International Journal of Heat and Mass Transfer 54 (2011) 413–423

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

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

Enhanced heat exchanger design for hydrogen storage using high-pressure metal
hydride: Part 1. Design methodology and computational results
Milan Visaria, Issam Mudawar ⇑, Timothée Pourpoint
Hydrogen Systems Laboratory, Purdue University, West Lafayette, IN 47907, United States

a r t i c l e i n f o a b s t r a c t

Article history: This study explores the use of a high-pressure metal hydride (HPMH), Ti1.1CrMn, to store hydrogen at
Received 24 February 2010 high pressures (up to 310 bar) and temperatures below 60 °C, conditions that are suitable for automobile
Received in revised form 18 August 2010 fuel cells. However, the exothermic reaction of hydrogen with this material releases large amounts of
Accepted 18 August 2010
heat, and the reaction rate depends on the metal hydride temperature, decreasing significantly if the heat
Available online 11 October 2010
is not removed quickly. Therefore, a powerful heat exchanger constitutes the most crucial component of a
HPMH hydrogen storage system. For automobiles, this heat exchanger must enable fueling 5 kg of hydro-
Keywords:
gen in less than 5 min. This is a formidable challenge considering the enormous amount of heat that must
Solid-state hydrogen storage
Heat exchanger
be released and the stringent limits on the heat exchanger’s weight and volume, let alone a host of man-
High-pressure metal hydride ufacturing requirements. Unlike conventional heat exchangers that are designed to exchange heat
Ti1.1CrMn between two fluids, this heat exchanger is quite unique in that it must dissipate heat between a reacting
powder and a coolant. In this first of a two-part study, a systematic heat exchanger design methodology is
presented, starting with a 1-D criterion and progressing through a series of engineering decisions sup-
ported by computations of fill time. A final design is arrived at that meets the 5-min fill time requirement
corresponding to minimum heat exchanger mass, supported by a 2-D computational model of the heat
exchanger’s thermal and kinetic response.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction (HPMHs) constitute one of the prime candidates for automotive

hydrogen storage systems.
With the increasing need for clean and sustainable energy op- HPMHs that are suitable for hydrogen storage have the highest
tions, interest in alternate fuels like hydrogen has never been higher. volumetric capacities compared to most other storage techniques
Hydrogen is both abundant in nature and has high gravimetric [4]. With these materials, both the hydriding process (producing
energy content. However, due to its low density, it poses major chal- metal hydride by reaction of hydrogen gas with metal alloy) and
lenges in both storage and transport. Hydrogen can be stored as a dehydriding process (releasing the hydrogen from the metal hy-
very high-pressure gas or very low temperature liquid. Alternately, dride) are reversible and achievable at temperatures and pressures
it can be stored in solid form by undergoing a chemical reaction into that are suitable for fuel cell use [2–4,6]. Although HPMHs have
chemical hydrides, metal hydrides, or cryo-adsorbents [1–4]. Each advantages over other storage techniques, they also pose several
of these storage techniques has its pros and cons, and the choice of challenges. First, their low gravimetric capacity increases storage
a particular technique is dictated by the application. For automo- system weight [4,6]. Second, HPMHs must be handled and stored
biles, the system must be as compact and lightweight as possible. with extreme care. This is because for HPMHs to absorb hydrogen,
Another important practical requirement is that the time required they must first be pre-processed – activated – by breaking them
to refuel a vehicle must be short. The Department of Energy (DOE) into very fine particles. Among HPMHs, Ti–Cr–Mn-based metal hy-
targets for onboard hydrogen storage is to charge 5 kg of hydrogen drides are the most widely studied because of their cold-starting
at the fueling station in less than 5 min by the year 2010 [5]; this capability. Activating these hydrides requires breaking them to
amount of hydrogen covers a travel distance of 300 miles before particles below 10 lm. However, once activated, they become
needing to be refueled. In solid-state hydrogen storage, it is addi- pyrophoric and react readily in presence of air or moisture. There-
tionally important that the hydrogen be quickly and repeatedly re- fore, they must be stored and handled with extreme care, and
leased in response to the demands of the fuel cell or internal sealed off from air or water before being packed into the storage
combustion engine. Presently, high-pressure metal hydrides system. Thirdly, and most importantly, the hydriding process is
highly exothermic, with a temperature-dependent reaction rate
⇑ Corresponding author. Tel.: +1 765 494 5705; fax: +1 765 494 0539. that decreases significantly at high temperatures. In fact, the most
E-mail address: [email protected] (I. Mudawar). crucial component of a HPMH storage system is a heat exchanger

0017-9310/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2010.09.029
414 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423

Nomenclature

Ca hydriding constant or activation rate (s1) wt% hydrogen to metal hydride mass ratio when completely
cp specific heat (J/kg K) hydrided
Ea activation energy (J/mol-H2) x coordinate (mm)
F fraction of completion of reaction y coordinate (mm)
DHr enthalpy of reaction (J/mol-H2)
h convective heat transfer coefficient (W/m2 K) Greek symbols
k thermal conductivity (W/m K) l viscosity
L metal hydride thickness (mm) q density
MW molecular weight u porosity
P pressure (N/m2)
Peq equilibrium pressure (N/m2) Subscripts
Po ambient pressure (N/m2) Al Aluminum
q_ 000 volumetric heat generation rate (W/m3) eq equilibrium
R universal gas constant (8.314 J/mol K) f coolant
Rtc contact resistance (mm2 K/W) H2 Hydrogen
DS entropy of reaction (J/mol-H2 K) MH metal hydride
t time (s) tot total length used in 1-D model
T temperature (K)

that must remove the large amounts of heat released during the (up to 500 bar) and ensure leak-proof operation. A practical objec-
hydriding process. tive of the heat exchanger is to achieve a hydrogen fill time under
Hydriding is a pressure and temperature dependent process 5 min using Ti1.1CrMn powder as storage material. In this part of
associated with high heat of reaction. The Ti1.1CrMn used in the this study, the detailed design methodology is discussed along
present study has a measured hydriding heat of reaction of about with the computational model and computational results. The sec-
14 kJ/mol-H2 [4] and dehydriding heat of reaction around 22 kJ/ ond part [13] will discuss the experimental methods used during
mol-H2 [7]. If at any time during the hydriding process the hydride the validation tests, present the experimental results, and compare
temperature exceeds the temperature corresponding to equilib- them with computational results.
rium pressure, the reaction will stop and cannot proceed further
until the temperature is brought once again below the equilibrium 2. Heat exchanger design
temperature. In addition, the rate of hydriding depends on the
temperature of the hydride, the lower the hydride temperature 2.1. 1-D and 2-D models
the faster the reaction. Thus, the key to achieving fast refueling
rates is to (1) quickly remove the heat released during hydriding The key tool in initiating the design of the heat exchanger is a
and (2) maintain low metal hydride temperatures. Since the metal 1-D model that is used to calculate the ‘‘maximum thickness of
hydride itself has low gravimetric capacity, it is important that the metal hydride layer,” which is the largest distance of hydride pow-
heat exchanger and its components possess both high gravimetric der from a liquid-cooled aluminum surface such that the heat can
efficiency and high volumetric efficiency (to maximize the volume be removed at a rapid rate that yields the desired reaction rate and
available for the hydride). The DOE target for the gravimetric den- a fill time of less than 5 min. Details of this 1-D model can be found
sity of automobile storage systems is 4.5 wt% by the year 2010 [5]. in a recent study by the present authors [4]. Due to thermal
To achieve this target, it is important to develop new and improved symmetry, a layer of hydride contained between two aluminum
materials along with highly optimized heat exchanger and storage surfaces can be assumed adiabatic along the centerline as shown
system designs. in Fig. 1. The model accounts for contact resistance between the
For decades, heat exchangers have been extensively used in hydride powder and the aluminum surface. A convective boundary
many diverse industrial applications. The most widely used heat characterized by a coolant temperature and a convective heat
exchangers are those intended to exchange heat between two fluids transfer coefficient is assumed on the other side of the aluminum
across a solid surface. These heat exchangers, like shell-and-tube, wall. As discussed in [4], the effects of hydride distance on hydride
cross-flow, plate-type, etc., have been studied extensively for many temperature, heat generation rate, and hydrogen fill time were
decades. However, not much attention has been given to heat ascertained by numerically solving the 1-D model using the prop-
exchangers involving heat flow between a reacting powder and erties and operating conditions listed in Table 1. Based on these
fluid across a solid surface. Design of this type of heat exchangers results, the maximum thickness for Ti1.1CrMn was estimated at
is becoming increasingly important with recent advances in about 10 mm.
nano-materials and powder metallurgy, let alone solid hydrogen Fig. 2(a) illustrates the relevance of the 1-D maximum thickness
storage. to 2-D (plate-type) heat exchanger design. Since one boundary of
A few recent heat exchanger studies related to metal hydride the hydride is assumed adiabatic in the 1-D model, the critical
hydrogen storage have been focused on low-pressure complex me- thickness is actually half the distance between two cooling sur-
tal hydrides [8–12], yet fewer studies address HPMH heat exchang- faces. Thus, a maximum thickness of 10 mm corresponds to a
ers [4,6]. In this paper, which is the first of a two-part study, the 20 mm distance between the cooling surfaces.
detailed process of designing a reacting powder-to-fluid heat ex- The high pressures required for HPMH storage systems preclude
changer for storing hydrogen using HPMHs is discussed. Since the use of the 2-D layout depicted in Fig. 2(a) or plate-type heat
HPMHs ignite spontaneously in the presence of air or moisture, exchanger designs. Instead, only a cylindrical outer shape is
their safe and reliable design is of utmost important. These heat permissible for this application since the heat exchanger must be
exchangers must also be designed to operate at high pressure inserted inside a cylindrical high-pressure vessel. Overall, the heat
 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423 415

Fig. 1. 1-D model setup for heat exchanger design.

Table 1 (3) Single coolant U-tube.

Metal hydride properties and coolant conditions used in model. (4) Two-dimensional shape that lends itself to cost-effective
Kinetic fabrication techniques such as extrusion.
Activation energy Ea = 20.7 kJ/mol of H2
Enthalpy of formation DHr = 14,390 J/mol of H2 Ideally, a highly efficient ‘‘extruded” design is desired in which
Entropy of formation DS = 91.3 J/mol of H2
Activation rate Ca = 150 s1
the hydride is housed in small ‘‘cells” that are cooled by a network
H2 storage capacity 1.5 wt% of fins that effectively channel the heat to the coolant U-tube as
Thermal
illustrated in Fig. 2(b). Assuming very high fin efficiency for the
Packing density qMH = 2500 kg/m3 time being, the usefulness of the 1-D maximum hydride thickness
Effective thermal conductivity kMH = 1 W/m K model lies in the sizing of the individual hydride cells. As indicated
Specific heat cp,MH = 500 J/kg K in Fig. 2(b) for two different cells, the center of the cell corresponds
Contact resistance Rtc = 2000 mm2 K/W
to the hydride point farthest from the aluminum surface and must
Coolant therefore satisfy the maximum thickness criterion as a staring
Temperature Tf = 0 °C
point in the heat exchanger design.
Convection coefficient hf = 2500 W/m2 K

2.2. Design progression

exchanger must adhere to four practical vehicle manufacturer de-
sign guidelines: Achieving the maximum hydride thickness criterion while
attempting to minimize heat exchanger weight and volume are
(1) Cylindrical external shape. two opposing design trends. The first requires using aluminum fins
(2) Aluminum construction. throughout the metal hydride, which tends to increase heat

Fig. 2. Relevance of 1-D maximum hydride layer thickness model to design of (a) 2-D plate-type and (b) cylindrical metal hydride heat exchangers.
416 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423

exchanger weight and volume. Hence, it is important that the fin er heat transfer rates. As a result, cells closer to the coolant tubes
structures be as thin and thermally efficient as possible. Following are made larger compared to the cells farther from the coolant
is a discussion of an iterative design progression that was used in tubes. The heat exchanger design in Fig. 3(c) occupies 33.1% of
pursuit of this goal. pressure vessel volume using a fin thickness of 3 mm and
Fig. 3(a)–(d) show the steps taken in the heat exchanger design achieves a fill time of 5 min.
progression, and the volume occupied by the heat exchanger as a Fin efficiency can be improved further by using radial fins orig-
fraction of the pressure vessel (PV) volume before reaching the fi- inating directly from the coolant tube rather than using branching
nal design. The procedure involves designing the heat exchanger fins. Fig. 3(d) shows the final heat exchanger design in which all
and calculating the fill time using a two-dimensional Fluent model the fins originate from the coolant tube. With this design, which
(details of this model will be discussed later). If the calculated fill satisfies the 5 min fill time, the heat exchanger occupies only
time is less than 5 min, the design is optimized by reducing fin 29% of the pressure vessel volume, 25% smaller than the design
mass. However, if the fill time of 5 min is not achieved, either in Fig. 3(a). This demonstrates the effectiveness of the design opti-
the fin mass is increased or the fin distribution altered. This process mization methodology.
continues until an optimum heat exchanger design is obtained Overall, there is a narrow range for optimum fin thickness. Be-
which achieves a fill time below 5 min. low this range, the fin cannot provide sufficient conduction path,
Fig. 3(a) shows an initial heat exchanger design using 5-mm and above this range, the size of the heat exchanger increases with-
thick fins, which occupies 39% of the pressure vessel volume. out a noticeable improvement in performance. These consider-
For this design, the Fluent model predicts a fill time of 4.5 min, ations lead to an optimum fin thickness between 2 and 3 mm.
which implies the heat exchanger performance can be improved Also, because metal hydride powder can expand during hydriding
by rearranging fins and/or reducing fin thickness. Fig. 3(b) shows and contract during dehydriding by as much as 30%, it can exert
a subsequent design in which the fin thickness is reduced to high stresses on the heat exchanger fins. The 2–3 mm thickness
2 mm and some fins are removed or repositioned. While the heat is deemed adequate to withstand these stresses.
exchanger now occupies only 24.9% of the pressure vessel vol-
ume, the fill time exceeds 5 min. Upon examining transient tem-
perature results, several locations were detected where 2.3. Final design
insufficient cooling caused the hydriding reaction to slow and
stop altogether, resulting in longer overall fill times. Those ‘hot The iterations and optimization that culminated in the design
spots’ were eliminated by strategic placement of fins and, more shown in Fig. 3(d) was modified to the one shown in Fig. 3(e) for
importantly, by compartmentalizing the hydride powder into experimental validation purposes. Both designs are identical ex-
cells surrounded by fins on all sides to provide a more uniform cept for six circumferential holes that were required for experi-
heat transfer path from all directions. This modification can be mental assembly. The following section discusses details of the
observed in the transition between the designs in Fig. 3(b) and prototype heat exchanger that was used in the experimental vali-
(c). Locations closer to the coolant tubes react faster due to high- dation tests.

Fig. 3. Progression of heat exchanger designs.

 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423 417

3. Prototype heat exchanger construction and instrumentation diameter of the collet taper is a few thousandths of an inch larger
than the diameter of the heat exchanger plate holes to achieve
3.1. Heat exchanger plates interference fit and hence good contact between the two.
During assembly, the collets are first slid over each leg of the
To assess the heat exchanger’s performance and compare it to coolant U-tube, and the heat exchanger plates pushed axially in-
computational results, a 260.3-mm (10.25-in.) long prototype wards towards the center the collets. The longitudinal slots allow
was fabricated. With a metal hydride density of 2500 kg/m3, the the collet to compress on the coolant tube as the hexagonal nuts
heat exchanger could store 2.65 kg of metal hydride powder. on the end of the collet press the plates axially towards the center.
As discussed earlier, one of the practical design guidelines of the This results in excellent mechanical and thermal contact between
heat exchanger is a two-dimensional construction that lends itself the heat exchanger plate and the collets and coolant tube.
to cost-effective fabrication techniques such as extrusion. Because
of the need to instrument the metal hydride cells with thermocou- 3.3. Cover plates
ples at several strategic locations, the solid extruded structure of
the heat exchanger was replaced by a series of 6.35-mm The heat exchanger is fitted with two cover plates, which are at-
(0.25 in.) thick aluminum 6061 plates – slices – having the identi- tached as shown in Fig. 6. Aside from sealing the two ends of the
cal cross-section and material as the extrusion. The plates were heat exchanger, the cover plates contain hydrogen and coolant fit-
fabricated by CNC machining. Fig. 4 shows the cross-section of tings as well as thermocouple feed-through. Made from 304 stain-
the heat exchanger plates. It has a network of fins radiating from less steel, the cover plates contain five fittings each: two for the
the coolant tube, forming cells where the metal hydride powder coolant U-tube, two for check valves, and one for a thermocouple
is filled. feed-through. The check valves play the dual role of (1) preventing
air from flowing into the containment vessel as the heat exchanger
is being transported from the glove box to the pressure vessel, and
3.2. Tapered collet
(2) allowing hydrogen gas to enter the containment vessel during
testing. A circular cavity within each cover plate provides space
Contact resistance between the heat exchanger plates and the
for a diffuser plate. This plate is a sintered disk made from 316
coolant tube was minimized by using tapered collets around the
stainless steel with 30% porosity and 10-lm pore size. Its purpose
coolant tubes. Fig. 5 shows the detailed construction of the collet.
is to distribute hydrogen gas uniformly to all the heat exchanger
It is a hollow, cylindrical tube with an inner diameter slightly lar-
cells as it flows through the two check valves, as well as to serve
ger than the outer diameter of the coolant tube, and is tapered on
as a filter for the metal hydride powder. Additional filtering is pro-
the outside. Three 1-mm slots run most of the collet length save for
vided by sheets of filter paper placed between the cover plate and
two threaded ends, where support hexagonal nuts are connected.
the adjacent heat exchanger plate.
The collet has a bidirectional taper of 1.2°, such that its diameter
increases from the two ends towards the center, where the diam-
eter is largest. Having a bidirectional taper instead of a unidirec- 3.4. Heat exchanger test assembly
tional one reduces the maximum diameter of the collet by half.
The tapered portion of the collet is 254-mm (10-in.) long and each Figs. 6 and 7 depict CAD renderings of the heat exchanger
half holds 20 heat exchanger plates. The holes in the heat exchan- assembly without the containment vessel. Some heat exchanger
ger plates through which the coolant tube and collets pass are also plates are omitted in Fig. 7 to show internal details. Notice that
tapered at the same angle to match the collet taper. The outer in addition to the 40 aluminum heat exchanger plates, a Teflon

Fig. 4. Cross-sectional view of heat exchanger plate.

418 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423

Fig. 5. Isometric view and detailed drawing of tapered collet.

Fig. 6. 3-D rendering of heat exchanger assembly.

plate similar in design to the other plates is added in the center. 4. Computational model
The softer Teflon material allows for any axial mismatch in the ta-
per of the collet and heat exchanger plate holes to be rectified. Two 4.1. Model description
heat exchanger plates are modified with stainless steel posts to
mount thermocouples at precise locations within the metal hy- Tests were performed that provided a fairly uniform fluid tem-
dride powder. These plates are placed one quarter and three quar- perature along the coolant tube. Most operating conditions pro-
ters along the length of the heat exchanger. duced a total temperature drop between the inlet and outlet of
 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423 419

Fig. 7. Detailed 3-D rendering of heat exchanger assembly.

the coolant tube of less than 0.5 °C with a worst case of 1.5 °C. are the effective thermal conductivity, density and specific heat of
These operating conditions produced a fairly 2-D temperature dis- the metal hydride powder, respectively, and q_ 000 is the rate of volu-
tribution within the heat exchanger’s cross-section, which also metric heat generation in the hydride. Eqs. (1) and (2) are solved
simplified the understanding and prediction of the metal hydride in Fluent to obtain the temperature distribution across the domain.
kinetics. The heat generation in Eq. (2) is the sum of heat of reaction and heat
The domain of the 2-D Fluent model employed in this study is a of pressurization (accounts for less than 10% of the total heat
cross-section of the heat exchanger shown as in Fig. 3(e). Since the generated).
upper and lower halves are thermally symmetrical, only one half is
dF ðwt%ÞqMH dP
modeled to reduce computing time. The heat exchanger is made of q_ 000 ¼ DH r þ / ; ð3Þ
aluminum 6061-T6 and the cells are filled with Ti1.1CrMn powder. dt MW H2 dt
The network of fins dissipates the heat generated from the metal
where F, MW H2 , DHr, / and P are the fraction of completion of reac-
hydride powder during hydriding to the coolant flowing through
tion, molecular weight of hydrogen, enthalpy of reaction, porosity
the U-tube. Since the heat exchanger plates are enclosed in 316
and pressure, respectively. As indicated in Eq. (3), the rate of heat
stainless steel containment vessel and encased in a low thermal
generation depends on the rate of hydriding, the faster the reaction
conductivity nylon sleeve before the entire assembly is finally in-
rate the higher the heat generation rate. The reaction rate is com-
serted in the high pressure vessel for testing, the model assumes
puted using an expression for hydrogen absorption in LaNi5 derived
an adiabatic outer boundary for the heat exchanger. Thus, in the
by Mayer et al. [14] which assumes 1st order kinetics typically ob-
model, all the heat generated during the hydriding is removed by
served in metal hydrides.
the coolant.
   
Assuming the hydride has constant and isotropic properties, the dF Ea P
¼ C a exp ln ð1  FÞ; ð4Þ
temperature at any location within the hydride can be defined by dt RT Peq
two-dimensional transient heat conduction with internal heat gen-
eration. The heat diffusion equations for the aluminum and the hy- where Ca, Ea, R and Peq are the hydriding constant, activation energy,
dride are given, respectively, by universal gas constant and equilibrium pressure, respectively. Eq.
! (4) helps explain the effects of pressure and hydride temperature
@2T @2T @T on the rate of reaction and, hence, heat generation rate. The reaction
kAl þ ¼ qAl cp;Al ð1Þ
@x2 @y2 @t cannot proceed until the pressure within the storage system ex-
ceeds the equilibrium pressure and stops once the pressure falls be-
and low equilibrium. Because Peq increases with increasing T, the
! reaction rate is slower at higher hydride temperature and, to accel-
@2T @2T @T erate the reaction rate, it is necessary to lower the hydride temper-
kMH þ þ q_ 000 ¼ qMH cp;MH ; ð2Þ
@x2 @y2 @t ature. This can be further explained by the metal hydride
temperature and equilibrium pressure plots in Fig. 8. The metal hy-
where kAl, qAl and cp,Al are the thermal conductivity, density and dride powder far from the cooling surface is at a relatively high
specific heat of the aluminum, respectively, kMH, qMH and cp,MH temperature and, therefore, has higher equilibrium pressure. The
420 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423

small difference between the pressure vessel pressure and the equi-
librium pressure causes reaction rate to be slow for metal hydride
in these locations; the reaction even stops when the two pressures
are equal. However, for metal hydride powder close to the cooling
surface, the difference between the two pressures is greater and,
therefore, the reaction rate is faster. Eq. (4) also indicates that hyd-
riding is a self-limiting reaction, that is, as the reaction progresses
its rate slows down.
Equilibrium pressure is the minimum pressure at a given tem-
perature at which hydriding can occur, and is given by the van’t
Hoff equation
 
DH r DS
Peq ¼ P o exp  ; ð5Þ
RT R

where Po and DS are the ambient pressure and entropy of reaction,

respectively. The equilibrium pressure increases rapidly with
increasing metal hydride temperature. It is therefore important to
maintain high pressures along with low hydride temperatures to Fig. 9. SEM images of Ti1.1CrMn (a) before activation and (b) after activation.
achieve fast fill rates.
The above Eqs. (1)–(5) are solved simultaneously in Fluent to
determine the transient variation of metal hydride temperature,
reaction progress, and volumetric heat generation rate. the average metal hydride particles size from 72 lm to less than
10 lm.
4.2. Metal hydride properties The properties of Ti1.1CrMn were measured at the Purdue Uni-
versity Hydrogen Systems Laboratory (HSL) at pressures ranging
In order for Ti1.1CrMn to absorb hydrogen, it must first be acti- from atmospheric to 253 bar. The kinetic and thermal properties
vated by breaking it into very fine particles. Activation is achieved of Ti1.1CrMn are listed in Table 1 and the thermal properties of alu-
by subjecting metal hydride powder to repeated cycles of high- minum 6061-T6 are given in Table 2. Experimental property mea-
pressure cooling and low-pressure (vacuum) heating, which is de- surements of HPMHs are extremely challenging because of the
scribed in detail in the second part of this study [13]. Fig.9 shows high pressures involved (300 bar), their pyrophoric behavior
SEM images of the Ti1.1CrMn powder before and after the activa- and sensitivity to impurities. Flueckiger et al. [15] used the tran-
tion process. Notice how the activation process drastically reduces sient plane source (TPS) technique to measure thermal properties
of Ti1.1CrMn. This technique employs a resistive element as both
a heat source and a temperature sensor, which is sandwiched be-
tween two identical hydride powder samples. The temperature re-
sponse and change in electrical resistance of the element in
response to a heat pulse are correlated to the thermal properties
of hydride powder [15–16]. Flueckiger et al. measured the thermal
conductivity and calculated the specific heat of activated Ti1.1CrMn
(with 70% porosity) at temperatures ranging from 14 to 18 °C and
pressure up to 275 bar from thermal diffusivity measurements
using the TPS technique. Before the onset of hydriding (170 bar
at room temperature), kMH increased from 0.3 to 0.7 W/m K with
increasing pressure. While during the hydriding phase (above
170 bar), kMH was constant at 0.7 W/m K. In general, kMH depends
on factors as particle size, packing density (porosity), and hydride
temperature, which should increase kMH under the operating con-
ditions of the present study compared to those of Flueckiger et al.
One reason for this increase is the higher packing density (60%
porosity) used in this study compared to the measured samples.
Another reason is the much higher metal hydride temperature dur-
ing the hydriding reaction for the present study (about 50 °C) com-
pared to 14–18 °C for the measured samples. It is also noteworthy
that Suda et al. [17] measured an increase in kMH for TiMn1.5 from
0.5 to 1.2 W/m K as hydrogen pressure was increased from 0.5 to
40 bar. Therefore, a constant value of kMH = 1 W/m K is used in
the present study, which lies between the two measurement
ranges.

Table 2
Properties of aluminum 6061-T6.

Density qAl (kg/m3) Specific heat cp,Al (J/kg K) Conductivity kAl (W/m K)
2719 871 202
Fig. 8. Variation of equilibrium pressure with metal hydride temperature.
 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423 421

Flueckiger et al. calculated cp,MH for Ti1.1CrMn samples in the 5. Computational results
range of 400 and 600 J/kg K up to 80% hydrided state. Above 80%,
they calculated an abrupt jump in cp,MH to 1000 J/kg K. They attrib- The heat exchanger is designed with the objective of achieving a
uted this jump to phonon transport at the hydrogen occupied fill time of less than 5 min. Fill time refers to the time necessary for
interstitials of the metal hydride lattice. Since cp,MH was calculated the entire metal hydride to complete 90% of the hydriding reaction
between 400 and 600 J/kg K for major part of the hydriding reac- on a volume-averaged basis. The computation model provides a
tion, in present model a constant value of cp,MH = 500 J/kg K is used. tool to compare various heat exchanger designs and predict the fill
Contact resistance is very difficult to measure or ascertain and time that can then be verified by experiment. The performance of
its magnitude depends on particle size, packing density, physical the final heat exchanger design in Fig. 3(e) is analyzed herein. Var-
properties of aluminum interface and pressure between the iation of temperature, T(x, y, t), reaction progress, F(x, y, t), and heat
power and the aluminum surface. Recently, contact resistance generated, q_ 000 (x, y, t) from the start of the pressurization until the
measurements were performed at the Purdue Hydrogen Systems end of hydriding reaction are discussed in this section. Fig. 10
Laboratory using the TPS technique. Contact resistances between shows color contour plots of these three parameters over time dur-
a Kapton-coated resistive element and a metal hydride powder ing the hydriding phase superimposed on a cross-section of the
bed ranged from Rtc = 400 to 1800 mm2 K/W with 30% uncertainty. heat exchanger.
However, these measurements cannot replicate the continuous Notice that for the first 20 s, no reaction takes place. Initially,
expansion and contraction of hydride powder during repeated cy- the metal hydride is at 20 °C and the corresponding equilibrium
cling in actual tests, which is expected to increase contact resis- pressure is 160 bar. Hence, unless the pressure increases above
tance with the aluminum surfaces. Therefore, a constant value of 160 bar or the hydride temperature decreases such that the equi-
Rtc = 2000 mm2 K/W is used in the present study. librium pressure is below the pressure vessel pressure, the hydrid-
Some of the kinetic properties of Ti1.1CrMn, like activation ing reaction will not start. Until then the only source of heating is
energy, Ea, were obtained from the literature while others, like pressurization, which results in a small hydride temperature raise
enthalpy of reaction, DHr, and entropy of reaction, DS, were mea- at 20 s. At 40 s, when the pressure is 200 bar, heat is generated
sured at the Purdue University Hydrogen Systems Laboratory. from the metal hydride due to the hydriding reaction. Up to
From the pressure composition isotherm (PCT), the maximum 2 MW/m3 of heat is generated in small cells and close to the alumi-
hydrogen storage capacity of Ti1.1CrMn powder to be used for num fins while the heat generation rate is closer to 1 MW/m3 in
experimental work was measured at 1.5 wt% corresponding to the interior of larger cells. Associated with this heat generation is
280 bar. an increase in the temperature of the metal hydride to 35 °C. At
In the absence of detailed kinetic property measurements for 60 s, the pressurization ramp is complete and peak heat generation
HPMHs, the activation energy of Ti1.1CrMn is estimated from mea- rate reaches 3.5 MW/m3. Notice how the metal hydride is now
surements by Suda et al. [18] for LaNi5 of Ea = 20.7 kJ/mol-H2. reacting faster around the coolant tubes and near the fins, where
For LaNi5, Suda et al. also measured an activation rate of heat transfer rates are highest and the temperatures lowest. The
Ca = 54.7 s1. However, experiments performed at the Hydrogen temperatures in the middle of the cells are at 48 °C while, near
Systems Laboratory suggest Ti1.1CrMn possesses better kinetics the fins, they are about 40 °C. Also at 60 s, the metal hydride
than LaNi5, with a Ca value closer to 150 s1. Nonetheless, the around the coolant tubes and near the fins has completed 30% of
authors of the present study have conducted a sensitivity analysis the reaction while, away from the fins, it has just begun hydriding.
of different kinetic and thermal properties of Ti1.1CrMn and dem- At 60 s, heat generation is highest during the hydriding process
onstrated that Ca has an unusually small influence on reaction rate since after 60 s the pressure remains constant and no additional
[4]. By increasing Ca from 150 to 10,000 s1, they calculated a heating is generated due to pressurization. Furthermore, with no
change of less than 1 min in reaction time. further increase in pressure after 60 s, the only way to increase
Using Eq. (5) over the entire range of the reaction, and the var- the reaction rate and hence the heat generated rate is by lowering
iation of the metal hydride temperature with pressure obtained the hydride temperature. Therefore, once the pressurization ramp
from previously performed experiments, the enthalpy of hydriding is completed, a drop in the reaction rate ensues. At 100 s, the metal
reaction of DHr = 14.4 kJ/mol-H2 and entropy change of hydride temperature is fairly uniform in all cells at 55 °C. The sec-
DS = 91.3 kJ/mol-H2 were measured. tions of fins close to the coolant tubes are at 38 °C while the sec-
The coolant used in the model is Dex-CoolÒ, which is commer- tions farthest from the coolant tube along the outer periphery are
cially available automotive antifreeze whose properties are listed 45 °C. Around the coolant tubes, heat generation rate is
in Table 3. In the model, coolant is assumed to be flowing at a con- 1.5 MW/m3 and in the longer cells less than 0.5 MW/m3; the me-
stant temperature of 0 °C. The chiller to be used in the experiments tal hydride located up to 10 mm from the coolant tubes has com-
has a maximum coolant flow rate of 15 lpm, from which the con- pleted 50% of the reaction.
vective heat transfer coefficient in the U-tube was calculated as At 300 bar, the equilibrium temperature of Ti1.1CrMn is 55 °C.
2500 W/m2 K, the value used in the Fluent model. It is further as- Hence at 55 °C the hydride stops reacting. As the metal hydride
sumed in the model that, initially the hydride is in a completely powder cools below 55 °C, it starts hydriding again, thereby gener-
dehydrided state and at 20 °C. The pressurization profile assumed ating heat and increasing in temperature. For further hydriding to
in the model is a linear ramp from atmospheric pressure to take place, the metal hydride needs to be cooled again to maintain
300 bar in 60 s, and a constant pressure of 300 bar thereafter until the temperature below equilibrium. This process continues until
the hydriding process is completed. the hydride completes the reaction. As time progresses from 100

Table 3
Properties of Dex-CoolÒ.

Temperature T (°C) Density qf (kg/m3) Specific heat cp,f (J/kg K) Conductivity kf (W/m K) Viscosity lf (Pa s)
0 1068 3460 0.415 7.26  103
20 1059 3570 0.428 3.39  103
422 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423

Fig. 10. Transient contours of temperature, fraction of reaction completion and volumetric heat generation rate during hydriding.

to 300 s, the coolant continues removing heat from the metal hy- deep within the cells causing slower reaction rates in these loca-
dride while the hydriding reaction progresses further from the tions. This demonstrates the importance of the metal hydride
periphery of the cells to the interior. At 300 s, the heat generated thickness criterion in the design of the heat exchanger. The second
ranges from 0.5 to 1 MW/m3 due to the reduced reaction rate. This reason for reduced reaction rate is the self-limiting nature of the
is primarily due to two reasons. The first is the still elevated hy- reaction that causes a decrease in the rate of reaction as it pro-
dride temperatures in the core of the larger cells. To achieve fast gresses. At 300 s, excepting the core region of a few large cells
reaction rates, it is very important that the hydride temperature where the hydride has completed 60% of reaction, the metal hy-
be significantly lower than the equilibrium temperature. Larger dride is completely hydriding everywhere else. At the end of
cell widths result in low heat transfer rates to the metal hydride 300 s, the metal hydride has stored 90% hydrogen of its maximum
 M. Visaria et al. / International Journal of Heat and Mass Transfer 54 (2011) 413–423 423

capacity. If the reaction is allowed to continue further by maintain- The second part of this [13] will discuss the experimental meth-
ing the pressure, the remaining hydride will eventually cool down ods used to evaluate the performance of the heat exchanger, and
to a uniform temperature, thus completing the hydriding process. compare the experimental and computational results.
While the computational model provides that the heat exchan-
ger design can achieve the desired fill time of 5 min, both the de-
Acknowledgement
sign and the model will require experimental validation. The
second part of the study [13] will discuss the experimental meth-
This study was partially supported by General Motors
ods used during the validation tests, present the experimental re-
Corporation.
sults, and compare the experimental and computational results.

References
6. Conclusions
[1] L. Schlapbach, A. Züttel, Hydrogen-storage materials for mobile applications,
This paper presented a methodology for the design of a heat ex- Nature 414 (2001) 353–358.
changer for hydrogen storage using high-pressure metal hydrides [2] J. Zhang, T.S. Fisher, P.V. Ramachandran, J.P. Gore, I. Mudawar, A review of heat
transfer issues in hydrogen storage technologies, J. Heat Transfer 127 (2005)
(HPMHs). A simple, yet powerful 1-D criterion for maximum metal 1391–1399.
hydride thickness served as a guide for initiating the heat exchan- [3] G. Walker, Solid-state Hydrogen Storage: Materials and Chemistry, Woodhead
ger design. This is followed by an iterative design approach to meet Publishing, Cambridge, England, 2008.
[4] M. Visaria, I. Mudawar, T. Pourpoint, S. Kumar, Study of heat transfer and
specific design goals, guided by a 2-D computational model. Key kinetics parameters influencing the design of heat exchangers for hydrogen
conclusions from the study are as follows: storage in high-pressure metal hydrides, Int. J. Heat Mass Transfer 53 (2010)
2229–2239.
[5] DOE, Hydrogen, fuel cells and infrastructure technologies program, multi-year
1. The heat exchanger is the most crucial component of a hydro- research, development and demonstration plan: planned program activities for
gen storage system utilizing a HPMH. The primary function of 2005–2015, Energy Efficiency and Renewable Energy, US Department of
the heat exchanger is to remove the large amounts of heat Energy, Washington, DC, 2007.
[6] D. Mori, K. Hirose, Recent challenges of hydrogen storage technologies for fuel
released from the hydriding reaction as the system is being cell vehicles, Int. J. Hydrogen Energy 34 (2009) 4569–4574.
filled with hydrogen at the station. The heat exchanger design [7] Y. Kojima, Y. Kawai, S. Towata, T. Matsunaga, T. Shinozawa, M. Kimbara,
must adhere to several design criteria including removing the Development of metal hydride with high dissociation pressure, J. Alloy Compd.
419 (2006) 256–261.
heat in less than 5 min, compact and lightweight design, and
[8] S. Mellouli, F. Askri, H. Dhaou, A. Jemni, S. Ben Nasrallah, A novel design of a
cost and ease of manufacturability. heat exchanger for a metal-hydrogen reactor, Int. J. Hydrogen Energy 32 (2007)
2. Using a single coolant U-tube and standard 101.6-mm (4-in.) 3501–3507.
pressure vessel for hydrogen storage, an iterative design culmi- [9] B.D. MacDonald, A.M. Rowe, Impacts of external heat transfer enhancements
on metal hydride storage tanks, Int. J. Hydrogen Energy 31 (2006)
nated in an extruded aluminum construction with thin fins 1721–1731.
radiating from the coolant tube and forming cells inside which [10] S. Mellouli, F. Askri, H. Dhaou, A. Jemni, S. Ben Nasrallah, Numerical study of
the HPMH powder is stored. The final design occupies 29% of heat exchanger effects on charge/discharge times of metal-hydrogen storage
vessel, Int. J. Hydrogen Energy 34 (2009) 3005–3017.
the pressure vessel volume, leaving the remaining volume for [11] A.K. Phate, M.P. Maiya, S.S. Murthy, Simulation of transient heat and mass
the metal hydride powder. Further enhancement in the thermal transfer during hydrogen sorption in cylindrical metal hydride beds, Int. J.
performance of the heat exchanger is achieved with the aid of Hydrogen Energy 32 (2007) 1969–1981.
[12] F. Laurencelle, J. Goyette, Simulation of heat transfer in a metal hydride reactor
tapered collets that help eliminate contact resistance between with aluminum foam, Int. J. Hydrogen Energy 32 (2007) 2957–2964.
the coolant tube and the heat exchanger parts. [13] M. Visaria, I. Mudawar, T. Pourpoint, Enhanced heat exchanger design for
3. A 2-D model provides a detailed understanding of the spatial hydrogen storage using high-pressure metal hydride – Part 2. Experimental
results, Int. J. Heat Mass Transfer 54 (2011) 424–432.
and temporal response of the metal hydride’s temperature, [14] U. Mayer, M. Groll, W. Supper, Heat and mass transfer in metal hydride
reaction rate and heat generation rate. The hydriding reaction reaction beds: experimental and theoretical results, J. Less-Common Metals
proceeds only when its temperature is lower than the equilib- 131 (1987) 235–244.
[15] S. Flueckiger, T. Voskuilen, T. Pourpoint, T.S. Fisher, Y. Zheng, In situ
rium temperature or the pressure is above the equilibrium pres-
characterization of metal hydride thermal transport properties, Int. J.
sure. Colder portions of the metal hydride close to the fin Hydrogen Energy 35 (2010) 614–621.
surface start hydriding earlier and finish hydriding before the [16] S. Gustafsson, Transient plane source techniques for thermal conductivity and
warmer portions away from the surface. The reaction rate and thermal diffusivity measurements of solid materials, Rev. Sci. Instrum. 62
(1991) 797–804.
heat generation rate peak at the end of pressurization ramp, [17] S. Suda, Y. Komazaki, N. Kobayashi, Effective thermal conductivity of metal
after which the reaction becomes localized to colder locations. hydride beds, J. Less-Common Metals 89 (1983) 317–324.
Thereafter, the reaction rate decreases due to the self-limiting [18] S. Suda, N. Kobayashi, K. Yoshida, Reaction kinetics of metal hydrides and their
mixtures, J. Less Common Metals 73 (1980) 119–126.
nature of the reaction and high metal hydride temperatures.