Version of Record: https://www.sciencedirect.

com/science/article/pii/S0360319918316008
Manuscript_c9e55217083588c86bb2ec10a111b406

Design of a large-scale metal hydride based hydrogen storage reactor:

simulation and heat transfer optimization
Mahvash Afzal, Pratibha Sharma*

Department of Energy Science and Engineering, Indian Institute of Technology Bombay,

Mumbai, India

Abstract

An optimized design for a 210 kg alloy, Ti-Mn alloy based hydrogen storage system for
stationary application is presented. A majority of the studies on metal hydride hydrogen systems
reported in literature are based on system scale less than 10 kg, leaving questions on the design
and performance of large-scale systems unanswered. On the basis of sensitivity to various design
and operating parameters such as thermal conductivity, porosity, heat transfer coefficient etc., a
comprehensive design methodology is suggested. Following a series of performance analyses, a
multi-tubular shell and tube type storage system is selected for the present application which
completes the absorption process in 900 s and the desorption process in 2000 s at a system
gravimetric capacity of 0.7 % which is a vast improvement over similar studies. The study also
indicates that after fifty percent reaction completion, heat transfer ceases to be the major
controlling factor in the reaction. This could help prevent over-designing systems on the basis of
heat transfer, and ensure optimum system weight.

Contents

1. Introduction ............................................................................................................................. 4
2. Design Objective ..................................................................................................................... 5
3. Mathematical modeling ........................................................................................................... 6
4. Results and discussions ......................................................................................................... 11
4.1 Effect of improvement in thermal conductivity ............................................................. 14
4.2 Addition of water jacket ................................................................................................. 17
4.3 Addition of cooling tubes ............................................................................................... 19
4.4 Multi-tube metal hydride reactor.................................................................................... 21

© 2018 published by Elsevier. This manuscript is made available under the Elsevier user license
https://www.elsevier.com/open-access/userlicense/1.0/
 4.5 System gravimetric capacities ........................................................................................ 29
4.6 Desorption behavior of the multi-tubular configuration (14 tubes) ............................... 30
5. Conclusions ........................................................................................................................... 31
References ..................................................................................................................................... 31

*Corresponding author email: [email protected]

Nomenclature

c Fraction of hydrogen absorbed

M Rate of absorption/desorption, kg/m3s
P Pressure, bar
T Temperature, K
u Velocity, m/s
R Universal gas constant, J/mol/K
ΔS Entropy of reaction, J/mol/K
ΔH Enthalpy of reaction, J/mol
ε Porosity
h Heat transfer coefficient, W/m2K
Cp Specific heat capacity, J/kg/K
k Thermal conductivity, W/m/K
Q Heat source/sink, W/m3
Subscripts
eq equilibrium
f final
e effective
m metal
g gas
s source/sink
a absorption
d desorption
Abbreviations
MH Metal hydride
ENG Expanded natural graphite
HTF Heat transfer fluid
PCM Phase changing materials
HPMH High pressure metal hydride
 1. Introduction

Metal hydrides are a favorable hydrogen storage alternative offering safe and efficient storage.
Hydrogen reacting with a metal or metal alloy, forms a metal hydride, and thus gets stored in
solid state. Of the various metal hydrides, AB2 type Ti-based metal hydrides have a promising
potential because of their comparatively high hydrogen storage capacity, fuel cell friendly
operating temperatures and pressures, and good reversibility[1]. However, the storage of
hydrogen in metal hydrides is primarily limited by heat transfer associated with the reaction.
While the absorption reaction is accompanied by release of heat which needs to be removed from
the system; for desorption to occur heat needs to be supplied to the system. This reaction heat has
a great bearing upon the system performance as it controls the MH bed temperature which in turn
controls the equilibrium pressure. The rate of absorption/desorption of hydrogen is a function of
the equilibrium pressure. If the equilibrium pressure is higher than the supply pressure of
hydrogen the absorption reaction will cease. Similarly if the equilibrium pressure is lower than
the discharge pressure of hydrogen desorption reaction will be stalled. Therefore, heat transfer
to/from the MH bed is an important factor for ensuring efficient hydrogen storage.

Multiple studies have addressed the issue of heat transfer in metal hydride based hydrogen
storage. Studies have suggested various designs for hydrogen storage systems to facilitate
effective heat transfer. These designs are as diverse as the heat transfer enhancements themselves
and include addition of fins, cooling tubes, cooling coils etc. Attempts have been made to
improve thermal conductivity of the MH bed by addition of foam and ENG (expanded natural
graphite) [2–10], improve heat transfer area by addition of fins and cooling tubes[11–19]. More
recently, combinations of these techniques have been employed, yielding favorable results.
Garrison et al studied the effect of a concentric cooling tubes equipped with a) transverse fins b)
longitudinal fins on a sodium alanate system[20]. The results pointed to a configuration
involving multiple cooling tubes at small distances as the most viable. Ferekh et al compared the
performance of metal fins versus metal foam and found that metal foam based design was
superior to the fin based design[10]. Helical coil based heat exchangers have also been shown to
improve system performances by reducing charging/discharging times significantly[21–24]. A
comparative study involving finned cooling tube reactor, helical coil reactor and shell-tube type
reactor was performed by Raju et al. Working with sodium alanate, for a filling time of 10.5
minutes they found that the helical coiled heat exchanger gave better volumetric and gravimetric
densities over the other designs[25]. Mazzucco et al studied three different heat exchanger
configurations: a) metal hydrides with embedded cooling tubes b) metal hydrides tubes
surrounded by heat transfer fluid in annular space and c) metal hydride tubes in a heat transfer
fluid shell [26]. Their results indicated that the last configuration was of most significant interest
for storage applications.

Clearly, there is no dearth of studies on design of metal hydride hydrogen storage systems but
the studies on large scale systems are far less than those on small-scale systems [27]. Further
investigation of large-scale practical systems is important keeping in view the issues that are
unique to these systems such as creation of hot spots (pockets of extremely high temperature
where heat transfer enhancements may not be able to reach), parasitic system weight, volumetric
efficiency etc. More importantly, the design methodology for large-scale systems is also unclear.
While every heat transfer enhancement results in improved system performance, it also leads to
reduced gravimetric and volumetric capacities. Therefore, the question of which heat transfer
enhancements should be preferred is very significant. In this study an attempt is made to present
a unified design policy which can serve as a road-map for the design of such systems. Using a
100 kWh Ti-Mn based (Hydralloy C5) hydrogen storage system as an example, the most
important parameters that affect the system performance are studied and a procedure outlined to
serve as a tool for the design of future systems. The gravimetric capacities of the final designs
are compared to ensure that there is a healthy trade-off between system performance and weight.

2. Design Objective

The objective of this exercise is to design an optimized Ti based, 100kWh metal hydride
hydrogen storage system for stationary applications and to outline a design methodology for the
same. To begin with, the volume required to store the metal alloy powder was evaluated on the
basis of the mass of alloy required to store the hydrogen and the density of the alloy (5500
kg/m3)[1] [28] . Assuming a porosity of 0.5, the volume was determined to be 0.09168 m3, which
includes a twenty per cent expansion volume, to allow for the expansion of the alloy. A
cylindrical design was chosen to ensure sufficient heat transfer area, ease in handling and
portability. To keep the design within a conformable limit for convenient storage and
transportation, the initial value of height of the reactor was limited to 1 m and the corresponding
radius was evaluated. The closest standard size pipe dimension was then taken and the
corresponding height was calculated. The thickness of the pipe was found to be suitable for the
pressure range under consideration on the basis of the ASME code for pressure vessels. In this
manner, the radius and height of the cylindrical reactor were found to be 0.1524 m and 1.047 m
respectively.

Table 1: Design constraints and variables

Variable Constraint

Reactor Volume H2 required to produce X kW power >> Alloy M required to

store H2 >> System volume required to store MH
Reactor diameter, D Standard pipe dimensions available

Reactor height, L Conformability in terms of storage space availability etc.

Reactor thickness, t Allowable stress for the tank

3. Mathematical modeling

The reactor of the said dimensions was first simulated under free convection conditions and its
performance was analyzed. The system was modeled as a 2-d axisymmetric domain. To model
the MH hydrogen storage system, mass, momentum and energy conservation equations are used.

The following assumptions have been used for modeling the system:

1) The solid and gas are at the same temperature (local thermal equilibrium).
2) Gas phase: hydrogen behaves as an ideal gas.
3) Solid phase: MH is isotropic and has uniform porosity.
4) The pressure inside the tank is uniform throughout.

The simulation domain consists of the metal hydride bed and the expansion volume. The reactor
is exposed to air at a temperature Tf. The alloy considered for the model is AB2 type, Ti-Mn
based Hydralloy C5, chosen for its good kinetics and reversibility, and fuel cell friendly
operating temperature and pressure. The model was developed in COMSOL 5.2. The details of
governing equations and boundary conditions are as follows:

Continuity:

The continuity equation for the expansion volume part which is the gas filled void space, and the
metal hydride bed is represented by equation (4). Here ρg is the density of hydrogen gas and ρm is
the density of the metal hydride. The density of the metal hydride depends upon M which is the
rate of absorption/desorption of hydrogen per unit volume. Note that the M will be negative in
equation (5) and positive in equation (6) for absorption and vice versa for desorption.

Expansion Volume: + ∇. . =0 (4)

MH bed: + ∇. . =− (5)

1− = (6)

Conservation of momentum:

The conservation of momentum is described by the Navier-Stokes equation for the expansion
volume, and the Brinkman equation for the metal hydride bed which reduces to Darcy law if
viscous effects and gas velocity are low. Here µ g, ε, and κ are the viscosity of the gas, the
porosity and the permeability of the bed respectively; F is the body force. The variable u
represents the velocity of hydrogen gas within the pores and the expansion volume domain.

Expansion Volume: + .∇ = −∇ + ∇. ∇ + ∇ − ∇. + (7)

! ! "# %
MH bed: + .∇ = −∇ + ∇. ∇ + ∇ − ∇. − + & + (8)
$

Conservation of energy:

Assuming thermal equilibrium between gas and solid, T represents temperature within the
system. The variables Cpg, kg and km are specific heat of the gas, thermal conductivity of the gas
and thermal conductivity of the porous bed. The effective thermal conductivity and the effective
thermal mass of the porous bed and the gas is represented by ke and (ρCp)e.
Expansion Volume: '( + . ∇) = ∇. * ∇) (9)

MH bed: '( + + '( ,. ∇) = ∇. *+ ∇) + -. (10)

where, '( + = '( + 1 − / '(/ (11a)

*+ = * + 1 − */ (11b)

The reaction kinetics that determine the amount of hydrogen absorbed/desorbed are expressed as:
456 (
Absorption: = '0 exp ln .. − / (12)
7 ∗ ;<=6

45@ (4;<=@
Desorption: = '> exp ?7 ∗
A? ;<=@
A / − + (13)

;<= F G
where, = DE − + 0.51 ∗ IJK L ∗ M − 0.5 + 0.15 (14)
;B<C 7 7∗

The equilibrium pressures are predicted with the help of van’t Hoff equation as stated in equation
(14). Please note that ‘c’ is the fraction of hydrogen absorbed/desorbed.

Initial and boundary conditions

The evolution of temperature, hydrogen absorption/desorption in the metal bed is calculated with
respect to certain initial and boundary conditions. Presently, the conditions for absorption only
will be discussed. For absorption the initial temperature of the tank is considered to be 298 K and
the pressure is 25 bar which is well above the equilibrium pressure of the alloy under
consideration. For the tank center line, the axisymmetric boundary condition is set. Along the
tank wall, a heat flux condition is set with constant heat transfer coefficient. A no-slip boundary
condition is set along the tank walls except for the inlet through which hydrogen is allowed to
enter. Continuity of temperature, density, velocity and other properties is established at the
boundaries between the MH bed and the expansion volume. The boundary and initial conditions
may be summarized as follows:

( ∗7Q
T(r,z,0)= 298 K; p(r,z,0)= 25 bar; u(r,z,0)= 0; ρm(r,z,0)= 5500 kg/m3 ; N, P, 0 =

(
At the center line, ,R = 0; T
=0; =0
U R
As can be seen from Fig.1, boundaries 1 and 3 are assumed to be thermally insulated, while 2 is
subjected to a heat flux through the cooling/heating fluid. Boundary 4 is exposed to the inlet
pressure.

At 1 and 3: −* V
= 0; =0

At 2: −* = W ) − )X
V

At 4: K. ∇) = 0; = YVZ+ /\] Z+

Model Validation

The mathematical model used is well established in literature [3,13,29–31], However, it has also
been validated in the present study as shown in Fig. 2. The results are compared to Laurencelle
and Goyette [3]. The grid independence and time independence are shown in Fig. 3 and 4
respectively.

4 1

Expansion volume

Metal Hydride bed

z

3
r

Figure 1: Boundary definition of the model MH tank

 1.4 Present model Experiment (Laurencelle, 2007)

1.2

0.8
H/M wt%

0.6

0.4

0.2

0
0 5 10 15 20 25 30
Time (mins)

Figure 2: Comparison of H/M weight per cent from the present model and experimental data

finer c(1) extra fine c(1) c (1)-1s c (1)-5s c (1)-10s

extremely fine c(1)
0.8 0.8
0.7 0.7
Reaction fraction 'c'

0.6 0.6
Reaction fraction 'c'

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1
0 800 1600 2400 3200 0 800 1600 2400 3200
Time (s) Time (s)

Figure 3: Grid independence- fraction Figure 4: Time step independence- fraction

of hydrogen absorbed at different grid of hydrogen absorbed at different time
 4. Results and discussions

The baseline mathematical model discussed in section 3.2 is based upon the above governing
equations and developed in COMSOL 5.2. The basic cylindrical reactor of radius 15.24 cm and
height 1.256 m was simulated operating at a pressure of 25 bar and a temperature of 298 K under
free convection, exposed to air at 298 K. Ordinarily, in the design of any heat exchanger, if the
heat load is known, starting with an initial value of heat transfer coefficient, the area of the
reactor can be found and further calculations can be made. But that approach cannot be taken
here because the surface area requirement in that case is impossibly large. Therefore, the reverse
procedure was followed: the performance of the system was studied under free convection
conditions, the key parameters influencing system behavior were identified and the design was
further improved. A parametric study was thus undertaken and the sensitivity of the system to
different design and operating parameters was noted.

Figure 5 shows the response of the system to change in different properties and operating
parameters. It is clear that thermal conductivity and pressure have the most significant impact on
the system performance. As the thermal conductivity is increased from 1 to 3 W/m/K, the system
performance improves by over 10%. However, when the thermal conductivity is increased
beyond 10 W/m/K, the system performance improves at a much slower rate. Similar trend can be
observed for supply pressure also. Increasing the supply pressure by 5 bar improves the system
performance by around 10%. The system performance is also sensitive to the temperature of the
heat transfer fluid, however, it doesn’t seem to have the same impact as pressure or thermal
conductivity. Variation of heat transfer coefficient from a value of 4.023 W/m2/K (natural
convection in air) to 500 W/m2/K, leads to less than 20% improvement in system performance.
Interestingly, the system shows a higher sensitivity to heat transfer coefficient at lower values as
can be noted in the transition from 4.023 to 50 W/m2/K W/m2/K. While the system performance
is highly sensitive to porosity, any change in porosity from a base value of 0.5 adversely affected
the system performance. Therefore, from the above analysis, it was realized that in order to
improve system performance, thermal conductivity and pressure were the two most important
parameters.
 0.8
0.6
0.5
0.4
0.2

500
300
150
100
50
4.023

278
Parameters

eps
ε
282
286 hcoeff (W/m^2/K)
290
294 Text(K)
298
Ps(bar)
50 k (W/m/K)
45
40
35
30
25

15
12
10
6
3
1
-50 -40 -30 -20 -10 0 10 20 30 40 50
Percentage variation in hydrogen absorption

Figure 5: Percentage variation in hydrogen absorption with change in thermal conductivity,

pressure, HTF temperature, heat transfer coefficient, porosity

A similar analysis for the system was also carried out for desorption. The reference pressure and
HTF temperature were 1 bar and 323 K respectively. The other parameters were varied in the
same manner as in case of absorption. Figure 6 shows the sensitivity of hydrogen desorption to
various parameters. Like absorption, the dependence on pressure and thermal conductivity seems
to be overshadowing the effects of other parameters. Other than porosity, the increase in which
appears to impact hydrogen desorption positively unlike hydrogen absorption, the other
parameters such as heat transfer coefficient and HTF temperature do not have a marked influence
on hydrogen desorption.

0.8
0.6
0.5
0.4
0.2

500
300
150
100
50
4.023

333 eps
ε
Parameters

328
323 hcoeff (W/m^2/K)
318
Text(K)
313
308
Pext (bar)
3
k (W/m/K)
2
1
0.5
0.1

15
12
10
6
3
1
-30 -20 -10 0 10 20 30 40 50 60

Percentage variation in hydrogen desorption

Figure 6: Percentage variation in hydrogen desorption with change in thermal conductivity,

pressure, HTF temperature, heat transfer coefficient, porosity
It is also important to point out that system scale can have a significant bearing on the results.
For instance, in case of smaller systems where diameter is less than 8 mm, a heat transfer
material such as aluminum foam may not be required[3], whereas in case of larger systems as
pointed out in the analysis, the same does not hold true. Similarly, heat transfer coefficient and
temperature of cooling fluid may have a greater significance in case of smaller diameter systems
where heat transfer from the core of the metal hydride bed is transferred to the wall without
much resistance[32][33].

4.1 Effect of improvement in thermal conductivity

In order to improve the thermal conductivity of the metal hydride bed, metal foam, ENG
compacts, or a network of fins can be used. While ENG compacts have exhibited better
performance than foam in terms of thermal conductivity however, the deterioration of ENG
compacts could be an issue [6]. Between fins and foam, it was observed by Ferekh et al [10], that
the improvement in heat transfer using metal foam is better than with fins. Moreover, foam is
more uniformly distributed within the alloy bed as compared to fins, therefore, can contribute to
a more uniform heat transfer. Therefore, in the present case aluminum foam will be used for
improvement of thermal conductivity. From the sensitivity analysis it was observed that any
improvement in thermal conductivity beyond 10 W/m/K was not very effective in improving
system performance. Therefore, aluminum foam of porosity 91% which leads to an effective
thermal conductivity of 11.45 W/m/K was used. The system with enhanced thermal conductivity
was observed to absorb hydrogen much faster, especially under forced convection conditions as
can be observed from Fig. 7. It should be noted that system performance here on is discussed in
terms of percentage reaction completion (hydrogen absorbed) with respect to time. The heat
transfer coefficient has been assumed on the basis of the known values of heat transfer
coefficients: natural convection in air 4.023 W/m2/K and around 1000 W/m2/K for forced
convection in water.
 c-enhanced k c-baseline k
c forced convection-enhanced k c forced convection-baseline k
0.59

0.54

0.49
Reaction fraction 'c'

0.44

0.39

0.34

0.29

0.24

0.19
0 400 800 1200 1600 2000 2400 2800 3200 3600
Time (s)

Figure 7: Comparison of fraction of hydrogen absorbed with and without thermal conductivity
enhancement under free and forced convection conditions

Figure 8 depicts the evolution of total heat flux streamlines, with the change in colour from blue
to red indicating an increase in magnitude of the heat flux. In case of natural convection with no
thermal conductivity enhancement, convection currents can be observed from Fig. 8 (a), which
rise due to the heat generated by reaction taking place on the alloy surface but for the most part
of the bed no significant heat transfer is taking place. At 500 s, however, heat transfer can also be
observed in the near wall region. Although the radial heat transfer in this case is far less
developed than the enhanced thermal conductivity cases i.e. Fig. 8 (e) and (f), where the
streamlines are originating from the core of the bed and the magnitude of flux is also much
higher. On comparing figures 8 (b) and (c), it can be noted that in case of forced convection the
reaction has not only moved closer to the core than natural convection, the magnitude of heat
flux is also much larger. Also in the figures at 50 s, strong convection currents can be observed
in the expansion volume regions which have weakened at 500 s. As discussed earlier, it is
because in the early stages of the reaction the metal hydride bed surface is a fast reaction zone
and the heat thus released is taken up by the gas present in the expansion volume and contributes
to the gas motion via thermal buoyancy.
 (a) (b) (c)

(d) (e) (f)

Figure 8: Heat flux streamlines at 50 s and 500 s in (a) and (d) natural convection with no k
enhancement; (b) and (e) natural convection with aluminum foam; (c) and (f) forced convection
with aluminum foam
4.2 Addition of water jacket

In the previous section it was observed that the system showed much better performance with
addition of aluminum foam under forced convection in water. In the next step, a water jacket is
added to the system, through which a fixed mass of water flows per unit time, determined by
simple energy balance. The width of the water jacket was decided on the basis of ASME
pressure vessel code (section VIII) and optimized on the basis of the system performance. Figure
9 (a) shows a half-sectional view of the reactor with jacket and aluminum foam. The
performance of the system is seen to depreciate upon increasing the jacket width as can be noted
from Fig. 9 (b). This could be attributed to the subsequent decrease in Reynolds’s number and
hence the heat transfer coefficient (since mass flow rate is kept constant). Therefore, it was
noted that 0.0278 m (2.78cm) was the most feasible jacket width.

Figure 9 (a) Half-sectional representation of the MH reactor with aluminum foam and external
water jacket
 0.64

0.59

0.54
Reaction fraction 'c'

0.49

0.44
0.0278 m
0.39
0.05162 m
0.34 0.07624 m

0.29

0.24

0.19
0 600 1200 1800 2400 3000 3600
Time (s)

Figure 9 (b): Comparison of fraction of hydrogen absorbed in MH reactors with aluminium foam
at different external water jacket widths

As can be seen from Fig.9 (b), the performance of the system does not meet the requirement of
80% to 90% reaction completion even in 3600 s. Therefore, further improvements in design are
required to meet that objective. Since heat transfer is known to be a major bottleneck in
hydrogen absorption/desorption, consequently we look at the methods through which heat
transfer can be further improved. Firstly, the thermal conductivity can be improved. Secondly,
heat transfer area can be added to the system such as fins, cooling tubes. Thirdly, the heat
transfer parameters such as heat transfer coefficient, heat transfer fluid temperature etc. can be
worked with. Decreasing the heat transfer fluid temperature below the ambient condition will
lead to the addition of cooling equipment, adding to auxiliary weight and costs. The first and the
third techniques have already been explored to some extent. The addition of heat transfer area in
the form of cooling tubes will now be investigated.
4.3 Addition of cooling tubes

The study was conducted with one, five and nine cooling tubes of three different sizes based
upon the standard dimensions. Since the embedded cooling tube model exceeds the
conformability limit of 1 m height significantly, therefore, it will have to be divided into two
reactors of radius 12.725 cm and length 1.004 m. The jacket width was also changed
accordingly. Fig. 10 (a) shows a section of the MH bed with nine cooling tubes, one at the centre
and two in each quadrant, uniformly spaced. Upon increasing the number of cooling tubes from
one to five a small improvement in the hydrogen absorption rate was noted as shown in Fig. 10
(b). However, when the number was increased from five to nine, the subsequent improvement
was even lesser. Increasing the diameter of the cooling tube also had a positive impact upon the
system performance. The reaction fraction achieved in 3600s even with 9 cooling tubes is only
around 66%.

Water jacket

Cooling tubes

MH bed with
aluminum foam

Figure 10 (a): MH bed with aluminum foam and nine cooling tubes surrounded by a water jacket
 0.69

0.64

0.59

0.54
Reaction fraction 'c' (1)

9 CT O.D.-0.0051 m

0.49 9 CT O.D.-0.006858 m
9 CT O.D.-0.0085725 m
0.44
5 CT O.D.-0.0051 m
0.39 5 CT O.D.-0.006858 m

0.34 5 CT O.D.-0.0085725 m

1 CT O.D.-0.0051 m
0.29
1 CT O.D.-0.006858 m
0.24 1 CT O.D.-0.0085725 m

0.19
0 400 800 1200 1600 2000 2400 2800 3200 3600
Time (s)

Figure 10 (b): Effect of addition of cooling tubes on fraction of hydrogen absorbed with one, five
and nine cooling tubes of three different diameters

Clearly, thermal conductivity, heat transfer coefficient, and heat transfer area are not the only
limiting parameters. The radius of the MH bed is pivotal in ensuring effective heat transfer as
was also pointed out by Gopal and Murthy [34,35], Visaria et al [36]. In order to study the effect
of bed radius, the temperature variation across different standard size bed radii was studied. As
can be seen from Fig. 11, the larger the bed radius, larger the range over which the temperature is
constant. As the bed radius decreases not only does the temperature decrease but the rate at
which it reduces also becomes sharper. Since opting for the smallest radius will not be the
optimum choice considering system size and cost. Therefore, middle range values of radii which
have a reasonably good temperature drop pattern were tested.
 0.0611 m 0.1524 m
0.0486 m 0.1215 m 0.1445 m
0.073165 m
0.04272 m

0.03683

Bed radius

Figure 11: Variation of temperature with bed radius for different MH bed radii

4.4 Multi-tube metal hydride reactor

As was observed from Fig. 10, increasing the cooling tubes did not improve upon the system
performance to the acceptable range of 80-90 %. Also Fig. 11 depicted that the temperature
within the core of the reactor remained constant for large diameter beds, therefore the bed radius
was further reduced. Clearly with a radius around half than that of the original tank, multiple
tubes will have to house the amount of metal, if the reactor height is to remain constant. Fixing
the height of the tank at 1 m, varying the radius of the tube, the number of tubes required can be
found since the volume of metal is known. The process is repeated till a feasible system height is
achieved. In the present case, with a radius of 0.073156 m (7.32 cm), the number of cooling
tubes was taken as 7, resulting in a metal bed height of 0.7134 m and a total height of 0.8561 m,
which is within the conformability limits. The system was designed along the lines of a shell and
tube exchanger, with the shell diameter given by [37] :

^_ ∗`a∗;7b ∗>cb
Ds = =0.635 m (15)
d.efg∗` ;

where, Nt = no. of tubes=7, CL=0.87 represents the incomplete coverage of the tubes by the
shell, CTP=0.93 is the constant for single-pass tubes, PR is the pitch to tube outer diameter ratio,
do= 0.16828 m (16.828 cm) is the outer diameter of the metal tube. It may be noted that a
triangular tube pitch with a PR value of 1.25 was chosen. Triangular tube pitch is more compact
and the value 1.25 falls within the recommended range of 1.25 to 1.5. With the shell diameter
known, a 2-D model of the tank was developed as shown in Fig. 12.

MH bed

Water

Figure 12: Shell and tube type metal hydride storage tank

The heat transfer coefficient for the aforesaid design was calculated by the Kern’s method and
was found to be 650.89 W/m2/K. Upon simulation it was observed that while system
performance improved over the single tank scenario, it was almost the same as the system with
cooling tubes. Hence, the tube radius was reduced further to 0.0486 m (4.86 cm). The
corresponding number of tubes was found to be 14 with a metal bed height of 0.8 m and system
height of 1 m. This resulted in a shell diameter Ds of 0.583 m and a heat transfer coefficient of
850 W/m2/K. A comparative analysis of the multi-tubular tank with 7 and 14 MH tubes, and the
single tank with nine cooling tubes, shown in Fig. 13, was done.
 I II III

Figure 13: Comparison of three different reactor configurations: i) 9 cooling tube

system; ii) 7- MH tube shell and tube type system; iii) 14-MH tube type shell and tube type
system

The evolution of heat flux streamlines in the reactors I i.e. with nine cooling tubes, II i.e.
with 7 MH tubes and III with 14 MH tubes is compared in Fig. 14. On the colour scale, blue
represents lower magnitude of heat flux and red the highest. In Fig. 14 (a) and (b), from the
orientation of the heat flux streamlines it can be seen that in the initial phases of the reaction
i.e. at 50 s, the external jacket seems to be a more active site of heat transfer as compared at
500 s. Each cooling surface has its zone of influence from which it draws heat. This would
lead to the conclusion that more number of cooling surfaces would improve heat transfer
leading to a higher reaction fraction; however, due to the fact that the contact area in case of
cooling tubes remains small, therefore, the impact may not be as significant. In Fig. 14 (c)
and (e), the highest heat flux can be observed near the wall while the core does not show
any major activity. At 500 s, in Fig. 14 (d) and (f), while the core registers higher heat flux
than before due to more reaction and the subsequent rise in temperature, however, the
magnitude is less than that at the wall. This may be due to the higher near wall temperature
gradient offered by the cooling fluid.
 (a) (b)

(c) (d)

(e) (f)

Figure 14: Heat flux evolution at 50 s and 500 s: (a), (b) 9 cooling tube configuration;
(c), (d) 7 metal hydride tubes; (e), (f) 14 metal hydride tubes
 T1 (K) T2(K) T3 (K) c1 c2 c3

320 0.7

0.6
315

Reaction fraction 'c'

0.5
Temperature (K)

310
0.4
305
0.3

300
0.2

295 0.1
0 400 800 1200 1600 2000 2400 2800 3200 3600
Time (s)
1) 9 cooling tube reactor; 2) 7 MH tubes multi-tubular; 3) 14 MH tubes multi-tubular

Figure 15: Comparative performance of the three different reactors at 25 bar supply pressure
Figure 15 shows the comparative performance of the three reactors in terms of temperature and
reaction fraction completed. The 14-MH tube multi-tubular reactor exhibits a superior
performance as compared to the other two reactors, reaching the maximum reaction fraction in
around 1600 s, whereas the 7-MH tube reactor and the nine cooling tube reactor take around
2700 s and 3400 s respectively. Also, observing T3 and the corresponding reaction fraction c3, a
very interesting point comes to light. Even though the average temperature is close to room
temperature and no heat transfer bottlenecks are present in the system, the reaction rate is still
very low. This phenomenon can be explained with the help of the equilibrium pressure equation
(14). It should be noted here that the modified Van’t Hoff equation has been used which includes
the MH concentration ratio term ‘c’. The effect of the MH concentration term may not be
dropped without loss of accuracy as was noted by Nam et al [38].

h+i j l n
= DE − + m + m\ ∗ IJK L ∗ M − 0.5 +
hR+X k k∗) 2
From the above equation, it is clear that once fifty per cent reaction completion is achieved, the
term ‘(c-0.5)’ in the above equation becomes positive leading to an addition in the exponential
index. This results in an increase in the equilibrium pressure not due to high temperature but as a
consequence of a higher hydride concentration in the system. This is quite simple even when
seen from an equilibrium reaction perspective.

E
+ l ↔ lq + lDJI
2

The above relation also indicates that as the hydride formation increases the reaction will be
stalled and the reaction rate will slow down. Therefore, while heat transfer may be an important
controlling parameter for a metal hydride formation reaction it is so only during the first half of
the reaction. The reaction kinetics may have a stronger impact on the reaction in later phase.
This, however, may not be true for all hydrides as the impact of the term ‘(c-0.5)’ would also
depend upon the coefficients m JKr m\ , which will vary with each hydride.

T1 T2 T3 c1 c2 c3

335 0.89

330 0.79

325
0.69 Reaction fraction 'c' (1)
Temperature (K)

320
0.59
315
0.49
310
0.39
305

300 0.29

295 0.19
0 500 1000 1500 2000 2500 3000 3500 4000
Time (s)

1) 9 cooling tube reactor; 2) 7 MH tubes multi-tubular; 3) 14 MH tubes multi-tubular

Figure 16: Comparative system performance of the three different reactors at 40 bar supply
pressure
In order to improve the performance through other means, the system was subjected to an
increased supply pressure and its behavior was analyzed. Figure 16, shows the response of the
three systems to an increased supply pressure of 40 bar. The rapid rise in temperature in all three
cases is consistent with the absorption of hydrogen, as fraction of hydrogen absorbed ‘c’
increases, releasing heat of reaction in process. At the supply pressure of 40 bar, a reaction
fraction of 80% can be achieved in around 900 s i.e.15 minutes in the 14-MH tube system,
whereas the same takes around 1800 s and 1900 s in 7-MH tube system and nine cooling tube
system respectively. The rapid fall of the temperature profiles is indicative of the effective heat
transfer in the 14- MH tube system.

Considering, the step-wise process undertaken so far, a design procedure for metal hydride
storage system design is suggested in Fig.17. First the system size is chosen on the basis of the
capacity, material etc. If the preliminary system diameter is small (less than 3 cm), then any
thermal conductivity enhancements may not be required. If the system is larger, using a thermal
conductivity enhancement will substantially improve the system performance. Once the method
of thermal conductivity enhancement has been decided upon, the next step is to check the bed
radius which is a critical parameter. While a small bed radius ensures efficient heat transfer, it is
difficult to maintain in large systems where system weight could be compromised by using
multiple small sized reactors instead of a single large one. There has to be a trade-off between
efficient heat transfer and lower system gravimetric capacity, as will be discussed in the next
section. After the selection of a suitable bed radius if the system performance is still not
satisfactory, additional heat transfer enhancements such as embedded cooling tubes etc. may be
included.
Figure 17: Design procedure for a metal hydride hydrogen storage system
4.5 System gravimetric capacities

Table 2: Comparison of empty reactor weights and performances:

9 Cooling tube
Mass (2 reactor system) 7-MH tube 14-MH tube
Mass of shell (kg) - 45.09161 48.48251836
Mass of tubes (kg) - 58.8738 77.77248356
Mass of tank (kg) 68.37446 - -
Mass of ECT (kg) 3.98 - -
Mass of mesh(kg) 20.4 22.61816 27.9078897
Total mass 92.75446 126.58357 149.5115767
System gravimetric
capacity 0.832% 0.749% 0.7%
80% absorption time
@ 40 bar (s) 1895 1805 900

The 14 tube multi-tubular system seems to outperform the other two systems as was seen from
Fig. 15 and Fig. 16. However, the inclusion of 14 cylinders would also add more weight to the
system. Any analysis undertaken for a practical system must weigh the benefits of efficient
performance against the additional weight that may be added to the system in such a case. Table
2 presents a comparison of the empty reactor weight of the three systems analysed in the last
section: a) system with 9 cooling tubes b) system with 7 MH tubes c) system with 14 MH tubes.
The 14-MH-tube weighs 1.6 times more than the nine cooling tube system. Also the 14-MH tube
reaches 80% reaction completion in less than half the time taken by the latter. Since the objective
of the present study is to cater to stationary applications, the 14-MH tube configuration will be
the most suitable design, as system weight is not a constraint. Also, in comparison to the large-
scale systems previously studied the system performance and gravimetric capacity in the present
case is quite promising considering that the alloy used in those cases was sodium alanate which
has an inherently higher gravimetric capacity compared to the Ti alloy [20,21,25]. However, the
hydrogen absorption capacity of sodium alanate diminishes upon cyclic usage and the kinetics
are also not as feasible as those of the Ti-Mn based alloys[39].
4.6 Desorption behavior of the multi-tubular configuration (14 tubes)

Any study on the performance of a metal hydride system will be inconclusive until and unless
desorption behavior of the system is also taken into consideration. Since the 14-tube shell and
tube configuration exhibited good absorption characteristics, desorption study for the same was
carried out. The mathematical model used for the same has been previously reported in literature
as discussed in section 3.4. Figure 18 shows the behavior of 14-tube multi-tubular system at a
desorption temperature of 313K and a pressure of 1 bar. At around time 2000 s, desorption
process is completed. It may be noted that the temperature and pressure of desorption are very
feasible for the operation of a PEM fuel cell.

T (K) 1-c
300 0.9

0.8
295
0.7

0.6
290
Temperature (K)

0.5

285 0.4 1-'c'

0.3
280
0.2

0.1
275
0

270 -0.1
0 400 800 1200 1600 2000 2400 2800 3200 3600
Time(s)

Figure 18: Desorption characteristics of 14-tube multi-tube hydrogen storage system

 5. Conclusions

On the basis of sensitivity to different design and operating parameters such as porosity, thermal
conductivity, pressure etc., the performance of a large-scale 210 kg Ti-Mn based MH hydrogen
storage system was analyzed. It was noted that the performance was most sensitive to changes in
thermal conductivity and pressure. With this as starting point an improvised design was
suggested. As a result of the series of studies involving different heat exchange configurations, a
design procedure was arrived at. Based on the analysis, a 14-MH multi-tube metal hydride
reactor was recommended as the best configuration for the present application. At a system
gravimetric capacity of 0.7 %, the absorption process was completed in 900 s and desorption
process in 2000 s demonstrating a superior performance. It was also found that after fifty percent
reaction completion, heat transfer ceases to be the major factor controlling system performance
as the effect of reaction kinetics becomes more significant.

Acknowledgement

The authors are thankful to Prof. Suneet Singh, Department of Energy Science and Engineering,
IIT Bombay for providing us the computational facility and Indian Space Research Organisation
for financially supporting the research vide project code 16ISROC001. The authors would also
like to thank Prof. P. Muthukumar for helpful discussions.

References:

[1] Capurso G, Schiavo B, Jepsen J, Lozano G, Klassen T, Dornheim M. Development of a

modular room-temperature hydride storage system for vehicular applications. Appl Phys
A 2016;236:1–11. doi:10.1007/s00339-016-9771-x.

[2] Nagel M, Komazaki Y, Suda S. Effective thermal conductivity of a metal hydride bed
augmented with a copper wire matrix. J Less-Common Met 1986;120:35–43.
doi:10.1016/0022-5088(86)90625-9.

[3] Laurencelle F, Goyette J. Simulation of heat transfer in a metal hydride reactor with
 aluminium foam. Int J Hydrogen Energy 2007;32:2957–64.
doi:10.1016/j.ijhydene.2006.12.007.

[4] Ron M, Gruen D, Mendelsohn M, Sheft I. Preparation and properties of porous metal
hydride compacts. J Less Common Met 1980;74:445–8. doi:10.1016/0022-
5088(80)90183-6.

[5] Lloyd G, Kim KJ, Razani A, Feldman TK. Thermal Conductivity Measurements of Metal
Hydride Compacts Developed for High-Power Reactors. J Thermophys Heat Transf
1998;12:132–7. doi:10.2514/2.6332.

[6] Sánchez AR, Klein H-P, Groll M. Expanded graphite as heat transfer matrix in metal
hydride beds. Int J Hydrogen Energy 2003;28:515–27. doi:10.1016/S0360-
3199(02)00057-5.

[7] Garrier S, Chaise A, De Rango P, Marty P, Delhomme B, Fruchart D, et al. MgH2

intermediate scale tank tests under various experimental conditions. Int J Hydrogen
Energy 2011;36:9719–26. doi:10.1016/j.ijhydene.2011.05.017.

[8] Lozano GA, Ranong CN, Bellosta Von Colbe JM, Bormann R, Hapke J, Fieg G, et al.
Optimization of hydrogen storage tubular tanks based on light weight hydrides. Int. J.
Hydrogen Energy, vol. 37, 2012, p. 2825–34. doi:10.1016/j.ijhydene.2011.03.043.

[9] Delhomme B, De Rango P, Marty P, Bacia M, Zawilski B, Raufast C, et al. Large scale
magnesium hydride tank coupled with an external heat source. Int J Hydrogen Energy
2012;37:9103–11. doi:10.1016/j.ijhydene.2012.03.018.

[10] Ferekh S, Gwak G, Kyoung S, Kang HG, Chang MH, Yun SH, et al. Numerical
comparison of heat-fin-and metal-foam-based hydrogen storage beds during hydrogen
charging process. Int. J. Hydrogen Energy, vol. 40, 2015, p. 14540–50.
doi:10.1016/j.ijhydene.2015.07.149.

[11] Guo Z. Enhancement of heat and mass transfer in metal hydride beds with the addition of
Al plates. Heat Mass Transf 1999;34:517–23. doi:10.1007/s002310050290.

[12] Muthukumar P, Maiya MP, Murthy SS. Experiments on a metal hydride-based hydrogen
 storage device. Int J Hydrogen Energy 2005;30:1569–81.
doi:10.1016/j.ijhydene.2004.12.007.

[13] MacDonald BD, Rowe AM. Impacts of external heat transfer enhancements on metal
hydride storage tanks. Int J Hydrogen Energy 2006;31:1721–31.
doi:10.1016/j.ijhydene.2006.01.007.

[14] Askri F, Ben Salah M, Jemni a., Ben Nasrallah S. Optimization of hydrogen storage in
metal-hydride tanks. Int J Hydrogen Energy 2009;34:897–905.
doi:10.1016/j.ijhydene.2008.11.021.

[15] Freni A, Cipitì F, Cacciola G. Finite element-based simulation of a metal hydride-based

hydrogen storage tank. Int J Hydrogen Energy 2009;34:8574–82.
doi:10.1016/j.ijhydene.2009.07.118.

[16] Muthukumar P, Singhal A, Bansal GK. Thermal modeling and performance analysis of
industrial-scale metal hydride based hydrogen storage container. Int J Hydrogen Energy
2012;37:14351–64. doi:10.1016/j.ijhydene.2012.07.010.

[17] Anbarasu S, Muthukumar P, Mishra SC. Tests on LmNi4.91Sn0.15 based solid state
hydrogen storage device with embedded cooling tubes - Part A: Absorption process. Int J
Hydrogen Energy 2014;39:3342–51. doi:10.1016/j.ijhydene.2013.12.090.

[18] Nyamsi SN, Yang F, Zhang Z. An optimization study on the finned tube heat exchanger
used in hydride hydrogen storage system - Analytical method and numerical simulation.
Int J Hydrogen Energy 2012;37:16078–92. doi:10.1016/j.ijhydene.2012.08.074.

[19] Ma J, Wang Y, Shi S, Yang F, Bao Z, Zhang Z. Optimization of heat transfer device and
analysis of heat & mass transfer on the finned multi-tubular metal hydride tank. Int. J.
Hydrogen Energy, vol. 39, 2014, p. 13583–95. doi:10.1016/j.ijhydene.2014.03.016.

[20] Garrison SL, Hardy BJ, Gorbounov MB, Tamburello DA, Corgnale C, Vanhassel BA, et
al. Optimization of internal heat exchangers for hydrogen storage tanks utilizing metal
hydrides. Int. J. Hydrogen Energy, vol. 37, 2012, p. 2850–61.
doi:10.1016/j.ijhydene.2011.07.044.
[21] Na Ranong C, Höhne M, Franzen J, Hapke J, Fieg G, Dornheim M, et al. Concept, design
and manufacture of a prototype hydrogen storage tank based on sodium alanate. Chem
Eng Technol 2009;32:1154–63. doi:10.1002/ceat.200900095.

[22] Dhaou H, Souahlia A, Mellouli S, Askri F, Jemni A, Ben Nasrallah S. Experimental study
of a metal hydride vessel based on a finned spiral heat exchanger. Int J Hydrogen Energy
2010;35:1674–80. doi:10.1016/j.ijhydene.2009.11.094.

[23] Wang H, Prasad AK, Advani SG. Hydrogen storage system based on hydride materials
incorporating a helical-coil heat exchanger. Int J Hydrogen Energy 2012;37:14292–9.
doi:10.1016/j.ijhydene.2012.07.016.

[24] Wu Z, Yang F, Zhang Z, Bao Z. Magnesium based metal hydride reactor incorporating
helical coil heat exchanger: Simulation study and optimal design. Appl Energy
2014;130:712–22. doi:10.1016/j.apenergy.2013.12.071.

[25] Raju M, Kumar S. Optimization of heat exchanger designs in metal hydride based
hydrogen storage systems. Int J Hydrogen Energy 2012;37:2767–78.
doi:10.1016/j.ijhydene.2011.06.120.

[26] Mazzucco A, Voskuilen TG, Waters EL, Pourpoint TL, Rokni M. Heat exchanger
selection and design analyses for metal hydride heat pump systems. Int J Hydrogen
Energy 2016;41:4198–213. doi:10.1016/j.ijhydene.2016.01.016.

[27] Afzal M, Mane R, Sharma P. Heat transfer techniques in metal hydride hydrogen storage:
A review. Int J Hydrogen Energy 2017;42:30661–82. doi:10.1016/j.ijhydene.2017.10.166.

[28] MacDonald BD, Rowe AM. Experimental and numerical analysis of dynamic metal
hydride hydrogen storage systems. J Power Sources 2007;174:282–93.
doi:10.1016/j.jpowsour.2007.09.021.

[29] Nishizaki T, Miyamoto K, Yoshida K. Coefficients of performance of hydride heat

pumps. J Less-Common Met 1983;89:559–66. doi:10.1016/0022-5088(83)90372-7.

[30] Muthukumar P, Madhavakrishna U, Dewan A. Parametric studies on a metal hydride

based hydrogen storage device. Int J Hydrogen Energy 2007;32:4988–97.
 doi:10.1016/j.ijhydene.2007.08.010.

[31] Chung CA, Ho CJ. Thermal-fluid behavior of the hydriding and dehydriding processes in
a metal hydride hydrogen storage canister. Int J Hydrogen Energy 2009;34:4351–64.
doi:10.1016/j.ijhydene.2009.03.028.

[32] Jiao K, Li X, Yin Y, Zhou Y, Yu S, Du Q. Effects of various operating conditions on the

hydrogen absorption processes in a metal hydride tank. Appl Energy 2012;94:257–69.
doi:10.1016/j.apenergy.2012.01.033.

[33] Bao Z, Yang F, Wu Z, Cao X, Zhang Z. Simulation studies on heat and mass transfer in
high-temperature magnesium hydride reactors. Appl Energy 2013;112:1181–9.
doi:10.1016/j.apenergy.2013.04.053.

[34] Gopal MR, Murthy SS. Studies on heat and mass transfer in metal hydride beds. Int J
Hydrogen Energy 1995;20:911–7. doi:10.1016/0360-3199(95)00026-A.

[35] Gopal MR, Murthy SS. Parametric Studies on Heat and Mass Transfer in Metal Hydride
Beds. Chem Eng Process Process Intensif 1993;32:217–23. doi:10.1016/0255-
2701(93)80003-Y.

[36] Visaria M, Mudawar I, Pourpoint T. Enhanced heat exchanger design for hydrogen
storage using high-pressure metal hydride: Part 1. Design methodology and computational
results. Int J Heat Mass Transf 2011;54:413–23.
doi:10.1016/j.ijheatmasstransfer.2010.09.029.

[37] Kakac S, Liu H, Pramuanjaroenkij A. Heat Exchangers: selection, rating and thermal
design. 3rd ed. Boca Raton: CRC Press; 2012.

[38] Nam J, Ko J, Ju H. Three-dimensional modeling and simulation of hydrogen absorption in

metal hydride hydrogen storage vessels. Appl Energy 2012;89:164–75.
doi:10.1016/j.apenergy.2011.06.015.

[39] Motyka T. Hydrogen Storage Engineering Center of Excellence Metal Hydride Final
Report No. SRNL-STI-2014-00226. Savannah River Site (SRS), Aiken, SC (United
States): 2014.
Graphical Abstract