International Journal of Hydrogen Energy 32 (2007) 3501–3507www.elsevier.com/locate/ijhydene

A novel design of a heat exchanger for a metal-hydrogen reactor

S. Melloulia,∗, F. Askria, H. Dhaoua, A. Jemnia, S. Ben Nasrallaha,b

aLaboratoire des Etudes des systèmes Thermiques et Energétiques (LESTE), ENIM, Route de Kairouan, 5019 Monastir, TunisiabCentre de recherche en Science et Technologies de l’Energie, Technopole de Borj Cédria, Tunisia

Received 18 December 2006; received in revised form 13 February 2007; accepted 13 February 2007Available online 30 April 2007

Abstract

A metal-hydride reactor equipped by a spiral heat exchanger is experimentally studied. The inserted exchanger provides significant insightsinto the problem of minimizing the total storage time by manipulating the operating parameters. Performance studies are carried out by varyingthe supply pressure, volume of the tank, absorption temperature, and overall heat transfer coefficient.

At any given absorption temperature, hydrogen absorption rate and storage capacity are found to increase with supply pressure. Coolingfluid temperature is found to have a significant effect on hydrogen storage time and higher values of the overall heat transfer coefficient yieldsbetter rates of absorption and desorption. Hydriding time mainly depends on the successful heat removal from the bed. A reactor equippedwith an exchanger that provides more heat transfer area significantly reduces hydriding time.� 2007 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.

Keywords: Heat exchanger; Metal-hydrogen reactor; Storage time

1. Introduction

Environmental problems related to the emission ofgreenhouse-effect gases and the depletion of fossil-fuel naturalresources have led to significant research effort on alterna-tive and cleaner fuels such as hydrogen which is especiallyattractive for electric vehicle use. Over the last decade, worldwide interest in the use of hydrogen has led to much researchinterest on its storage and usage [1–3]. In recent years, therehas been increasing interest in using metal hydrides for hy-drogen storage thanks to advantageous characteristics such aslow operating pressure and high volumetric density. One keyissue for metal hydride storage systems is limited heat trans-fer within the hydride bed. Often, the heat transfer rate is thecontrolling variable that determines the rate at which hydrogengas can be extracted from a hydride tank [4]. Recently, manymathematical and experimental works have concentrated onmetal hydride beds to be used for hydrogen storage medium.Muthukumar [5] analysed the effects of various operating

∗ Corresponding author. Tel.: +216 97 644 090.E-mail address: mellouli_sofiene@yahoo.fr (S. Mellouli).

0360-3199/$ - see front matter � 2007 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.ijhydene.2007.02.039

conditions in a metal hydride-based hydrogen storage deviceusing AB5 alloys and found that cold fluid temperature has asignificant effect on hydrogen storage capacity at lower supplypressures, and higher values of overall heat transfer coefficientsyield better rates of absorption and desorption. Jemni et al. [6]conducted an experimental and numerical study to determinethe effective thermal conductivity, the equilibrium pressure,and reaction kinetics. The optimization of hydrogen storage inmetal hydride beds was investigated by Kikkinides et al. [7].In these two references, cooling design options are investi-gated by introducing additional heat exchangers at a concentricinner tube and annular ring inside the tank. The problem ismathematically formulated as a dynamic optimization problemwhere the control variables are the cooling fluids flow ratesand/or temperatures and also the hydrogen charging profile.Design decisions include, among others, the radius of the innertube and the radial position of the concentric annular ring inthe tank. Optimization results indicate that significant improve-ments in the total storage time can be achieved under a safeand economically attractive operation. Demircan [8] experi-mentally examined the hydrogen absorption in two LaNi5–H2reactors. They found that hydriding time mainly depends onthe successful heat removal from the bed, and a bed geometry

3502 S. Mellouli et al. / International Journal of Hydrogen Energy 32 (2007) 3501–3507

that provides more heat transfer area significantly reduces hy-driding time. Pons et al. [9] pointed out that the wall heat trans-fer coefficient is an important parameter for characterizing andoptimizing the heat transfer efficiency of metal hydride bed.Different attempts to improve the effective thermal conductivityof metal hydride beds have been made: insertion of aluminiumfoam [10], integration of copper wire net structure [11],packingin a multilayer waved sheet structure, microencapsulated metalhydride compact [12], and compacting metal hydride powderwith an expanded graphite [13].

For effective design of metal hydride reactor, a mathematicalmodel is needed which describes correctly the physical andchemical process in a metal hydride bed. A number of pertinentand effective mathematical models for the process have beendeveloped [14–21].

Metal hydride reactors have been designed and manufactured[5–8,10,22,23]. The structure of almost all reactors was basedon the tubular type heat exchanger.

In this paper, heat transfer characteristics of a metal hydridereactor based on the spiral-type heat exchanger were experi-mentally investigated. Options of a novel cooling design are in-vestigated by introducing a spiral heat exchanger in the reactor.The inserted heat exchanger provides significant insight intothe problem of minimizing the total storage time by manipulat-ing the spiral coolant flow rate and temperature. The spiral heatexchanger is tested for its performance at various supply pres-sures, volume of the tank, cooling fluid temperature and overallheat transfer coefficient. The experimental results indicate thatsignificant improvements of the time of storage can be carriedout if the operation parameters of the heat exchanger are wellselected.

2. Design of the reactor equipped with a spiral heatexchanger

Due to the curvature of the tube, a centrifugal force is gener-ated as fluid flows through the curved tubes. Secondary flowsinduced by the centrifugal force have significant ability to en-hance the heat transfer rate. Helical and spiral coils are theknown types of curved tubes which have been used in a widevariety of applications like heat recovery processes, air condi-tioning and refrigeration systems, chemical reactors, food anddairy processes. Heat transfer and flow characteristics in curvedtubes have been widely studied by researchers both experimen-tally and theoretically [23–26], but they are not studied in ametal-hydride reactor. The heat exchanger consists of a steelshell and a spirally coiled tube unit. The spiral-coil unit consistsof one layer of spirally coiled tubes with a 50 mm diameter.Each tube is made by bending a 6 mm diameter straight stain-less tube into a spiral coil of 11 turns (Fig. 1). Water is used asa working fluid. The chilled water entering the outermost turnflows along the spirally coiled tube and flows out at the inner-most turn. The hydride enters the heat exchanger at the centreof the shell and surrounds radially across spiral tubes to theperiphery.

Fig. 1. Geometrical configuration of a spiral heat exchanger.

Fig. 2. Geometrical configuration of a reactor. (a) Burst sight of the reactor;(b) cross-section of the reactor

Fig. 3. Photograph of the reactor.

The reactor is manufactured with stainless 316-L; it is com-posed of two parts: (Figs. 2, 3)

• A cylindrical body: internal diameter = 80 mm, externaldiameter = 110 mm, interior height = 110 mm, thickness ofthe bottom = 15 mm.

• A lid: diameter = 180 mm, thickness = 10 mm.

S. Mellouli et al. / International Journal of Hydrogen Energy 32 (2007) 3501–3507 3503

Four tubes are connected on the lid by a tight passage. Thefirst, in the centre, allows the input–output of hydrogen andthe setting vacuum. A centred stainless steel filter (of 14 �mpore size) is used to prevent the metal grains from leaving thetank during desorption. The second allows a pressure pick-uprelative to measure the gas pressure in the reactor.

The two others carry tight passages which allow the assem-bly of the spiral heat exchanger. A joint ring in Viton ensures the

Fig. 4. The distribution of the thermocouples in the reactor.

Fig. 5. Synoptic scheme of the installation.

sealing. Three others carry tight passages which make it possi-ble to measure temperature with three thermocouples (type K,sheath stainless 0.5 mm in diameter). Their distribution in thebed is described in Fig. 4. Their positions are measured beforethe closing of the reactor.

3. Experimental set-up

In order to determine in experiments the performance of thespiral heat exchanger at various supply pressures, cooling fluidtemperatures and overall heat transfer coefficients, we carriedout the device shown in Fig. 5. The principal element of thisdevice is a simple cylindrical form (reactor) containing 1 kg ofLaNi5. The reactor is equipped with a spiral exchanger whichallows heat exchange between the coolant and the hydride bed.A bath equipped with a thermostat and a refrigerator at con-stant temperature are used to provide the hot and cold liquidsto the flows extending to 13.5 g/s. The range of temperature ofdelivered water lies between 10 and 60 ◦C. The heat transfercoefficient is varied from 750 to 1150 W/m2 K by varying thefluid flow from 7 to 13.33 g/s. Four cylindrical tanks are usedfor the hydrogen. The used hydrogen gas is of the UP type

3504 S. Mellouli et al. / International Journal of Hydrogen Energy 32 (2007) 3501–3507

(99.99%); the tanks and the reactor are manufactured out ofstainless steel.

The vacuum in the reactor and the tanks is ensured by avacuum pump. The different elements of the device are boundby stainless tubes via tight stainless valves. The experimentaldevice is instrumented by a pressure pick-up, thermocouplesand a chart of acquisition (Agilent) connected to a computer.

4. Results and discussion

The performance of a hydrogen storage device is mainly de-noted by the hydrogen storage capacity and the rates of absorp-tion and desorption. Hence, the foregoing discussion is basedon these parameters.

4.1. Activation of hydrogen storage metal

Fig. 6 shows the progressive activation of the LaNi5 alloy.Before letting hydrogen gas into the reactor, it is evacuateddown to about 1 mbar. Then, the whole system was flushedthree times using hydrogen gas to remove any other trace gases.Then hydrogen at 10 bar pressure was allowed to the reactorand the pressure was measured. During the first cycle of oper-ation, hydrogen absorption of about 0.1 wt% was observed in30 min. Then the reactor is heated up to 50 ◦C and the hydro-gen was completely desorbed. This was repeated till the maxi-mum storage capacity of 1.3 wt% was attained after 25 cyclesas seen in Fig. 6.

4.2. Effect of the overall heat transfer coefficient

Fig. 7 proves that the overall heat transfer coefficient does nothave any effect on the storage capacity of hydrogen. However,a higher value of overall heat transfer coefficient (importantwater flow) reduced the time of absorption. To demonstrate theeffect of the heat exchanger on the absorption process, we havepresented the time evolution of the hydrogen mass absorbedwhich obtained with and without cooling (Fig. 8).

Fig. 6. Activation of the metal (LaNi5).

Fig. 7. Effect of the water mass flow in the heat exchanger on the absorptionmass.

Fig. 8. Time evolution of stored mass of hydrogen.

From this figure we note that the time at which 80% of theamount of hydrogen has been stored (i.e., 9 g ) is 1000 s withthe cooling and 5000 s without the cooling. In this extremelysimplified example the filling time has already been reducedby a factor 5 approximately.

4.3. Effect of the initial amount of hydrogen

In order to evaluate the effect of the initial amount of hy-drogen used for the absorption process, the experiment wasconducted for different numbers of hydrogen tanks (Fig. 5). Inthis experiment, the initial pressure P0 is constant and equal to10 bar.

S. Mellouli et al. / International Journal of Hydrogen Energy 32 (2007) 3501–3507 3505

Fig. 9. Effect of the initial amount of hydrogen on the average bed temperature.

Fig. 10. Effect of the initial amount of hydrogen on the storage capacity.

As depicted in Fig. 9, the maximal value of the temperaturein the reaction bed is not significantly sensitive to the initialamount of hydrogen and the reactor needs more time to reachthe equilibrium state for higher amount.

From Fig. 10, it can be noticed that the total mass absorbedincreases with the initial amount of hydrogen for the sameinitial pressure. Besides, it can be seen that the increase of thisamount above a certain value has no considerable effect on thetotal mass absorbed.

4.4. Effect of supply pressure

The reactor absorbs hydrogen with various pressures bykeeping a constant temperature of the coolant (20 ◦C). Theflow rate of cold fluid is maintained constant in order to fixthe overall heat transfer coefficient. Fig. 11 shows the varia-tion of the rate of absorption and the temperature within thereactor. It is observed that for a given supply pressure, the rate

Fig. 11. Effect of the supply pressure on the bed temperature and the ab-sorption rate.

of absorption reaches a peak at the beginning and decreasesgradually towards zero at the end of the absorption process.This is due to the fact that, at the beginning of the absorptionprocess, the potential of mass transfer (difference between thesupply pressure and the equilibrium pressure) is high. Whiletime progresses, the equilibrium pressure increases followingthe increase in the temperature of the bed under the effect ofthe absorption. This reduces the potential of the kinetics andthereafter the deceleration of the absorption rate. Because ofthe low thermal conductivity of the bed of hydride, the pro-duced heat of absorption cannot be transferred starting fromthe bed with the cold fluid at the initial period from fast ab-sorption and consequently excessive heat is stored in the bedof hydride itself, having for result a rise in the temperature ofthe bed.

It is also observed that for the pressures indicated, the risein the temperature bed during the initial stage of absorption ishigher with the higher supply pressures due to faster kinetics.Fig. 12 shows that the total mass absorbed increases with thesupply pressure.

4.5. Effect of cooling fluid temperature

To study the effect of cooling fluid temperature on the ab-sorption rate, three values are considered (10, 15 and 20 ◦C)and the other operating parameters are kept the same for thethree cases. The obtained results shown in Fig. 13 indicate thatlow temperature can accelerate the reaction but it has no effecton the total mass absorbed. Also, it is important to note that

3506 S. Mellouli et al. / International Journal of Hydrogen Energy 32 (2007) 3501–3507

Fig. 12. Effect of the supply pressure on the absorption mass.

Fig. 13. Effect of the cooling temperature on the absorption mass.

decreasing the cooling temperature below certain value has noimportant effect.

4.6. Characteristics of desorption

The desorption characteristics presented in Fig. 14 show thatdesorption rate is improved at higher heating fluid temperaturesdue to higher mass transfer-driving potential. Due to poor ther-mal conductivity of the hydride bed, the necessary amount ofheat is not being transferred to the bed at the initial stage ofrapid desorption, and hence the hydride bed is taking the des-orption heat from the bed itself, resulting in a sudden fall intemperature.

From Fig. 15 we note that the necessary time to desorb 2.5 gof H2/kg of metal is of 750, 950 and 2400 s, respectively, for aheating temperature of 50, 40 and 30 ◦C. On the one hand thisresult indicates that higher heating temperature accelerates the

Fig. 14. Effect of the heating temperature on the bed temperature and thedesorption rate.

Fig. 15. Effect of the desorption temperature on the desorbed mass.

desorption process, on the other hand, we can conclude fromthis profile that for a given pressure, there is a heating temper-ature value beyond which there is no important improvementon the desorption time. Fig. 15 indicates also that heating fluidtemperature circulating in the exchanger has an effect on thetotal mass desorbed.

S. Mellouli et al. / International Journal of Hydrogen Energy 32 (2007) 3501–3507 3507

5. Conclusion

The present work is a contribution to the improvement ofheat transfer which controls sorption processes within metal-hydrogen reactor. A spiral-type heat exchanger was inserted inthe hydride bed and an experimental set-up was designed tostudy its effect on absorption/desorption times and hydrogenmass stored or discharged. The experimental results show thatthe charge/discharge times of the reactor are considerably re-duced, when heat exchanger is used. In addition, the effect ofdifferent parameters (flow mass and temperature of the coolingfluid, applied pressure, and hydrogen tank volume) has beendiscussed and obtained results show that a good choice of theseparameters is important.

References

[1] Momirlan M, Veziroglu TN. Current status of hydrogen energy.Renewable Sustainable Energy Rev 2002;6:141–79.

[2] Mohanty KK. The near-term energy challenge. AIChE J 2003;49(10):2454–60.

[3] Dogan A, Kaplan Y, Veziroglu TN. Numerical investigation of heat andmass transfer in a metal hydride bed. Appl Math Comput 2004;150:169–80.

[4] MacDonald BD, Rowe AM. Impacts of external heat transferenhancements on metal hydride storage tanks. Int J Hydrogen Energy2006;31:1721–31.

[5] Muthukumar P. Experiment on a metal hydride-based hydrogen storagedevice. Int J Hydrogen Energy 2005;30:1569–81.

[6] Jemni A, Ben Nasrallah S, Lamloumi J. Experimental and theoreticalstudy of metal hydrogen reactor. Int J Hydrogen Energy 1999;24:631–44.

[7] Kikkinides ES, Georgiadis MC, Stubos AK. On the optimization ofhydrogen storage in metal hydride beds. Int J Hydrogen Energy2006;31:737–51.

[8] Demircan A. Experimental and theoretical analysis of hydrogenabsorption in LaNi5–H2 reactors. Int J Hydrogen Energy 2005;30:1437–46.

[9] Pons M, Dantzer P, Guilleminot JJ. A measurement technique and a newmodel for the wall heat transfer coefficient of a packed bed of (reactive)powder without gas flow. Int J Heat Mass Transfer 1993;36:2635–46.

[10] Lévesque S, Ciureanu M, Roberge R, Motyka T. Hydrogen storagefor fuel cell systems with stationary applications—I. Transientmeasurement technique for packed bed evaluation. Int J Hydrogen Energy2000;25:1095.

[11] Nagel M, Komazaki Y, Suda S. Dynamic behaviour of paired metalhydride systems—I. J Less Common Met 1986;120:45–53.

[12] Kim KJ, Lloyd G, Razani Jr A, Feldman KT. Development ofLaNi5/Cu/Sn metal hydride. Powder Technol 1998;99:40.

[13] Kim KJ, Montoya B, Razani A, Lee K. Metal hydride compacts ofimproved thermal conductivity. Int J Hydrogen Energy 2001;26:609.

[14] Gopal MR, Murthy SS. Prediction of a heat mass transfer in annularcylindrical metal beds. Int J Hydrogen Energy 1992;17(10):795–805.

[15] Gopal MR, Murthy SS. Studies on heat and mass transfer in metalhydride beds. Int J Hydrogen Energy 1995;20(11):911–7.

[16] Mayer U, Groll M, Super W. Heat and mass transfer in metal hydridereaction beds: experimental and theoretical results. J Less Common Met1987;131:235–44.

[17] Ben Nasrallah S, Jemni A. Heat and mass transfer model in metal-hydrogen reactor. Int J Hydrogen Energy 1994;20:197–203.

[18] Jemni A, Ben Nasrallah S. Study of two dimensional heat and masstransfer during desorption in a metal hydrogen reactor. Int J HydrogenEnergy 1995;20:881–91.

[19] Shmalkov UYF, Kolosov VI, Solovey VV, Kennedy LA, ZeleponguSA. Mathematical simulation of heat and mass transfer processes inmetal hydride-hydrogen-gas impurities systems. Int J Hydrogen Energy1998;23(6):463–8.

[20] Mat D, Kolpan Y. Numerical study of hydrogen absorption in an Lm–Ni5hydride reactor. Int J Hydrogen Energy 2001;26:957–63.

[21] Aldas K, Mat D, Koplan Y. A three-dimensional mathematical modelfor absorption a metal hydride bed. Int J Hydrogen Energy 2002;27(10):1063–9.

[22] Nishimura K, Inazumi C, Oguro K, Uehara I, Itoh Y, Fujitani S. et al.Simulation and evaluation of a hydrogen storage system using hydrogenstorage alloy for a chemical CO2 fixation and utilization system. Int JHydrogen Energy 2000;25:1087.

[23] Fleming WH, Khan JA, Rhodes CA. Effective heat transfer in a metal-hydride-based hydrogen separation process. Int J Hydrogen Energy2001;26:711.

[24] Garimella S, Richards DE, Christensen RN. Experimental investigationof heat transfer in coiled annular ducts. J Heat Transfer 1988;110:329–36.

[25] Prabhanjan DG, Raghavan GSV, Rennie TJ. Comparison of heat transferrates between a straight tube heat exchanger and a helically coiled heatexchanger. Int Comm Heat Mass Transfer 2002;29:185–91.

[26] Bolinder CJ, Sunden B. Numerical prediction of laminar flow and forcedconvective heat transfer in a helical square duct with finite pitch. Int JHeat Mass Transfer 1996;39:3101–15.