You are on page 1of 13

This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy


i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

System simulation models for on-board hydrogen


storage systems
Sudarshan Kumar a,*, Mandhapati Raju a,b, V. Senthil Kumar c
a

Chemical Sciences and Materials Systems Lab, General Motors Global R&D, Warren, MI 48090, USA
Optimal CAE Inc., Plymouth, MI 48170, USA
c
India Science Laboratory, General Motors Global R&D, Creator Building, International Technology Park, Bangalore 560066, India
b

article info

abstract

Article history:

System simulation models for automotive on-board hydrogen storage systems provide

Received 22 November 2010

a measure of the ability of an engineered system and storage media to meet system

Received in revised form

performance targets. Thoughtful engineering design for a particular storage media can

8 April 2011

help the system achieve desired performance goals. This paper presents system simulation

Accepted 21 April 2011

models for two different advanced hydrogen storage technologies e a cryo-adsorption

Available online 12 June 2011

system and a metal hydride system. AX-21 superactivated carbon and sodium alanate are
employed as representative storage media for the cryo-adsorbent system and the metal

Keywords:

hydride system respectively. Lumped parameter models incorporating guidance from

Hydrogen storage

detailed transport models are employed in building the system simulation models.

Cryo-adsorption

Simulation results to test the storage systems ability to meet fuel cell demand for

Sodium alanate

different drive cycles and varying operating conditions are presented. Systems are engi-

System simulation models

neered to provide the ability to refuel a vehicle in a short time guided by DOE targets.
Gravimetric and volumetric hydrogen densities are computed for the engineered systems
and compared to the DOE system goals.
Copyright 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.

1.

Introduction

System level models for two hydrogen storage systems were


developed as part of a DOE funded project for evaluating the
performance of cutting-edge hydrogen storage technologies.
This paper presents the system level performance of the two
systems e metal hydride and cryo-adsorbent systems. The
system design and dynamic performance of the two systems
is presented along with a brief literature review for each of
these systems. The performance of the two systems is
compared with respect to the DOE targets. The drive cycle
simulations are tested on an integrated vehicle level model
framework. This vehicle level model framework [1] consists of

three primary modules e a vehicle level module, a fuel cell


module and a storage system module. Different storage
system models can be used in this integrated framework for
evaluation on a consistent basis. The fuel cell model used in
this framework is adapted from the fuel cell modeling work
of Pukrushpan et al [2]. and the vehicle level model is an
Excel based model integrated into the vehicle level module
of the framework.
In this paper we report on the system level models and
system simulations using this integrated framework for two
separate systems e a cryo-adsorbent system using the activated carbon AX-21 and a metal hydride system using sodium
alanate. Both systems are designed to carry w 5 kg of usable

* Corresponding author. Tel.: 1 586 986 1614; fax: 1 586 986 1910.
E-mail address: sudarshan.kumar@gm.com (S. Kumar).
0360-3199/$ e see front matter Copyright 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2011.04.182

Author's personal copy


2863

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

hydrogen. System level dynamics during drive cycle simulations are presented for the two systems and the ability of the
storage system to deliver the required amount of hydrogen
demanded by the fuel cell during drive cycle simulations is
examined under different conditions. Finally, performance of
these two systems in meeting the DOE gravimetric and volumetric energy density targets is evaluated.

mo

Anode

H2

Coolant
Radiator

Fuel Cell
Air

mf , T
Qh

mo : Net H2 output to fuel-cell

Cryo bed

2.
AX-21 based cryo-adsorbent storage
system
On-board storage of hydrogen by adsorption at low temperatures and moderately high pressures (77K, up to 60 bar) is
considered viable and competitive with other storage technologies including liquid hydrogen, compressed gas, and
metallic or complex hydrides [3]. At these conditions,
superactivated carbons like AX-21 offer good gravimetric
capacity and fast and reversible kinetics. For example, AX-21
has a reversible hydrogen storage capacity of about 5.8 wt%
at 77 K and 35 bar [4]. AX-21 as an adsorbent material has
been studied extensively and has been considered to assess
the tank performance in previous studies [5e7]. An allied
technology is the cryo-compressed storage, wherein
hydrogen is stored inside a pressure vessel at 250e360 atm
pressures and 50e300 K temperatures, without any
adsorbent material [8,9].
Consider a fuel-tank with an initial operating condition of
35 bar and 80 K, and a fuel cell operating at 3 bar. The four
processes occurring in a cryo-adsorber fuel-tank are refueling,
discharge, dormancy, and venting. These fuel-tank processes
occur over different time scales: refueling over a few minutes,
discharge over a few hours, dormancy over a few days, and
venting over a few weeks. In our previous studies [10,11] it was
shown that refueling, the fastest process is quasi-static i.e.
local equilibrium conditions prevail. Hence, the slower
processes are also quasi-static. When the molecular processes
are fast, slow processes are expected to have negligible
internal gradients and are generally amenable for a lumped
parameter analysis. Hence, a quasi-static lumped parameter
model for the cryo-adsorber fuel-tank was developed in [10].
That model is used in the current work to study the drive
cycle discharge simulations for a cryo-adsorption hydrogen
storage tank. However, during a drive cycle discharge, the
hydrogen demand fluctuates rapidly. Therefore, in this
work, the quasi-static approximation is relaxed and an
adsorption kinetic model is developed and employed.
During discharge hydrogen is desorbed from the adsorbent
bed. Since desorption is an endothermic process, we need to
add heat during discharge to avoid very low tank temperatures
and maintain fast desorption kinetics. Heat can be added into
the tank by heating a part of the recirculating gas, as shown in
Fig. 1. Since the gas is in intimate contact with the bed, this
mode of heating is expected to be efficient. An alternate way
of adding the heat is through the use of a jacketed or
embedded electrical heater. Although such external heating
will be an electrical load penalty on the fuel cell, it might be
beneficial in terms of gravimetric/volumetric capacities of
the system, since hot gas recirculation loop, along with the
recirculation pump etc., are expected to be bulkier and

mo ,T

H2

Cathode

m f ,Tf
Fig. 1 e Schematic of a cryo-adsorber bed with hot gas
recirculation.

heavier than this alternative option. At the level of lumped


parameter description, hot gas recirculation and electrical
heater are mathematically equivalent. Hence, the lumped
parameter adsorption kinetic model developed in the current
work can be used to include hot gas circulation or external
heating during discharge for on-board implementation.

2.1.

Adsorption system model development

This model uses the mass balance, energy balance and


adsorption kinetics to develop the time evolution of pressure,
temperature and adsorbate concentration. The hydrogen
content in the bed at any time is the sum of the gaseous and
adsorbed hydrogen i.e. mH2 t ms qt Vb 3t rg T; P:

2.1.1.

Transient mass balance

The rate of change of hydrogen content of the tank balances


the net flow into the tank. Hence the transient mass balance
for hydrogen is given by
dmH2
_f m
_
m
dt
i:e: ms

(1)

drg
dq
_ f  m:
_
m
Vb 3t
dt
dt

(2)

The time derivative of density is expressed in terms of the


temperature and pressure time derivatives as drg =dt
rg aPg dT=dt rg kTg dP=dt. Using this result the transient mass
balance simplifies to
A11

dT
dP
dq
A12 A13
B1 ;
dt
dt
dt

where A11 Vb 3t rg aPg ,


_ f  m.
_
B1 m

2.1.2.

(3)
A12 Vb 3t rg kTg ,

A13 ms ,

and

Transient energy balance

The thermal masses associated with the fuel-tank are the gas
phase, adsorbed phase, adsorbent, pressure vessel including
the bed restrainers and other bed internals, insulation layer,
outer shell and ambient, as shown in Fig. 2. The insulation
layer isolates the inner thermal masses (gas, adsorbed
phase, adsorbent, and pressure vessel) from the outer ones
(shell and ambient). The transient energy balance for the

Author's personal copy


2864

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

carbon at near liquid nitrogen temperature [14] and also


note that the adsorbate diffusivities could vary with
temperature as in [15].
A Simulink model was developed to compute the transient
temperature, pressure and weight fraction by solving the
mass balance, energy balance and the adsorption kinetics. The
property correlations used in this model are functions of
temperature and pressure, described in [16]. The bulk density
and the skeletal density of AX-21 are taken as 0.27 g/cc
and 2.2 g/cc respectively [17]. The heat of adsorption is taken
as 6.0 kJ/mol [17]. The vessel material is assumed to
be aluminum; properties of aluminum at cryogenic
temperatures are taken from Marquardt and Radebaugh [18].

Flow in
Outer shell
Header

Pressure
vessel

Adsorbent
bed
Insulation
Collector

Ambient

2.2.

Flow out

Fig. 2 e Sectional view of a cryo-adsorber bed.

inner thermal masses (system), with the assumption of


a constant average heat of adsorption is
 dH
dHw
dHs 
dq
dP
g
ms
ms q Vb 3t rg
ms DHa  Vb 3t
dt
dt
dt
dt
dt
Q_ h Q_ l :

mw

(4)

The time derivative of gas enthalpy is written in terms of


temperature and pressure time derivatives as dHg =dt
CPg dT=dt vg 1  aPg TdP=dt. Similarly, for the solid phases
(pressure vessel and adsorbent), neglecting the thermal
expansion of the material, gives dHw =dt CPw dT=dt vw dP=dt
and dHs =dt CPs dT=dt vs dP=dt. Using these equations and
rearranging the transient energy balance simplifies to
A21

dT
dP
dq
A22 A23
B2 ;
dt
dt
dt

(5)

where
A21 mw Cpw ms Cps ms q Vb 3t rg CPg ,
A22 mw vw ms vs ms qvg Vb 3t 1  aPg T  Vb 3t
and A23 ms DHa ,and B2 Q_ h Q_ l .
The assumed heat leak has the form Q_ l TN  T=Reff . A
typical value of Reff 74:0K=W is used so that the heat leak
into the tank during typical dormancy conditions is about 3
Watts, as in [10].

2.1.3.

Adsorption kinetics

In our earlier study [10], the quasi-static kinetics i.e.


qtzq Tt; Pt was used. A constant discharge is typically
quasi-static [10]. Hence, actual desorption kinetics is not very
important for constant discharge case. However, in a real
drive cycle hydrogen demand variations occur at time scales
of a second or less and quasi-static conditions may not
prevail. Therefore, in this paper we use the Linear Driving
Force (LDF) model with Glueckaufs approximation [12,13]:
dq 15Da 
2 q  q
R
dt

(6)

Sircar and Hufton [13] show that the LDF model can be
used to capture adsorption transients. We have used a
representative value of Da =R2 z1:52  102 s1 for activated

Drive cycle simulations

In general, there is no unique relationship between the


amount of hydrogen discharged and heat input because it is
possible to discharge a particular amount of hydrogen with or
without heating the bed. However, in order to remove most of
the hydrogen within the storage system with the final pressure above the fuel cell pressure, it is necessary to heat up the
adsorbent material. It is possible to assume a constant heat
input or time varying heat input proportional to the hydrogen
demand depending on the scenario being studied. In this
paper, we assume a constant heat input and performed
simulations for both the FTP75 and US06 drive cycles [19]. The
drive cycle simulations are performed for a compact vehicle in
the vehicle level framework developed by the HSECoE team.
Fig. 3 shows the vehicle speed and fuel consumption rate
(g/s) for the two drive cycles.
The FTP75 cycle is a mild & short duration cycle consuming
only 159.36 g of H2 in 1874 s (or 31.25 min), with an average
hydrogen demand of 0.085 g/s. Assuming that a significant
fraction of the discharged hydrogen is desorbed from the
adsorbed phase, the added heat must supply the heat of
adsorption of the desorbed hydrogen. For AX-21 with an
average heat of adsorption of 3.0  106 J/kg, and
0.085  103 kg/s average discharge rate, the necessary heating rate is 0.204 kW assuming 80% of the discharged hydrogen
is desorbed. The US06 cycle is a shorter but more aggressive
cycle than FTP75. It consumes 155.15 g of H2 in 601 s with an
average demand of 0.258 g/s. For AX-21, with an average heat
of adsorption of 3.0  106 J/kg and 0.258  103 kg/s average
discharge rate, the heating rate is calculated to be 0.62 kW,
again assuming that 80% of the discharged hydrogen is desorbed hydrogen.
Fig. 4 shows drive cycle simulation for a single cycle of
FTP75. Fig. 4(a) shows the net fuel consumption rate in g/s.
The oscillations in the fuel cell demand causes oscillations
in the bed pressure. Comparing the gaseous and adsorbed
phase loads for the single drive cycle (Fig. 4(d)), it is seen
that the gas phase responds to the demand fluctuations and
the adsorbed phase responds to the steady demand. As
noted earlier, the heating rate mainly needs to target the
heat of desorption. Hence, the heating rate need not be
altered in shorter time scales to meet the fluctuating
demand. It needs to be changed only if the average demand
changes over longer periods of time, as long as there is
sufficient hold-up in the gas phase. Model formulation and

Author's personal copy


i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

2865

Fig. 3 e Speed and fuel cell consumption for FTP75 and US06 drive cycles.

results apply equally to both the recirculation gas heating and


electrical heating of the bed.
To study the discharge of a tank with about 5 kg useable H2,
the cycle is repeated continuously. The drive cycle simulations are presented for both FTP75 and US06 drive cycles. The
hydrogen demand, temperature, pressure and tank load
evolutions for such a sequence of FTP75 and US06 cycles are
shown in Figs. 5 and 6 respectively.

2.3.

Hot gas recirculation versus electrical heating

Hot gas recirculation takes advantage of the intimate contact


between gas and solid in a porous bed. Since porous beds tend
to have low thermal conductivity, such an intimate gas-solid
contact is an efficient way of heating up the bed. However, gas
recirculation requires additional elements including piping,
insulation, a blower or compressor on the recirculation loop,
along with a heat exchanger, and valves. These components
add to the capital cost and lower the gravimetric capacity at
system level. In addition, minimizing the heat leak into the
tank through the recirculation loop, while recirculation is not
on, could be an engineering challenge for cryogenic systems.
Electrical heating, on the other hand, requires few additional
components. Hence, it may be possible to achieve better
gravimetric capacity, easier control, and probably lower
capital cost. However, there is an electrical penalty on the fuel
cell system which could be at least partially obviated by
thermal integration of the cold hydrogen from the storage unit

with the fuel cell cooling system or by energy recovery


through expanding the high pressure gas from the cryoadsorber to fuel cell feed pressure. Heat leaks into the tank, or
leaks from the piping should be significantly lower in this
design.

2.4.

System weight and volume

A viable on-board hydrogen storage system must have high


gravimetric and volumetric storage densities. A heavy storage
system results in the so-called mass compounding effect as
heavier supporting components are needed to fit the system
in the vehicle. In addition, a bulkier system results in lower
passenger or trunk space. The DOE has specified system
gravimetric and volume density targets for hydrogen storage
systems - for 2010 these targets are 0.045 kg H2/kg of system
mass and 0.028 kg/L of system volume. In the following, we
calculate approximate system gravimetric and volumetric
densities for a base case design. The cryo-adsorbent system
considered is a relatively low-pressure system and the storage
vessel can be made of a hydrogen compatible aluminum alloy.
Because of low temperatures, the storage vessel will need to
be insulated with multi-layer vacuum insulation enclosed in
an outer vessel. We consider a system that can deliver 5 kg of
usable hydrogen, with the empty conditions specified to be
135 K and 3 bar Table 1 gives information on weights and
volume of various parts and components of the system and
shows that the gravimetric density is 3.3 wt% and the

Author's personal copy


2866

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

1.1

36

0.9
35.5

0.8

Pressure (bar)

Net H2 to fuel Cell (g/s)

0.7
0.6
0.5
0.4
0.3

35

34.5

0.2
0.1
0

0.2

0.4

34

0.6

0.1

0.2

84

0.5

2.4

0.6

3.4

83

82

81

80

0.4

adsorbed H2
gaseous H2

gaseous H2 (kg)

Temperature (K)

0.3

time (hr)

time (hr)

0.1

0.2

0.3

0.4

0.5

0.6

2.35

2.3

3.35

0.1

0.2

0.3

0.4

0.5

adsorbed H2 (kg)

3.3
0.6

time (hr)

time (hr)

Fig. 4 e Variation of (a) Net hydrogen demand from the fuel cell (b) pressure (c) temperature and (d) hydrogen content in the
bed for a single FTP75 drive cycle.

volumetric density is 13.1 kg/L. We have used this base case as


an illustration. When we consider different designs, we need
to revise the different masses (adsorbent, inner and outer
vessel masses, etc.). Note that a lumped parameter cryoadsorber model distinguishes different tank designs just
through the masses involved.

3.

Sodium alanate based storage system

The absorption and desorption of Ti-doped sodium alanate


[20] can be described as a two-step reaction [21] given below
NaAlH4 41=3Na3 AlH6 2=3Al H2

(7)

1=3Na3 AlH6 4NaH 1=3Al 1=2H2

(8)

First stage is the decomposition of NaAlH4 (sodium


aluminum tetrahydride, or the tet phase) and the second stage
is the decomposition of Na3AlH6 (sodium aluminum hexahydride, or the hex phase). The theoretical capacity of sodium
alanate is 5.6 wt% but its practical storage capacity is much
smaller than this. Luo and Gross [22] report that the maximum
hydrogen weight percent in their sample is 3.9%. The present
paper incorporates the kinetics presented by Luo and Gross [22].
The system level implementation of sodium alanate based
hydrogen storage system is different from that of high pressure metal hydride storage systems. The primary reason is

that the heat of absorption/desorption for sodium alanate is


much higher compared to high pressure metal hydrides like
Ti1.1CrMn. The performance of the high pressure metal
hydride system has been demonstrated [23] using a system
level model on a Matlab/Simulink platform. The advantage
of the high pressure metal hydride systems lies in their
operation near the fuel cell stack temperature. Hence the
heating of the bed can be achieved by using the same
radiator fluid used for cooling the fuel cell. However for
sodium alanate system, high temperatures are required for
decomposition. Temperatures around 180e200  C [24] are
required to decompose the hex phase to meet a practical
drive cycle. The bed is heated to this high temperature by
passing a portion of the hydrogen to the combustor to heat
up the heat exchanger fluid, which in turn heats up the bed.
A buffer tank is also needed to supply H2 during periods
when the bed is not able to supply sufficient H2 to the fuel
cell. We have assumed a buffer tank capacity of 100 g. This
number was arrived at by considering the hydrogen needed
for vehicle operation under conditions of cold start and low
tank pressure. Gas phase hydrogen is needed to warm up
the hydrogen storage system, and to supply H2 to the fuel
cell until the storage system is warm enough to desorb
hydrogen from the sodium alanate in the tank. Earlier
efforts for system level modeling for sodium alanate
considered only the low temperature decomposition [25] of
the tetrahydride phase. However this limits the storage
capacity of the system to a maximum theoretical capacity of

Author's personal copy


i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

3.7% by weight. Recent modeling efforts [24,26,27] of alanate


storage bed include a catalytic burner and incorporate both
the tet and the hex phase decomposition. Dedrick et al. [26]
considered a shell and tube heat exchanger with the alanate
in the tubes and the cooling fluid in the shell while Raju and
Kumar [24] considered a shell and tube heat exchanger with
the alanate in the shell and the cooling fluid in the tubes.
Results for drive cycle simulations for a Chevrolet Equinox
vehicle were presented. This present work is an extension of
the same with a full scale system of w5 kg usable H2 using
dual bed strategy. The drive cycle simulations are run within
the framework of the vehicle level model developed by the
DOE Hydrogen Storage Engineering Center of Excellence
(HSECoE) team. In addition, the system level targets are
evaluated for this system.

3.1.

Description of the storage system

Fig. 7 shows a schematic flow sheet of alanate storage system


in a fuel cell vehicle. The storage system consists of two beds,
each of approximately 5 kg usable hydrogen. Fig. 8 shows the
cross-sectional view of the storage bed, which consists of
alanate in the shell and coolant through the tubes. The
tubes are interconnected by fins to provide efficient heat
transfer.
The details of the system level modeling for sodium alanate storage system including bed design, bed properties and
alanate properties are presented in [24]. Both refueling and
drive cycle simulations have been studied. Here only a brief

2867

synopsis of the system level modeling strategy is provided.


The emphasis in this paper is to evaluate the performance
of the system dynamics of a dual bed system that provides
w5 kg of total usable hydrogen. The dual bed system is of
interest because a single bed system can be quite large and
difficult to accommodate in a vehicle. In addition, the
control system necessary for a dual bed system is more
complex than that for a single bed. The control system used
in the simulations includes a recharging of the buffer tank
to 150 bar when the bed is hot and is able to deliver
hydrogen at rates higher than those demanded by the fuel cell.
During refueling, both hydrogen and coolant are supplied
at the refueling station. The coolant is passed through the
tubes to provide for efficient cooling during refueling. Since
the kinetics of absorption is slow for sodium alanate system, it
is not possible to achieve the refueling of the bed in the DOE
target refueling time of 4.2 min. Instead, a refueling time of
10.5 min (based on 40% refueling rate of the target value) is
chosen. A two-dimensional model is developed in COMSOL to
simulate the refueling of the bed [24]. In addition, overall heat
transfer coefficients are extracted from the two-dimensional
COMSOL model, which can be incorporated into the lumped
parameter model for desorption. During refueling, it is
ensured [24] that the local temperature within the bed does
not shoot above 500 K to avoid sintering of the bed due to
melting of alanate.
The drive cycle simulation is performed using a lumped
parameter model in Matlab/Simulink. All the components
including the storage beds, buffer tank, catalytic combustor,

Fig. 5 e Variation of (a) Net hydrogen demand from the fuel cell (b) pressure (c) temperature and (d) hydrogen content in the
bed for FTP75 drive cycle.

Author's personal copy


2868

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

Fig. 6 e Variation of (a) Net hydrogen demand from the fuel cell (b) pressure (c) temperature and (d) hydrogen content in the
bed for US06 drive cycle.

and oil loop are included in the model. Various component


level equations are described in detail in [24] and the initial
state of the bed is taken as the state at the end of refueling.
During driving, the bed is heated by passing hot fluid
through the tubes. The fluid is in turn heated by a catalytic
combustor. A small amount of hydrogen is burnt in

Table 1 e System weight and volume for the cryoadsorbent system.


System Temp & Pressure
Final pressure
Adsorbent volume (L)
Total usable H2
Adsorbent mass (kg)
Total inner volume (L)
Cylindrical part L (cm)
2 Hemispheres D (cm)

77 K, 35 bar
3 bar
250
5 kg
67.5
275
59.5
59.5

INNER VESSEL & OUTER VESSEL Material


Inner vessel mass (kg)
Outer vessel mass (kg)
Insulation mass (kg) e MLVSI (1 thick)
BOP components (kg)
Total mass (kg)
Outer volume(L)
Gravimetric capacity (kg/kg)
Volumetric density (kg/L)

Aluminum 6061
46.1
11.1
12
15
151.7
380.3
0.0330
0.0131

a catalytic combustor. Minimum amount of oil that needs to


be carried by the vehicle is calculated based on the total
volume occupied by the cooling fluid flow path in the bed.
To account for the volume of any connecting tubes outside
the bed, we include an additional 50% oil volume. Based on
this volume estimation, the vehicle needs to carry 13 kg of
heating oil, which is part of the storage system. During
driving, the oil is pumped through the storage bed tubes at
a flow rate of 2 LPM per tube. The oil passes through
a catalytic burner, where it is heated. The oil temperature is
set to a maximum of 450 K during the tet phase decomposition
and 470 K during the hex phase decomposition. A 12 kW
catalytic burner is provided for heating the oil. The efficiency
of the burner is assumed to be 90%. A buffer tank carrying
100 g of hydrogen at 150 bar and 300 K is provided.
A flow control strategy for the dual bed storage system is
adopted to control the flow of hydrogen between different
components. Hydrogen is supplied to the fuel cell and burner
either by the storage bed or by the buffer tank depending on
the state of these storage components. The strategy employed
in the current storage system is shown in Fig. 9. Initially the
first storage bed is given control to supply the hydrogen to
the fuel cell and burner. It will supply hydrogen as long as
the bed pressure does not fall below 1.1 times the fuel cell
cut-off pressure. When the bed pressure reaches below this
limit, the control is transferred to the second bed to supply
the hydrogen from its gas phase. Note that the heating fluid
at this time is flowing through the first bed. The heating

Author's personal copy


i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

2869

Buffer Volume
Anode

H2
Cathode

H2

Coolant
Radiator

Fuel Cell
Air

Catalytic heater
Alanate bed

Heating fluid
Oil tank

Fig. 7 e Schematic of sodium alanate based dual bed hydrogen storage system.

fluid is routed to the second bed only when the first bed is
almost empty. When the control is transferred to the second
bed, the second bed tries to supply the hydrogen from its
gas phase as it cannot supply the absorbed hydrogen since
there is no heating. If the gas phase hydrogen in the second
bed is unable to supply the fuel cell demand, then the
control is transferred to the buffer tank to supply
the hydrogen demand. Meanwhile pressure builds within
the first bed due to hydrogen desorption reactions. When
the bed pressure exceeds twice the cut-off pressure, the
control is again transferred back to the first bed. When
the first bed is almost empty, the heating fluid is rerouted to
the second bed and the control is transferred to the second
bed to supply the hydrogen. At this stage, whenever the
pressure in the second bed falls below 1.1 times the fuel cell
cut-off pressure, the control is transferred to the buffer
storage tank to supply hydrogen. Because of heating and
hydrogen desorption, pressure in the second bed starts
increasing and as soon as the bed pressure exceeds twice

Fig. 8 e Cross-section of the alanate storage bed.

the fuel cell cut-off pressure, control is transferred back to


the second bed. This strategy is chosen to ensure continuous
supply of hydrogen to the fuel cell and at the same time
extract most of the absorbed hydrogen from the storage bed.

3.2.

Drive cycle simulations

The vehicle level model has different drive cycle options to


evaluate the performance of the storage system. FTP75 and
US06 drive cycles are chosen to evaluate the dynamic
performance of the storage system during real driving conditions. The cycles are periodically repeated to run a full tank to
empty tank simulation. Simulations start with a nearly full
tank based on a refueling time of 10.5 min and bed temperature set at 390 K. The tank is considered empty and the
simulation stops when the pressure in each of the beds and
the buffer falls below the fuel cell cut-off pressure.
Fig. 10 shows the system level dynamic performance during
FTP75 drive cycle simulation. Fig. 10(a) shows the variation of
the bed pressures and the buffer pressure. In the beginning,
hydrogen is extracted from the first bed. Heating fluid is
supplied to the bed. While the bed is getting heated up (see
Fig. 10(b)), the gas phase hydrogen in the first bed supplies
the hydrogen to the fuel cell. This leads to a drop in the
pressure of the bed. As the bed gets heated up, the rate of
desorption (tetrahydride phase) increases and hence the
pressure in the bed starts rebuilding after a short while. The
fluctuations in the bed pressure are due to the fluctuations in
the fuel cell demand. Once the tetrahydride phase is almost
converted to the hexahydride phase, the decomposition of
hexahydride phase begins. The kink in the red line of
Fig. 10(c) at 5 h is due to this transition. The temperature
of the heating oil is increased to 470 K during this transition.
This causes a rise in bed temperature (Fig. 10(b)) at 5 h. The
bed pressure drops to the equilibrium pressure (Fig. 10(a)) of
the hexahydride phase at the current bed temperature. Even
at this high temperature the rate of decomposition is small.
Eventually when the hexahydride phase decomposition is
almost complete, the pressure in the first bed drops to the

Author's personal copy


2870

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

if bed 1 is not empty


heating fluid flows through bed 1
if Pbed 1 > 1.1Pcut _ off
First bed supplies the H 2 to the fuel cell and burner
else
Control is transferred to eith er bed 2 or buffer
In the meanwhile, pressure is building up in bed 1 due to heating
if Pbed 2 > 1.1 Pcut _ off
Second bed supplies the H 2 from its free volume to the fuel cell and burner
No heating is supplied to this bed
else
Buffer supplies the H 2 to the fuel cell and burner
end
Once the first bed pressure rea ches 2Pcut _ off , the control is transferred back to first bed
elseif bed 2 is not empty
heating fluid flows through bed 1
if Pbed 2 > 1.1 Pcut _ off
Second bed supplies the H 2 to the fuel cell and burner
else
Control is transferred to buffer
In the meanwhile, pressure is building up in bed 2 due to heating
end
Once the second bed pressure reache s 2Pcut _ off , the control is transferred back to first bed
elseif bed 1 and bed 2 are empty
Buffer supplies the H 2 to the fuel cell and burner till buffer is emptied
end

Fig. 9 e Control system for dual bed system.

Fig. 10 e System performance for FTP75 drive cycle.

fuel cell cut-off pressure. Once the bed pressure drops, then
the second bed is called in. The heating oil is rerouted to the
second bed. Hence the temperature of the second bed starts
rising. Similar dynamic behavior is observed for the second
bed as that of the first bed. Eventually when the hexahydride
phase decomposition is near completion in the second bed,
buffer supplies the hydrogen to the fuel cell. The total driving
time for the FTP75 is approximately 16 h.
Fig. 11 shows the system level dynamic performance
during US06 drive cycle simulation. US06 is an aggressive
drive cycle compared to FTP75. Fig. 11(a) shows the variation
of the bed pressures and the buffer pressure. The
fluctuations in the bed pressure for US06 drive cycle are
larger compared to those for the FTP75 cycle. At first,
hydrogen is extracted from the first bed. Heating fluid at
450 K is supplied to the first bed but the bed does not heat
up to that temperature. This is due to the cooling produced
by excess hydrogen demand which prevents the bed from
heating quickly. Once the tetrahydride phase is almost
converted to the hexahydride phase, the decomposition of
hexahydride phase begins. Since the rate of hexahydride
phase decomposition is low, the bed cannot supply the
hydrogen demand. Consequently, the control is switched
from the first bed to the second bed when the pressure in
the first bed falls below 1.1 times the fuel cell cut-off
pressure. Note that the heating oil is still being supplied to
the first bed and the second bed is not being heated up. The
gas phase hydrogen in the second bed now supplies the
hydrogen to the fuel cell. This results in a drop in second

Author's personal copy


2871

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

H2. The vessel is made of an inner liner and an outer layer of


carbon composite. The thickness of the materials are chosen
to withstand pressures of 150 bar and temperatures of 180  C.
The cooling tubes and the fins are made of aluminum. The
gravimetric capacity for this system is roughly 0.012 kg H2/kg.
It is quite far below the DOE 2010 target of 0.045 kg H2/kg.
There has to be a significant improvement in the hydrogen
absorption capacity of metal hydrides in order to meet the
DOE target. The volumetric capacity for this system is
0.0148 kg H2/L, which is also below the DOE 2010 target value
of 0.028 kg H2/L.

4.

Relative merits of the two storage systems

The two storage systems operate at entirely different operating conditions. Each system has its relative merits and
demerits. Overall the performance of the cryo-adsorbent
system is much better in terms of gravimetric capacity as
compared to the metal hydride system. The volumetric
capacities for the two systems are nearly identical. In addition, there are some important distinguishing features of the
two systems that should be noted.

4.1.

Fig. 11 e System performance for US06 drive cycle.

bed pressure as well as some desorption in the second bed.


Correspondingly, the temperature and weight fraction of
absorbed hydrogen drop slightly. Meanwhile, the first bed
gets heated up and the pressure in the bed builds up. If the
pressure in the first bed exceeds twice the fuel cell cut-off
pressure, control is shifted to the first bed. This shifting of
control back and forth continues till the first bed becomes
almost empty. This results in fluctuations in the first and
second bed pressures during the range of 2e3.5 h of driving
time. Once the first bed is almost empty, the second bed is
called in. The heating oil is rerouted to the second bed.
Hence the temperature of the second bed starts rising.
Eventually when the hexahydride phase decomposition
starts, the second bed takes the help of buffer to supply the
hydrogen demand as explained by the control strategy. The
excess hydrogen demand eventually leads to an empty
buffer even before the second bed is completely empty. The
second bed can no longer supply the hydrogen demand even
though there is absorbed hydrogen still present in the
hexahydride phase. Hence for aggressive drive cycles, it is
difficult to extract all the absorbed hydrogen.

3.3.

System weight and volume

Table 2 below shows the preliminary estimation of gravimetric


and volumetric density of the current sodium alanate storage
system. Based on the FTP75 drive cycle simulation, it is
assumed that each bed will deliver roughly 2.75 kg of usable
H2. Two such beds are used to deliver a total of 5.5 kg usable

Cold start capability

Cryo-adsorbent system can handle cold start at very low


temperatures. However, in the case of metal hydride systems
like sodium alanate, cold start is a challenge. If the car has

Table 2 e System weight and volume for the sodium


alanate system.
Bed specifications
Number of beds
deliverable hydrogen
Length (alanate packing)
Actual length of the bed
Diameter of the bed (inner)
Diameter of the bed (outer)
Shell material
No of cooling tubes
Diameter of cooling tubes (inner)
Weight of alanate
weight of shell include liner
weight of tubes and fins
accessories (manifolds, end
plates etc)
pump/HEX/burner
pump/HEX/burner volume
BOP mass
Oil mass
Buffer
Buffer volume
Total weight of the bed
Total volume of the beds
Total system volume
Total system mass (tubes,
plates, shell/insulation, alanate)
Gravimetric density
Volumetric density

units

Value

mm
kg
kg
kg
kg

2
5.5
1000
1292.0
416.0
436.9
Composite
carbon
24.0
20.0
200.00
44.00
137.00
33.70

kg
liters
kg
kg
kg
liters
kg
liters
liters
kg

8.00
8.00
16.85
13.00
5.05
11.30
381.00
351.21
370.51
457.60

kg/kg
kg/liter

0.012
0.0148

kg
mm
mm
mm
mm

Author's personal copy


2872

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

been parked for a long time during peak winter days, the bed
will cool down. During start up, the bed may not be able to
supply the hydrogen and the buffer tank will need to supply
the required hydrogen demand. Size of the buffer tank will
decide whether the system will be able to handle cold start
conditions. Fig. 12 shows cold start simulation results for the
FTP75 drive cycle. The bed and the heating fluid is assumed
to be at ambient temperature assumed to be 20  C. Since
the bed is cold, there is no desorption in the beginning. Gas
phase hydrogen in the first bed supplies the fuel cell
demand. Hence the pressure in the first bed falls rapidly as
shown in Fig. 12(a). In the meantime, the bed is heated. The
bed takes a long time (Fig. 12 (b)) to heat up for two reasons the low initial bed temperature and the heat needed for the
endothermic desorption reaction. As shown in Fig. 12(a), the
bed pressure almost falls to cut-off pressure before the bed
pressure starts increasing. If the bed pressure falls below
cut-off, then the gas phase hydrogen in the second bed will
supply the fuel cell demand before hydrogen can be
desorbed from the first bed to supply the hydrogen demand.
As the temperature increases, the bed pressure slowly rises.
Once the bed temperature reaches 450 K, the system will
perform normally as shown in Fig. 12.

4.2.

Hydrogen overhead

In the case of cryo-adsorbent system, hot gas recirculation can


be used to heat up the bed during the discharge cycle. A small
heat exchanger to warm up the cold hydrogen using heat

exchange with the ambient would be sufficient. However, as


the storage system warms up, venting may be necessary
resulting in a loss of hydrogen to the atmosphere. The amount
of hydrogen vented depends on the total heat leak into the
system. In the case of sodium alanate storage system,
hydrogen has to be burnt for heating up the bed during driving
for supplying the heat of desorption and in transient heating
of the bed from the initial temperature of the bed to the
desorption temperature. The amount of hydrogen burnt can
be substantial because of the significant enthalpy (w40 kJ/mol
H2) of hydrogen desorption and the need to keep the system at
a high temperature (140  C) for speeding up the kinetics to
supply the hydrogen demand.

5.

Summary

Lumped parameter system simulation models are developed


for the cryo-adsorption and metal hydride hydrogen storage
systems. For the cryo-adsorbent system, the model solves the
mass, energy balances and adsorption kinetics to compute
temperature, pressure and adsorbate concentration. The
adsorption kinetics included is the linear driving force model
with Glueckaufs approximation. Simulations for the FTP75
and US06 drive cycle demand are performed and the
temperature, pressure, adsorbate concentration, adsorbed
and gaseous hydrogen content in the tank are presented.
Simulation results show that the gas phase responds to the
demand fluctuations and the adsorbed phase responds to the
average demand. Hence, the heating rate need not be altered
in shorter time scales to meet the fluctuating demand. In
a cryogenic adsorption storage unit, an electrical heater could
be more optimal (in the sense of heat leak, gravimetric and
volumetric capacities and cost) than a hot gas recirculation
system, since the heating rate needs to change on longer time
scales than the fluctuating demand.
For the metal hydride based system, a dual bed storage
system is considered to supply w 5 kg of usable hydrogen. The
system performance of the dual bed storage system is shown
for the FTP75 and US06 drive cycle demands. It is shown that
the usable hydrogen for a given system depends on the drive
cycle, with aggressive cycles like US06 resulting in lower
usable hydrogen. The gravimetric and volumetric capacities of
the two storage systems are evaluated and the relative merits
and demerits of the two systems are presented.

Acknowledgments

Fig. 12 e Cold start simulation for FTP75 drive cycle.

This work was performed under DOE contract DE-FC3609GO19003 as GMs contribution to the DOE Hydrogen Storage
Engineering Center of Excellence (HSECoE). The authors would
like to acknowledge the support of Ned Stetson, Monterey
Gardiner and Jesse Adams of DOE and Don Anton of SRNL. The
authors would like to thank Lincoln Composites for supplying
data on shell design and thickness for the given operating
conditions of the storage systems. The authors also
acknowledge Mei Cai and Scott Jorgensen of General Motors

Author's personal copy


i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 8 6 2 e2 8 7 3

for their valuable suggestions and HSECoE team members for


contributing to the development of vehicle level model.

Nomenclature

T; P
L; R
ms ; Vb
mH2
mw ; vw
_ f;m
_o
m

Temperature and Pressure K, bar


Length and radius of the adsorbent bed, m
Mass and volume of the adsorbent bed kg, m3
Mass of hydrogen in the bed, kg
Mass and specific volume of outer shell kg, m3/kg
Mass flow rate of H2 in the feed and outlet
streams, kg/s
Porosity of the bed, m3/m3
3t
rb
bed densities kg/m3
rg ; vg
Gas density and specific volume, kg/m3, m3/kg
Gas viscosity, Pa s
mg
aPg ; kTg Isobaric thermal expansion coefficient and
isothermal compressibility, 1/K, 1/bar
Hg ; Hq ; Hs ; Hw Specific enthalpy of gas, adsorbate, adsorbent
and outer shell, J/kg
CPg ; CPs ; Cpw Specific heat capacity of gas, adsorbent and outer
shell, J/kg/K
Q_ h ; Q_ l
Heat flux supplied, and heat flux leak into the
system, W
Excess adsorbate concentration and its equilibrium
q; q
value, kg, H2/kg adsorbent
Heat of adsorption, J/kg H2 adsorbed
DHa
Effective diffusivity of the adsorbate in the adsorbent
Da
particle, m2/s

references

[1] Pasini JM, van Hassel BA, Mosher DA, and Veenstra MJ.
System modeling methodology and analyses for materialsbased hydrogen storage. Int J Hydrogen Energy, in press.
[2] Pukrushpan J, Peng H, Stefanopoulou A. Control-oriented
modeling and analysis for automotive fuel cell systems.
J Dynamic Systems Measurement Control 2004;126:14.
[3] Zhou L. Progress and problems in hydrogen storage methods.
Renewable Sustainable Energy Reviews 2005;9:395e408.
[4] Benard P, Chahine R. Determination of the adsorption
isotherms of hydrogen on activated carbons above the
critical temperature of the adsorbate over wide temperature
and pressure ranges. Langmuir 2001;17:1950e5.
[5] Richard MA, Benard P, Chahine R. Gas adsorption process in
activated carbon over a wide temperature range above the
critical point. Part 1: modified DubinineAstakhov model.
Adsorption 2009;15:43e51.
[6] Richard MA, Benard P, Chahine R. Gas adsorption process in
activated carbon over a wide temperature range above the
critical point. Part 2: conservation of mass and energy.
Adsorption 2009;15:53e63.

2873

[7] Ahluwalia RK, Peng JK. Automotive hydrogen storage system


using cryo-adsorption on activated carbon. Int J Hydrogen
Energy 2009;34:5476e87.
[8] Ahluwalia RK, Peng JK. Dynamics of cryogenic hydrogen
storage in insulated pressure vessels for automotive
applications. Int J Hydrogen Energy 2008;33:4622e33.
[9] Ahluwalia RK, Hua TQ, Peng JK, Lasher S, McKenney K,
Sinha J, et al. Technical assessment of cryo-compressed
hydrogen storage tank systems for automotive applications.
Int J Hydrogen Energy 2010;35:4171e89.
[10] Senthil Kumar V, Raghunathan K, Kumar Sudarshan. A
lumped-parameter model for a cryo-adsorber hydrogen
storage system. Int J Hyd Energy 2009;34:5466e75.
[11] Senthil Kumar V, Kumar Sudarshan. Generalized model
development for cryo-adsorber and 1-D results for the
isobaric refueling period. Int J Hyd Energy 2010;35:3598e609.
[12] Ruthven DM. Principles of adsorption and adsorption
processes. NY: John Wiley & Sons; 1984.
[13] Sircar S, Hufton JR. Why does the linear driving force model
for adsorption kinetics work? Adsorption 2000;6:137e47.
[14] Changpeng Li, GM R&D, Internal communication.
[15] Saha D, Wei Z, Deng S. Equilibrium, kinetics and enthalpy of
hydrogen adsorption in MOF-177. Int J Hydrogen Energy 2008;
33:7479e88.
[16] Senthil Kumar V, A generalized cryo-adsorber model and 2-D
refueling results, GM R&D Internal report, submitted for
publication to Int J Hydrogen Energy.
[17] Richard MA, Cossement D, Chandonia PA, Chahine R, Mori D,
Hirose K. Preliminary evaluation of the performance of an
adsorption-based hydrogen storage system. AIChE J 2009;
55(11):2985e96.
[18] Marquardt E, Le J, Radebaugh R. Cryogenic material
properties database. Cryocoolers 11. US: Springer; 2002.
p. 681e687.
[19] DieselNet website: http://www.dieselnet.com/standards/
cycles/.
[20] Bogdanovic B, Brand R, Marjanovic A, Schwickardi M, Tolle J.
Metal-doped sodium aluminum hydrides as potential new
hydrogen storage materials. J Alloys Compd 2000;302:36e58.
[21] Sandrock G, Gross K, Thomas G. Effect of Ti-catalyst content
on the reversible hydrogen storage properties of the sodium
alanates. J Alloys Compounds 2002;339:229e308.
[22] Luo W, Gross KJ. A kinetics model of hydrogen absorption
and desorption in Ti-doped NaAlH4. J Alloys Compounds
2004;385:224e31.
[23] Raju M, Ortmann JP, Kumar S. System simulation model for
high-pressure metal hydride hydrogen storage systems. Int J
Hydrogen Energy 2010;35:8742e54.
[24] Raju M, Kumar S. System simulation modeling and heat
transfer in sodium alanate based hydrogen storage systems.
Int J Hydrogen Energy 2011;36(2):1578e91.
[25] Ahluwalia RK. Sodium alanate hydrogen storage system for
automotive fuel cells. Int J Hydrogen Energy 2007;32:
1251e61.
[26] Dedrick DE, Kanouff MP, Larson RS, Johnson TA, Jorgensen
SW. Heat and mass transport in metal hydride based
hydrogen storage systems. Proceedings of HT 2009, ASME
summer heat transfer conference, July 19-23, San
Francisco, CA.
[27] Raju M, Kumar S, Optimization of heat exchanger designs in
metal hydride based hydrogen storage systems, Int J
Hydrogen Energy 2012;37:2767e78

You might also like