EFFICIENCY AND PERFORMANCE MEASUREMENTS OF A PDC …

STAGE DIAPHRAGM HYDROGEN COMPRESSOR
In Partial Fulfillment
Master of Science
August, 2009
EFFICIENCY AND PERFORMANCE MEASUREMENTS OF A PDC INC. SINGLE STAGE
DIAPHRAGM HYDROGEN COMPRESSOR
HUMBOLDT STATE UNIVERSITY
Dr. Peter Lehman, Major Professor Date
Dr. Charles Chamberlin, Committee Member Date
Dr. Christopher J. Dugaw, Committee Member Date
Dr. Christopher J. Dugaw, Graduate Coordinator Date
Dr. John Lyon, Dean Date Research & Graduate Studies
ABSTRACT
EFFICIENCY AND PERFORMANCE MEASUREMENTS OF A PDC INC. SINGLE STAGE
DIAPHRAGM HYDROGEN COMPRESSOR
Andrea Leticia Allen
In this thesis I used measured data from Humboldt State University’s hydrogen fueling
station, and ideal gas thermodynamic models, to calculate the specific energy (kWh/kg)
and 2nd Law efficiency for a hydrogen compressor. I used adiabatic, isothermal, and poly-
tropic thermodynamic models and found that the measured specific energy was substan-
tially greater than that predicted using the models. The measured specific energy was
approximately 8 kWh/kg compared to 1-1.5 kWh/kg for the thermodynamic models. The
2nd Law efficiencies for the models varied from approximately 17% relative to the adia-
batic model to approximately 11% relative to the polytropic model. I speculate that the
large discrepancy between the measured and calculated specific energies, and the low effi-
ciencies, is due to the details of the process the compressor uses to compress gas; it is not
well modeled by an ideal gas process. I propose a preliminary model that more accurately
represents the compressor’s operation, based on the Bernoulli equation for fluid flow.
While collecting data for this analysis, my colleagues and I noticed that the power used
by the compressor varies depending on the time of day. We speculated that this was caused
by variation in the incoming line voltages for the compressor. I monitored the line voltages
for ten days and found that they do vary depending on the time of day. The compressor
power variation is correlated with that of the line voltage.
iii
ACKNOWLEDGEMENTS
There are many people who helped me along the way. Many thanks to my committee
members. Peter, thank you for letting me take on this project and putting up with all my
run-on sentences. Charles, thank you for your helpful data insights and all your thoughtful
comments. Chris, thank you for all the LATEX help and introducing me to the wonders of
find and replace.
Thanks to everyone at SERC for providing such a supportive work environment. I’d
especially like to thank Greg Chapman for all the hard work as project manager for the
fueling station, and for the insight about the compressor operation. Scott and Marc, thank
you for your help in all things data acquisition.
Thanks to PGE’s Pacific Energy Center and their tool lending library for use of the
ElitePro data logger. The lending library has been a valuable resource for this and other
projects.
To my colleague, husband and best friend Peter Johnstone, I don’t have enough space
to thank you for everything you’ve helped me with so I’ll keep it short. Thanks for all the
MATLAB help and letting me hack apart you code for my own uses. Thank you for all the
emotional support, shoulder rubs, cooked dinners, washed dishes and thesis conversations.
Most importantly thank you for letting me love you and loving me back.
Last, but certainly not least, thanks to Jumping Bean for motivating me, like no one else
could, to get this thing finished. I can’t wait to meet you.
iv
METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Performance and Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
APPENDIX B: SAMPLE DATA FILE . . . . . . . . . . . . . . . . . . . . . . . . . 75
APPENDIX C: EQUIPMENT DATA SHEETS . . . . . . . . . . . . . . . . . . . . 76
APPENDIX D: MASS FLOW TRANSDUCER CALIBRATION . . . . . . . . . . 97
APPENDIX E: PROGRAM LISTING . . . . . . . . . . . . . . . . . . . . . . . . . 102
vii
Figure Page
1 Ribbon cutting ceremony. Peter Lehman, SERC director cuts the ribbon
while, left to right, congressman Mike Thompson, HSU president Rollin
Richmond, and SERC engineer Greg Chapman, P.E. look on. . . . . . . . . 3
2 HSU hydrogen fueling station with hydrogen powered Prius parked in fu-
eling bay. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3 A schematic of the fueling station. . . . . . . . . . . . . . . . . . . . . . . 6
4 Schematic drawing of Humboldt State University’s Hydrogen Fueling Station 8
5 Specific work curve for an adiabatic compression process with a constant
suction pressure of 190 psig, suction temperature of 12 C, and discharge
pressure varying from 200- 6000 psig. . . . . . . . . . . . . . . . . . . . . 15
6 Specific work curve for an isothermal compression process with a constant
temperature of 12 C, suction pressure of 190 psig, and discharge pressure
varying from 200- 6000 psig. . . . . . . . . . . . . . . . . . . . . . . . . . 17
7 Specific work curves for a polytropic compression process with various
polytropic exponents. For all the curves the suction pressure is 190 psig,
suction temperature is 12 C, and discharge pressure varies from 200- 6000
psig. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
viii
8 Diaphragm compressor is shown at BDC at the start of compression stroke.
At this part of the compression stroke there is low oil pressure and low
hydrogen pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
9 Diaphragm compressor shown mid compression stroke. The piston is trav-
eling from BDC to TDC resulting in intermediate oil pressure and interme-
diate hydrogen pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
10 Diaphragm compressor shown at TDC at the end of compression stroke.
The oil pressure is high enough to flow past the relief valve resulting in
high pressure hydrogen being pushed into storage. . . . . . . . . . . . . . . 24
11 Measured compressor power for filling of both tanks . . . . . . . . . . . . 33
12 Measured compressor power for versus time of day for filling both tanks . . 35
13 Measured hydrogen flow versus compressor discharge pressure for filling
of both tanks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
14 Suction, discharge temperatures, and their difference versus compressor
discharge pressure for filling of both tanks. The dips in temperature occur
when the compressor is shut down. The temperature rises as the compressor
head heats up while running. . . . . . . . . . . . . . . . . . . . . . . . . . 39
15 Measured compressor suction pressure versus compressor discharge pres-
sure for filling of both tanks. Notice the y-axes do not contain the origin. . . 41
ix
16 Calculated ideal adiabatic power versus compressor discharge pressure for
filling of both tanks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
17 Calculated ideal isothermal power versus compressor discharge for filling
of tank A. The curves are nearly identical due to the small temperature
difference between the suction and discharge temperatures. . . . . . . . . . 45
18 Calculated ideal isothermal power versus compressor discharge for filling
of tank B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
19 Calculated polytropic exponent versus compressor discharge pressure for
filling of both tanks. The average value was use to calculate the polytropic
power. Notice the y-axes do not contain the origin. . . . . . . . . . . . . . 48
20 Calculated ideal polytropic power versus compressor discharge pressure for
filling both tanks. For tank A, n=1.027. For tank B, n=1.026 . . . . . . . . 50
21 All power curves, calculated and measured, versus compressor discharge
pressure for filling of both tanks. Note the polytropic and isothermal (using
the suction temperature) power curves are difficult to read because they are
all very similar and overlap each other. . . . . . . . . . . . . . . . . . . . . 52
22 Specific energy as measured and calculated for each ideal thermodynamic
model for both tanks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
23 2nd Law efficiency relative to each ideal thermodynamic model for both tanks. 55
x
24 Percentage of energy used for compression compared to the LHV of hydro-
gen as measured and for each ideal thermodynamic model for each tank. . . 56
25 Compressor motor efficiency curve based on the manufacturer’s data. . . . . 61
26 2nd Law efficiency as versus flow rate for compressor motor efficiencies of
55 and 65% . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
27 Incoming compressor line voltages, with respect to ground, versus time of
day. Notice the y-axis does not contain the origin. . . . . . . . . . . . . . . 65
28 Compressor power versus line 1 voltage. Notice the y-axis does not contain
the origin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
the origin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
the origin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
31 Schematic of experimental setup used to calibrate the mass flow transducer . 98
32 Apparent hydrogen flow versus mass flow transducer voltage. This curve
was used to find the voltage offset at zero. . . . . . . . . . . . . . . . . . . 100
33 Hydrogen flow measured by mass flow transducer versus measured flow.
This plot was used to find the calibration curve for the MFT. . . . . . . . . 101
xi
1 Data collected by the DAQ system and units . . . . . . . . . . . . . . . . . 25
2 Data acquisition system measurements, transducers, and write thresholds . . 27
3 Calculated power for different oil flow rates and motor efficiencies and 2nd
Law efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
LIST OF VARIABLES
Symbol Definition Units k Ratio of specific heats Dimensionless, specific to gas
(1.41 for hydrogen) m Mass flow rate kg/sec n Polytropic exponent Dimensionless P0 Initial pressure (Suction) Pa or psig P1 Final pressure (Discharge) Pa or psig R Gas constant (specific to gas) 4.1243 kJ/kg-K for hydrogen T Temperature C or K T0 Initial (Suction) temperature C or K T1 Final (Discharge) temperature C or K V Velocity m/s v0 Initial specific volume m3/kg W Work J w Specific work J/kg W Power W Z Elevation m
1
INTRODUCTION
Hydrogen based transportation is becoming a feasible alternative to petroleum based trans-
portation. A major advantage of hydrogen over petroleum is that it can be produced from a
variety of fuel feedstocks, including renewable energy. When hydrogen is used as a vehicle
fuel, the only emissions are water, and in the case of an internal combustion engine, NOx.
For hydrogen to become a mainstream fuel, the economics of the refueling infrastruc-
ture need to be favorable. Hydrogen has to be able to compete economically with petroleum
for auto manufacturers, drivers, and fueling station owners to make the switch. An impor-
tant aspect of the economics is the energy required to produce, compress and dispense
hydrogen fuel. This thesis will focus on the compression fraction of the energy required,
by studying the compressor at Humboldt State University’s hydrogen fueling station. I
will calculate the efficiency of the compressor and the energy required, on a mass basis, to
compress hydrogen produced at the station to our storage pressure. The goal of this anal-
ysis is to provide efficiency and performance values for hydrogen compressors for use in
economic modeling of hydrogen transportation.
Economic models can play an important part in determining the most cost effective
ways to build a hydrogen infrastructure. Currently most economic models rely on assump-
tions, rather than empirical data, for the costs and energy requirements associated with
hydrogen refueling infrastructure. There are now hundreds of fueling stations worldwide
(Fuel Cells 2000, 2009) that are in operation and these stations should be used to provide
empirical values to refine economic models. To date there have been very few studies that
use empirical data to determine the energy requirements to make and compress hydrogen.
2
3
System Overview
On September 4, 2008 Humboldt State University (HSU) celebrated the grand opening of
its hydrogen fueling station, Figure 1. The fueling station was designed by engineers at
the Schatz Energy Research Center (SERC) at HSU, and built largely by SERC and HSU
Plant Operations personnel. Funding for the station was provided by Chevron Technology
Ventures, CalTrans, North Coast Air Quality Management District, HSU and SERC. The
hydrogen station is the 23rd and currently the northernmost and only rural link in Califor-
nia’s Hydrogen Highway Network (CaH2Net), Figure 2.
Figure 1: Ribbon cutting ceremony. Peter Lehman, SERC director cuts the ribbon while, left to right, congressman Mike Thompson, HSU president Rollin Richmond, and SERC engineer Greg Chapman, P.E. look on.
4
Figure 2: HSU hydrogen fueling station with hydrogen powered Prius parked in fueling bay.
5
Hydrogen for the fueling station is generated, compressed, and stored on site. A simple
schematic of input and output flows can be found in Figure 3. Hydrogen is generated by
a Proton Energy Systems1 HOGEN S40 proton exchange membrane electrolyzer capable
of producing 2.3kg of hydrogen per day at a pressure of 200 pounds per square inch gauge
(psig). The medium pressure hydrogen is pressurized to 6000 psig for storage by a PDC2
single stage diaphragm compressor. High pressure gas is stored in CPI3 storage tanks that
hold a total of 12 kg of hydrogen at 6000 psig. FTI4 manufactured the dispenser, which
is capable of filling a vehicle storage tank to 5000 psig. A more detailed schematic of the
fueling station can be found in Figure 4.
Currently the station is serving one vehicle, a Toyota Prius converted by Quantum Tech-
nologies to run on hydrogen5. We estimate the fueling station has the capacity to serve 3-4
total vehicles. An important part of all of SERC’s projects is data collection and dissem-
1www.protonenergy.com 2www.pdcmachines.com 3www.cp-industries.com 4www.fuelingtech.com 5The Prius uses an internal combustion engine, not a fuel cell like most of new hydrogen powered vehicles.
6
7
ination. Because the design work was done by SERC, we were able to incorporate data
collection into the system. Our data acquisition (DAQ) system measures and records:
• Power used by the electrolyzer
• Power used by the compressor
• Pressure of the hydrogen exiting the electrolyzer, which is also the compressor suction pressure
• Pressure of each storage tank. The discharge pressure of the compressor corresponds to the pressure of the tank with the lowest pressure. If both tanks are at the same pressure the compressor discharge pressure corresponds to the uniform tank pressure6
• Hydrogen flow, either from the electrolyzer (hydrogen generated) or to the compressor (hydrogen compressed), depending on a valve setting.
• Two temperatures. The thermocouples have the ability to be placed in four different locations see Figure 4. The location depends on the data desired; for this report they were located on the compressor suction and discharge lines.
Our DAQ system enables SERC to report accurate information regarding the energy
requirements for producing and delivering high purity, high pressure hydrogen.
6The station was designed for cascade fueling and bulk storage tank filling. This means that when a car is fueled, gas is taken from one tank at a time, for our station, tank A followed by tank B. This allows one tank to end up at a higher pressure than it would have been if fueling took place from both tanks simultaneously. This is useful when more than one vehicle fuels before the storage tanks are fully replenished with hydrogen.
Bulk filling of the tanks means that both tanks are filled simultaneously. However, because of check valves in the lines from the compressor discharge to the tanks’ inlets, the tank with the lower pressure is filled exclusively until it reaches the same pressure as the higher pressure tank. Once the tanks are at the same pressure they are filled simultaneously. This is not relevant to this analysis as the tanks were filled independently to provide more data. For this analysis the compressor discharge pressure corresponded to the pressure in the tank being filled.
8
Fi gu
re 4:
Sc he
m at
ic dr
aw in
g of
H um
bo ld
tS ta
te U
ni ve
rs ity
’s H
yd ro
ge n
Fu el
in g
St at
io n
LITERATURE REVIEW
Hydrogen Transportation
There has been interest in hydrogen based transportation for as long as there has been in-
terest in the internal combustion engine. The first internal combustion engine was built by
Francois Isaac de Rivaz in 1807 using hydrogen for fuel (Holland and Provenzano, 2007).
Modern automakers have made cars that use hydrogen as a fuel since at least the 1940’s
with several vehicles manufactured from the late 1960’s through modern times (H2 mobil-
ity, 2009). SERC built the first highway legal proton exchange membrane (PEM) hydrogen
fuel cell vehicle in 1998 (Schatz Energy Research Center, 2009). Currently many major
automobile companies are developing hydrogen vehicles. Honda has begun leasing 200
PEM fuel cell vehicles, called the Clarity, in Southern California (American Honda Motor
Co., 2009). Daimler also has a fuel cell vehicle under development, the Mercedes-Benz
F-Cell, expected to be available to the public in 2010 (Mercedes-Benz, 2009). American
auto manufacturers are also developing hydrogen cars. General Motors launched “Project
Driveway” in 2008, deploying a test fleet of fuel cell Chevrolet Equinox vehicles in New
York City, Washington D.C. and Southern California (General Motor Company, 2009).
Ford has a plug-in hybrid fuel cell vehicle, the Ford Edge, being road tested, as well as 30
fuel cell Focus vehicles that have accumulated over 300,000 miles as of this writing (Ford
Motor Company, 2009).
Even though hydrogen as a transportation fuel has been around for as long as there
has been internal combustion, it has never threatened to replace gasoline as the dominant
transportation fuel. Petroleum, and therefore gasoline, has been inexpensive7 since World
7The real price of gasoline at the pump has generally decreased from the 1940s until the terrorist attacks on September 11, 2001 and the start of the Iraq and Afghanistan wars. The major exceptions to this trend occurred in 1970s and 80s and were caused by the OPEC embargo and the Iraq/Iran wars (Energy Information
9
10
War II (Vatter and Walker, 1996), and we did not know the long term consequence of fos-
sil fuel use, namely global climate change, until relatively recently. A major hurdle for
switching to a new transportation fuel is building new infrastructure; there is currently a
vast infrastructure in place to deliver gasoline to drivers, but very little for delivering hy-
drogen (Department of Energy, 2009). Switching to an alternative fuel, such as hydrogen
would pose a ‘chicken and egg’ problem. Auto manufacturers do not want to make hy-
drogen vehicles if there is no infrastructure to support them and investors do not want to
build hydrogen fueling stations if there is not enough demand for the fuel due to a lack of
available vehicles.
To help solve this problem, the California state government initiated the California
Hydrogen Highway Network (CaH2Net) in April 2004 by Executive Order S-07-04. The
stated purposes of the order are to encourage the transition to hydrogen based transportation
in California, reduce dependence on foreign oil, reduce greenhouse gas emissions, improve
air quality, and stimulate the economy (Schwarzenegger, 2004).
For a transition to a hydrogen based transportation system to be successful, it has to
be economically feasible. A major aspect to the economics of hydrogen transportation is
the production of hydrogen itself. Hydrogen, though the most abundant element in the
universe, does not occur naturally on earth in its elemental state. It is always found in
molecules with other elements. To produce elemental hydrogen, it must be cracked out of
molecules that contain it, either chemically or electrolytically. Thus, hydrogen is an energy
carrier rather than a fuel source like fossil fuels. This means that there is more energy
associated with producing hydrogen than energy contained in the hydrogen.
Since hydrogen is an energy carrier rather than an energy source, the amount of energy
required to produce it affects the economics. For my analysis I will use the term specific
Agency, 2009). These spikes also created a demand for fuel efficient vehicles.
11
energy to quantify the energy intensity of producing hydrogen. Specific energy is the en-
ergy required to produce a unit of mass of hydrogen, with the units of kWh/kg. Specific
energy can be calculated for any step, or combination of steps, in the production chain,
i.e. the specific energy for electrolysis, compression, or the entire generation, compression
and dispensing process. The specific energy for the entire production chain needs to be
accounted for to get a realistic view of the economics of hydrogen transportation. For the
purpose of this analysis, I will focus on one aspect of the chain, compression.
Hydrogen in its elemental state at normal everyday conditions (temperature and pres-
sures) is a diatomic gas. For hydrogen to have a large enough volumetric energy density
to be useful as a vehicle fuel it must be compressed. Gasoline has the advantage of be-
ing a liquid with a high volumetric energy density. The average volumetric energy density
for regular gasoline is 8,700 Wh/liter. Hydrogen at atmospheric pressure and temperature
contains less than 35 Wh/liter (Fuel Cells for Power, 2009). Thus compression is necessary.
There are several studies that estimate the specific energy to compress hydrogen gas,
most of those coming out of the Institute for Transportation Studies at the University of
California, Davis (ITS-Davis). ITS-Davis has been a leader in studying the transition to
hydrogen transportation. Yang and Ogden (2007) estimated that the energy for hydrogen
compression is 0.7-1 kWh/kgH2, approximately 2-3% of the lower heating value (LHV)8
of the the hydrogen. The authors do not state the assumed type of compressor, suction and
discharge pressures, or the mass flow rate of hydrogen through the compressor. In another
ITS-Davis paper, Weinert et al. (2007) state that, for a diaphragm compressor with an
isentropic efficiency of 65% operating between 200 psig and 6250 psig, the specific energy
for compression is 2.1 kWh/kgH2, approximately 6% of the LHV. Again the assumed mass
flow rate of hydrogen is not stated. However, the suction and discharge pressures cited by
8The lower heating value of hydrogen is 33.39 kWh/kg
12
Wienert et al. are very close to the operating pressures of the HSU compressor, and both are
diaphragm compressors. Amos (1998) states that the energy to compress hydrogen from
atmospheric pressure to 2800 psig can be 8-10% of the energy content of hydrogen 9.
Each of these studies use thermodynamic models (to be discussed in greater detail
below) to calculate the specific energy, and not empirical data from actual compressors.
One company, Linde, has an ionic compressor, as part of a hydrogen refueler package,
that they state uses 2.3 kWh/kg. It was introduced into US markets at the 2009 National
Hydrogen Association conference (Linde Group, 2009). I was unable to find similar data
from other compressor manufacturers.
Definitions of Efficiency
In this analysis the definition of efficiency that I will be using is 2nd Law efficiency based
on the second law of thermodynamics. This definition of efficiency compares the energy
of an ideal (reversible) process to the energy required by a real process. For a compressor,
or any energy consuming device, it is given below (Cengel and Boles, 1998).
ηII = Wideal
Wideal is the ideal process work requirement (J)
Wactual is the work required for the real process (J)
I will use this definition to find the efficiency of the compressor based on various ideal
gas thermodynamic models.
9Amos did not state whether the LHV or HHV was used for this calculation.
13
Thermodynamic models
In this section I present the three thermodynamic models of compression used for my anal-
ysis: adiabatic, isothermal, and polytropic. Each model will be used to calculate the mini-
mum required specific work. Power can then be found from the specific work by multiply-
ing by the mass flow rate of hydrogen, as shown in equation (2).
W = wm (2)
m is hydrogen mass flow (kg/s)
For all the models I assume that hydrogen is an ideal gas. This is a good assumption for
our compressor at the suction pressure and temperature and low discharge pressures and
temperatures. The compressibility factor, Z, varies from 1.0088 at the suction temperature
and pressure (approximately 35 C and 200 psig, to 1.04511 at 1160 psig and 65 C (Perry
et al., 1997). At higher discharge pressures hydrogen acts less as an ideal gas; Z is 1.2359 at
5800 psig and 65 C. Assuming that hydrogen is an idea gas affects the 2nd Law efficiency
of each of the models. However, the deviation between treating hydrogen as ideal versus a
real gas is less than 10%. Appendix A shows the analysis I used to compare the effects of
treating hydrogen as a real rather than ideal gas.
Adiabatic Compression
An adiabatic process is one in which there is no heat transfer into or out of the system. For a
reversible process this would indicate that it is also isentropic. An ideal gas that undergoes
14
adiabatic compression has a temperature increase dependent on the compression ratio, or
the ratio of inlet pressure to the outlet pressure. The equation for finding the specific work
required by an adiabatic process is given below (Cengel and Boles, 1998).
w = kRT0
k − 1
where w is the specific work required (J/kg)
k is the ratio of specific heats of the gas (1.41 for hydrogen)
R is the gas constant (4.1243 kJ/kgK for hydrogen)
T0 is the initial Temperature (K)
P0 is the initial pressure (Pa)
P1 is the final pressure (Pa)
An example adiabatic compression process is shown in Figure 5 below. To calculate
this curve I assumed conditions similar to our compressor:
• Suction pressure of 190 psig,
• Suction temperature of 12 C,
• Discharge pressure varies from 200 psig to 6000 psig.
Notice that as the discharge pressure increases the specific work required for compression
also increases.
15
Figure 5: Specific work curve for an adiabatic compression process with a constant suction pressure of 190 psig, suction temperature of 12 C, and discharge pressure varying from 200- 6000 psig.
16
Isothermal Compression
An isothermal process is one in which the temperature remains constant throughout the
process. The equation for specific work for an ideal gas undergoing isothermal compression
is given by equation (4) below (Cengel and Boles, 1998).
w = RTln ( P1
where T is the process temperature (K)
A lower process temperature leads to a smaller specific work. An actual isothermal
compressor would require some type of cooling to keep the gas at a constant temperature
throughout the process. This adds complication, and hence cost, both initial and ongoing, to
the compressor. Large compressors, however, often have some type of cooling mechanism
to improve their efficiency. An example curve for an isothermal compression process is
given in Figure 6 below. The temperature is assumed to be 12 C with a suction pressure of
190 psig and discharge pressure varying from 200-6000psig. Notice that the specific work
is lower than for the adiabatic curve. At 6000 psig discharge pressure, the specific work
for the adiabatic process is approximately 7000 kJ/kg while for the isothermal process, the
specific work is approximately 4750 kJ/kg, 32% less.
17
0
1000
2000
3000
4000
5000
6000
7000
8000
Sp ec
ific W
or k
(k J/
Discharge Pressure (psig)
Figure 6: Specific work curve for an isothermal compression process with a constant temperature of 12 C, suction pressure of 190 psig, and discharge pressure varying from 200- 6000 psig.
18
A polytropic process is one defined by equation (5) below.
PV n = C (5)
n is an exponent that depends on the process.
If n = 1 the process is isothermal; if n = k the process is adiabatic.
The equation for specific work for a polytropic compression process is given below
(Cengel and Boles, 1998).
− 1
(6)
Example curves for several polytropic exponents are given in Figure 7 below for n =
1.1, 1.2, and 1.3. All of the curves assume:
• Suction temperature of 12 C
• Suction pressure 190 psig
19
0
1000
2000
3000
4000
5000
6000
7000
8000
S pe
cif ic
W or
k (k
J/ kg
n=1.1
n=1.2
n=1.3
Figure 7: Specific work curves for a polytropic compression process with various polytropic exponents. For all the curves the suction pressure is 190 psig, suction temperature is 12 C, and discharge pressure varies from 200- 6000 psig.
20
There was limited literature available concerning hydrogen compressor efficiency. Amos
(1998) states that large compressors have an efficiency of 65-70%, while small compressors
have an efficiency of 40-50%, though there is no quantification for “big” and “small”.
The Department of Energy’s Hydrogen Analysis (H2A) Project, developed a Microsoft
ExcelTM based economic analysis tool to estimate the costs of different types of hydrogen
fueling stations. For compressors the default assumption is an isentropic (adiabatic) model
with an efficiency of 70% (H2A Group, 2006). The default can be changed by the user if
sufficient data are available.
An adiabatic process is a common assumption for positive displacement compressors.
Positive displacement compressors, of which a diaphragm compressor is one type, are usu-
ally considered adiabatic (which for a reversible process would also be isentropic) (Marks,
1996).
PDC Inc. Single Stage Diaphragm Compressor
The compressor under study is a single stage diaphragm compressor configured specifically
for our application by Pressure Dynamic Consultants (PDC Machines, Inc.). The compres-
sor can compress hydrogen with an inlet pressure range of 75-200 psig to a maximum
discharge pressure of 6000 psig.
The diaphragm compressor has a head consisting of two chambers separated by a flex-
ible metal diaphragm as shown in Figure 8. Hydrogen gas is confined to one chamber and
oil to the other. The hydrogen suction and discharge lines have one-way poppet valves (a
type of check valve) to prevent reverse hydrogen flow. The oil injection line has a check
valve allowing oil flow in one direction. Oil flow out of the oil side occurs through a pres-
21
sure relief valve set to open once the desired oil pressure has been reached, 6800 psig for
our compressor. Oil pressure in the head is increased with a piston, causing the diaphragm
to flex, compressing the hydrogen.
Figure 8: Diaphragm compressor is shown at BDC at the start of compression stroke. At this part of the compression stroke there is low oil pressure and low hydrogen pressure.
A compression cycle works as follows. During the inlet stroke the piston moves to its
bottom dead center (BDC) position and the pressure on the gas side of the cavity decreases
allowing the suction valve to open and gas to flow into the cavity as shown in Figure 8. Gas
flows into the cavity until the piston reaches BDC. As the piston starts to move upward, the
22
suction check valve closes and the gas compression begins. As the piston moves towards
top dead center (TDC), pressure inside the gas cavity increases until the pressure inside the
cavity is greater than the storage pressure and the discharge check valve opens allowing
gas to flow to storage as shown in Figure 9. Flow will continue out of the cavity until
the piston reaches TDC. At TDC the diaphragm is completely flexed ensuring most of the
gas in the cavity is pushed into storage as shown in Figure 10. The piston then returns to
BDC stopping the discharge flow. During the downward stroke to BDC, pressure inside
the hydrogen cavity decreases allowing gas to flow in through the suction valve starting the
cycle again. To replace the oil that flows out of the cavity through the pressure relief valve,
make-up oil is injected at the beginning of each stroke.
23
Figure 9: Diaphragm compressor shown mid compression stroke. The piston is travel- ing from BDC to TDC resulting in intermediate oil pressure and intermediate hydrogen pressure.
24
Figure 10: Diaphragm compressor shown at TDC at the end of compression stroke. The oil pressure is high enough to flow past the relief valve resulting in high pressure hydrogen being pushed into storage.
METHODS
The data for my analysis were recorded during the initial filling of the fueling station stor-
age tanks between June 28, 2008 and July 7, 2008. The tanks were filled independently to
gather more data and have replicate measurements. I extracted the data during the times
when the system was in standby to eliminate standby losses in my analysis. For my data
analysis I used The MathworksTM MATLAB R© programming language to write analysis
scripts for the data files created by the data acquisition (DAQ) system. A more detailed
account of data analysis is given below in the Data Analysis section.
Data Acquisition
Our DAQ system records the information shown in Table 1.
Table 1: Data collected by the DAQ system and units
Measurement Range Units Time Stamp 00:00:00 -1:00:00 HH:MM:SS Program Run Hours none Hours Compressor Suction Temperature -100−260 C Compressor Discharge Temperature -100−260 C Tank A Pressure 0−7,000 psig Tank B Pressure 0−7,000 psig Hydrogen Flow 0−100 slm Suction Pressure 0−250 psig Electrolyzer Power 0−20,000 Watts Compressor Power 0−6,000 Watts
Data acquisition is controlled by an AdvantechTM field computer running a LabviewTM
program written specifically for our application. Data are measured on a digital back plane
with separate differential channels for each transducer. The channels are scanned at 1000
25
26
Hz. To reduce noise in the signals, 100 sequential points from each channel are averaged
together resulting in a time average sampling rate of 10 samples per second for all the
channels. To reduce the amount of data actually written to a data file, a sample (a set of
values for all 10 channels) must meet one of the following conditions to be written to a data
file.
• The data acquisition program is reset.
• A new data file is created (this happens at midnight or if there is a safety shutdown).
• A sample differs from the previously written data point by an amount equal to or greater than that channel’s threshold. Thresholds are shown in Table 2.
• More than 60 seconds have passed since the last data write.
• A minimum interval of 2 seconds has passed.
If none of the conditions are met, the sample is discarded and the next sample is tested
against the conditions. Table 2 lists the transducers we are using for the measurements
needed for this thesis, the write thresholds for each channel and the accuracy and response
time for the transducers. An example data file is shown in Appendix B and specification
sheets for components of our DAQ system are given in Appendix C.
Data Analysis
I used the MATLAB R© programming language for all my data processing. For each row of
data in the data files I calculated the power that would be required for each of the thermo-
dynamic models, using equations (7) - (10) presented below, assuming the same operating
conditions as those measured . I then used numerical integration to estimate the total energy
actually used by the compressor, and the minimum required energy for each of the models
27
28
using equation (11). To find the specific energy, I needed to estimate the total amount of
hydrogen that was compressed. I again used numerical integration for the hydrogen flow
rate using equation (12). I calculated the specific energy based on compressor measure-
ments and for each of the models by dividing the total energy consumption by the total
mass of hydrogen compressed, see equation (13). To find the 2nd Law efficiencies I divided
the specific energies predicted for each of the models by the measured specific energy, as
shown in equation (15).
Thermodynamic models
To find the ideal power consumption of the compressor, I used three different thermody-
namic models: adiabatic, isothermal, and polytropic. The inputs to the models were the
measured operating temperatures and pressures of the suction and discharge lines, as well
as the measured mass flow into the compressor. Each of the inputs was converted to appro-
priate units within the MATLAB R© scripts.
Adiabatic Power
To find the ideal adiabatic power I used the following equation.
W = mwadiabatic (7)
m is the measured hydrogen flow rate (kg/s)
wadiabatic is the specific work (J/kg) required for an ideal adiabatic process with
the same operating parameters, i.e. suction and discharge pressures.
29
The equation for power of an isothermal compression process is
W = mwisothermal (8)
wherewisothermal is the specific work required for an ideal isothermal process with the same
operating parameters (J/kg)
Polytropic Power
For the polytropic case I first needed to calculate the polytropic exponent n. According
to Marks’ Handbook (1996), if the suction and discharge temperatures and pressures are
known, n can by found by using equation (9) below.
n = 1
) (9)
I found n for each row in the data file and then found the average for the all the data. I
used the average to calculate the power for the polytropic process. The power required for
a polytropic compression process is then given by:
W = mwpolytropic (10)
where wpolytropic is the specific work required for an ideal polytropic process with the same
operating parameters (J/kg)
Specific Energy and Efficiency
To find the total energy required for each of the models, as well as the measured data, I
used the trapezoidal Riemann sum, given in equation (11) to numerically integrate power
over time. The power for each time step (a row in the data file) was either measured
or found using equations (7) - (10). I then found the energy for each time step in the
data file and summed them to get the total energy, in kWh, for filling each tank. To find
the total hydrogen mass compressed, I again used the trapezoidal Riemann sum, given in
equation (12). The mass flow measured by the mass flow transducer for each row of data
was converted from slm to kg/hr. I assumed standard conditions of 1 atmosphere and 20C
to do the conversion, based on the mass flow transducer’s calibration. I calculated the
mass compressed for each row in the data file and summed them to estimate the total mass
of hydrogen compressed for each tank filling. I then used equation (13) to calculate the
specific energy. I also found the specific energy as a percentage of the of energy contained
in the hydrogen on a LHV basis using equation (14).
E = N∑
j=0
where E is the total energy used for compression (kWh)
tj is the time stamp in the jth row of the data file (hr)
Wj is the power (measured or calculated) in the jth row of the data file (W)
mtotal = N∑
j=0
where mtotal is the total mass of hydrogen compressed (kg)
tj is the time stamp in the jth row of the data file (hr)
31
mj is the mass flow in the jth row of the data file (kg/hr)
w = E
mtotal
(13)
where w is the specific energy for the given process (kWh/kg)
E is the total energy required for compression (kWh)
mtotal is the total mass of hydrogen compressed (kg)
%LHV = w
LHV ∗ 100% (14)
where %LHV is the percentage of energy contained in 1 kg of hydrogen, on a LHV
basis, that is used in the compression process
w is the specific energy for the give process (kWh/kg)
LHV is 33.39 kWh/kg for hydrogen
To find the 2nd Law efficiency relative to each of the models I divided the theoretical
specific energy for each by the measured specific energy as shown in equation (15) below.
ηII = wcalc
where ηII is the 2nd Law efficiency
wcalc is the ideal specific work for the thermodynamic model (kWh/kg)
wmeasured is the measured specific work of the fill (kWh/kg)
RESULTS
This chapter presents the results of my analysis. It is divided into three sections. The first
section presents the data measured during the tank fills, the second section presents the
predicted power results for the various thermodynamic models using the measured data as
inputs, and the third section presents the specific energy and 2nd Law efficiencies of each
of the models for both tank fills.
Measured Data
This section presents the data measured from the compressor. The suction and discharge
temperatures and pressures, and the hydrogen flow rates were used as input to the thermo-
dynamic models that I investigated.
Power
Figure 11 presents the measured power used to fill tanks A and B. The fluctuations seen are
caused by fluctuations in the incoming line voltage and are discussed in more detail in the
Discussion chapter.
Measured Power
Po we
r ( W
Po we
r ( W
Figure 11: Measured compressor power for filling of both tanks
34
Looking at the measured compressor energy versus pressure, note that there are sur-
prising fluctuations in the power. When the same data are looked at in a different way a
more obvious pattern emerges: the power fluctuates as a function of the time of the day as
seen in Figure 12.
35
600
650
700
750
800
850
900
950
1000
0 2 4 6 8 10 12 14 16 18 20 22 24
Compressor Power
Po we
r ( W
600
650
700
750
800
850
900
950
1000
0 2 4 6 8 10 12 14 16 18 20 22 24
Compressor Power
Po we
r ( W
Hour of Day
(b) Tank B
Figure 12: Measured compressor power for versus time of day for filling both tanks
36
Hydrogen Flow
Figure 13 presents measurements of hydrogen flow versus discharge pressure for each tank.
The spikes in the flow rate occur when the compressor is first turned on and the ballast tank
is full. The system was initially powered down at the end of the day so that it did not
run unsupervised before we were confident that everything was running properly. Because
of this the compressor was cycled on and off several times during both fills causing the
observed spikes. Oscillations in the flow rate result as the electrolyzer goes into and out
of standby mode. The flow decreases until the ballast tank pressure drops to 185 psig,
at which point the electrolyzer comes out of standby and flow increases. This cycling is
clearly seen at higher pressures (>5000 psig).
37
0
10
20
30
40
50
60
Hydrogen Flow
Hy dr
og en
F lo
w (s
Hydrogen Flow
Hy dr
og en
F lo
w (s
Figure 13: Measured hydrogen flow versus compressor discharge pressure for filling of both tanks.
38
Temperature
Figure 14 presents the measured temperature for the suction and discharge lines of the
compressor, and the difference between them. The suction and discharge temperatures
generally track each other and the difference between them is relatively constant when the
compressor is running.
Te m
pe ra
tu re
( o C)
Te m
pe ra
tu re
( o C)
(b) Tank B
Figure 14: Suction, discharge temperatures, and their difference versus compressor discharge pressure for filling of both tanks. The dips in temperature occur when the compressor is shut down. The temperature rises as the compressor head heats up while running.
40
Suction Pressure
This section presents the relationship between the suction pressure and the discharge pres-
sure for both fills. The suction pressure steadily increases until the it reaches 200 psig, the
maximum pressure output of the electrolyzer. When the ballast tank reaches this pressure,
the electrolyzer goes into standby and does not resume hydrogen production until the bal-
last pressure decreases to 185 psig. This is the reason for the cycling at higher discharge
pressures that is observed for both tanks. Spikes in the suction pressure result when the
compressor is stopped and the electrolyzer has a chance to fill the ballast tank before the
compressor is turned on again. This is the case when the system is shut down overnight.
Dips in the suction pressure happen when the electrolyzer is not producing hydrogen while
the compressor is running.
Suction Pressure
Su ct
io n
Pr es
su re
(p sig
Suction Pressure
Su ct
io n
Pr es
su re
(p sig
42
Thermodynamic Models
Below are the theoretical minimun power requirements, as a function of discharge pressure,
for each of the models investigated using the measured inputs shown above.
Adiabatic
Figure 16 shows the results for the adiabatic model for filling both tanks. The spikes and
dips in the power curves are caused by spikes and dips in the hydrogen flow and temperature
observed in Figures 13 and 14 above.
43
0
50
100
150
200
250
Adiabatic Power
Id ea
Adiabatic Power
Id ea
Figure 16: Calculated ideal adiabatic power versus compressor discharge pressure for filling of both tanks.
44
Isothermal
For the isothermal model I produced ideal power curves using both the suction temperature
and the discharge temperature. Figure 17 shows the predicted power curves for tank A
for both the suction and discharge temperatures. Figure 18 shows the predicted power
curves for tank B for both the suction and discharge temperatures. For both tanks the
predicted power curves are nearly identical because of the small difference in the suction
and discharge temperatures.
Isothermal Power (Suction Temperature)
Isothermal Power (Discharge Temperature)
(b) Discharge Temperature
Figure 17: Calculated ideal isothermal power versus compressor discharge for filling of tank A. The curves are nearly identical due to the small temperature difference between the suction and discharge temperatures.
46
0
50
100
150
200
250
Isothermal Power (Suction Temperature)
Isothermal Power (Discharge Temperature)
Figure 18: Calculated ideal isothermal power versus compressor discharge for filling of tank B.
47
Polytropic
Figure 19 shows the instantaneous polytropic exponents calculated at each time step for
filling both tanks. Rather than using the instantaneous polytropic exponent to calculate the
power at each time step, I used the average value for each tank seen as the horizontal line in
Figures 19(a) and 19(b). The average value for tank A is 1.026 and 1.027 for tank B. The
exponent for each tank fill is close to 1, indicating the process is close to isothermal. This
is why the power predicted by the polytropic model, shown below in Figure 20, is similar
to the power predicted by the isothermal model shown in Figures 17 and 18.
48
0.98
1
1.02
1.04
1.06
1.08
n
n
n ave = 1.026
(b) Tank B
Figure 19: Calculated polytropic exponent versus compressor discharge pressure for filling of both tanks. The average value was use to calculate the polytropic power. Notice the y-axes do not contain the origin.
49
Figure 20 shows the results for the polytropic model for filling both tanks. The spikes
and dips in the power curves are caused by spikes and dips in the hydrogen flow and tem-
perature observed in Figures 13 and 14 above.
50
0
50
100
150
200
250
Polytropic Power
Id ea
Polytropic Power
Id ea
51
Measured and Theoretical Power
In this section the measured power and ideal calculated powers are presented together for
comparison. All the models investigated predicted smaller power requirements, by a factor
of 4-5 depending of the model, than the compressor actually used. In the following chapter
I discuss reasons I think this discrepancy exists.
52
0
200
400
600
800
1000
Measured Power Adiabatic Power Polytropic Power Isothermal Power (Suction Temperature) Isothermal Power (Discharge Temperature)
Po we
r ( W
Measured Power Adiabatic Power Polytropic Power Isothermal Power (Suction Temperature) Isothermal Power (Discharge Temperature)
Po we
r ( W
(b) Tank B
Figure 21: All power curves, calculated and measured, versus compressor discharge pressure for filling of both tanks. Note the polytropic and isothermal (using the suction temperature) power curves are difficult to read because they are all very similar and overlap each other.
53
Performance and Efficiency
This section presents the measured and calculated specific energies, the 2nd law efficiencies,
and the energy used for the compression process as a percentage of energy contained in the
hydrogen. These values facilitate comparisons to other compressors, and are the most
widely cited performance measures in the literature.
Figure 22 shows the specific energy as measured and as calculated for each of the ideal
thermodynamic models, for both tanks. The actual specific energy used by the compressor
is much higher than what was predicted in all of the models, indicating a low 2nd Law effi-
ciency, as seen in Figure 23. Figure 24 provides a different way of presenting the efficiency;
the power required for compression is given as a percentage of the energy contained in the
hydrogen that is compressed.
Sp ec
ific E
ne rg
y (k
W h/
kg )
Figure 22: Specific energy as measured and calculated for each ideal thermodynamic model for both tanks.
55
0
5
10
15
20
2nd L
aw E
ffi cie
nc y
(% )
Figure 23: 2nd Law efficiency relative to each ideal thermodynamic model for both tanks.
56
0
5
10
15
20
25
% L
HV
Figure 24: Percentage of energy used for compression compared to the LHV of hydrogen as measured and for each ideal thermodynamic model for each tank.
DISCUSSION
As can be see from Figure 21, none of the ideal thermodynamic models closely approx-
imates the actual behavior of the compressor. The adiabatic model predicts the highest
specific energy and therefore has the highest 2nd Law efficiency. However, the predicted
curve (Figure 16) does not approximate the measured power curve of the compressor (Fig-
ure 11). One reason the compressor is not well modeled by an adiabatic process is that
the compressor head is very large compared to the compressor cavity and since hydrogen is
such an effective heat transfer medium, the temperature increase caused by the compression
is quickly dissipated to the compressor head and across the diaphragm to the oil.
The polytropic and isothermal models provide similar predictions of specific energy
and efficiency. This is because the estimated polytropic exponent for both tank fills is
close to 1, which is characteristic of an isothermal process. Though there is a temperature
difference between the gas at the suction and discharge of the compressor, the difference is
small, approximately 10% or 30 K. The process has a small temperature difference, again
because hydrogen is an effective heat transfer medium. The gas is able to quickly transfer
heat its surroundings, as noted above.
The measured specific energy is larger than those stated in the literature. Values in the
literature range from 0.7-2.1 kWh/kg. I measured a specific energy of 8.3 kWh/kg for both
tank fillings. This is 400 -1200% greater than literature values. Some factors that may
contribute to this large discrepancy are the size of compressor, the size of the compressor
motor, and the fact that the compressor is single stage.
Our compressor has a small capacity; it was sized to match the electrolyzer which can
produce about 2 kgH2 a day. This is small for a fueling station. Larger compressors in
general are more efficient than smaller ones (Amos, 1998) and this could affect the energy
57
58
requirements.
The motor that runs the compressor is grossly oversized. It has a rating of 5.6 kW (7.5
horse power) but is seldom run over 1 kW. This affects the motor’s efficiency (discussed in
more detail below).
The compressor is single stage which can affect the energy consumption. The compres-
sion ratio is the ratio of the discharge pressure to the suction pressure. The compression
ratio affects how much energy is needed to compress a given mass of gas as can be seen in
each of the ideal thermodynamic models; a lower compression ratio requires less energy.
Multi stage compressors take advantage of this by bringing suction pressure gas up to an
intermediate pressure. The intermediate pressure gas is then compressed to the discharge
pressure, or a higher intermediate pressure depending on the number of stages. The total
energy required for all of the stages is less than if the gas had been compressed from a low
suction pressure to a high discharge pressure as it is in the case of a single stage compressor.
However, multiple stages add complexity and cost to a compressor. For the HSU fueling
station, SERC was trying to minimize both cost and complexity and chose to use a single
stage compressor, which likely affects the specific energy required for compression.
Another reason the measured power differs significantly from the calculated power is
because the PDC compressor is not well modeled by ideal gas thermodynamic models. This
is because of how the compressor works. In order for the oil to reach sufficient pressure to
compress the gas to the desired output pressure, an oil pressure relief valve is used. This
valve causes the motor to work to bring the oil pressure up to 6800 psig (or the setpoint
of the given compressor, ours is 6800 psig) for every stroke regardless of the discharge
pressure. If the oil pressure relief valve were somehow variable so that the oil pressure
were maintained only slightly above the storage pressure, the compressor would be more
efficient and perhaps better modeled by the thermodynamic models presented. As is, a
59
better approach to modeling the compressor would be to use a fluid dynamics model, such
as the Bernoulli equation, to find the work done on the oil rather than on the gas.
For a reversible, steady-state process, the specific work required to increase the pressure
of an incompressible fluid can be found using:
w = − ∫ P1
2 + g (Z1 − Z0) (16)
where w is the specific energy for the given process (kWh/kg)
υ is the specific volume (m3/kg)
P1 − P0 is the change in pressure for the process (Pa)
V1 is the velocity of the oil out of the pressure relief valve (m/s)
V0 is the velocity of the oil injected into the oil cavity (m/s)
Z1 is the elevation of oil exiting the pressure relief valve (m)
Z0 is the elevation of oil entering the oil cavity (m)
g is the acceleration due to gravity (9.8 m/s2)
To simplify the equation I will assume that the difference in velocity of the oil into and
out of the system is zero, the change in elevation is negligible, and that the specific volume
is constant. After performing the integration the equation simplifies to
w = −υ (P1 − P0) (17)
The negative sign indicates that work is done on the system rather than by the system.
The power can then be found by multiplying the specific work by the mass flow rate of the
oil in through the system. I performed some preliminary calculation using the density of the
oil (6.92 lbs/gallon or 0.830 kg/L) in our compressor and estimated oil flow rates. Obtaining
accurate measurements of the oil flow rate would require plumbing modifications for the
60
compressor and is beyond the scope of this analysis. I contacted PDC Inc. to inquire about
the oil flow rates and was informed that they are proprietary, but I was given a range that
they could fall into, which I used for this analysis. The oil flow rate range I used is 0.1-0.5
liters per minute. Because the flow rate is dependent on the oil relief valve setting, it is
impossible to know the flow rate without measuring it. I also wanted to take into account
the efficiency of the compressor motor because it is running at a low efficiency. After
looking up data that make up the performance curve for our motor, see Figure 25, I found
the motor was oversized by a factor of 5-9 for the application it is being used for, leading to
a low efficiency. I estimate that the efficiency varies from 55-65% for the range of power we
obeserve. I based this estimate range off the high and low power measurements observed
for the compressor. At the low range of compressor power, 650 W, the compressor motor
is running at approximately 11% of its rated full load of 5590 W (7.5 hp). At the high
range of measured compressor power, 1100 W, the motor is running at approximately 20%
of its rated full load. Table 3 below gives calculated power requirements for different oil
flow rates and different motor efficiencies. It also shows the calculated 2nd Law efficiencies
assuming that the nominal compressor power is 700 W, approximately the modal value of
compressor power. The 2nd Law efficiencies are a strong linear function of the flow rate,
see Figure 26, note that the two line have identical intercepts and the slopes are inversely
proportional to the motor efficiency. The 2nd Law efficiencies calculated with this method
are higher than those calculated using the ideal thermodynamic models, which indicates
that this might be a better method for evaluating our compressor’s 2nd Law efficiency. A
more thorough analysis should be done that carefully measures the flow rate and takes into
account other variables such as oil compressibility and friction losses.
61
0
20
40
60
80
100
Motor Efficiency
M ot
or E
ffi cie
nc y
Load (% of Rated)
Figure 25: Compressor motor efficiency curve based on the manufacturer’s data.
62
55% Motor Efficiency 65% Motor efficiency
y = 0.019997 + 197.6x R= 1
y = 0.02 + 167.2x R= 1
2 n d L
Oil Flow Rate (L/min)
Figure 26: 2nd Law efficiency as versus flow rate for compressor motor efficiencies of 55 and 65%
64
Time of Day Variations
My colleagues and I speculated that the time of day variations observed in Figure 12 are
caused by fluctuations in the incoming line voltages. To verify this I used a DENT Instru-
ments ELITEpro10 data logger to monitor each incoming line voltage. Each of the three
line voltages (three phase power is required by the compressor) were measured with respect
to ground. The voltages were measured every three seconds and averaged over ten minutes.
The ten minute averages were recorded by the data logger. The voltages were measured
upstream of the power transducer that measures the compressor power. The text data files
generated by the data logger did not need any further processing.
To analyze whether the compressor power was varying in a pattern similar to the line
voltage, I looked at data when the compressor was running and the line voltages were mea-
sured. In total 65 hours of compressor run time were included in the analysis. The hours
were non-consecutive, but represented four separate tank filling episodes that occurred be-
tween June 8 and June 18, 2009. To plot the compressor power data versus the line voltages
I created 10 minute averages of the compressor power for the same time frame the voltages
were monitored using a MATLAB R© script. I then plotted the compressor power versus the
line voltage for each of the three lines.
The line voltages do fluctuate during the day in a similar pattern to our compressor
power fluctuations as shown in Figure 27. All three incoming voltages fluctuate in a similar
manner, but one leg, line 1, varies more than the other two.
10http://www.dentinstruments.com/detailsElitePro.htm
65
Compressor power is well correlated to the each the line voltages as shown in Figures
28-30. Line 1 showed the least amount of scatter. Why the voltages vary diurnally, how the
three lines interact with each other and the compressor, and the mechanism by which this
might affect the compressor power are questions beyond the scope of this analysis.
Figure 27: Incoming compressor line voltages, with respect to ground, versus time of day. Notice the y-axis does not contain the origin.
66
650
700
750
800
850
900
950
1000
1050
Compressor Power
Po we
r ( W
1.654238.126m2 NA6.5594e+05Chisq NA0.75979R
Figure 28: Compressor power versus line 1 voltage. Notice the y-axis does not contain the origin.
67
650
700
750
800
850
900
950
1000
1050
Compressor Power
Po we
r ( W
Voltage (V)
68
650
700
750
800
850
900
950
1000
1050
Compressor Power
Po we
r ( W
Voltage (V)
CONCLUSION
In this thesis, I have shown that the diaphragm compressor at HSU’s hydrogen refueling
station is not well modeled by ideal gas thermodynamic models. The calculated power and
specific energy for the models I investigated (adiabatic, isothermal and polytropic) were
all lower than the measured values by a factor of 4-5 depending on the ideal model. The
adiabatic model had the highest 2nd Law efficiency of the models, but it was still low, as
were the efficiencies of the other models.
The compressor is not well modeled by the thermodynamic models because of its op-
erating principles. The compressor motor works to bring the compressor oil to a set 6800
psig for each piston stroke, regardless of the gas discharge pressure. Because of this the
power is relatively constant throughout the the range of discharge pressures. All of the ther-
modynamic models predict an increase in the power requirements as the discharge pressure
increases.
A better way to model the compressor would be to look at the oil side of the diaphragm
rather than the gas side. Using principles of steady-state fluid flow, I developed a prelimi-
nary mechanical model that can predict the power requirement and more closely matches
the power actually measured.
Our compressor power exhibits an interesting dependence on the time of day. This phe-
nomenon has been noted since start up and continues as of this writing. The incoming line
voltages to the compressor were monitored for 10 days to see if fluctuations in the voltage
could be causing the fluctuations in compressor power. There is a correlation between the
line voltages and compressor power consumption, suggesting that the motor performance
depends strongly on the line voltage.
69
70
Suggestions for further research would be to carefully measure the flow rate of the oil
through the compressor to obtain an improved prediction for the required power using fluid
dynamics. Also, a more thorough model of the oil side of the compressor would include
parameters such as the compressibility of the oil and the efficiency of the motor and the
injection pump. A model such as this would be very useful in determining the specific
energy requirements for a diaphragm compressor. Further study of the effects of voltage
on power consumption and methods to avoid them would also be interesting.
The results of this analysis suggest that more work needs to be done to characterize
hydrogen compressor efficiency for economic modeling. The HSU compressor is only one
data point for compressor efficiency, but the efficiency is based on empirical data and not
models. Our data suggest that some of the assumptions commonly used for economic mod-
eling may be incorrect. To validate or invalidate the assumptions behind economic models,
more empirical analyses should be performed on compressors, and other equipment in the
production chain. If most of the fueling stations currently in operation contributed opera-
tional data, the economic models currently in use could be made more robust, benefiting
those making decisions about a hydrogen based transportation system.
BIBLIOGRAPHY
American Honda Motor Co. 2009. Honda fcx clarity - hydrogen fule cell vehicle - official web site. URL http://automobiles.honda.com/fcx-clarity/.
Amos, W. A. 1998. Costs of storing and transporting hydrogen. Tech. Rep. 570-25106, National Renewable Energy Laboratory.
Cengel, Y. A. and M. A. Boles. 1998. Thermodynamics: An Engineering Approach. WCB/McGraw-Hill, San Francisco, 3rd ed.
Department of Energy. 2009. Hydrogen, fuel cells and infrastructure. URL http://www1.eere.energy.gov/hydrogenandfuelcells/delivery/current technology.html.
Energy Information Agency. 2009. Short-term energy outlook - real gasoline prices. URL http://www.eia.doe.gov/emeu/steo/pub/fsheets/real prices.html.
Ford Motor Company. 2009. Hydrogen fuel technology and research. URL http://www.ford.com/innovation/environmentally-friendly/hydrogen.
Fuel Cells 2000. 2009. World wide hydrogen fueling stations. URL http://fuelcells.org/info/chartsh2fuelingstations.pdf.
Fuel Cells for Power. 2009. Fuel cells for power - energy density. URL http://www.fuelcellsforpower.com/Energy-Density.html.
General Motor Company. 2009. The best emissions strategy is a zero-emissions strategy. URL
H2 mobility. 2009. Hydrogen vehicles worldwide - timeline. URL http://www.netinform.net/h2/H2Mobility/H2MobilityStart.aspx?CATID=1.
H2A Group. 2006. H2A Delivery Components Model Version 1.1: Users Guide.
Holland, G. B. and J. J. Provenzano. 2007. The Hydrogen Age: Empowering A Clean- Energy Future. Gibbs-Smith.
Linde Group. 2009. Linde unveils ground-breaking technology for fueling hydrogen- powered vehicles. URL http://www.lindeus.com/international/
web/lg/us/likelgus30.nsf/0/AB77A765F8A64F68C1257582005DC0F7.
Marks, L. 1996. Mark’s Standard Handbook for Mechanical Engineers. McGraw- Hill: New York, 10th ed.
71
72
Mercedes-Benz. 2009. F-cell vehicle. URL http://www.mbusa.com/mercedes//greenFuelCell/.
Perry, R. H., D. W. Green, and J. O. Maloney, eds. 1997. Perry’s Chemical Engineers’ Handbook. McGraw- Hill: New York, 7th ed.
Schatz Energy Research Center. 2009. Projects- real world appications. URL http://schatzlab.org/projects/real world/vehiclefactsheet.html.
Schwarzenegger, A. 2004. Executive order s-7-04.
Vatter, H. G. and J. F. Walker. 1996. History of the U.S. economy since World War II. M.E. Sharpe.
Weinert, J. X., L. Shaojun, J. M. Ogden, and M. Jianxin. 2007. Hydrogen refueling station costs in shanghai. International Journal of Hydrogen Energy, 32:4089–4100.
Yang, C. and J. Ogden. 2007. Determining the lowest-cost hydrogen delivery mode. Inter- national Journal of Hydrogen Energy, 32:268–286.
APPENDIX A: INVESTIGATION OF THE IDEAL GAS ASSUMPTION
For my analysis I assumed that hydrogen is an ideal gas. This assumption is good at low
pressures and high temperatures but becomes poor at higher pressures and low tempera-
tures. Here I investigate one scenario treating hydrogen as a real gas using the Z compress-
ibility factor to see what affect it has on the total specific energy and 2nd Law efficiency
predicted by thermodynamic models. For this analysis I used a real gas adiabatic model
shown in equation (18), (Marks, 1996) to calculate the power predicted for a real gas ther-
modynamic model. For the Z factors I used tabulated data published in Perry’s Handbook
(1997). There was only a limited number of calculated Z factors available ranging from
10 bar (145 psi) to 600 bar (8700 psi) for temperatures of 300 K and 400 K. I wanted Z
factors in more finely spaced pressure ranges so I assumed that they varied linearly and
used linear interpolation to find Z factors for the pressures and temperatures observed in
our compressor and not available in Perry’s Handbook. For simplification I assumed a con-
stant suction temperature and pressure of 300 K and 200 psig (15 bar) respectively, and a
constant discharge temperature of 330 K.
W = mkRT0
k − 1
where W is the power predicted (W)
k is the ratio of specific heats of the gas (1.41 for hydrogen)
R is the gas constant (4.1243 kJ/kgK for hydrogen)
T0 is the initial Temperature (K)
P0 is the initial pressure (Pa)
P1 is the final pressure (Pa)
Z0 is the initial compressibility factor
73
74
Z1 is the final compressibility factor
For this investigation I looked at the filling of tank A. I broke my data file into blocks
with pressure ranges of 20 bar (290 psi) increments, i.e., a 20-40 bar block, 40-60 bar
block up to 400-420 bar. I used bars for the pressure increments rather than psi because the
compressibility factors were given in bars, I then converted from bars to psi. For all the data
I assumed Z0 was constant, 1.0088, and within each 20 bar pressure range I assumed Z1
was constant. I then calculated the Z correction factor Z0+Z1
2Z0 . I multiplied the Z correction
factor by the ideal gas adiabatic power I previously calculated. I then used numerical
integration to find the total energy predicted by the real gas adiabatic model. For tank A,
I calculated a total energy requirement of 9.76 kWh or a specific energy of 1.58 kWh/kg.
The ideal adiabatic model predicted a total energy requirement of 8.92 kWh/kg or a specific
energy of 1.44 kWh/kg. The real gas model results differs from the ideal gas model by 0.84
kWh/kg or 9.4%.
Because the real gas model predicts higher energy consumption the 2nd Law efficiency
for the compressor would increase. In this example it would increase from 17.4% to 19.0%.
This does not affect my conclusion that the compressor is not well modeled with thermo-
dynamic gas models.
Test Started: 7/1/2008
Compressor
75
Features
All product specifications are subject to change without notice Last updated : 1-Mar-2007
Onboard Intel® ULV 600 MHz processor
Intel 82852GM Chipset
One DIMM socket supports up to 1 GB DDR 200/266 SDRAM
2-CH LVDS, DVI
Four COM, six USB 2.0, 16-bit GPIO
RoHS COMPLIANT 2002/95/EC
AIMB-251 Fanless Mini-ITX Motherboard Supports Dual Display for CRT, LVDS, DVI and TV-Out
Specifications
Processor System
CPU Onboard Intel ULV Celeron® 600 MHz Max. Speed 600 MHz Front Side Bus 400 MHz L2 Cache 512 KB Chipset Intel 82852GM + ICH4 BIOS Award™ 4 Mb FWH
Expansion Slot PCI 32-bit/33MHz, 1 slot Mini-PCI 32-bit/33MHz, 1 slot
Memory Technology DDR 200/266 SDRAM Max. Capacity 1GB Socket One 184-pin DIMM socket
Graphic
Controller Chipset Integrated VGA Controller VRAM Shared system memory up to 64 MB video memory LVDS Single channel 18-bit/ Dual-channel 36-bit LVDS
TV-Out Supports both S-video and composite video Chrontel CH7009A TV encoder supports both NTSC/PAL
DVI Chrontel CH7009A DVI transmitter up to 135M pixels/second Dual Display CRT + LVDS, or DVI/TV-out + LVDS or CRT + DVI
Ethernet Interface 10/100/1000Base-T Controller 1 LAN 1 Realtek™ RTL8110S Gigabit LAN (PCI) Connector 1 (RJ45)
EIDE Mode 2 x EIDE (Ultra DMA 100) Channel 2
Rear I/O
VGA 1 Ethernet 1 USB 2 (USB 2.0 ports) Audio Mic-In, Line-In, Line-Out (VIA VT1616 supports 5.1 CH AC97 Audio) Parallel 1 Serial 1 (RS-232/422/485), supply 5 V or 12 V via jumper PS/2 2 (keyboard and mouse)
Internal Connector
LVDS 1 TV-Out 1 DVI 1 USB 4 (USB 2.0 ports) Serial 3 (RS-232) IDE 2 (40/44 Pin) Compact Flash 1 IrDA 115k bps, IrDA 1.0 compliant FDD 1 DIO 16-bit General Purpose I/O for DI and DO
Watchdog Timer Output System reset Interval Programmable 1 ~ 255 sec
Power Requirement Typical Celeron 600 MHz, 1 GB DDR SDRAM
+5 V +3.3 V +12 V +5 VSB 1.58 A 4.66 A 0.05 A 0.4 A
Environment Operating
Temperature 0 ~ 60° C (32 ~ 140° F) Physical Characteristics Dimensions 170 mm x 170 mm (6.69" x 6.69")
NEW
76
77
Ordering Information Part Number Specification
AIMB-251F-00A1E Intel® ULV Celeron 600 MHz Mini ITX Motherboard with VGA, 2-CH LVDS, DVI, TV-out, 5.1 CH Audio, Gb LAN, CF, PCI, Mini PCI, 4 COM, 6 USB 2.0 & GPIO
Bracket View
FSB 400
2 IDE Ports (UltraDMA100)
Gb NIC Realtek
1 PCI Slot
Packing List Description Quantity AIMB-251 SBC x 1 IDE HDD cable (40 pin) x 1 IDE HDD cable (44 pin) x 1 FDD cable x 1 CPU Cooler x 1 I/O port bracket x 1 Startup Manual x 1 Driver CD x 1 Serial cable (RS-232) x 3
Accessory Part Number Description 1700003434 TV-Out cable 1700003433 USB cable 1700003435 DVI cable
AIMB-251F-00A1E
78
Manual Print History
The print history shown below lists the printing dates of all revisions and addenda created for this manual. The revision level letter increases alphabetically as the manual undergoes subsequent updates. Addenda, which are released between revisions, contain important change information that the user should incorporate immediately into the manual. Addenda are numbered sequentially. When a new revision is created, all addenda associated with the previous revision of the manual are incorporated into the new revision of the manual. Each new revision includes a revised copy of this print history page.
Revision A (Document Number 141-0999) ..........................................................................September 1999 Revision B (Document Number 141-1199)............................................................................November 1999 Revision C (Document Number 141-102002) ...........................................................................October 2002 Revision D (Document Number 141-082005) .............................................................................August 2005 Revision E (Document Number 141-062008).................................................................................. June 2008 Revision F (Document Number 141-092008)........................................................................September 2008 Revision F (Document Number 141-022009).......................................................................... February 2009
Visit www.teledyne-hi.com for WEEE disposal guidance.
Hastings Instruments reserves the right to change or modify the design of its equipment without any obligation to provide notification of change or intent to change.
The instruments described in this manual are available with multiple pin-outs.
Ensure that all electrical connections are correct. CAUTION:
The instruments described in this manual are designed for Class 2 installations
in accordance with IPC standards
CAUTION:
CAUTION:
The instruments described in this manual are designed for INDOOR use only.
80
Table of Contents
4. MAINTENANCE.............................................................................................................................................................. 15 4.1. AUTHORIZED MAINTENANCE..................................................................................................................................... 15 4.2. TROUBLESHOOTING ................................................................................................................................................... 15 4.3. ADJUSTMENTS ........................................................................................................................................................... 16 4.4. END CAP REMOVAL:.................................................................................................................................................. 17 4.5. PRINTED CIRCUIT BOARD REPLACEMENT.................................................................................................................. 17 4.6. SENSOR REPLACEMENT: ............................................................................................................................................ 17 4.7. ORIFICE CHANGES: .................................................................................................................................................... 17 4.8. REPLACEMENT PARTS................................................................................................................................................ 18
5. WARRANTY .................................................................................................................................................................... 20 5.1. WARRANTY REPAIR POLICY ...................................................................................................................................... 20 5.2. NON-WARRANTY REPAIR POLICY ............................................................................................................................. 20
81
141-022209 - 201/203 Series Page 4 of 20
1. General Information The Hastings HFM-201/HFC-203 series Mass flow meter (HFM-201) and controller (HFC-203) are designed to accurately measure and control mass flow over the range of 30 slm to 500 slm, without corrections or compensations for gas pressure and temperature with an accuracy of better than ±1% from the mean (±2% FS for 500 slm). Hastings mass flow instruments do not require any periodic maintenance under normal operating conditions with clean gases. No damage will occur from the use of moderate overpressures (~500 psi/3.45MPa) or overflows. Instruments are normally calibrated with the appropriate standard calibration gas (nitrogen) then a correction factor is used to adjust the output for the intended gas. Special calibrations for other gases, such as oxygen, helium and argon, are available upon special order.
1.1. Features
• LINEAR BY DESIGN. The HFM-201/HFC-203 series is inherently linear (no linearization circuitry is employed). Should recalibration in the field be desired (a calibration standard is required), the customer needs to simply set the zero and span points. There will be no appreciable linearity change of the instrument when the flowing gas is changed.
• MODULAR SENSOR. The HFM-201/HFC-203 series incorporates a removable/replaceable sensor module. Field repairs to units can be achieved with a minimum of production line downtime.
• METER SETTLING TIME. Changes in flow rate for the HFM-201 are detected in less than 2 seconds when using the speed up circuitry.
• LOW TEMPERATURE DRIFT. The temperature coefficient of span for the HFM-201/HFC-203 series is typically less than 0.05% of full scale/°C from 15-45°C. The temperature coefficient of zero is typically less than 0.1 % of reading/°C from 0-50°C.
• FIELD RANGEABLE. The HFM-201/HFC-203 series is available in ranges from 30 slm to 500 slm. For HFC-203 controller’s, an orifice change is required as well. Calibration is required after all changes.
• CURRENT LOOP. The 4-20 mA option gives the user the advantages of a current loop output to minimize environmental noise pickup.
82
1.2. Specifications
Repeatability ..............................................................................................................<±0.1% of F.S.
Operating temperature ........................................................0-50°C in non-condensing environment
Temperature coefficient (span) ..................................................±0.1 ppm/°C (±0.05%/0C typical)
Zero drift .............................................................................................................................±0.1%FS
Flow ranges .............................................................................. 30, 50, 100, 300, 500, 600 slm (N2)
Output ............................................................................................................................... 0-5 VDC
Power requirements ........................................................................................ ±(15) VDC @ 50 mA
Wetted materials ...................................................................... 304 & 316 stainless steel, nickel 200,
.......................................................................................................................... Viton, Au13Ni braze
Attitude sensitivity of zero .................................................. < ±6.5% F.S. for 90° without re-zeroing
..............................................................................................................{N2 at 19.7 psia (135 KPa)}
Controller weight ......................................................................................................5.6 lb (2.54 kg)
83
1.3. Optional 4-20 mA Current Output
An option to the standard 0-5 VDC output is the 4-20 mA current output that is proportional to flow. The 4 - 20 mA signal is produced from the 0 - 5 VDC output of the flow meter. The current loop output is useful for remote applications where pickup noise could substantially affect the stability of the voltage output.
The current loop signal replaces the voltage output on pin 6 of the “D” connector. The current loop may be returned to either the power supply common or the -15 VDC connection on the power supply. If the current loop is returned to the power supply ground, the load must be between 0 and 600 ohm. If it is returned to the -15VDC, the load must be between 600 and 1200 ohm. Failure to meet these conditions will cause failure of the loop transmitter.
The 4-20 mA I/O option can accept a current input. The 0-5 VDC command signal on pin 14 can be replaced by a 4-20mA command signal. The loop presents an impedance of 75 ohms and is returned to the power supply through the valve common.
1.4. Other Accessories
1.4.1. Totalizer (TR-1J) The Hastings Flow Totalizer integrates the 0-5 VDC signal generated by the flow meter to give a total flow reading. Count rates from 0 to 999 counts per minute are selectable by internal setting.
1.4.2. Hastings Model 40/200/400 Power Supply Hastings power supplies are available in either two or four channel versions. They convert 115 or 230VAC to the ±15 VDC required to operate the flow meter. Interface terminals for the ±15 VDC input and the 0-5 VDC linear output signal are located on the rear of the panel. Also, a cable can be supplied with the power supply that provides the +15 VDC on pin 11 of a 15-pin “D” connector and the 0 - 5VDC output measurement on pin 6. Pins 5, 7 and 12 are common, and pin 7 is chassis ground. Throughout this manual, when reference is made to a power supply, it is assumed the customer is using a Hastings Model 200/400/40 supply. Hastings power supplies do not meet CE standards at this time.
84
PHONE (614) 889-6152 TECH. ASSISTANCE (614) 876-8308
FAX # (614) 876-85386625 McVey Blvd. Columbus, Ohio 43235 Div. Morlan & Associates, Inc.
94
• Accurate regardless of variations in voltage,current, power factor, or load.
• Available with 1, 2, 2 1/2, or 3 element configurations. Provides bi-directional operation.
• Accuracy maintained over wide temperature range, calibration traceable to NIST.
• Equipment monitoring for process control.
• Integration into energy management systems, or a variety of sub-metering applications.
• Measurement using direct-connection, current and/or potential transformers.
PRECISION AC WATT TRANSDUCER
To calculate full scale Watts when using potential and/or current transformers: a = initial transducer calibration (from table above) b = current transformer ratio (e.g. 100:5, or 20) c = potential transformer ratio (e.g. 600:120, or 5) F.S. WATTS = a x b x c NOTE: UL recognized current tra

Documents

EFFICIENCY AND PERFORMANCE MEASUREMENTS OF A PDC …