approximately one tenth of the maximum power. The extreme
variability in the power demands suggests that hybridization would be useful, with a battery
providing surges of extra power during bursts of acceleration and also the capability to store
braking energy.
The physical model shows that the power needed, under the TMDC, is an average 566 W of electric
power out of the fuel cell without parasitics. A complete analysis of fuel economy, however,
requires a polarization curve of efficiency versus net power, and an understanding of the parasitic
power. This is in section 4.5.
4.3.4.1 Battery powered scooter
The parasitic power requirements of the fuel cell have not yet been calculated but there is enough
information here to calculate performance of a scooter running on just a single battery. The power
output is the same as the fuel cell power output except without the parasitic requirement. A total of
4.1 kWh (output) are needed to store enough electricity for 200 km of range.
Average power (electric output)
Mileage in terms of
electric output
Maximum power output
Total energy storage for 200 km at 30 km/h
4.1 kWh
The battery requirements, given the different USABC battery technology predictions and a 4.1 kWh
storage capacity, are compared to today’s lead-acid batteries.
Table 4.9: Various battery-powered designs for Taiwan scooter
Lead-acid:
today’s battery
Mid-Term
advanced battery
Long-Term
advanced battery
Weight
117 kg
62 kg
25 kg
Volume
58 L
36 L
16 L
$245-$735
(current)
$735
(mid-term)
$490
(long-term)
Cost
For the lead-acid battery, the critical assumptions were 35 Wh/kg and 70 Wh/L. Note that in the
mid-term and long-term cases, the battery weight and volume are determined by the maximum
energy requirement, not the power requirement. In fact, a mid-term battery of this size would offer
7.7 kW of maximum power, while the long-term battery would output a maximum of 8.2 kW! The
limiting factor in battery size and weight for these batteries is energy storage, not power storage.
Section 4.7 on hybrid design shows how decoupling the energy and power functions of the battery
improve the system.
185
Today’s ZES-2000 scooter with far shorter range than the systems described above gives an idea of
how much room is available in the scooter: at least 44 kg and 15 L (not including the electric motor
and controller). The next step is to design a fuel cell power system within these size and weight
parameters that can output a continuous 610 W for 30 km/h cruising for 200 km (or 6.7 hours);
produce a 5.6 kW maximum output; and generate 3.2 kW of continuous hill climbing power.
4.4 Fuel Cell System Design and Integration
4.4.1 Design tradeoffs
The fuel cell stack is specified by only two independent variables plus the polarization curve and the
maximum power:
1. Maximum power
2. The polarization curve
3. Power density
4. Number of cells in the stack
(Note that, instead of power density and number of cells, we could have equivalently chosen area
per cell and total active area of all cells)
186
4.4.1.1 Maximum power and the polarization curve
A maximum gross fuel cell power of 5.9 kW is assumed from parasitic requirements calculated in
section 4.4.5. This is slightly over the sum of the maximum TMDC requirement of 5.6 kW net
power plus an initial estimate of 300 W for parasitic power losses. The parasitic requirements are
discussed in greater detail later on, but for now it is enough to say that they are approximately linear
with gross power output so that maximum parasitic power occurs at maximum gross power.
The polarization curves used are those derived from Energy Partners for single atmosphericpressure cells; these are extrapolated to be equal to future stack performance.
Figure 4.12 Polarization curve
1000
0.9
voltage
voltage (V)
0.8
900
power
800
0.7
700
0.6
600
0.5
500
0.4
400
0.3
300
0.2
200
0.1
100
0.0
power density (mW/cm^2)
1.0
0
0
200 400 600 800 1000120014001600
current density (mA/cm^2)
Data is from Barbir is for a single cell running on hydrogen/air, with a Gore
MEA and operating temperature of 60(C. Air-side stoichiometry is 2.5.26
187
4.4.1.2 Power density
The first choice to make is to determine the power density to operate at, under conditions of
maximum power. If maximum output is arranged to take place at low power density, the left portion
of the polarization curve is used (since power density scales almost linearly with current density);
voltage, and thus efficiency, are high in this portion of the curve. However, a large total active fuel
cell area is needed and much electrolyte membrane and platinum will then be required. The stack
size will be larger.
If a low total area is used, then power density per area is high. The power density curve peaks at
high current densities, with correspondingly low voltages and low efficiencies. (The low total area
can be achieved by a combination of low cell area and/or few cells). Total price is a monotonic
function of total area, so the high power density approach results in a smaller stack and cheaper
stack, although efficiency would suffer and consequently hydrogen storage requirements for a given
range would be higher.
Power density also controls the flow rates of the air and hydrogen reactants. The flow rates are
functions of the designed hydrogen utilization and oxygen stoichiometric ratio, and also of the
efficiency of the fuel cell; low efficiency means that more reactants must be flowed for a given
power output. Thus, for a given power level, high power densities mean higher flow rates because
they operate in the low voltage (low efficiency) portion of the polarization curve. The sizes and
costs of air management equipment are determined in part by flow rate, but it should be noted that
the scooter application demands flow rates far below those normally available for vehicle systems so
most systems for scooters based on automotive designs would be oversized anyway.
188
To maximize efficiency and produce a small, cheap fuel cell, a high power density is selected.
According to the polarization curves presented, power is maximized at about 1300 mA•cm-2; to be
conservative and leave room for unusual bursts of speed, a point below this power peak is chosen:
1000 mA•cm-2. At 1000 mA•cm-2, voltage is 0.614 V, and power density is thus 614 mW•cm-2. For
5.9 kW at this point, 9.6 x 103 cm2 of active area are needed.
4.4.1.3 Number of cells
The number of cells is a function of the desired operating voltage. The electric scooter industry in
Taiwan is standardizing on 48 V electric motors, so the number of cells is chosen so that the stack
operates in the vicinity of 48 V at the most common power demand; note that in a fuel cell, as the
total power output changes, the voltage varies as well.
The DC-to-DC conversion is assumed to be performed by the motor controller, and is included in
the 77% drivetrain efficiency. (In comparison, standalone DC-to-DC converters from Vicor offer
approximately 410 W/L and 381 W/kg, with efficiencies of 80%-90%; prices are on the order of 1
$/W.27,28) All parasitic power – fans, pumps, blowers, etc. – should be DC to avoid the large
additional expense of a DC-to-AC inverter.
The average TMDC power demand is 566 W, and later modeling in shows that the final result
including parasitic power (after iterative calculation) is 674 W. With a 9.6x103 cm2 total membrane
area, the power density is 70 mW•cm-2. To obtain this power density on the polarization curve, 79
mA•cm-2 and 0.870 V are needed - only 8% of the maximum current density, and high up on the
efficiency curve. The system is designed to run at 48 V at this point, and dividing 48 V by the 0.870
V per cell gives a minimum of 56 cells. Area per cell is then approximately 170 cm2.
189
Table 4.10: Fuel cell design parameters at maximum power
maximum
power
hill climbing
power
average
TMDC power
5.9 kW
3.2 kW
665 W
1000 mA•cm-2
448 mA•cm-2
79 mA•cm-2
172 A
76 A
13 A
0.614 V
0.751 V
0.870 V
Power density
614 mW•cm-2
336 mW•cm-2
69 mW•cm-2
Stack voltage
34.2 V
42.0 V
49.0 V
Power with parasitic load
Current density
Stack current
Cell voltage
Open-circuit voltage
(occurs at minimum power;
parasitics not included)
56.0 V
9600 cm2
Total active area needed
Total number of cells
56
Active area per cell
170 cm2
(Note that this stack could also have been designed as two electrically parallel stacks of 56 cells
each 85 cm2 in area and with half the maximum current (86 A), or various other configurations).
4.4.1.4 Flow rate parameters
It is further assumed that the air flow rate is 2.5 times stoichiometric (to reproduce the performance
of the Energy Partners polarization curve) and that the hydrogen consumption is 100% due to deadended operation. The surplus air lessens the effects of oxygen depletion in the cathode as oxygen is
consumed by the fuel cells and also helps to push out product water. The exhaust gas flow ratewas
calculated by summing the input gas streams of air (21% oxygen) and hydrogen and then
subtracting the hydrogen and oxygen consumed. The water is assumed to emerge as liquid so is not
included in the exhaust flow rate. From this set of data, the following flow rate characteristics are
190
derived:
Table 4.11: Flow rate parameters at maximum power
volumetric
molar
mass
hydrogen intake rate
2.6 CFM
0.05 mol•s-1
0.1 g•s-1
air intake rate
15.6 CFM
0.30 mol•s-1
8.6 g•s-1
total exhaust gas rate
16.9 CFM
0.27 mol•s-1
7.8 g•s-1
(liquid) water
production rate
0.9 mL•s-1
0.05 mol•s-1
0.9 g•s-1
4.4.2 Gas subsystem
As discussed previously, the pressure drop in fuel cell stacks has been estimated at 0.5 - 2 psi. With
a 50% blower efficiency, worst-case 2 psi air drop, and the calculated 15.6 cfm of air flow, this is a
maximum theoretical power consumption of 200 W. In theory, this scales down linearly with
decreasing flow rate, but in practice the pressure drop also decreases as a function of flow rate so
the decrease is somewhat sharper than linear. A heavy-duty blower that could be used to provide the
required output is the Ametek 5.7" BLDC three-stage blower, model 116638-08. Its volume flow
capacity is much higher than the 16 cfm required, but it is the smallest model capable of the
relatively high 2 psi needed.29 Retail cost is $430.
The blower power was modeled in the simulation as a linear 50-250 W load, for a gross power
output of 50-5850 W. This is a conservative calculation based on comparison with a reported
parasitic power draw of 105 W for a 24 VDC Ametek blower at 1.3 psi for a 4 kW nominal power
fuel cell.30
On the hydrogen side, a pressure regulator expands the hydrogen from either the 1-10 atm partial
191
pressure of a metal hydride system, or a 3600 psi (260 atm) pressure of a compressed gas system.
4.4.3 Water subsystem
A low fuel cell operating temperature of 50(C is chosen to minimize evaporation losses and
eliminate the need for external humidification (and complex control of that humidification). The
maximum allowable fuel cell temperature is set to 65(C.
According to Mazda Demio documentation, external humidification requires an additional 15% in
stack volume, but thin wicking polymer membranes can allow water to be backdiffused from the
cathode to the anode to keep the membranes humidified without external humidification. 31 The
wicking polymers could alternately transfer water from a reservoir to pre-humidify incoming air if
humidification turns out to be necessary after all.
4.4.4 Cooling subsystem
There are several heat flows to consider in fuel cell systems.
1. Waste heat must be removed from the stack with a liquid coolant loop.
2. The liquid coolant must have heat removed at the cold side with a fan.
3. Air entering the system may be preheated in order to retain more water from any humidification
4. The humidification water, if any, can be preheated as well
5. Reformers, if present, require temperatures of at least 300(C to operate, and may produce net
heat
6. Heat must be supplied to the metal hydride system, if one exists, in order to desorb the hydrogen.
192
To optimize the system, some of the flows can be combined. For example, the coolant loop is a
convenient source of heat for a metal hydride storage system, while the blower could be designed to
draw its input air from behind the coolant radiator.
All heat flows in the system are functions of the instantaneous power, since the waste heat,
hydrogen demand, and air demand all scale according to the efficiency curve with respect to power
in almost the same way. The maximum heating load occurs at maximum power in the driving cycle
- at 5.9 kW of gross electric output, efficiency is 41.2% and heat output is 8.4 kW. However, the
average heat load during typical TMDC driving is much lower, only 742 W. Given sufficient
thermal mass, temperatures in the stack can be kept near the design point without the use of a large
radiator. The ultimate concern is to keep the temperature low to avoid evaporating too much water
from the membrane.
In comparison, a 20% efficient 5 kW internal combustion engine outputs 20 kW of waste heat. The
difference is that this load is produced at high temperatures and thus is easily rejected to the
environs by air blowing over cooling fins
There are two beneficial cooling effects that are not quantified here. First, fuel cell efficiency is
calculated on a higher heating value basis, meaning that the product water is assumed to emerge
solely as liquid water. In fact, some is vaporized and this removes heat from the system, making the
cooling design somewhat conservative.32
Second, some heat is “used up” by the intake air. At a maximum flow rate of 8.6 g/s, an intake
temperature of 30(C, and a heat capacity for air of approximately 1 J•g-1•K-1, 170 W of heat are
required if the incoming air heats up to the stack temperature of 50(C. This small amount of cooling
193
is not included below but is noted for the sake of completeness.
4.4.4.1 Cooling from storage system
Here, cooling from metal hydride adsorption is discussed, along with more conventional cooling.
The temperature of the fuel cell stack is modeled over the entire driving cycle to ensure that the fuel
cell remains within its designed limits of 50(C and 65(C.
TiFe metal hydride systems consume 28 kJ of heat per mole of hydrogen desorbed. In the system,
the maximum heat production is 8.4 kW and occurs at the maximum gross power of 5.9 kW. Here,
hydrogen consumption is 0.05 moles•s-1, so the hydride takes up 1.4 kW (16.7%) of the waste heat.
This percentage increases as the fuel cell is turned down to lower powers, because the number of
moles of hydrogen per heat output increases due to the greater efficiency; over the full range, the
metal hydride system eliminates 16.7% - 30.2% of the waste heat. This reduces the size of the
radiator and decreases the parasitic power required for cooling.
194
Figure 4.13 Metal hydride cooling vs. power
average
hill climbing
maximum
9000
8000
heat or power (W)
7000
6000
5000
4000
gross waste heat
net waste heat
3000
2000
1000
TiFe hydride cooling
0
0
1000 2000 3000 4000 5000 6000
net power (W)
For this reason, metal hydrides have a significant advantage over gas cylinders. Combining a water
cooling system for the fuel cell with the metal hydride allows the transfer of this heat, although a
backup heating system might be necessary for startup heating and active control of the metal
hydride temperature. This backup system could be a resistance heater wrapped around part of the
metal hydride, connected to the startup battery. (The pressure of hydrogen gas over the hydride at
room temperature would be greater than atmospheric, so there would be some hydrogen at startup.
However, with waste heat slow to reach the hydride, an extra heater could provide a faster flow rate
for immediate high power).
195
4.4.4.2. Active cooling
There are several possibilities for liquid cooling.
1. Cooling is provided by a closed loop water coolant circulated at constant flow rate through the
stack by way of cooling plates. The hot water exiting the stack is circulated to a heat exchanger
exposed to the atmosphere, where a fan enhances heat transfer. The fan speed can be increased to
provide additional cooling. Alternately, a constant-speed fan can be used in a thermostat mode
(allowed to switch on and off as needed), allowing the coolant temperature to fluctuate.
2. Variable coolant flow rate, fixed fan speed. As stack output power increases, the coolant
circulation speed is increased. The change in temperature of coolant as a function of power is
dependent on the heat exchanger properties. A variable-flow pump is needed.
3. Variable coolant flow rate, variable fan speed.
4. The pump is eliminated, reducing weight, cost, and parasitic power; instead, a refrigerant
designed to boil at the fuel cell operating temperature is used in conjunction with a check valve, and
the expanding gas drives circulation in the coolant loop.
In all cases, water circulation should be stopped altogether when the engine is warming up to
operating temperature, so some kind of thermostat (if not variable) control should be employed for
the coolant loop.
The heat exchanger cooling factor, in terms of watts of heat dissipated per degree of temperature
196
difference between the coolant and ambient temperature, provides an upper limit on how much heat
can be removed from the stack at a constant output power. Charts of cooling factor (W/K) as a
function of air flow rate and coolant water flow rate for various heat exchangers, in conjunction
with charts of pressure drops, were used to determine parasitic loads for the radiator fan and coolant
pump, respectively. This information was obtained from Lytron, a heat exchanger manufacturer
(see Appendix C)
Continuous hill climbing determines maximum continuous power output (intermittent higher power
output, like that of the TMDC, is interspersed with many periods of low power output). To satisfy
the hill climbing requirement of 18 km/h at a 12( slope, 3200 W (gross output with parasitics) are
required. Efficiency at this output point is 50.7%, waste heat is 3.1 kW and hydrogen consumption
is 0.022 moles/s. The 28 kJ/mol of the hydrides eliminates 20% of the waste heat, leaving 2.5 kW,
or 100 W/K that must be dissipated for a fuel cell at the design maximum of 65(C and a worst-case
ambient temperature of 40(C ( T=25(C). An extra 10 W/K was added for a design specification of
110 W/K, which is met by an M14-120 radiator at a retail cost of $240.
Due to the lack of space at the front of the scooter, the radiator would likely have to be installed at
the sides of the rear compartment, where flowing air would pass over the radiator pipes.
Here, the assumption is that the coolant pump and cooling fan draw power all the time. This is not a
realistic assumption in that the power should be turned off when cooling is not needed (to ensure the
fuel cell is above the minimum operating temperature). However, it is a safe assumption in that it
overestimates (slightly) the power needed for parasitics, especially since pump and fan power draw
needs are uncertain.
197
4.4.4.3. Heat generation under the TMDC
Maximum heat dissipation is determined by continuous hill climbing, but to ensure that the short
spikes of high heat generation in the TMDC do not push the fuel cell temperature past 65(C, the
temperature of the stack was simulated using a simple model. Heat generation under the TMDC
simulation has the same general shape as the power output graph, but the peaks and valleys are
exaggerated because efficiency decreases with increasing power output.
Figure 4.14 Heat generation as a function of time in TMDC
198
Note that average heat generated is 742 W for the 5.9 kW stack. After the metal hydride heat
absorption is included, this decreases to an average load of only 393 W. The net heat generated was
calculated for each step of the TMDC and used to calculate a change in temperature for the stack.
The purpose of this further test was to make sure that the cooling factor (sized for continuous load)
would be enough to keep stack temperature below the design limit of 65(C through the peaks and
spikes of the TMDC.
The heat capacity equation used was
Q = M C ∆T
for heat Q, stack mass M, average heat capacity T, and change in temperature T. This was
discretized for each time step, and the mass and heat capacity were separated into the different
materials found in the stack, resulting in the following equation:
Q
∆T
=
M 1C 1 + M 2 C 2
τ
Q is the heat power generated in a given time interval; M is the mass of the parts of the stack that
make up the bulk of the total weight and are closest to the membranes (i.e. “1" for the stainless steel
separator plates and “2" for the polypropylene gaskets); C is the heat capacity of the stack
materials; and - is the length of time of a single time step of the model.
199
Table 4.12 Stack temperature model parameters
parameter
value
note
Mass of 316 stainless steel
separator plates
2.0 kg
0.5 J•g-1•(C-1 specific heat capacity
Mass of polypropylene gaskets
0.8 kg
2 J•g-1•(C-1 specific heat capacity
Heat capacity of stack as a unit
2.6 kJ•(C-1
Specific heat capacity of stack
0.93 J•g-1•(C-1
Heat exchanger cooling factor
150 W•K-1
Ambient temperature
Heat removed by metal hydride
40(C
28 kJ•mol H2-1
Weighted sum of steel and polypropylene
c.f. water at 4.2 J•g-1•K
maximum cooling
worst case
minimum 17% of waste heat
The assumption made was that most of the heat would be trapped inside the plastic housing
(designed for electrical insulation), removed mainly by the active cooling system. The membrane
itself is negligible because it is so light. This simple model does not include the effects of heat
conductivity - just heat capacity.
The cooling system was designed to turn on only above a temperature of 50(C in the stack (as
detected by thermistors in the stack), with a maximum allowable temperature of 65(C. The
following temperature patterns were recorded for the TMDC for cooling factors of 110 W/K as
designed, and 30 W/K for comparison.
200
Figure 4.15 Stack temperature as a function of time in TMDC
The smaller cooling factor is perfectly adequate to keep the maximum temperature below 65(C but,
as calculated previously, the full 110 W/K is needed for sustained hill climbing.
As discussed, either the cooling fan or the pump could be switched on and off to produce cooling. In
practice, the pump should be the unit controlled, because some cooling is still derived from pumping
the water through the externally-exposed radiator, even if the fan blowing over the radiator is off.
This seems like a benefit, except that excessive cooling would lower the fuel cell stack temperature
below its designed operating point of 50(C; also, when the fuel cell is first started, it needs to warm
up as quickly as possible.
201
A more sophisticated temperature simulation would include a time lag between the temperature
measurement inside the stack and the control (turning the pump on or off). A more sophisticated
design would vary the pump speed in order to reduce the power taken up by the pump, especially
since adequate control could be maintained at just 20 W/K to achieve the result graphed above. This
would decrease parasitic load and extend range, but, again, this was not assumed here due to the
uncertainty in actual cooling power needs.
4.4.4.4 Selection
To achieve the upper limit of 110 W•K-1, a Lytron M14-120 radiator was selected. The cooling
system includes a metal heat exchanger, pump operating at 1 gallon per minute (0.06 L/s), and fan
blowing over the heat exchanger. For 110 W•K-1 of cooling, a fan speed of 375 cubic feet per
minute (cfm) is necessary.33
According to performance charts (Appendix C), the coolant pressure drop in the radiator is 1.0 psi,
while the fan-blown air decreases in pressure by 0.16 inches of water (40 Pa). Assuming a 50%
pump efficiency and 50% fan efficiency, these translate into power demands of 25 W for the pump
and 14 W for the fan. (Note that this takes the 1.0 psi coolant water pressure drop in the radiator
and adds an estimated 2 psi drop from the fuel cell cooling cell flow fields)
It might seem that air supplied by ram-effect from the motion of the scooter would be enough to
cool the exchanger. This is true in most cases, making the heat exchanger fan speed conservative,
but false for the case of low-speed hill climbing, where power demands are high but air speed low.
The design is also conservative because parasitic power is calculated as if the pump were on all the
time, whereas this is actually not true as discussed previously.
202
This heat exchanger weighs 8.9 kg and takes up 15.2 L of space, including the fan, and accounts for
a significant fraction of the total weight and volume.
4.4.5. Overall parasitics
Note that the specifications for pumps and blowers and radiators are for industrial, and often
stationary, AC power components. This was done to get an idea of real world performance and cost
without designing specifically for the scooter application. In practice, DC components would be
needed to avoid an expensive DC-to-AC inverter, and components could likely be optimized for the
application at hand. Prices, described in Chapter 5, are retail; cost to the scooter manufacturer
would be as little as a quarter of retail.
The system requires a blower to push the air in, but omits the bulkier compressor and possibility of
expander motor power recovery at the exhaust of the fuel cell.
The maximum parasitics estimated for the scooter system are 25 W for the coolant pump and 14 W
for the fan blowing over the radiator, as discussed previously, plus a power draw from the Ametek
blower power which is assumed to scale linearly with flow rate (i.e. power), from a minimum of 50
W of power at no load to 200 W at maximum fuel cell output of 6 kW. The power is calculated as
if the pump and fan were on all the time, with the blower always requiring 50 W, for a total
parasitic load of 89 W – 239 W over the output range of the fuel cell; in practice, power is saved by
turning off – or slowing down – the fan and pump when not used.
The results are compared with those obtained by a study published by the Schatz Energy Research
Center.34 The Schatz report calculates the following total parasitic power requirements for a
203
nominal 4 kW vehicle (“Personal Utility Vehicle” - i.e. golf cart) fuel cell stack operating at
atmospheric pressure.
In the Schatz system, the parasitic power comes from the water coolant pump, atmospheric-pressure
blower, and the cooling fans. The system operates at approximately 40(C, and the air flow rate at
1.74 kW is 4.48 cfm (approximately 60% higher at the same power than the system presented here).
So parasitic demands vary from 100% of gross power down to 5% above 2000 W, a fairly small
fraction.
Figure 4.16 Parasitics as a function of power
fraction of gross power
350
300
250
linear estimate of
parasitic power
presented here
200
150
Schatz parasitic
power
100
50
0
1000
2000
3000
4000
5000
6000
net power (W)
The next graph compares power usage as a percentage of gross power for both systems.
204
Figure 4.17 Parasitics as a percentage of power
100%
fraction of gross power
90%
linear estimate of
parasitic power
presented here
80%
70%
60%
50%
40%
30%
Schatz
parasitic power
20%
10%
0%
0
1000
2000
3000
4000
5000
6000
net power (W)
Finally, the parasitic power reduction is represented as a voltage reduction in the polarization curve.
The result is a combined efficiency that has net electricity output as its numerator. The peak
efficiency point is shifted to higher current densities and efficiency is reduced below 50% at all
points.
205
100%
10
80%
8
efficiency
gross efficiency
60%
6
net efficiency
with parasitics
40%
4
20%
net power (kW)
Figure 4.18 Effect of parasitics on efficiency
2
net power
0%
0
0
1000
2000
3000
4000
5000
6000
gross power (W)
4.5 Integrated Model
4.5.1 System performance
The complete model takes the vehicle physical model described at the beginning of this chapter, and
integrates the efficiency of the motor/controller subsystem, parasitic power demands, and the fuel
cell polarization curve to determine overall efficiency: the amount of hydrogen consumed for a given
travel distance under both the Taipei Motorcycle Driving Cycle and steady state 30 km/h driving
conditions. The overall performance is used to identify the sizing of subcomponents like the fuel
206
storage supply and thus to determine the overall system weight and size.
Essentially, the driving cycle model is run, and at each discrete point in time, the power needed at
the wheels is calculated and divided by the 77% drivetrain efficiency. An auxiliary power
component of 60 W is added. Parasitic power is determined as a function of this total, and added.
(This is an iterative process because the parasitic power is included in the total power, from which
the parasitic power is calculated.)
Next, this total electrical output power is divided by the efficiency at that power demand; this
efficiency is determined from the voltage on the polarization curve for the power required. The
result is the amount of hydrogen consumed at the time interval, in terms of higher heating value
energy units.
The results, over the driving cycle, are the maximum and average power (including parasitics); and
the fuel economy in terms of hydrogen consumed per kilometer traveled. The latter is readily
converted to miles per gallon of gasoline equivalent. An overall efficiency is calculated for the
conversion process.
A battery-powered option is considered using the same basic model, but with no parasitic power to
consider because fans, blowers, and pumps are not needed.
As before, the total scooter curb weight was set at 130 kg, and a 75 kg driver was added. Later
results show that the vehicle weight is approximately the same as this assumption.
207
Table 4.13 System Performance under TMDC and at cruising speed
TMDC
30 km/h cruise
5.91 kW
725 W
Average fuel cell output power
674 W
725 W
Overall efficiency
46.7%
58.5%
0.527 km/g H2
0.807 km/g
344 mpge
522 mpge
Hydrogen storage for 200 km range
380 g
248 g
Average output power without parasitics
(battery powered scooter)
566 W
614 W
Fuel economy of battery powered scooter
35.5 km/kWh
48.8 km/kWh
Battery energy storage for 200 km range
6.5 kWh
4.1 kWh
Maximum power from fuel cell
(includes drivetrain and parasitics)
Fuel economy relative to hydrogen
Equivalent “on-vehicle” fuel economy
Note that these battery energies are given in terms of total energy output. The energy that must be
put into these batteries is higher due to less-than-100% charging and discharging efficiency.
However, this additional energy is not included here because price and performance figures for the
zinc-air batteries are given in terms of energy output.
4.5.2 Size and weight of power system
The detailed analysis done in Appendix B and described in section 3.1.3.3 estimates a stack size and
volume of 7.6 kg and 7.8 L respectively. This is a power density of 0.78 kW/kg and 0.76 kW/L for
the stack alone, slightly less than 1996 Ballard stacks at 1 kW/L.
Assuming a factor of two extra for air and heat and water management subsystems gives power
densities of 0.39 kW/kg and 0.38 kW/L. (The PNGV Technical Roadmap requirements are 0.4
kW/kg and 0.4 kW/L.35) That is a simple estimate;; the following paragraphs produce a more
208
detailed analysis of the subsystems. First, the heaviest and bulkiest subcomponents are analyzed and
listed in Table 4.14. These are the blower that supplies air to the fuel cell stack, the pump that
circulates cooling air, and the radiator. Prices are projected for mass production, with more details
in section 5.1.3.
Table 4.14 Subcomponent summary
Brand
Model
Dimensions
(in cm)
Size
Weight
Cost
(long term)
Fuel cell stack
–
–
–
7.8 L
7.6 kg
$220
Starter battery
Yuasa
GRT
YT4L-BS
11 x 7 x 9
0.7 L
1.3 kg
$10
Coolant pump
generic
–
8 x 12 x 12
1.2 L
1 kg
$20
Radiator with fan
Lytron
M14-120
15.2 L
8.9 kg
$60
Blower
Ametek
116638-08
15 diameter
x 17 length
2.9 L
2.7 kg
$110
plumbing,
wiring, etc.
generic
–
–
2.0 L
3.0 kg
$50
coolant water
–
–
–
–
0.64 kg*
–
TOTAL STACK
WITH AUXILIARIES
–
–
–
29.8 L
24.6 kg
$470
The “generic” pump has pressure requirements of 2.5 psi and a
pumping flow rate of 1 gallon per minute. This is adequately supplied
by an aquarium-type pump. Electric cabling, air manifolds, water
plumbing were estimated to add 3 kg and 2 L and a cost of $100
*Note that the radiator holds 320 mL of water when full, and with an
estimated total of twice that amount of water in the entire system, this
adds an additional 0.64 kg of weight.
The total performance figures for the stack with auxiliaries are 0.24 kW/kg and 0.20 kW/L. The
stack proper takes up 27% of the mass and 26% of the volume.
In comparison, the overall fuel cell stack weight of the Schatz 4 kW system described previously
was 75 lbs or 34 kg; the entire power system weighed 200 lbs or 90 kg. Stack volume was 10" x
209
11" x 21", or 39 L.36 With the greater room of a golf cart, engineering for minimum volume is not
so critical, but these results still indicate that reduced size and weight have not yet been
demonstrated
The sizes and weights with the hydrogen storage system included are listed in Table 4.15 below.
Both a current Ergenics metal hydride system and predicted FeTi performance are included, along
with a Dynetek cylinder and the ZES-2000 battery-powered scooter for comparison purposes.37
Table 4.15 Size of various storage designs
storage system
energy
stored
energy storage
(battery or H2)
complete
drive system
range at
30 km/h
(km)
range
under
TMDC
(km)
weight
volume
weight
volume
ZES-2000
lead-acid batteries
1.34
kWh
60
40*
44 kg
15 L
60 kg
24 L
DTI TiFe hydride
250 g H2
200
132
21 kg
4L
61 kg
43 L
Ergenics hydride
(aluminum body)
204 g H2
165
108
27 kg
14 L
67 kg
53 L
Dynetek cylinder
(compressed gas)
350 g H2
282
184
11 kg
31 L
51 kg
70 L
“complete drive system” refers to the motor and controller from Table
2.2 (15.5 kg and 9.1 L together), the fuel cell stack, auxiliaries, and
hydrogen storage.
* Note that the ZES-2000 can not actually sustain the TMDC, as it
lacks the power necessary for the high-speed accelerations; its range is
given as if it had could produce the required maximum power using its
current batteries. For comparison, according to an unspecified pattern
of “urban driving”, it is listed at a range of approximately 30 km;
reported data shows that it reaches 80 km on a single complete
discharge at 30 km/h.38
Also, although the ZES-2000 does not actually use the Unique
Mobility motor / controller system specified, these numbers are similar
to those of other motor/controller systems and were used in calculating
total size and weight for the battery-powered scooters.
210
This comparison has used the criteria of size and weight, but it should also be noted that cooling
loads are higher for compressed gas hydrogen storage systems, due to the lack of desorption
cooling. This results in larger cooling loads and higher parasitic demands from the cooling system,
not shown here. The results of Table 4.15 are discussed in the subsection following.
4.5.3 Evaluation
Average two-stroke scooters weigh 70 kg and up, but the overall ZES-2000 prototype mass is 105
kg. A Honda CUV-ES electric scooter weighs 130 kg. Fuel cell powered scooters would weigh
about as much as the CUV-ES; recall that one of the concerns in the consumer satisfaction survey
was the extra weight of the electric scooter. While this might remain an item of difficulty in terms of
handling while the system is off, the extra performance provided by the 6 kW fuel cell system
partially compensates for the high mass.
The fuel cell vehicle offers more than three times the range of the ZES-2000, with roughly the same
weight of drive system. The fuel cell systems requires 18 to 36 additional liters of storage space
over that of the ZES-2000. There is a helmet storage chamber in the design that could be
commandeered for additional fuel storage; this is approximately 10-15 L of space. Additional
volume would have to come from a redesigning of the scooter body to make more room available,
but note that current electric scooters already use redesigned large-capacity bodies.
Current laboratory-scale metal hydrides, as exhibited by the Ergenics system, are almost 30 L larger
than the ZES-2000 power system, so progress must be made in improving metal hydride technology.
The following charts describe the breakdown of size and weight for the various subsystems, for a
211
FeTi hydrogen storage system
Figure 4.19 Weights of subsystems
FC stack (10.9%)
TiFe storage (34.8%)
auxiliaries (29.1%)
motor/controller (25.2%)
Figure 4.20 Volumes of subsystems
TiFe storage (8.7%)
FC stack (18.3%)
motor/controller (21.4%)
auxiliaries (51.6%)
A subsequent chapter deals with the costs of the various systems, but the next section is a
discussion of whether operating the fuel cell at higher pressures (with consequent greater efficiency)
might improve performance and reduce the size of the fuel cell required.
212
4.6 Pressurized fuel cell option
One of the two options considered for improving the performance of the base design was to operate
the fuel cell above atmospheric pressure. To quantify this benefit, the higher voltage output
obtainable was compared to the additional power needed to drive the compressor. Also included was
the fact that a turbine running of the fuel cell exhaust would reduce the compressor load.
Calculations were based on a adiabatic, non-isentropic system with the 68% efficiency assumed
previously.
Power out for an adiabatic expander (or compressor):
γ
γ
γ P1 −1
− 1
= n ∫ V d P = n R T
γ − 1 P2
P1
P2
Pad iab atic
n is the flow rate in moles•s-1; R is universal ideal gas constant, 8.314 J·mol-1·K-1. The specific heat
ratio
is the ratio of Cp to Cv for the working fluid. Assumptions: the intake air is an ideal gas
consisting of nitrogen and 20.95% oxygen; the exhaust (mainly nitrogen, with some water vapour
and unused hydrogen) is an ideal gas with the stoichiometric amount of oxygen removed by the fuel
cell reaction, and no water vapour.
The benefit of compression is calculated using the two Energy Partners polarization curves
discussed previously (3 atm and atmospheric). The atmospheric power is net of blower parasitic
power, which is a linear function from 50 W to 250 W over the 5.9 kW operating range of the fuel
cell stack. The pressurized stack power is net of compressor and expander powers, as calculated for
a 68% efficiency expander and 68% efficient compressor. (same as the DOE’s projected automotive
213
system). Note that this fails to capture the lowered efficiency expected at lower-than-nominal flow
rates, while the blower’s linear function does effectively reduce efficiency at low flow rates.
The mechanical power required by the compressor is always greater than the mechanical power
recovered by the expander, and any supplement is calculated to be made up from the fuel cell at a
90% conversion efficiency from electricity to shaft power through an electric motor.
The figure below compares the power outputs at given current densities. Note that the powers, here,
are proxies for efficiency because at a given current density both options are operating on the same
amount of hydrogen per second.
Figure 4.21 Atmospheric power versus 3 atm power
7000
3 atm power, net
of compressor /
expander power
6000
5000
power (W)
atmospheric
pressure power,
net of blower power
4000
3000
2000
1000
0
0
200
400
600
800
1000
current density (mA/cm^2)
214
1200
The difference expands to 350 W at 1200 A•cm-2, so the pressurized fuel cell does have benefits at
high pressure. However, at the maximum operating point of 1000 A•cm-2, the advantage of the 3
atm fuel cell is only 105 W, or 1.8% of the gross power. Although there is an unquantified benefit
for the pressurized fuel cell due to the lesser need for humidification and easier water removal, the
gain is quite limited and the scooter cannot accept the additional cost and weight of an additional
turbocompressor system. Also, unlike the case for automobiles, the fuel cell stack size would not be
appreciably reduced by operating at 3 atm.
If the DOE goals of 3 kg and 4 L for a turbocompressor can be realized, and efficiency is good at
the under-10 g/s flow rates required in scooters, then the slight performance benefit will be worth a
small price premium.
4.7 Hybrid option design
A second optimizing option was considered. Due to the extremely irregular nature of scooter
driving, with average power only 8% of maximum power, peaking power provided by a secondary
power device might be an efficient way to reduce the fuel cell size. The fuel cell would provide
baseload power and charge up the secondary (peaking) power source for later use. A peaking power
device, which could be an advanced high-power battery, an ultracapacitor, or a flywheel, would also
enable regenerative braking.
The targets of hybridization, for the scooter are to:
215
1. Reduce system cost by reducing fuel cell size; peaking power batteries are cheaper than
additional fuel cell capacity, at least for the next several years.
2. Increase total fuel economy with regenerative braking and lower curb weight. For the same
hydrogen storage, greater fuel economy means greater range.
3. Improve vehicle handling by reducing curb weight.
The major drawback is that hybrid systems require more complex controls and power conditioning
systems. To optimize the system, a balance must be struck between secondary power source size
and main power source size. An important consideration is whether weight and volume of
auxiliaries will decrease significantly if a smaller fuel cell is used.
4.7.1. Types of hybrids
There are several types of hybrid vehicle designs. Most of these are defined for internal combustion
engines and used for the purposes of (i) increasing engine efficiency, (ii) reducing emissions, and
(iii) reducing cost if the main engine is expensive, like a fuel cell. Hybrids are divided into parallel
and series systems.
For a combustion hybrid, a parallel system allows both the engine and the peaking battery (via
electric motor) to drive the wheels using two separate systems. This system is more mechanically
complex due to the two driveshaft attachments, but allows simultaneous use of the battery and the
combustion engine.
216
In contrast, a series hybrid routes all power electrically. The combustion engine drives a generator
(not required in the parallel version), which can supply energy to the wheels and the peaking battery.
Output from the generator and the peaking battery go to the wheels. Generally, the engine can be set
to operate only at its most efficient speed, recharging the battery as needed and avoiding transients
which might produce more emissions.
The more specific operating policies listed below are from the DOE HEV program guidelines.39 The
word “battery” is used interchangeably with “secondary energy storage system,” although this
system could actually be a flywheel or ultracapacitor.
1. Thermostat series. The vehicle is run on battery power by default, with the primary power source
activated (at the optimum power level) when the battery drops below a predefined minimum. The
primary power source acts as a thermostat, turning on and off to maintain battery state of charge.
This policy allows the primary power source to run at its peak efficiency point.
2. Electric-assist parallel. The primary source provides power by default, and the battery is
discharged only in high-power situations. The battery is recharged by regenerative braking, but in
the case of combustion engine hybrids, cannot be recharged by the engine because there is no
generator. The primary power source can be sized below the maximum power required since it is
assisted by the battery.
3. Primary power-assist parallel. Identical to the previous system, but with the battery providing
power, and the internal combustion engine supplying peaking power. Very similar to a “rangeextender” system in that the battery functions as the main energy storage device.
217
4. Load-leveler series. The average power is matched to the output of the primary power source,
while logic attempts to maintain the state of charge of the battery at a predefined medium level.
Peaking power is supplied by the battery, but the primary power source load is allowed to vary as
well to maintain the state of charge of the battery. (This contrasts with the thermostat series option,
where the combustion engine stays at its optimal point).
The different policies show that the design is motivated by a desire to keep the combustion engine at
its optimum point (maximizing fuel economy and minimizing emissions) for the series options, and
reducing the engine size in the parallel options.
A fuel cell hybrid is technically a series hybrid because there is a single drive system (electric). The
fact that all power is electric means that, if designed properly, the control logic can allow both fuel
cell and battery to drive the wheels (like an electric-assist parallel), or let the fuel cell recharge the
battery (like a load-leveler series). Several axes of optimization remain - for example, how large to
make the fuel cell, when to use the battery and when to recharge it, and how large to make the
battery.
4.7.2. Fuel cell sizing
In order to calculate how to divide the power needs between the fuel cell and the battery, two criteria
were specified. First, maximum power (5.6 kW net of parasitics) had be sustained for at least ten
seconds. Users would be very disappointed to find that, after a few seconds of pushing their scooters
to the maximum output, power started to fall off because the battery was used up. Second, the
scooter had to sustain - indefinitely - the slope-climbing requirements. In other words, it had to
generate enough power to climb a 12( slope at 18 km/h.
218
The physical model showed that these results can be obtained with a fuel cell stack with a maximum
output of 3.2 kW gross power (3.0 kW net). Total cell active area was reduced by reducing the area
of each cell to 110 cm2, while the number of cells was kept the same in order to retain the 48 V
motor voltage. To reach the maximum power output level thus required the addition of a 2.6 kW
peaking battery. This is the first hybrid scooter case.
One other case was considered. This one is targets the short term where fuel cells are extremely
expensive and the fuel cell size must be minimized at all costs: a fuel cell sized to output only 1.0
kW of net power (1.1 kW gross power), with a large battery for the remaining 4.6 kW.
This scooter would not be capable of indefinitely sustaining hill climbing as specified at the
beginning of the chapter. The specified performances can only be achieved for under three minutes
until the batteries drain down to their 20% limit, and the continuously sustained hill climbing speeds
are lower than the requirements, as shown in the table below.
Limited amount of time at
this speed as battery is drained
140 s
70 s
Limited distance at this speed
380 m
360 m
4.7 km/h
5.8 km/h
Required performance
Sustainable maximum speed
(after battery depleted; fuel cell only)
Cooling load is significant in hybrid vehicle design as well as in the original pure fuel cell design;
again, the critical situation is hill climbing requiring 3.2 kW of continuous output. In the 5.9 kW
full-sized fuel cell, continuous operation at 3.2 kW gross output takes place at 51% efficiency, and
219
heat production is consequently low. A 3.2 kW maximum power fuel cell, however, is only at 43%
efficiency when it is continuously operated at 3.2 kW, producing more heat. The greater heat also
necessitates higher pump and fan loads for the radiator, meaning greater parasitic loads.
Parasitic power loads were calculated with an iterative process. The first step was to estimate a
parasitic load. Then the maximum gross output power including parasitics was calculated for hill
climbing. A maximum heat output was then solved, and the parasitic power required to sustain this
cooling load was calculated. The process was repeated until the results converged. In the example
here, after the results had settled down to stable values, the following results were obtained:
For the first hybrid, the maximum load was 3020 W, with parasitic power costs of 66 W for pump
and fan, and 50-250 W for the blower depending on load. The total fuel cell output was 3205 W at
this net power output, for a waste heat load of 4240 W. With hydride cooling at 17.2% at this
hydrogen flow rate, the remaining heat load was 3510 W, or 140 W/K at the pre-defined T of
25(C, which was supplied by fans and pumps with the total of 66 W.
The 1.1 kW system was easier to solve because the output was simply capped at 1.0 kW of net
output, because no attempt had to be made to find the total power at which the hill climbing
requirements could be met. A smaller M10-080 radiator was used. Under maximum conditions of
1.1 kW gross power, heat generation was 1430 W, or 1180 W after metal hydride desorption, for a
cooling load of 47 W/K.
220
Table 4.17 Hybrid fuel cell stack designs
Pure FC
Hybrid 1
Hybrid 2
Maximum gross power
5.9 kW
3.2 kW
1.1 kW
Stack current at maximum power
172 A
89 A
31 A
Efficiency at maximum current density
41.2%
43.2%
43.6%
Total active area needed
9600 cm2
5600 cm2
2000 cm2
Total number of fuel cells
56
56
56
Active area per cell
170 cm2
100 cm2
35 cm2
Maximum fuel cell heat generation
8.4 kW
4.2 kW
1.4 kW
Maximum heat generation
(after metal hydride cooling)
7.0 kW
3.5 kW
1.2 kW
Cooling factor needed at
maximum power ( T=25(C)
280 W/K
140 W/K
47 W/K
Maximum net fuel cell power
5.9 kW
3.0 kW
1.0 kW
–
2.6 kW
4.6 kW
Battery power needed
4.7.3 Peaking battery and operating policy
The peaking power battery model is described in section 2.2.3.1 and has the following properties.
Note that the specific power and power density are much higher for the battery than the fuel cells
(see Table 4.24), so that hybridization makes sense in reducing total weight and volume.
Table 4.18 Peaking power battery characteristics
Specific power at 43 second discharge
836 W/kg
Specific energy
17 Wh/kg
Power density by volume
853 W/L
Energy density by volume
17 Wh/L
The heat generated by the battery is not included in the system modeling. It is on the order of 10%
221
of the output power, low compared to the fuel cell. At maximum output power (4.6 kW for the 1.1
kW stack hybrid) this could be almost 500 W in addition to the fuel cell, but this heat load is
infrequently reached. The operating policy of the battery is defined as follows.
1. The state of charge of the battery is allowed to vary between 80% and 20% over the driving
cycle. The state of charge is kept away from these “hard limits” in order to reduce the risk of large
excusrions (overcharging or draining to zero), either of which could permanently damage the
battery.
2. The state of charge is set at the beginning of the driving cycle is set to 50%
3. The battery is activated (discharged) whenever the fuel cell maximum power is not enough for the
driving cycle power plus auxiliary and parasitic loads, and makes up the entire difference.
4. Regenerating always recharges the battery as long as the maximum charging current is not
exceeded, up to the 80% limit
5. The battery is charged up from the fuel cell at a specified rate whenever (i) the state of charge
dips below 55% and (ii) power demand at that instant is less than 400 W. The charging rate is equal
to 400 W minus the instantaneous power demand from the wheels, auxiliaries, and parasitics.
4.7.4 Simulation results
First, fuel economy was determined under steady-state, 30 km/h driving conditions.
222
Table 4.19 Hybrid performance at 30 km/h
5.9 kW
pure FC
3.2 kW
hybrid
1.1 kW
hybrid
fuel cell conversion efficiency
58.5%
56.4%
50.0%
average fuel cell output power
725 W
751 W
710 W
fuel economy in terms of hydrogen
0.807 km/g
0.751 km/g
0.703 km/g
“on-vehicle” fuel economy
522 mpge
486 mpge
455 mpge
change from pure FC
–
-4%
-15%
hydrogen for 200 km
248 g
266 g
284 g
Parameters
Efficiency decreased as the fuel cell size decreased, because the smaller fuel cells were operating
closer to their maximum output. For the results below, the hydrogen storage system was scaled
linearly to keep range at 200 km. The driving cycle was more interesting because its decelerations
created the possibility of regenerative braking.
Table. 4.20 Hybrid performance under TMDC
Parameter
5.9 kW
pure FC
3.2 kW
hybrid
1.1 kW
hybrid
Fuel cell maximum power
5.91 kW
3.24 kW
1.11 kW
–
2.61 kW
4.63 kW
Average total power output,
battery + fuel cell
674 W
709 W
726 W
Average fuel cell power
674 W
698 W
577 W
344
316
343
–
-8.2%
-0.3%
56%
53%
47%
Braking energy absorbed
–
62 kJ
82 kJ
Average power absorbed
–
65 W
86 W
Braking energy recovered as a fraction
of theoretical maximum braking losses
–
51%
68%
Battery maximum power needed
Fuel economy (mpge)
Change in fuel economy
Overall conversion efficiency
223
The 1.1 kW fuel cell puts out 100 watts less average power than the 5.9 kW fuel cell, because the
86 W (average) that is recovered from regenerative braking can supplement the peaks. However,
conversion efficiency from hydrogen heating value to electricity is lower at 47%, because the fuel
cell is operating more frequently near its maximum power (minimum efficiency). More hydrogen
must be used to produce the 577 W of power than if the larger, pure fuel cell were used with the
same batteries. The result is a net fuel economy that is the same as the pure fuel cell.
The 3.2 kW hybrid actually consumes more power in the TMDC due to the higher parasitic losses,
and for the same reason of operation near maximum power given above.
The regenerated fractions of 51% and 68% are close to the 70% theoretical maximum that can be
regenerated. The differences are due to the limiting maximum charging rate and the fact the at some
points the battery is already full when regenerative braking is possible. In terms of total power,
though, these figures are equivalent to 9% – 14% of the total fuel cell output.
224
Figure 4.22 Division of power between fuel cell and
battery during TMDC, 3.2 kW stack
In this diagram, the dark area above the horizontal line represents the
energy supplied by the peaking power battery, while the light gray area
below the line is the energy supplied by the fuel cell. The same scheme
is used in Figure 4.24.
225
Figure 4.23 State of charge of battery over TMDC, 3.2 kW stack
The careful observer will note that battery energy does not return to initial levels over the driving
cycle - in fact, there is a net gain from regeneration of 58.3 kJ, or an average of 60 W over the 950
second cycle. There is a not insignificant amount of surplus energy that could be applied to the
wheels, if the battery were employed using a more sophisticated policy (for example, using the
battery even if power demand is not greater than the maximum fuel cell output), and fuel economy
would be slightly improved.
226
Since the object was to minimize component cost and not necessarily produce the maximum
mileage, the design of more efficient hybrid strategies will have to await future studies.
Figure 4.24 Division of power between fuel cell
and battery during TMDC, 1.1 kW stack
227
Figure 4.25 State of charge of battery over TMDC, 1.1 kW stack
4.7.5 Hybrid power system designs
The subsystem performance requirements and sizes/weights are listed for the two hybrid designs,
with the base pure fuel cell system for comparison.
228
Table 4.21 Hybrid system design
Parameter
5.9 kW
pure FC
3.2 kW
hybrid
1.1 kW
hybrid
Maximum hydrogen flow rate
2.6 cfm
1.4 cfm
0.5 cfm
Maximum air flow rate
15.6 cfm
8.1 cfm
2.8 cfm
Worst-case cooling requirement
110 W/K
150 W/K
50 W/K
Cooling fan power requirement
14 W
28 W
4W
Coolant pump power requirement
25 W
38 W
21 W
Fuel cell stack weight
6.7 kg
5.4 kg
4.1 kg
Fuel cell stack size
7.8 L
5.3 L
3.2 L
Fuel cell cost (manufacturing cost)
$220
$160
$140
Battery maximum power needed
–
2.6 kW
4.6 kW
Battery weight
–
3.1 kg
5.6 kg
Battery size
–
3.0 L
5.4 L
Battery cost (retail)
–
$195
$340
The same DTI model was used to calculate fuel cell stack sizes, weights, and costs. The designs are
described in greater detail below.
4.7.5.1 Design for 3.2 kW fuel cell
For the 3.2 kW fuel cell, fuel economy actually goes down with hybridization. There are three
reasons: (i) the cooling system parasitic load is larger (ii) the fuel cell is operating more frequently
near its maximum load (iii) the battery is rarely used to output energy.
At maximum cooling load and 50% efficiency, the cooling fan runs at 500 cfm (0.24 inches of water
pressure drop) and uses 28 W of power while the pump pushes coolant at 1 gpm (2.5 psi pressure
drop) and uses 38 W. Thus, the base load power needed for the cooling system is 66 W.
229
The lower air intake flow requirements might make a smaller and cheaper blower feasible.
With auxiliaries the power system totals are 21.9 kg and 26.6 L, for power densities of 0.15 W/kg
and 0.12 W/L.
Table 4.22 Component breakdown for 3.2 kW scooter
Component
Weight
Volume
Long-term cost
Fuel cell stack
5.4 kg
5.3 L
$165
Radiator and fan
8.9 kg
15.2 L
$60
Coolant pump
1.0 kg
1.2 L
$10
Blower
2.7 kg
2.9 L
$110
plumbing,
wiring, etc.
2.0 kg
2.0 L
$50
coolant water
0.9 kg
–
–
peaking power battery
3.1 kg
3.0 L
$195
TOTAL STACK
WITH AUXILIARIES
25.1 kg
29.6 L
$590
motor
11.4 kg
5.0 L
$125
controller
4.1 kg
4.1 L
$150
hydride for 266 grams H2
22.8 kg
4.0 L
$200
TOTAL DRIVE SYSTEM
63 kg
43 L
$1065
Due to the lower fuel economy of the 3.2 kW system at 30 km/h, slightly more metal hydride is
needed: 266 grams for 200 km. This requires a 22.8 kg system taking up 4.0 L of space.
The total size and weight are very close to the original, 5.9 kW fuel cell. Mainly this is because the
subsystem sizes remain the same even though the fuel cell proper becomes smaller. Smaller blowers
and coolant pumps could be used, reducing weight slightly, but the largest subcomponent (the
radiator) has to stay large in order to handle the higher cooling factor of 140 W/K.
230
4.7.5.2. Design for 1.1 kW fuel cell
In the second hybrid, the battery is used much more often and contributes frequently to the total
power output. However, for the same reason, the battery must be frequently replenished, and due to
the small size of the 1.1 kW fuel cell, the fuel cell often runs near its inefficient maximum output.
The net fuel economy is almost identical to that of the pure fuel cell.
Cooling is different. The system is sized for cooling when it is running flat out at the maximum 1
kW of net output, rather than for sustained (3020 W) hill climbing. After metal hydride cooling, the
worst-case cooling load is less than 50 W/K. This is achievable with a fan at 140 cfm with 0.125
inches of water air pressure drop (4 W of power), and a pump at 1 gallon per minute with 0.5 psi
water pressure drop (21 W of power), for a total of 25 W. The lower cooling load of 50 W/K means
that a smaller radiator can be used here – the Lytron M10-080.
With auxiliaries, the power system totals are 21.9 kg and 26.6 L, for power densities of 0.15 W/kg
and 0.12 W/L.
231
Table 4.22 Component breakdown for 1.1 kW scooter
Component
Weight
Volume
Long-term cost
Fuel cell stack
4.1 kg
3.2 L
$135
Radiator and fan
3.1 kg
2.3 L
$30
Coolant pump
1.0 kg
1.2 L
$10
Blower
2.7 kg
2.9 L
$110
plumbing, wiring, etc.
2.0 kg
2.0 L
$50
coolant water
0.6 kg
–
–
peaking power battery
3.1 kg
3.0 L
$340
TOTAL STACK
WITH AUXILIARIES
20.1 kg
17.0 L
$675
motor
11.4 kg
5.0 L
$125
controller
4.1 kg
4.1 L
$150
hydride for 285 grams H2
24.3 kg
4.3 L
$215
TOTAL DRIVE SYSTEM
60 kg
30 L
$1165
Due to the lower fuel economy of the 1.1 kW system at 30 km/h, slightly more metal hydride is
needed: 284 grams for 200 km. This requires a 24.3 kg system taking up 4.2 L of space.
The extra battery size is compensated for by the smaller fuel cell, while the smaller radiator makes a
large difference in reducing the total volume.
4.7.5.3 Hybrid zinc-air scooters
If fuel cells can be hybridized with peaking power batteries, why not battery powered scooters?
Regenerative braking would extend the range of the vehicle, while separating the power and energy
functions might allow a smaller, less expensive power system. In the fuel cell, energy and power are
decoupled. The fuel cell engine is sized for a maximum power output, and the amount of hydrogen
232
carried can be varied to meet range requirements. With the battery, a larger power necessarily
corresponds to a larger energy storage, and sizing for power may result in overdesigned energy (or
vice versa). The hybrid battery system reduces this dependence by using batteries with two different
power-to-energy relationships.
The hybrid battery design presented here uses zinc-air batteries (high energy density) for range, plus
the peaking power lead-acid batteries (high power density) described above. The two sets of
batteries operate in the same way as the fuel cell hybrid, so that the peaking power of the high
power lead-acid batteries can be added to the zinc-air baseload battery’s power, and the peaking
batteries can be used to absorb regenerative braking energy.
The power and energy requirements are specified so that the scooter has enough power to sustain
the maximum accelerations of the TMDC, and enough energy to drive 200 km at a constant speed
of 30 km/h (or 145 km under the TMDC). 5.6 kW of output power and 4.1 kWh of energy storage
are needed, and the following results are obtained:
Table 4.24: Hybrid battery configuration for Taiwan scooter model
Zinc-Air
baseload battery
Spiral lead-acid
peaking battery
Total
Weight
20.5 kg
4.5 kg
25.0 kg
Energy Capacity
4.1 kWh
0.07 kWh
4.1 kWh
Peak Power
1.8 kW
3.8 kW
5.6 kW
Specific Energy
200 Wh/kg
17 Wh/kg
168 Wh/kg
Specific Power
90 W/kg
836 W/kg
210 W/kg
27.1 L
4.4 L
31.5 L
80 $/kWh
75 $/kW
–
$330
$285
$615
Parameter
Volume
Long term massproduction price
Total price
233
The zinc-air volumetric energy density is calculated to be 150 Wh/L from published data; cost for
high-power lead acid battery is estimated at 166 $/kW based on a scaled up version of Bolder
batteries.40 This is reduced to 75 $/kW in mass production (The Rebel battery’s density is
approximately 1 kg/L and the future goal is $150/kWh for long term USABC batteries and
advanced lead-acid batteries. Somewhat arbitrarily, power systems are assumed to be half as
expensive per kilowatt as base load batteries.41)
Note that the continuous power output of 1.8 kW is not sufficient for the continuous hill climbing
needs; rather, the baseload battery is sized for the 200 km range and 4.1 kWh required. Thus, the
hybrid battery vehicle is more comparable to the 1.1 kW fuel cell system than the 3.2 kW system.
(The Electric Fuel company has its own, more conservative hybrid battery scooter theoretical
design; it is specified for 4 kW maximum power and 3.0 kWh energy storage, for a claimed range of
200 km. This 3.0 kWh storage is underspecified, according to the calculations earlier in this
chapter. Their power source consists of two 8 kg zinc-air batteries, each 9.9 L, 7.5 kg, and with an
output of a maximum of 0.75 kW. The peaking power source must thus supply 2.5 kW, and is not
described in the Electric Fuel report although it likely will be a NiCd battery.42
4.7.6 Hybrid results
The hybridization study reveals that peaking power batteries do not significantly reduce the mass
and volume of the fuel cell system, because the auxiliary cooling and fluid management systems
require a certain minimum space and mass which does not decrease rapidly with size.
Fuel economy did not improve significantly, due to the fact that with a 3.2 kW fuel cell, only a few
234
excursions required use of the battery and parasitic loads were higher than for the pure fuel cell.
The 1.1 kW version produced fuel economies equal to the original pure fuel cell because of its lower
efficiencies; on the other hand, it reduced the size of the fuel cell stack by a factor of more than five,
which is important for the short term cost.
Table 4.25 Hybrid power system summary
5.9 kW
fuel cell
3.2 kW
hybrid
1.1 kW
hybrid
1.8 kW battery
hybrid
ultimate stack cost
$244
$161
$124
–
stack cost/kW
$42
$47
$103
–
stack size
7.8 L
5.3 L
3.2 L
–
stack weight
7.6 kg
5.4 kg
4.0 kg
–
stack with auxiliaries, size
(excl. peaking battery)
29.8 L
26.6 L
11.6 L
–
stack with auxiliaries, weight
(excl. peaking battery)
24.6 kg
21.9 kg
14.6 kg
–
baseload battery
(zinc-air) volume
–
–
–
27.1 L
baseload battery
(zinc-air) weight
–
–
–
25.0 kg
power source
power density
0.20 kW/L
0.12 kW/L
0.09 kW/L
0.07 kW/L
power source
specific power
0.24 kW/kg
0.15 kW/kg
0.08 kW/kg
0.07 kW/kg
In terms of the useful performance measurements of fuel economy total power system size/weight
(including fuel cell or baseload battery, peaking power battery, and hydrogen storage, but not motor
and controller), the following results are obtained:
235
Table 4.26 Performance metrics
5.9 kW
fuel cell
3.2 kW
hybrid
1.1 kW
hybrid
1.8 kW battery
hybrid
21.4 kg
22.8 kg
24.3 kg
–
3.7 L
4.0 L
4.3 L
–
peaking power battery weight
–
3.1 kg
5.6 kg
4.5 kg
peaking power battery size
–
3.0 L
5.5 L
4.4 L
total drive system weight
61 kg
63 kg
60 kg
45 kg
total drive system size
43 L
43 L
30 L
41 L
30 km/h fuel consumption
0.807 km/g
0.751 km/g
0.703 km/g
48.8 km/kWh
TMDC fuel consumption
0.527 km/g
0.484 km/g
0.525 km/g
36.2 km/kWh
30 km/h fuel economy
522 mpge
486 mpge
455 mpge
1053 mpge
TMDC fuel economy
344 mpge
320 mpge
363 mpge
780 mpge
Range at 30 km/h
200 km
200 km
200 km
200 km
Range under TMDC
131 km
129 km
149 km
148 km
hydrogen storage (FeTi hydride)
weight for 200 km range
hydrogen storage
(FeTi hydride) volume
“total drive system” includes fuel cell and hydrogen storage, or zincair battery; motor / controller; peaking power battery. Costs are
discussed in the next chapter.
Note that the ZES drive system weight is 44 kg of lead-acid batteries plus 15.5 kg of motor and
controller (assuming similar characteristics as the UQM motor/controller). This is about 60 kg, the
same as the fuel cell systems.
System size does not decrease significantly with hybridization except in the case of the 1.1 kW
hybrid where much lower cooling requirements mean a much smaller radiator. The smaller fuel cell
weight is roughly compensated for by the added weight of the batteries, and the larger hydrogen
storage tank. TMDC performance is better for the 1.1 kW hybrid because the frequent decelerations
allow regenerative braking gains; the same is true for the 1.8 kW battery hybrid.
236
On the other hand, certain other factors decrease more or less linearly with fuel cell size, foremost
among them membrane area and platinum cost. Significant cost reductions might be possible under
such a system for the near future while fuel cells remain extremely expensive; unfortunately, these
costs are not reflected in the price calculations above, which rely upon ultimate price estimates for
mass-produced stacks.
More complex hybrid operation policies could be used to try to better predict power demands; these
would optimize battery usage by controlling more cleverly when to recharge, how quickly to
recharge, and the state of charge to maintain in the battery.
4.7.7 Near-term possibilities
To compare these designs with what is available today, portable stacks from H-Power and Ballard
were inserted into the 1 kW fuel cell design. A single PS-250 fuel cell unit commercially available
from H Power produces 250 W net power, weighs 10.3 kg, and has a volume of 16 L.43 This does
not include storage. (The system is air-cooled, and with the correct geometry this could be possible
for a scooter version) The existing design can actually output 330 W; to be conservative, retail units
are sold derated to 250 W. Using three units to supply the 1.0 kW net output desired produces a
power system weight of 31 kg and volume of 48 L. Currently costs are on the order of $6,000 for a
unit like the PS-250, with mass-production costs expected to drop to $1,000.44 For the required
three, then, the current cost would be on the order of $18,000 and $3,000 in the long run.
On the other hand, Ballard Power Systems has developed a 1 kW (net power) stack that weighs 18
kg and has a volume of 33 L including all packaging.45
237
Table 4.27: Near term 1 kW fuel cell hybrid designs
Parameter
as designed
Ballard stack
H-Power stack
fuel cell stack weight
4.1 kg
–
–
fuel cell stack volume
3.2 L
–
–
stack power density, gravimetric
95 W/kg
–
–
stack power density, volumetric
76 W/L
–
–
fuel cell system weight
14.6 kg
18 kg
31 kg
fuel cell system volume
11.6 L
33 L
48 L
fuel cell system power density, gravimetric
75 W/kg
61 W/kg
35 W/L
fuel cell system power density, volumetric
65 W/L
33 W/L
23 W/L
The fuel cell system size and weight exclude peaking power battery in
all cases.
So in the short term, an unoptimized fuel cell hybrid based on Ballard performance figures would
have a drive system at 63 kg and 51 L. This is 3 kg and 27 L more than the ZES-2000, which is
technically feasible in a scooter.
The results here for the various types of hybrids show that they offer the opportunity to reduce fuel
cell stack size and power, and thus save money, since peaking power batteries are currently
cheaper (per kW) than additional fuel cell capacity. In the short term, hybrid scooters are
overwhelmingly favoured. The first steps to take would be to reduce volume by optimizing the
subcomponents for the scooter body.
However, in the long run, the decreasing cost of fuel cell stacks is expected to make hybrids less
and less economical. A larger fuel cell also improves fuel economy at constant driving speeds,
since the fuel cell is more frequently operating at its most efficient output levels.
238
A calculation of costs, both for the scooter and for the fuel it uses, is found in the next chapter.
Infrastructure and overall fuel economy are also discussed.
References for Chapter 4
1. Jet P. H. Shu, Wei-Li Chiang, Bing-Ming Lin, Ming-Chou Cheng. Mechanical Industry Research
Laboratories, Industrial Technology Research Institute. “The Development of the Electric Propulsion
System for the ZES2000 in Taiwan”. (Date unknown; not the same as the SAE paper in reference 7, even
though it appears to be written by the same authors. October 1997 or later.)
2. Yoshihiro Nakazawa, Chiaki Humagai, Mikio Kato. “Development of an electric scooter for practical
use” in JSAE Review 15 (1994) 373-377
3. Chien-Tung Liu, Bing-Ming Lin, Jyh-Sheng Pan. “Design and development of a zero-emission scooter
for Taiwan” in Journal of Power Sources 59 (1996) 185-187
4. Peter A. Lehman, Charles E. Chamberlin. “Design and Performance of SERC’s Fuel Cell Powered
Vehicle Fleet” in Fuel Cell Seminar Abstracts 1998. 714-717
5. Mowick Limited Golf & Leisure Vehicles. “E-Z Go 4Caddy” http://www.mowick.com/ezwhite.htm
Accessed July 30, 1999
6. Chien-Tung Liu, C. C. Kuo, Jyh-Sheng Pan, Bing-Ming Lin. “Development of electric motor cycle
technologies in Taiwan” J. Power Sources 48 (1994) p. 244
7. P. H. Jet Shu, Wei-Li Chiang, Bing-Ming Lin, Ming-Chou Cheng. “The Development of the Electric
Propulsion System for the Zero Emission Scooter in Taiwan”. (1997) SAE 972107
8. ibid
9. Highway Tire Committee, SAE. Revised by the Rolling Resistance Subcommittee. “Measurement of
Passenger Car, Light truck and highway truck and bus tire rolling resistance” (March 1997) SAE
Information Report J1270. Section 8.1
10. Frank Rowland Whitt, David Gordon Wilson. Bicycling Science (MIT Press: Cambridge, 1974) p. 93
11.Wei-Li Chiang, Power Machinery Division Director, ITRI-MIRL (Industrial Technology Research
Institute, Mechanical Industry Research Laboratories. Personal communication, September 11 1998
12. CALSTART web site. “DARPA Consortia EV and Hybrid EV Technology Projects - (PNGV) Target
Attributes” http://www.calstart.org/about/pngv/pngv-ta.html
13. Energy and Environment Analysis. “Analysis of Fuel Economy Boundary for 2010 and Comparison to
Prototypes” (November 1990) Prepared for Martin Marietta Energy Systems, Contract No. 11X-SB0824.
239
p. 4-11.
14. M. Ross and W. Wu. “Fuel Economy Analysis for a Hybrid Concept Car Based on a Buffered FuelEngine Operating at a Single Point” SAE Paper 950958, presented at the SAE International Exposition,
Detroit, MI, February 27 - March 2 1995
15. C. E. (Sandy) Thomas, Brian D. James, Frank Lomax, Ira F. Kuhn, Jr. Directed Technologies,
Inc.“Integrated Analysis of Hydrogen Passenger Vehicle Transportation Pathways”, draft final report. For
NREL, subcontract AXE-6-16685-01. March 1998. p. 61
16. Bernward E. Bayer, “Motorcycles”. Chapter 10 of Aerodynamics of Road Vehicles. Fourth Edition.
Ed. Wolf-Heinrich Hucho. p. 502
17. Arne LaVen. “Driving Cycle Analysis of the Sun ComTM Battery Electric Scooter” (November 1998)
Desert Research Institute. p. 26
18. Wei-Li Chiang, Power Machinery Division Director, ITRI-MIRL. Personal communication,
September 11 1998
19. Owner’s manual for Honda CH250 Elite250 1986, (Honda Motor Company: 1985) p. 68
20. T. C. Pong, personal communication. October 29, 1998.
21. Arne LaVen. Energy and Environmental Engineering Center, Desert Research Institute. “Driving
Cycle Analysis of the Sun ComTM Battery Electric Scooter” November 1998
22. Chien-Tung Liu, Bing-Ming Lin, Jyh-Sheng Pan. p. 186
23. Toshiharu Sawada, Minoru Wada, Masanori Noguchi, Buhei Kobayashi. Komatsu Zenoah Co.
“Development of a Low Emission Two-Stroke Cycle Engine” SAE 980761 (1998)
24. C.T. Liu, C. C. Kuo, J. S. Pan, B. M. Lin.“Development of electric motor cycle technologies in
Taiwan”
25. Christopher S. Weaver, Lit-Mian Chan.“Motorcycle Emission Standards and Emission Control
Technology” Revised Final Report, submitted toThe World Bank. (Engine, Fuel, and Emissions
Engineering, Inc.: August 1994) page v.
26. Frano Barbir. Energy Partners. “Operating Pressure and Efficiency of Automotive Fuel Cell Systems”
No date. 1997 or later.
27. Vicor product data sheet. “The MegaPAC family” http://www.vicr.com/pdf/ds_megapac.pdf Accessed
July 9, 1999
28. Vicor sales representative. Personal communication. July 14 1999.
29. Lisa Fawcett, AMETEK. Personal communication. April 26 1999.
30. Charles Chamberlin, Schatz Energy Research Center, Personal communication. May 20 1999
31. Mazda Australia web page. “Mazda”. http://www.mazda.com.au/corpora/460.htm. Accessed June 17,
1999
240
32. If the entire 0.9 g/s were vaporized (with a latent heat of vaporization of 2.2 kJ/g for 100(C water)
approximately 2 kW of cooling would be realized, or about 25% of the total heat generated.
33. Lytron product catalog, “Standard OEM Coils” at http://www.lytron.com/catalog/oemframe.htm. Last
accessed April 11, 1999
34. Charles E. Chamberlin and Peter A. Lehman. “Design and Performance of SERC’s Prototype Fuel
Cell Powered Vehicle”
35. National Research Council. Review of the Research Program of the Partnership for a New Generation
of Vehicles. Second Report (National Academy Press, Washington: 1996) p. 53
36.Susan Ornelas, Research Engineer, Schatz Energy Research Center. Personal communication February
17 1999.
37. No similar size and weight data was available for the CUV-ES scooter. Also, the Dynetek cylinder is
one that holds about 40% more hydrogen than the DTI model, as no Dynetek cylinder currently exists for
the designed 250 gram size.
38. Jet P. H. Shu et al, “The Development of the Electric Propulsion System for the ZES2000 in Taiwan”
39. Department of Energy Hybrid Electric Vehicle Program. “Energy Management and System Control”.
http://www.hev.doe.gov/components/energman.html. Accessed May 10, 1999.
40. Bolder “Rebel” battery pack product sheet; see specifications in section 2.2.3.1
41. ALABC [Advanced Lead-Acid Battery Consortium] web site. “About ALABC”
http://www.alabc.org/about.html
42. Jonathan Whartman, Ian Brown. Electric Fuel, Ltd. “Zinc Air Battery-Battery Hybrid for Powering
Electric Scooters and Electric Buses” November 1, 1998
43. Rene Dubois, H Power. Personal communication June 7 1999
44. Arthur Kaufman, H Power, personal communication May 21 1999
45. Ballard product data sheet. “1 kW fuel cell generator”
241
~
Chapter Five
Implementation and
Conclusions
~
242
The main purpose of the thesis is to analyze the technical merits of the proposed fuel cell scooter.
However, the scooter will never be practical unless it can be affordable as well. This chapter
attempts to assess various issues contributing to the eventual adoption of commercial fuel cell
scooters: price of the scooter itself; operating costs, in terms of fuel; and a brief survey of
infrastructure and hydrogen distribution. The results of this chapter and the previous chapters is
summarized in the final section.
5.1 Scooter cost
This first section of this chapter estimates costs of the fuel cell scooter after commercialization
and mass production. This identifies the long term prospects of the hydrogen scooter, which are
fair.
Three different types of prices must be defined. The cost-to-manufacture is the cost most often
quoted by developers of new technology like zinc-air batteries. Estimated at double this
manufacturing cost is the cost of the component as part of the vehicle – this is defined as the sale
or retail vehicle price.1 Finally, after-market replacement automotive components or those
developed for separate purposes (like industrial blowers or radiators) are four times the
manufacturing cost. Note that, depending on the part in question and the particulars of
manufacturing and marketing and distribution, this factor of four may be quite different, so the cost
totals should be taken with a grain of salt.
The method used here is to take existing scooter prices, remove the cost of the power system, and
add back the new hydrogen fuel cell power system.
243
5.1.1 Base cost by subtraction
Current 50 cc two-stroke scooters retail for approximately $1,000 in Taiwan (1999), while the
ZES-2000 has a target sale price of $1,850.2 Costs for the major power system components are
shown below and used to estimate a “base” cost for the vehicle body (with wheels and frame and
controls and electronics) plus assembly.
Table 5.1 Internal combustion engine scooter parts
Part
Description
Cost
50 cc two-stroke engine
includes carburetor, transmission
$300
Exhaust system
Muffler, exhaust pipe
$60
Fuel tank
5 L of fuel
$40
Starting battery
Yuasa-Exide
$10
TOTAL
$410
Prices are retail prices listed by an American parts dealer3, and thus
divided by two for cost as part of the complete vehicle sale price. The
sole exception is the engine cost, which is based on a $150
manufacturing cost and a factor of two multiplier for the price it would
cost as part of a vehicle’s total sale price.4
Subtracting the total from a two-stroke’s $1,000 sale price leaves a $590 base cost for the vehicle
shell, assembly, internal electronics, controls, and assorted other ancillaries.
To verify this result by comparing with the cost of existing electric scooters, the most important
electric scooter components – battery, motor, and controller – are added. With lead-acid batteries
currently costing about $80/kWh retail, and a typical electric scooter having a stored energy of 1.2
kWh for 50 km of range at 30 km/h, a total battery cost of $96 is obtained. As previously quoted,
Unique Mobility predicts motor and controller prices of $250 and $300 respectively for mass
244
production today, although this is expected to decrease over time.5 This is assumed to be in-vehicle
retail cost.
Table 5.2 Battery-powered electric scooter parts
Part
Description
Cost
DC motor
UQM brushless SR121/1.5 L
$250
Controller
UQM CD05-100A
$300
lead-acid batteries
1.2 kWh
$100
“base” cost
retail
$590
TOTAL
$1,240
This electric scooter retail total is close to cost of the cheapest electric scooters, which now retail
for about $1,500, and is lower than the $1,850 previously quoted for the ZES-2000. The difference
is likely due to the current cost of electric motors.
5.1.2 Cost of hydrogen storage system
Metal hydride hydrogen systems are based on the calculations of Chapter 3. The long-term cost
projection for the FeTi metal hydride was based on an $8.80/kg materials price, and an extra factor
of 50% for systems and packaging.6 This resulted in a storage cost of $0.75 per gram of hydrogen
stored. The quantity of hydrogen, as previously discussed, was set to allow 200 km of travel at 30
km/h.
Table 5.3 Metal hydride storage costs
5.9 kW
3.2 kW
1.1 kW
Hydrogen stored
248 g
266 g
285 g
Manufacturing
cost of hydride
$187
$201
$215
245
5.1.3 Fuel cell system cost based on parts predictions
The cost estimate of Appendix B, first described in section 3.1.3, was repeated for the two other
hybrid alternatives. For each of the three fuel cell options and the hybrid battery, a long term price
prediction is made based on mass production of all parts – especially the fuel cell and metal
hydride storage system. These cost predictions are difficult to make, given the uncertainty involved
with this emerging technology and the several years before the mass produced prices can be
realized, so should be treated as a rough estimate only.
Retail prices for industrial parts like blowers, radiators, starting batteries, and coolant pumps are
divided by a factor of four to include them in the sum below. Other costs, like the fuel cell, zinc-air
battery, and metal hydride, are already given as cost-to-manufacture. Peaking battery costs are
assumed to decrease to 75 $/kW.
The fuel cell stack costs are for the ultimate long-term prices predicted by an automobile analysis
done by Directed Technologies, Inc. These estimates are valid for the larger membrane sizes of 170
cm2 and 100 cm2 for the 5.9 kW and 3.2 kW stacks, but the 1.2 kW stack has membranes only 34
cm2 in area, less than the minimum of 116 cm2 employed in the DTI study, so the cost estimates
are much less certain for this size. More details are in Appendix B. In contrast, the zinc-air battery
cost was based on an 80 $/kWh projection for “large-scale production” made by Electric Fuel.7
The component parts are added to the base vehicle cost of $590 retail (or $295 to manufacture) in
the chart below.
246
Table 5.4 Long-term scooter cost to manufacture
Part
Description
pure FC
(5.9 kW)
hybrid
(3.2 kW)
hybrid
(1.2 kW)
DTI model; long-term cost
$220
$165
$135
Yuasa-Exide
$10
Hydrogen storage
DTI metal hydride
model; long-term cost
$190
$200
$215
Storage batteries
Zn-air, 4.1 kWh
Heat exchanger
Lytron M10-080
FC stack
Starter battery
zinc-air
hybrid
$330
$30
Lytron M14-120
$60
$60
generic
$10
$10
$10
Ametek 116628-E
$110
$110
$110
Water, air pipes
$50
$50
$50
DC brushless motor
UQM SR121/1.5L
$125
$125
$125
$125
Controller
UQM CD05-100A
$150
$150
$150
$150
Bolder lead-acid
–
$195
$340
$285
body and misc. parts
$295
$295
$295
$295
TOTAL
$1,220
$1,360
$1,455
$1,185
Coolant pump
Blower for 1-2 psi
Plumbing
Peaking battery
Vehicle shell
The hybrid battery appears to be a very competitive option in terms of capital cost, if Electric
Fuel’s predictions of long-term zinc-air battery cost are correct. Due to the low fuel cell prices
predicted for the long run, the (relatively) high expense of peaking power batteries eradicate the
benefit of smaller fuel cells. Right now this is certainly not the case, as peaking power batteries
currently sell for approximately $166 per kW, while fuel cells are as high as $3000 per kW.
The greatest uncertainties in these cost estimates are in the most important components: the fuel
cell stack itself, the metal hydride storage unit, and in the case of the electric hybrid, the zinc-air
battery. The peaking power lead-acid battery is also relatively novel technology, although it has
been demonstrated in prototype vehicles. The other components are all “off-the-shelf” industrial
247
parts and are not expected to decrease dramatically due to advancing technology. On the other
hand, better engineering integration and design specific to the scooter application might reduce
costs.
Doubling the costs gives a rough estimate of retail cost - the price for the consumer.
Table 5.5 Summary of cost estimates
type
price
pure FC (5.9 kW)
$2,440
hybrid (3.2 kW)
$2,720
hybrid (1.1 kW)
$2,910
hybrid battery
$2,370
There is some room for cost reductions in the motor and controller prices of $250 and $300,
respectively. The peaking power battery price may also drop further, while there is great
uncertainty in the zinc-air battery price. However, the fuel cell cost is quite low and it is unlikely
that it could go much lower. In addition, the fuel cell stack itself makes up a relatively small
portion of the total cost.
The results suggest that margins will be very low since the manufacturing cost is very close to the
current prices for small scooters. The scooter as designed, with 5.9 kW of output, might be better
targeted against the low end of the 125 cc scooter range, rather than the 50 cc two-stroke scooter.
This would give more freedom in terms of higher sale price and larger size to store the various
subcomponents. By the same token, resizing for smaller-power and lower-performance scooters
(say 3 kW) would bring a fair cost reductions – a simple calculation shows that the 3.2 kW hybrid
scooter stripped of its peaking power batteries would cost $1160 to make. While it would be able
248
to perform the basic performance criteria of acceleration and hill climbing, it would not be as quick
to accelerate as comparable two-strokes.
5.1.3.1 The short term
In the short term, hybridization with peaking power batteries drastically reduces the price of the
scooter. In the long run when fuel cells are less expensive, the added complexity of batteries (and
their lack of performance advantage over comparably-sized fuel cells) make them unnecessary.
However, there is an intermediate stage as the price of the fuel cell drops to meet the cost of the
peaking power batteries, where at a rough estimate, fuel cells will cost about $500/kW and
batteries are $100/kW. The hydrogen storage and Zn-air batteries might be twice as much as the
ultimate costs.
Table 5.6 Short term bridging to the future
pure FC
(5.9 kW)
hybrid
(3.2 kW)
hybrid
(1.1 kW)
DTI model; long-term cost
$2950
$1600
$600
Hydrogen storage
DTI metal hydride
model; long-term cost
$380
$400
$430
Storage batteries
Zn-air, 4.1 kWh
Peaking battery
Bolder lead-acid
Part
Description
FC stack
zinc-air
hybrid
$660
$260
$460
$380
Rest of system
$810
$800
$770
$570
TOTAL
$4140
$3060
$2260
$1610
The ordering of the costs is reversed for the fuel cell hybrids here, and illustrates how hybrids
might be required for the next several years in order to bridge the gap to inexpensive fuel cells.
Once again, hybrid batteries prove to be an able competitor, although it should be recalled that
neither the 1.1 kW hybrid nor the zinc-air hybrid can sustain the original hill climbing
249
requirements. Whether the zinc-air hybrid scooter’s lower capital cost is mirrored by lower
operating costs is a subject for the next two sections, which deal with overall efficiency and fuel
costs.
5.2 Wells-to-wheels efficiency
The on-vehicle fuel economy does not account for the entire story. To obtain complete cycle
efficiencies, the inefficiencies in producing hydrogen or electricity, and in distributing the “fuel,”
must be considered. To be consistent with previous results and with the standard for large power
plants in the United States, the higher heating value efficiency is considered here.
Steam reforming of natural gas in large plants is 84% efficient, with another 87% efficiency for
distribution of hydrogen (including losses due to hydrogen compression). With a driving cycle fuel
cell net conversion efficiency of 47.7% and 77% drivetrain efficiency, the final result is 27% from
natural gas to road work.8,9
This should be compared to a scooter engine with gasoline distribution from the refinery to the
filling station at 95% efficiency, a thermal efficiency of at most 20%, and transmission efficiency
of 77%, for a total of 15%.10
In the case of the electric scooter, a factor of 40% is used to account for electricity production: the
product of 90% electricity distribution efficiency and 45% electricity generation efficiency from a
very good combined-cycle coal plant. The efficiency involved in electrowinning zinc from solution
250
and then discharging it again in the battery is the ratio of the electrowinning voltage (2.2 V) to the
output voltage (1.16 V), or 52.7%. On the vehicle there is a 77% drivetrain efficiency, for a total
efficiency of 16% from coal to road work.
(Admittedly, these three efficiencies of 27%, 15%, and 16% technically do not start from the same
starting point. Hydrogen will almost certainly be produced from natural gas, scooter combustion
engines will run on gasoline, and coal is the major source of electricity in Taiwan and would be
used (indirectly) for battery powered scooters.)
The internal combustion engine performs its most wasteful conversion step onboard, and cannot
take advantage of economies of scale to produce high efficiencies. On the other hand, the zinc-air
battery converts coal to electricity at a large power plant and loses (relatively) little there, but the
electrowinning process is energy-hungry. So the hydrogen fuel cell system is actually the most
efficient in terms of converting chemical energy to road work. However, even if it is the most
efficient, it may not be the cheapest.
5.3 Fuel cost and infrastructure
In addition to the cost of the scooter itself, the fuel cost must also be accounted for. This is the
electricity in the case of the battery-powered scooter, the hydrogen for the fuel cell scooter, and the
gasoline in the standard internal combustion engine scooter.
Energy prices for Taiwan and the United States (4th quarter1997) are compared below. Note that
251
there is regional variation in American prices which is not reflected here.
Table 5.7 Taiwan vs. USA energy prices, 1997 USD
Taiwan
U.S.A.
Unleaded premium gasoline
65.1 ¢/liter
36.9 ¢/liter
Natural gas (industrial price)
7.71 $/GJ GCV
3.53 $/GJ GCV
Natural gas (household price)
10.81 $/GJ GCV
6.26 $/GJ GCV
83.97 $/tonne
36.03 $/tonne
Electricity (industrial price)
6.69 ¢/kWh
4.07 ¢/kWh
Electricity (household price)
10.02 ¢/kWh
8.31 ¢/kWh
Coal (steam coal)
Data is from the International Energy Agency.11 GCV stands for Gross
Calorific Value, which is the same as higher heating value (“Net
Calorific Value” is equivalent to lower heating value).
When converted to common units of GJ of thermal energy (higher heating value), this is:
Table 5.8 Fuel costs for Taiwan in $/GJ HHV
Taiwan
U.S.A.
Gasoline (premium, at the pump)
16.7 $/GJ
9.5 $/GJ
Natural gas (industrial price)
7.7 $/GJ
3.5 $/GJ
Coal (industrial price)
2.6 $/GJ
1.1 $/GJ
18.6 $/GJelec
11.3 $/GJelec
Electricity (industrial price)
This calculation assumes that the coal energy content is 14,000
BTU/lb (HHV, and average for an American coal), gasoline is an
average 140,000 BTU/gallon, and that the quoted price is industrial
pricing.12,13 The gasoline price “at the pump” includes a markup for
retail which was not listed in the source data. All energy values are
higher heating value.
Taiwan prices are about twice as high as American prices, due to the island’s lack of natural
resource. Also, the higher gasoline price may be due to a different taxation policy.
252
5.3.1 Zinc-air battery “fuel” costs
Traditional storage batteries are punished for their short lifetimes (on the order of 600 cycles, or
approximately 2 years at the present); the need to purchase replacement batteries adds to the
lifetime cost. The zinc-air case is different, though, since it is not electrically recharged. Instead,
after the zinc anode is oxidized into ZnO, it is switched for a fresh zinc anode. The depleted anode
is sent back to a factory where it is electrolyzed an converted back into zinc.
Electric Fuel’s proposed zinc air infrastructure involves refilling stations (gas stations equivalents)
where depleted anodes are exchanged for fresh zinc anodes, and “regeneration centers” which are
centralized factories where the anodes are regenerated. The refilling stations essentially act as
distributors and installers for the regeneration centers. The refilling stations could use automated
machinery to switch the anodes in a short time (comparable to gasoline refilling) although capital
costs might be high.
The regeneration centers are more complex. The “used up” zinc oxide is removed from the current
collector plates, and dissolved in an alkaline (potassium hydroxide) solution. The solution is then
electrolyzed (“electrowinning”) to restore the original zinc, which collects on the cathode and is
scraped off and allowed to sink to the bottom of the electrolyte solution. The resulting zinc and
potassium hydroxide slurry is periodically drained off, strained, and pressed against a current
collector frame to produce a new anode assembly.
First, the theoretical minimum cost from the energy required is calculated. The maximum
efficiency of the process is 52.7%, based on the ratio between the 1.16 V discharge voltage and the
2.2 V for electrowinning. So a 4.1 kWh output from the batteries is equal to 7.8 kWh at the
253
electrowinning plant, and the electricity cost of 6.7 ¢/kWh gives a price of 52¢ for a single
recharge. During daily driving, the Taipei Motorcycle Driving Cycle gives a fuel economy of 36.2
km/kWh. At 40 km/day (two hours of driving at TMDC speeds), and 300 days of travel per year,
this is a total annual mileage of 12,000 km. The annual driving cost, taken from the 52¢ per
recharge cost given previously for just the electricity needed to recreate the zinc anodes, is $42 per
year. This is a net electricity cost of just 0.35 ¢/km.
A CSC (China Steel Corporation) study of costs for zinc-air battery replacement, however,
calculated a total zinc-air driving cost over ten times higher, 4.3 ¢/km.14 The assumptions are
slightly different, with 4,300 km a year driving and a fuel economy of 40.3 km/kWh of output
rather than 12,000 a year and 43.5 km/kWh:
Table 5.9 Comparison of assumptions for zinc-air electrowinning costs
CSC study
this study
zinc-air battery size (output)
3.6 kWh
4.1 kWh
energy to recharge battery
4.9 kWh
7.8 kWh
single-charge mileage
145 km
200 km
(at 30 km/h)
fuel economy
(km/kWh-recharged)
29.6
km/kWh
25.7 km/kWh
(at 30 km/h)
electricity cost
4.7 ¢/kWh
6.7 ¢/kWh
electricity cost per km
0.16 ¢/km
0.35 ¢/km
The electricity cost assumption is lower, but the energy price does not tell the whole story. The
CSC study goes on to calculate costs for the complete refueling infrastructure: labour, periodic
electrode and electrolyte replacement materials costs, spare batteries kept at the service station for
exchange purposes, and land and building fees. Each station supplies approximately 2,400 scooters
a day. This more comprehensive study results in annual fuel costs of $185 per year at 4,300 km
254
per year, or $516 per year at the 12,000 km per year assumed here – for a total “fuel” cost of 4.3
¢/km. This is extremely high but it is possible that costs may come down with time.
(Note that the electricity cost is insignificant, only 3.6% of the total cost of refueling the vehicle.)
When converted to a cost per kWh for comparison to the hydrogen and gasoline cases, this is $2.1
per kWh of output. In other words, a full recharge of the 4.1 kWh battery costs $8.60. By taking
the cost as a function of energy recovered, this allows the original assumptions about annual
driving and fuel economy to be used.
Pilot regeneration plants currently exist in Italy and Israel to service fleets of zinc-air
demonstration vehicles, but one major drawback for zinc-air scooters is that car makers are not
considering this technology and that eliminates one of the major players in technology advancement
and cost reduction. This is different from PEMFCs, which are seeing broad development for not
only vehicles but also portable power and stationary generation.
5.3.2 Hydrogen costs and infrastructure
In comparison, the 5.9 kW pure fuel cell system with 250 g of storage has a fuel economy of 0.527
km per gram of hydrogen (344 mpge) under TMDC driving. Hydrogen in Taiwan would likely be
produced by imported natural gas converted at local hydrogen filling stations using steam
reformers.
A study by Ogden et al. calculated that hydrogen produced by on-site conventional steam
reformers would cost 12-40 $/GJ based on a Los Angeles-area natural gas price of 2.8 $/GJ.15 The
255
range of costs is a function of how large each reforming station is; a large reformer capable of
producing 2 million standard cubic feet of hydrogen per day could handle 13,000 automobiles at a
cost of 11.5 $/GJ, while a small 100,000 SCF/day station would handle 650 cars at 40 $/GJ. (Note
that these calculations assume the existence of a natural gas distribution network, which may not
be the case in Taipei.)
The same driving pattern that was calculated for the zinc-air version is assumed here: 45 km of
travel per day. Taiwan prices for natural gas are 7.7 $/GJ, so prices for hydrogen increase by 5
$/GJ to 17-45 $/GJ. At the smallest station size (100,000 SCF/day), an area of 4,050 scooters
running at 12,000 km per year could be serviced at a cost of 45 $/GJ. The fuel cost of operating a
scooter turns out to be $145 a year or 1.21 ¢/km.16 If a larger plant capable of servicing an area of
72,000 scooters was built, costs would drop to 17 $/GJ for a cost per vehicle per year of $55 and a
driving cost of 0.46 ¢/km.
More advanced reformers would reduce the cost to 29 $/GJ for a 100,000 SCF/day 4,050-scooter
plant, but larger stations would not be much cheaper. $29 $/GJ is the cost assumed here. Note that
the raw natural gas cost is only 27% of the total delivered hydrogen cost; the rest is for labor,
reformer construction, electricity, hydrogen storage and compressor.
While direct hydrogen is not currently being considered for first-generation fuel cell automobiles,
buses are being demonstrated that store hydrogen in large compressed gas cylinders, and scooters
could “piggyback” on a hydrogen distribution infrastructure for public transportation. In this case,
there would not have to be 72,000 scooters within the operating area of a single refueling station
plant.
256
Refueling metal hydrides essentially is a matter of filling at pressures of about 10 atm; the rate of
adsorption is dependent on how quickly the excess heat of adsorption can be removed, and liquid
coolants in the nozzle design can be used to effect this. The process could be done in just 5-15
minutes.17
5.3.3 Combustion scooter gasoline costs
The 100 mpg fuel economy of the gasoline-powered scooter is scaled down to 65 mpg for driving
cycle performance (the same ratio as for the hydrogen powered scooter; note that actual
performance will be different because of the efficiency-versus-power characteristics of the
combustion engine).
Assuming the same travel distance of 12,000 km a year, and 65.1 ¢/liter for gasoline yields an
annual fuel cost of $105, or 1.5¢ per kilometer.
5.3.4 Fuel cost summary
The advanced reformer for small service areas, at 24 $/GJ, was used for the hydrogen case; the
China Steel Corporation cost analysis was used for the zinc-air battery to obtain a per-kWh price
of $2.1, which was applied to driving 12,000 km per year under the lower mileage of the TMDC.
257
Table 5.10 Fuel cost summary
zinc-air
hybrid
gasoline
5.9 kW
pure FC
3.2 kW
hybrid
1.1 kW
hybrid
refueling cost
583 $/GJelec
(2.1 $/kWh)
16.7 $/GJHHV
(65.1 ¢/L)
24 $/GJHHV
(0.34 ¢/g)
24 $/GJHHV
(0.34 ¢/g)
24 $/GJHHV
(0.34 ¢/g)
TMDC mileage
36.2 km/kWh
65 mpg
0.527 km/g
0.484 km/g
0.525 km/g
on-vehicle mileage
780 mpge
65 mpg
344 mpge
320 mpge
363 mpge
cost per distance
under TMDC
5.8 ¢/km
1.5 ¢/km
0.65 ¢/km
0.70 ¢/km
0.65 ¢/km
$696
$184
$78
$84
$78
$4,275
$1130
$480
$515
$480
annual cost
present value of fuel
over 10-year lifetime
Present value costs were calculated over a ten-year scooter lifetime,
with a 10% discount rate (meaning future fuel costs are discounted
heavily compared to the up-front capital cost)
The zinc-air battery’s energy cost is actually very low, but infrastructure costs, spare battery
expense, and depreciation of the stock of batteries all add up to a very expensive per-km cost. Over
time, infrastructure costs must reduce drastically if zinc-air batteries are to be competitive.
The gasoline-powered scooter reflects current efficiencies, and improvements in air pollution
technology like fuel injection will likely improve fuel economy by a small amount and reduce
driving costs.
Hydrogen production at the infrastructure levels assumed here results in hydrogen costs that are
less than half the price of gasoline, due to the high efficiency of the fuel cell scooter. Even with the
extremely small-scale hydrogen reforming station assumed here, the cost is low enough to make
hydrogen fuel cell scooters a cheaper option to drive than gasoline-powered scooters.
258
Under these assumptions, the fuel cost is roughly on the order of the vehicle cost for the gasoline
vehicle, much more expensive for the zinc-air hybrid due to its low cost, and significantly cheaper
for the hydrogen powered scooter due to its high efficiency. However, the comparison is uncertain
due to the great uncertainties in the fuel cell and zinc-air technology costs. Also, maintenance and
repair costs are not yet quantified for the advanced technologies. These two reasons explain why
the capital and present value of fuel were not directly summed.
5.4 Final Conclusions
Advanced fuel cell powered scooters could produce more than three times the 100 mpg of current
gasoline-powered scooters, with zero tailpipe pollution. In the long run, a rough cost estimate
predicates that they would cost about $1,200 – $1,300 to produce, although prices for the
consumer would be as much as twice this amount. In comparison, more advanced combustionpowered scooters (like four-stroke scooters) could offer pollution reduction of about 75% of
hydrocarbons and 50% of carbon monoxide for an additional cost over two-stroke scooters of
under $200. However, one methodology shows that there is significant health benefit to even
removing that last stage of emissions, so if the proper dollar value were assigned to the air
pollution reductions, a value of several hundred dollars would apply to the zero-emission scooters.
This study arrived at this conclusion through an analysis of current scooter performances and
pollution trends, a general examination of the health benefits of zero-pollution scooters over fourstroke scooters, a discussion of electric vehicle technology including battery, fuel cell, and
hydrogen storage options, and detailed modeling of scooter driving and fuel cell performance that
had not been done before for this type of vehicle.
259
A fuel cell design was presented for the scooter that focused on simplicity on all fronts: pure
hydrogen operation, low temperature. In addition, hybrid designs were examined in an effort to
accelerate fuel cell scooter adoption by reducing the size of the fuel cell stack needed. Hybrids
reduce price in the short term, but in the long run fuel cells should come down in price by enough
to make peaking power batteries unnecessary.
Finally, another option, the zinc-air battery, was examined. This technology showed good technical
performance, and a zinc-air drive system would be half the weight of a fuel cell system. Questions
remain about the expense of developing an infrastructure for zinc anode regeneration and
electrolyte / cathode replacement, which currently are projected to be very expensive. So low
capital cost is traded for high fuel costs
The original work done and results obtained are summarized below.
5.4.1 Background
The focus was placed on Taiwan because of its extremely high vehicle density, the large number of
scooters, and its pre-eminence as a scooter manufacturing center. High pollution levels, especially
in cities, are a concern among the populace and government and one of the primary motivators for
cleaning up cheap two-stroke scooters. Four-stroke scooters will likely expand to fill the role of
these two-stroke scooters, and provide significant reductions in emissions.
A sketch of cost and benefits shows that hundreds to thousands of dollars of health benefit per
vehicle could be realized by switching from four-stroke engines to zero-emission vehicles. The
government’s chosen solution, the battery powered scooter, currently lacks adequate range.
260
Proton exchange membrane fuel cell systems offer greater range, with the same zero tailpipe
emissions since they produce electricity electrochemically, with water as the only exhaust. As is the
case with battery-powered scooters, pollution emissions would be shifted to central (in this case,
hydrogen-generating) plants.
Hydrogen storage is best supplied with hydrogen stored onboard in cylinders or in the form of
metal hydrides. The latter offers excellent synergy with the cooling system due to its endothermic
hydrogen release, and greater safety due to the far lower pressures (1-10 atm), and is recommended
here.
5.4.2 Modeling results
Modeling shows that maximum power required is just under 6 kW for performance comparable to
combustion two-stroke scooters. Average power demands are a tenth, at 670 W for urban driving.
The following performance specifications were required:
This thesis presents a fuel cell design for a scooter of approximately 5 or 6 kW, a niche that was
previously unexplored. Size, weight, and cost restrictions force the design to be simple and to
remove any unnecessary systems. Where data were not well known and assumptions were
necessary, performances were assumed to be worse than expected to ensure a feasible design.
A compressor/expander system to provide 3 atm pressurized air at the cathode was rejected as the
benefit did not outweigh the parasitic power losses and additional complexity, weight, and expense.
The advantages of high pressure for water removal were not considered.
Cooling of the fuel cell system proved to be a significant problem. A radiator was chosen that
could handle the maximum continuous cooling load (produced by slope climbing requiring 3020 W
of power for an indefinite period). Benefits in cooling derived from “ram effect” air flowing over
the radiator were not included so the system was somewhat overdesigned. Integration of a metal
hydride hydrogen storage system provided a useful method of extracting 17% to 30% of surplus
heat. The thermal mass of the system and the low average heat production meant that over the
TMDC, cooling is not a problem
The complete drive system configuration (fuel cell, battery, motor, controller, hydrogen storage in
DTI-modeled metal hydride) is summarized below.
262
Table 5.12 System design results
today’s
ZES-2000
pure FC
(5.9 kW)
hybrid
(3.2 kW)
hybrid
(1.1 kW)
zinc-air
battery
Sustained net power
~ 3 kW
5.6 kW
3.0 kW
1.0 kW
1.8 kW
Range (30 km/h)
65 km
200 km
200 km
200 km
200 km
Range (TMDC)
< 35 km
131 km
129 km
149 km
148 km
Drive system (size)
24 L
43 L
43 L
30 L
41 L
Drive system (weight)
60 kg
61 kg
63 kg
60 kg
45 kg
The fuel cell auxiliary systems make the drive system heavy, even though the weight of the fuel cell
itself is relatively low. To go from today’s ZES-2000 to a 1.1 kW hybrid requires 6 L of additional
space which is easily found in the current body frame, but the other systems will require a redesign
of the body (although this is not be a “show-stopping” requirement).
The relatively poor energy density of the hybrid zinc-air battery causes the high volume
requirements of that design, whereas the 5.9 kW and 3.2 kW fuel cells feature very large radiators
that account for the high volume.
Hybrid power systems with a combined peaking power battery and hydrogen prime energy source
offer significant reductions in cost because fuel cells are so expensive right now, although in the
long run this situation is expected to reverse in favour of pure fuel cell scooters. The costs are
detailed in the next section.
5.4.4 Costs and infrastructure
The following costs-to-manufacture are listed below for various cases.
263
Table 5.13 Long-term cost of hybrid fuel cell scooters
pure FC
(5.9 kW)
hybrid
(3.2 kW)
hybrid
(1.1 kW)
hybrid
battery
$1,220
$1,360
$1,455
$1,185
The sale cost may be as much as double this price, which would be significantly more than today’s
two-stroke internal combustion scooters at $1,000 and electric battery-powered scooters (albeit
with only one third the range) at $1,500.
Assuming hydrogen costs of 24 $/GJ from small reforming stations running on pipelined natural
gas obtains the following comparison:
Table 5.14 Fuel cost summary
zinc-air
hybrid
on-vehicle mileage
cost per distance
under TMDC
annual cost
present value of fuel
over 10-year lifetime
gasoline
5.9 kW
pure FC
3.2 kW
hybrid
1.1 kW
hybrid
780
mpge
65 mpg
344 mpge
320 mpge
363 mpge
5.8 ¢/km
1.5
¢/km
0.65
¢/km
0.70
¢/km
0.65
¢/km
$696
$184
$78
$84
$78
$4,275
$1,130
$480
$515
$480
Fuel prices for the zinc-air hybrid battery scooter are incredibly high according to the one study
cited in this study. Hydrogen fuel cell scooters have very good range and mileage, and can operate
cheaply even if hydrogen is produced at the relatively small scales assumed here. It is not only the
pollution benefits that make hydrogen scooters better than current gasoline-powered scooters; the
fuel savings are also significant.
264
5.4.5 Parting words
Fuel cell scooters face many of the same problems as fuel cell automobiles: the technology is new,
and thus expensive, and distribution infrastructure does not exist for hydrogen delivered to the end
user. On the other hand, the Taiwan situation features somewhat high fuel prices, relatively high
income levels, and extremely poor air quality, two drivers for more efficient and cleaner vehicles.
The situation in China, Japan, and other Asian countries is similar, with varying degrees of
pollution and wealth. Some European countries have high numbers of scooters as well and could be
markets for this technology.
In the long run, the hydrogen fuel cell scooters could cost approximately $1,200 to manufacture.
This is dependent upon significant reductions in metal hydride and fuel cell costs, and these
reductions would occur most quickly if they piggybacked off other markets that called for large
numbers of these two core components.
Ordinary electric battery scooters are not projected to have the energy densities required, and a
recharging infrastructure is a problem in Taiwan where indoor vehicle storage off the street is not
always guaranteed. Zinc-air scooters are likely to cost less than fuel cell scooters, but are projected
to have, at least in the short term, extremely expensive refueling costs based on more complex
battery exchanging stations. Hydrogen scooters would be cheaper to drive than combustion
scooters.
Future work on this topic would move the conceptual design and feasibility test presented here to
more applied design work and prototype construction. The performance presented here is unlikely
to be far from actual results, and assumptions made here have always erred on the side of more
265
waste and worse performance. So building a prototype fuel cell scooter – or even only the fuel cell
power system – and obtaining more detailed data on the parasitic power would be an excellent way
to measure real-world performance, which for the reasons outlined above is expected to be higher
than the conservative estimate here. System integration of the various heat flows and physical
assembly of the variously-shaped parts also needs to be demonstrated.
Also, “smart” hybrid power management algorithms should be designed to optimize hybrid
scooters that may be used in the next several years while fuel cells are still too expensive to be the
sole power source for vehicles. Detailed research specific to Taiwan and other Asian locations is
needed on the subject of hydrogen distribution. Finally, as reformer and direct methanol fuel cell
technology advance, they may become feasible for this application and this must be kept in mind.
Recommendations: in the short run, getting rid of two-strokes and speeding a transition to fourstroke engines is an inexpensive path to deep reductions in emissions. The Taiwan government is
following the right track in legislating emissions performance requirements rather than enforcing
technology. A hydrogen fuel cell scooter offers additional air pollution reduction benefits over fourstrokes that could justify its increased costs, however, and is worth investment in research and
development. Peaking power batteries for upcoming scooters are highly recommended to allow the
use of smaller and less expensive fuel cells without the performance compromises that would cause
public rejection of advanced scooters, and metal hydride technology is the best fuel storage
strategy.
FIN
266
References for Chapter 5
1. Mark DeLuchi. Institute of Transportation Studies, University of California, Davis. Hydrogen Fuel-Cell
Vehicles. Research Report UCD-ITS-RR-92-14 September 1, 1992. p. 147
2. Jet P. H. Shu, Wei-Li Chiang, Bing-Ming Lin, Ming-Chou Cheng. Mechanical Industry Research
Laboratories, Industrial Technology Research Institute. “The Development of the Electric Propulsion
System for the ZES2000 in Taiwan” Internal paper, date unknown (October 1997 or later)
3. Randy Knudson, Scooter Therapy. Personal communication May 21, 1999
4. $150 estimate of cost from Dr. Philip G. Felton, Princeton University Department of Mechanical and
Aerospace Engineering, personal communication July 26 1999
5. Jeffrey Ho. Deputy Director, Taiwan UQM. Personal communication May 24, 1999.
6. Brian D. James, George N. Baum, Franklin D. Lomax, Jr., C. E. (Sandy) Thomas, Ira F. Kuhn, Jr.
Directed Technologies, Inc. “Comparison of Onboard Hydrogen Storage for Fuel Cell Vehicles” Task 4.2
Final Report. Prepared for Ford Motor Company under Prime Contract DE-AC02-94CE50389 “Direct
Hydrogen Proton-Exchange-Membrane (PEM) Fuel Cell System for Transportation Applications” to the
U. S. Department of Energy, pp. 4-56, 4-63, 4-52
7. Jonathan Goldstein, Ian Brown, Binyamin Koretz. “New developments in the Electric Fuel Ltd. zinc /
air system” Journal of Power Sources 80 (1999) pp. 171-179
8. Joan M. Ogden, Margaret M. Steinbugler, Thomas G. Kreutz. “A comparison of hydrogen, methanol
and gasoline as fuels for fuel cell vehicles: implications for vehicle design and infrastructure
development” Journal of Power Sources 79 (Elsevier: 1999), pp. 143-168
9. Note that the overall conversion efficiency is actually 55.7% from hydrogen heating value to electricity.
However, about 10% of that electricity is needed for parasitic purposes and is wasted in terms of useful
output like road motion or auxiliary power. The efficiency net of parasitics is 46.7%
10. Matthew Brekken and Enoch Durbin. “An Analysis of the True Efficiency of Alternative Vehicle
Powerplants and Alternative Fuels.” Society of Automotive Engineers 981399, 1997.
11. International Energy Agency. “Key World Energy Statistics”
http://www.iea.org/stats/files/keystats/stats_98.htm. Accessed May 8, 1999
12. World Bank. “World Bank - Typical Coals of the World”
http://www.virtualglobe.com/html/fpd/em/power/sources/coal_tcw.htm Accessed August 19, 1999
13. World Bank. “World Bank - Typical Analyses and Properties of Fuel Oils*”
http://www.virtualglobe.com/html/fpd/em/power/sources/oil_tapfo Accessed August 19, 1999
14. China Steel Corporation study for chemTEK. Received August 21, 1999 from chemTEK.
15. Joan Ogden, Margaret Steinbugler, Thomas Kreutz. “Hydrogen as a fuel for fuel cell vehicles: a
technical and economic comparison”
16. ibid
17. Joan Ogden, Princeton University. Personal communication July 22 1999
267
Appendix A current and prototype electric scooters
Country
ROC
ROC
ROC
Japan
Japan
Japan
USA
maker
ITRI
Shangwei
ENTER
TOKYO R&D
HONDA
YAMAHA
DORAN (a)
model
ZES 2000
SWAP
City Bike II
ES-600
CUV-ES
MEST
ECO-SCOOT
maximum speed
(km/h)
50
80
55
55
60
50
33
climb capability
(tan )
16 km/h on
18( slope
30 km/h on
30( slope
15( (b)
18( (b)
12((b)
10((b)
12((b)
range (km)
at 30 km/h
60
80
75
60
60 (35 km in
“urban mode”)
30
40
dry weight, kg
105
130
95
117
130
80
89
DC
brushless
DC brushless
DC brushless
(3.25 kW), 48 V
DC brushless
DC brush
motor
battery
DC
DC
brushless, 48 V brushless, 48V
four sealed
lead-acid
sealed leadacid
lead-acid
maintenancefree
lead-acid
Ni-Cd
maintenancefree
lead-acid
lead-acid
28/12
40/12
52/24
30/48
20/86.4
17/48
46/24
charging time
(hr)
no data
6~8
5~8
8
8
8
~10
transmission
CVT (?)
CVT
single-stage
reduction
CVT
CVT
no data
single-stage
reduction
acceleration
0-30 m, 4.5 s
no data
no data
no data
0-200 m, 17.3 s
no data
no data
no data
2000
1460
4750
8150
no data
1900
demo only
750 W rated
power
in
development
limited
fleet sales
limited fleet
sales
introduction
on market
battery capacity
(Ah/V)
price (US$)
notes
a
b
The Doran vehicle was bought out by the Sun Cat Motor Company in 1995 and renamed the Sol Gato
Speeds on these “climb capability” slopes were not given
The data were based on tables from the following sources:
P. H. Jet Shu, Wei-Li Chiang, Bing-Ming Lin, Ming-Chou Cheng. “The Development of the Electric
Propulsion System for the Zero Emission Scooter in Taiwan” Japan Society of Automotive Engineers.
1997. JSAE 9734403, SAE 92107
Shang Wei web site. “Shang Wei SWAP vs. Other Electric Scooters”.
http://www.shangwei.com/compar-e.htm. Accessed August 30, 1999
Appendix B
detailed stack cost/size analysis
The DTI model outlined in Detailed Manufacturing Cost Estimates for Polymer Electrolyte
Membrane (PEM) Fuel Cells for Light Duty Vehicles (August 1998) was used to calculate size,
weight, and cost of a scooter fuel cell stack. The ultimate purpose was to use the methodology
described in that report to produce reasonable estimates of size, weight, and cost that were more
accurate than simply linearly scaling down automotive fuel cells.
In their report, DTI studied several different sizes of fuel cells and calculated manufacturing costs
for the various components of the cell stack (auxiliaries like compressors and cooling pumps were
not examined). Two sets of analyses were done, and in each case the area of the membranes in each
cell was varied between six different sizes ranging from 116 cm2 to 697 cm2. In the first set, “equal
voltage,” the number of cells was held constant at 420 cells and thus power varied with membrane
area. In the second set, “equal power,” the number of cells was varied to keep total power at 70
kWelec. Options for the cell unit design include “unitized” bipolar metallic separator plates stamped
with flow fields on both sides, three-piece unipolar metallic plates, carbon-polymer composite
plates, and amorphous carbon plates. The more conservative three-piece metallic separator plates
were chosen here (see section B.2 for details)
269
Note: A May 1999 DTI study has examined a range of small fuel cells (among them, 3-5 kW
stacks) for electricity and heat for individual residences. For a production quantity of 10,000 units,
this study found the installed cost per kW to be about $4,500.1 These figures are based on topdown cost analyses, and other companies who are designing home fuel cell systems have quoted
figures on the order of $3,500 to $5,000 for a residential fuel cell system of a few kilowatts using
batteries for peak power.2
These high prices for this size of fuel cell would seem prohibitive for the scooter, except for the
fact that stationary power fuel cells must be designed very differently from automotive fuel cells;
they must operate 24 hours, unlike vehicle engines which are only run a few times daily. Also,
vehicle engines rarely reach top output. All in all, stationary fuel cell lifetimes must be longer and
they must be designed more robustly. Another factor cited in the DTI report was that the larger
production quantities for vehicles would help to drive down costs. So despite these recent high
projections of cost for stationary fuel cell systems of similar size as the scooter studied here,
predictions of fuel cells for vehicles are still on target.
B.1 “Blind” curve-fitting
Curves were fit to the original DTI results for automotive fuel cell, based on an equal-voltage study
with 420 cells and stack power varying with different membrane area. Each of the six different cell
1
C. E. (Sandy) Thomas, Jason P. Barbour, Brian D. James and Franklin D. Lomax, Jr. Directed
Technologies, Inc. “Analysis of Utility Hydrogen Systems & Hydrogen Airport Ground Support
Equipment”. Prepared for the Proceedings of the U.S. DOE Annual Hydrogen Program Review. May
1999.
2
Plug Power and American Power Corporation. Quote is from Ronald J. Wolk. “Fuel Cells for Homes and
Hospitals” IEEE Spectrum May 1999 p. 45
270
membrane areas is listed in Table B.1 below.
Table B.1 DTI automotive stack parameters
Membrane
area (cm2)
Stack power
(kW)
Cost
($/gross kWe)
Weight
(kg)
Volume
(L)
116
31.5
$36
22.0
23.7
181
49.0
$29
29.4
32.5
258
70.0
$26
38.5
42.7
348
94.5
$23
48.1
54.2
452
122.5
$22
59.5
67.1
568
154.0
$21
72.4
81.3
697
188.9
$20
86.4
96.9
Note: the total weight was not given in the study, and was summed in
the same way the stack weights were calculated later in section B.3.
Volumes were given in terms of stack dimensions and converted here
to liters.
The DTI automotive stack costs per kilowatt were plotted against cell active membrane area in
Figure B.1, and fit with a hyperbola of the form
y = A / (x-B) + C
for cost per kilowatt “y” and active membrane area “x”, and fitting parameters A, B, and C.
271
Figure B.1 Cost as a function of cell membrane area
90
price ($/kWe-gross)
80
70
60
50
40
30
20
10
0
0
100
200
300
400
500
600
700
membrane area (cm^2)
Similarly, per-kilowatt data for weight and volume were plotted and regressed against curves of the
same form as for the cost in Figure B.2:
Figure B.2 Weight and volume as a function of cell membrane area
These results were extrapolated to the membrane areas required for the three fuel cell sizes studied
here: 1.1 kW, 3.2 kW, and 5.9 kW (corresponding to stack membrane areas of 35 cm2, 100 cm2,
and 170 cm2 as discussed in Chapter 4). This produced the following results:
Table B.2 Curve-fitting versus bottoms-up model
Stack cost
Stack weight
Stack volume
Stack power
Curve-fitting result
1.1 kW
$96
3.2 kW
$125
5.9 kW
$176
1.1 kW
1.5 kg
3.2 kW
2.4 kg
5.9 kW
3.6 kg
1.1 kW
1.6 L
3.2 kW
2.6 L
5.9 kW
4.0 L
B.2 Size and volume
To improve upon this estimate, bottom-up calculations of size, weight, and cost were made by
following the DTI procedure. First, the total size of the three fuel cells were calculated by first
calculating the sizes of three-piece stainless steel cooler cells and active cells, and assuming a 2:1
active-to-cooler cell ratio:
The active cell requires one metal separator plate and two separate, unipolar plates etched with
flow fields and gaskets separating the flow fields from the MEA, for a total thickness of 2.27 mm
per active cell:
273
51 µm separator plate
76 µm anode flow field
1000 µm anode gasket
70 µm MEA
76 µm cathode flow field
1000 µm cathode gasket
[repeat with next separator plate]
The cooler cells are thinner, at 1.13 mm each:
51 µm separator plate
76 µm coolant flow field
1000 µm gasket
[repeat with next separator plate]
With 56 active cells and 28 cooler cells, this produced a total thickness of 15.9 cm. The
arrangement of cells is described in Figure B.3 following:
274
Figure B.3 Diagram of stack assembly
cooler cell
active cell
current collector
insulator
endplate
275
Following the DTI procedure, the dimensions of the repeat components were calculated by taking
the active area needed per cell, selecting a height and width, and adding 2.54 cm of inactive
membrane area to each dimension to obtain a total membrane area. Next, a 5.1 cm manifold space
was added to the width of the membrane (including inactive margin) to obtain the total stack face
area, and then 1.27 cm was added to each side of the stack to account for the thickness of the
plastic housing and produce the fuel cell stack’s overall dimensions as listed in Table B.3 below.
(Note that the original DTI stack designs contained two parallel strings of 210 cells, while the
design presented here uses a single 56-cell stack.)
Table B.3 Stack dimensions
5.9 kW
3.2 kW
1.1 kW
170 cm2
100 cm2
35 cm2
active membrane
dimensions,
10.0 cm x 17.0 cm
(170 cm2)
10.0 cm x 10.0 cm
(100 cm2)
5.0 cm x 7.0 cm
(35 cm2)
total membrane dimensions,
including inactive margin
12.5 cm x 19.5 cm
(244 cm2)
12.5 cm x 12.5 cm
(156 cm2)
7.5 cm x 9.5 cm
(71 cm2)
total dimensions of stack face
-- includes manifolding
17.6 cm x 19.5 cm
(340 cm2)
17.6 cm x 12.5 cm
(220 cm2)
12.6 cm x 9.5 cm
(120 cm2)
overall dimensions with
plastic housing
20.2 cm x 22.1 cm
x 17.5 cm
20.2 cm x 15.1 cm
x 17.5 cm
20.2 cm x 12.1
cm
x 17.5 cm
7.8 L
5.3 L
3.2 L
membrane active area
total volume
B.3 Weight of stack
B.3.1 Weight of non-repeat components
The weights of the non-repeat components (plastic stack housing, two endplates, two insulators,
two current collectors, sixteen tie-rods) were calculated based on their dimensions and
276
extrapolation from the DTI study.
The stack housing mass was based on previously calculated stack housing volume and a density of
0.576 g•cm-3 for the plastic housing material.
The sixteen tie-rods were estimated at 2.05 kg for DTI’s default equal-voltage 52.5 cm stack. Since
length is only 15.9 cm in this fuel cell, the tie-rod mass was pro-rated down to 0.6 kg.
The insulator and current collectors cap the ends of the array of cells and thus have an area equal
to the total membrane area. They were regressed against total membrane area. The endplates,
containing reactant and exhaust ports, tie rod holes, and serving as structural support for the stack,
cover the entire face area of the stack. The endplate weights were regressed against stack face area.
However, the regressions for the current collectors and endplates became negative at the low stack
areas designed for the hybrid, so they were not used.
So, in the case of the 100 cm2 and 35 cm2 stacks, the current collector and endplate weights were
simply taken to be the same as the weights for the DTI 116 cm2 stack. For the 170 cm2 stack, the
current collector weight was interpolated between the weights for the collectors obtained for the
DTI 116 cm2 and 181 cm2 stacks. The endplate, of area 344 cm2, was set equal to the weight of
the endplate for the DTI 116 cm2 (452 cm2 face area) stack.
The linear regression was retained for the insulator weights. The results for all of the parts were:
The repeat components were modeled as follows. First, the masses of the repeat units (separator
plate, anode flowfield, cathode flowfield, gasket, MEA) were taken from DTI’s values for the
rubber gasket and MEA, and calculated from dimensions and densities for the separator plates and
flowfields. A stainless steel density of 8 kg/L was used. This was done for each of the sizes studied
in the DTI report; for example, the weights were 28.2 g for each active cell and 16.4 g for each
cooling cell for a 116 cm2 active area.
Then, the total weights of the repeat units were summed for all the 56 cooler cells and 28 active
cells, and expressed as a function of total membrane area for the membrane sizes studied in the
DTI study:
278
Figure B.4 Regression of total repeat unit weight
10
total repeat unit weight, kg
9
8
7
6
5
4
3
2
1
0
0
100 200 300 400 500 600 700 800 900
total membrane area (cm^2)
A straight line regression fit well to the repeat masses; because the majority of the weight comes
from the stainless steel plates, and these plates’ weights vary linearly with changing membrane area
(because thickness is constant)
Table B.5 Total weight of stack repeat units
total membrane dimensions
(including inactive)
total membrane area
total repeat unit weight
5.9 kW
3.2 kW
1.1 kW
12.5 cm x 19.5 cm
12.5 cm x 12.5 cm
7.5 cm x 9.5 cm
245.0 cm2
157.3 cm2
71.9 cm2
2.8 kg
1.8 kg
0.8 kg
279
B.4 Summary of stack weight and volume
The total stack weight and volume are summarized below with power densities.
Table B.6 Summary of size and weight results
5.9 kW
3.2 kW
1.1 kW
170 cm2
100 cm2
35 cm2
non-repeat mass
5.1 kg
3.6 kg
3.3 kg
repeat mass
2.8 kg
1.8 kg
0.8 kg
total mass
7.6 kg
5.4 kg
4.0 kg
volume
7.8 L
5.3 L
3.2 L
specific power
0.78 kW/kg
0.62 kW/kg
0.27 kW/kg
power density
0.76 kW/L
0.62 kW/L
0.34 kW/L
membrane active area
In comparison, Ballard reported in a 1995 press release a stack power density of 0.7 kW/L.
B.5 Cost
The costs were summed from minimum values of component costs as a function of cell membrane
total area. In the case of this study, the total membrane areas were 244 cm2, 156 cm2, and 71 cm2
for the 5.9 kW, 3.2 kW, and 1.1 kW hybrid scooters respectively. The insulators were the only
parts actually regressed because the others became negative at the low membrane areas involved.
1. Again following DTI reported figures, stack housing prices were calculated at $1.16 per kg of
material (based on the previous weight calculations) plus $15 assembly cost.
280
2. The insulators regressed to $0.24 each for the 5.9 kW stack, $0.20 each for the 3.2 kW stack,
and $0.16 for the 1.1 kW stack.
3. The cost of the endplates became negative when regressed versus total membrane area, so the
endplates were assumed to be the same cost ($4.02) as those for the smallest (116 cm2 ) DTI case.
For the same reason, “floor” values of $1.23 were chosen for the current collectors in the 100 cm2
and 35 cm2 stacks. The sixteen tie rods were calculated at $1.00 each.
MEA costs were simply the total membrane area multiplied by a cost of 52.3 $/m2 predicted by
DTI for mass-produced, low-cost technology. All subcomponents – MEA and cell hardware
(gasket, separator plates, flow field plates) – were summed and then regressed against total
membrane area for the repeat unit cost:
Figure B.6 Stack repeat unit regression
2.5
cost per unit, $
2.0
1.5
1.0
0.5
0.0
0
100
200
300
400
500
total membrane area, cm^2
cooler cell
281
active cell
600
700
Finally, $10.00 was added for assembly, and a 10% contingency cost added at the end. The prices
broke down as follows:
Table B.7 Cost summary
price per component (number)
5.9 kW
3.2 kW
1.1 kW
56 active cells
56.04
37.75
29.55
56 MEAs
83.56
34.10
11.94
28 cooler cells
17.63
11.82
9.21
assembly line machine cost per cell
(total 84 cells)
8.40
8.40
8.40
165.63
92.07
59.11
2 current collectors
3.39
2.46
2.46
16 tie rods
16.00
16.00
16.00
2 insulators
0.47
0.40
0.33
2 endplates
8.04
8.04
8.04
1 stack housing
16.53
16.14
15.77
TOTAL
44.44
43.04
42.59
10
10
10
11% contingency cost
24.21
15.96
12.29
total stack cost
$244
$161
$124
REPEAT
TOTAL
NON-REPEAT
SUMMARY
final assembly and inspection
The costs per kilowatt for the stack are $103/kW for the 1.1 kW stack, $47/kW for the 3.2 kW
stack, and $42/kW for the 5.9 kW stack.
These results were compared with the curve-fitting results listed in Table B.2:
282
Table B.8 Curve-fitting versus bottoms-up model
Stack cost
Stack weight
Stack volume
Stack power
Curve-fitting result
Bottoms-up result
1.1 kW
$96
$124
3.2 kW
$125
$176
5.9 kW
$176
$244
1.1 kW
1.5 kg
4.0 kg
3.2 kW
2.4 kg
5.4 kg
5.9 kW
3.6 kg
7.4 kg
1.1 kW
1.6 L
3.2 L
3.2 kW
2.6 L
5.3 L
5.9 kW
4.0 L
7.8 L
The simplistic curve fits consistently overestimate stack performance (light weight, low cost, small
volume). The more detailed model, which admittedly uses curve fitting in its numerous elements,
more accurately captures the fact that parts not only increase in size, weight, and cost non-linearly
as cell membrane size decreases, but that they increase more than expected from the initial fit.
283
Appendix C radiator performance data
The radiators used in the model are OEM coils produced by Lytron. The series has the following
properties:
Table C.1 Lytron OEM coil weight and contained liquid volume
radiator
model
dry (empty)
weight
volume of
coolant inside
M05-050
0.9 kg
115 mL
M05-100
1.8 kg
188 mL
M10-080
2.3 kg
320 mL
M10-160
3.6 kg
549 mL
M14-120
4.5 kg
606 mL
M14-240
7.3 kg
1090 mL
Figure C.1 shows performance curves of cooling factor (W/K) versus air flow rate (cfm) and
coolant flow rate (gpm). Note that the M14-120 curve was accidentally printed twice on the data
sheet from Lytron; the bottom figure is incorrect.
The source of the data in this appendix is the Lytron web site, “Lytron OEM Heat Exchangers
(Radiators) Performance Curves” http://www.lytron.com/Catalog/oemperf.htm and “Lytron
Manufacturers of Thermal Transfer Solutions”, http://www.lytron.com/Catalog/techwght.htm.
Both sites were last accessed June 1999.
284
photocopy not included in PDF;
please see
http://www.lytron.com/Catalog/oemperf.htm
285
Appendix D
conversion factors
1 gallon
1 gallon per minute
1 cubic foot
1 cubic foot per minute
3.785 L
0.06308 L/s
28.32 L
0.472 L/s
1 mole
22.5 L
1 atm
1.013 bar
(standard conditions)
760 mmHg
14.7 psi
101.3 kPa
407 in H2O
1 calorie
4.18 J
1 BTU
1055 J
1 horsepower
746 W
1 ampere per ft2 (“ASF”)
Appendix E
AIV
BDC
CAFE
CFM
CSC
DoD
DFI
DMFC
ECE
FHDS
1.0764 mA•cm-2
1 foot
30.48 cm
1 pound
0.454 kg
1 mile
1.609 km
acronyms and abbreviations
Aluminum Intensive Vehicle
Bottom Dead Center
Corporate Average Fuel Economy
Cubic Feet per Minute
China Steel Corporation
Depth of Discharge
Direct Fuel Injection
Direct Methanol Fuel Cell
Economic Commission for Europe
Federal Highway Driving Schedule
Federal Test Procedure
Federal Urban Driving Schedule
Gross Calorific Value (used in natural gas industry for HHV)
Higher Heating Value
Industrial Technology Research Institute [Taiwan]
Lower Heating Value
Liquefied Natural Gas
Membrane Electrode Assembly
Mechanical Industry Research Laboratory [ITRI]
Molten Carbonate Fuel Cell
Miles Per Gallon (Equivalent)
Net Calorific Value (lower heating value)
New Generation Motors
Nickel Metal Hydride
New Taiwan Dollars
Original Equipment Manufacturer
Partnership for a New Generation of Vehicles
Phosphoric Acid Fuel Cell
Proton Exchange Membrane Fuel Cell
(also Polymer Electrolyte Membrane Fuel Cell)
Particulate Matter
Polytetrafluoroethylene
Republic of China
Society of Automotive Engineers
State of Charge [of batteries]
Solid Oxide Fuel Cell
Top Dead Center
Total Hydrocarbons
Taipei Motorcycle Driving Cycle
Total Suspended Particulates
Unique Mobility
United States Dollars
Volts Alternating Current
Vehicle-Kilometers Traveled
Vehicle-Miles Traveled
Zero Emission Scooter
287
Appendix F
MATLAB simulation program listing
This is a listing of the MATLAB m-file program used to simulate scooter performance over
various driving cycles, launch14.m
%
%
%
%
%
%
%
launch14.m
Test bed to run various configurations through a specified driving cycle.
Version 11 revises the efficiency calculation and changes baseparasitics
Version 12 includes parasitic power in the final plot
Version 13 refills the state of charge frequently - not just by regen
Version 14 cleans up the code
more off; clear
%% STACK HEAT PARAMETERS
%
% stackmass is in kg, divided by two for the part in thermal contact
% with the cells; heatcapacity is in kJ/kg/C (water is 4.19,aluminum
% is 0.900, copper 0.386. We estimate 1.0 for stack) cooleff is in W/C
% and is different depending on which hybrid design is used
ambientTemp=40; initTemp=50;
stackmass=2.8; specheatcap=0.929;
%% CHOOSE TYPE OF HYBRIDIZATION (IF ANY)
%
% Select one of four configurations; adjust parameters accordingly
% The various hybrid versions reuqire different fuel cell areas,
% numbers of peaking power battery cells, and different parasitic power
% loads. "kickin" defines the power (watts) at which the battery kicks in.
disp(sprintf('Pick:\n 1 for pure FC\n 2 for 3.3 kW\n 3 for 1.1 kW\n 4 for
elec hybrid'));
hybridtype=input(' ?');
switch hybridtype
case 1,
cooleff=110; baseparasite=39.7; cellarea=170; numbolder=40;kickin=99000;
case 2,
cooleff=150; baseparasite=66.0; cellarea=100; numbolder=27; kickin=3020;
case 3,
cooleff=50; baseparasite=25.3; cellarea=35; numbolder=47; kickin=1000;
case 4,
cooleff=35; baseparasite=0; cellarea=90; numbolder=38; kickin=1830;
otherwise,
disp('Unknown option')
keyboard
end
%% SET SCOOTER PHYSICAL PARAMETERS
%
% Crr
(coefficient of rolling resistance, dimensionless)
% Af
(frontal area, m^2)
% Cd
(drag coefficient, dimensionless)
% mass
(total mass of vehicle + driver)
% effd
(drive train efficiency - about 70%)
% paux
(auxiliary power, W)
288
g=9.81; rho=1.23;
Crr=0.014; mass=130+75;
Af=0.6; Cd=0.9;
effd=0.77; paux=60;
%% POLARIZATION CURVE
%
% Set number of cells in stack; area, in cm^2, of each stack. We define the
% polarization curve once, here, so we don't have to calculate efficiencies
% eachtime_inside_ the loop. Polarization curve is modeled with an
% analytic formula based on a least-squares fit to experimental data from
% Energy Partners (see reference in main body of thesis)
%
% power_for_density calculates gross power output for a given current
% density. vatmo is the voltage under atmospheric pressure for a given
% current density.
numcells = 56;
max_current_density=1800;
jaxis=[1:1:max_current_density];
for j=1:1:max_current_density,
vatmo(j)=1.00-0.0260*log(j)-(2.015e-4)*j-(1.113e-5)*exp((6.00e-3)*j);
power_for_density(j)=vatmo(j)*j*cellarea*numcells/1000;
end
%% DEFINE PEAKING BATTERY PARAMETERS
%
% We define efficiency and power here. The only reason we care about
% current is that we want to make sure the maximum charge/discharge
% current is not exceeded. Max current is in amps.
%
% numbolder
number of bolder peaking power cells
% capacity
maximum energy storable (J)
% batteryweight weight of batteries (kg)
% effregen
fraction of kinetic energy available
% battenergy
current charge
% initSOC
initial state of charge (battenergy/capacity)
% currSOC
current state of charge
% regenerated
total energy regenerated so far (J)
% friction
total energy lost to friction in braking if no regen (J)
capacity=1*12*(numbolder/6)*60*60;
batteryweight=0.7173*(numbolder/6);
effregen=0.7;
initSOC=0.5; currSOC=initSOC;
battenergy=capacity*currSOC;
regenerated=0; friction=0;
%% DRIVING CYCLE DEFINITION
%
% Load in a driving cycle; uncomment to choose driving cycle.
% Note that FUDS must be converted to km/h as it is in mph.
% Timestep is defined here because different cycles have
% different time intervals.
%load ftp75.txt -ascii; vinput=ftp75; clear ftp75;
%timestep=1;
%load v2.txt -ascii; vinput=v2; clear v2;
%timestep=0.1;
%load v3.txt -ascii; vinput=v3; clear v3;
%timestep=1;
load realtmdc.txt -ascii; vinput=realtmdc; clear realtmdc;
289
timestep=1;
%load fuds.cycle -ascii; vinput=fuds; clear fuds;
%for i=1:1:size(vinput,1); vinput(i,1)=vinput(i,1)*1.609; end
%timestep=1;
%load j1082.txt -ascii
%vinput=j1082; clear j1082
%timestep=0.1;
%load ece40.txt -ascii; vinput=ece40; clear ece40;
%timestep=0.1;
v(1)=0; t(1)=0; cyclelength=size(vinput,1);
%% SMOOTH THE DRIVING CYCLE (BOX SMOOTH)
%
% (uncomment to use. Essentially we define temporary velocity vx
% and then overwrite the original vinput with vx when done smoothing)
%
% vx=vinput;
% vx(1,2)=vinput(1,2); vx(2,2)=vinput(2,2);
% vx(3,2)=vinput(1,2); vx(4,2)=vinput(2,2);
% vx(cyclelength,2) =vinput(cyclelength,2);
% vx(cyclelength-1,2)=vinput(cyclelength-1,2);
% vx(cyclelength-2,2)=vinput(cyclelength-2,2);
% vx(cyclelength-3,2)=vinput(cyclelength-3,2);
% for i=3:1:cyclelength-1
% vx(i,2) = ( 0.50*vinput(i,2) + ...
%
0.30*vinput(i-1,2)+0.20*vinput(i-2,2) );
% end
% vinput=vx;
%% CONVERT KM/H to M/S
for i=1:1:cyclelength
t(i)=vinput(i,1)-timestep;
v(i)=vinput(i,2)/3.6;
if v(i)(40/3.6), v(i)=40/3.6; end
end
%% PAUSE TO ALLOW PLOTTING OF DRIVING CURVE
%
disp('plot your graph now')
keyboard
%% SMOOTH THE DRIVING CYCLE (LOW-PASS)
%
% note that the cutoff frequency - or rather cutoff period % occurs at To = 2pi/s0; periods below To are attenuated
% the smaller the To, the less smoothing. smaller s0 = more smoothing.
%
% If I wanted to close the loop I would use this
% [numc,denc]=feedback(num,den,gclose,1);
s0=1.5; k=1; gclose=1;
num=k;
den=[1 s0];
v2=transpose(lsim(num,den,v,t))*s0;
v=v2; clear v2;
%% EXAMINE DRIVING CYCLE
%
% Evaluate driving cycle to calculate acceleration from finite
290
%
%
%
%
%
%
%
%
%
%
differences of velocity; also, determine what fraction of the
driving cycle is spent accelerating, decelerating, etc.
plustime
minustime
steadytime
if positive acceleration
deceleration
steady velocity, no acceleration
We also accumulate acceleration values to separate out
average of positive accelerations and average of
negative accelerations.
plustime=0; minustime=0; steadytime=0;
for i=2:1:cyclelength
a(i)=(v(i)-v(i-1))/timestep;
if a(i)>0,
aplus(i)=a(i);
aminus(i)=0;
plustime=plustime+1;
elseif a(i)
Последние комментарии
10 часов 35 минут назад
12 часов 4 минут назад
13 часов 17 секунд назад
1 день 11 часов назад
1 день 11 часов назад
1 день 12 часов назад