INTERNATIONAL JOURNAL OF ENERGY RESEARCH, VOL.

20,965-978 (1996)

HIGH PRESSURE AMMONIA DISSOCIATION

EXPERIMENTS FOR SOLAR ENERGY TRANSPORT AND
STORAGE
KEITH M. LOVEGROVE
Energy Research Centre, Depaflment of Engineering, Australian National University, Canberra, ACT 0200, Australia

SUMMARY
Work relating to the application of the ammonia dissociation reaction to the thermochemical transport and storage
of solar energy is reported. A two-dimensional pseudo-homogeneous packed-bed catalytic reactor model has been
adapted to predict the behaviour of ammonia dissociation and synthesis reactors. A series of steady-state experi-
ments, with a high-pressure ammonia dissociator operated in an ‘open-loop’ experimental configuration, have been
used to verify the model. These experiments involved operating pressures up to 16 MPa and temperatures up to
720°C. The results indicated an activation energy of 192 kJ mol-’ and a pre-exponential factor of 3.0 X lO’mol s - ’
’
cm-3 atm - for the nickel-on-alumina dissociation catalyst used. The experiments also indicated that the experimen-
tal configuration is suitable for development into a ‘closed-loop’energy storage experiment operating at a power level
of approximately 1kWchem.

KEY WORDS: thermochemical; energy storage; ammonia dissociation; reactor model; solar energy

1. INTRODUCTION
A fluid-phase thermochemical energy storage system consists of a closed system of reactants passing
alternately to endothermic and exothermic reactors. Each reactor is operated in conjunction with a
counterflow heat exchanger that allows the reactants in the intervening storage/transport system to be
maintained at ambient temperature. In this way, the storage and transport of energy from a thermal
source can be achieved potentially without thermal losses.
A number of gas phase reactions have received attention for the storage and transport of solar energy.
Systems based upon the reforming of methane, either with steam or carbon dioxide, have been the
subject of most recent work. This has included the development of closed-loop experiments (Levy et al.,
1993); and receiver designs for solar thermal collectors (Buck et al., 1991; Diver et al., 1992; Levitan et al.,
1989).
The ammonia dissociation reaction:
1 3
NH, + AH2 2 N 2 + 7 H 2 (1)
has also been considered, primarily by a group at the Australian National University. Early work included
theoretical (Carden and Williams, 1978, 1979a) and experimental (Williams and Carden, 1979b) investiga-
tions. A detailed description of a solar thermochemical system, in which the exothermic reaction products
themselves become the working fluid in a Brayton-cycle-like heat engine, has been given by Carden
(1987). A group at the University of Colarado has also presented the results of a receiver design study
(Lenz et al., 1978) and catalyst tests (Nandy and Lenz, 1984). More recently, the thermodynamic limits on
the efficiency of an ammonia based system (Lovegrove 1993a) and an approach to its optimisation
(Lovegrove 1993b) have been explored.
A major advantage of the ammonia reaction is the complete absence of undesirable side reactions.
This is in contrast to the methane-based reactions, for which great care is required in the choice of

CCC 0363-907X/96/ 110965-14 Received 21 February 1995

0 1996 by John Wiley & Sons, Ltd. Revised 3 May 1995
966 K. M. LOVEGROVE

catalyst and operating conditions to avoid reactions that result in carbon deposition within the system.
The ammonia reaction’s relatively low characteristic temperature has a detrimental effect on exothermic
reaction rates and on the amount of work that can be produced from the energy output. However the use
of high system pressures, although placing extra restrictions on component design, shifts the equilibrium
towards higher temperatures and helps to make the system feasible. The ammonia-based system also has
the unique feature that when a mixture of reactants (i.e. various amounts of H,, N, and NH,) is cooled
to close to 300 K, the majority of the ammonia component condenses 2nd spontaneously separates from
the mixture. This means that the reactant feedstock for both endothermic and exothermic reactors can be
stored in the same vessel. In designing the system, the composition of the reaction products of each
reactor can be chosen independently.
This paper presents results obtained with a high pressure ammonia dissociation reactor that is
intended to operate as part of a closed-loop thermochemical energy storage experiment.

2. MODELLING OF REACTOR PERFORMANCE

The ammonia reaction, in common with all gas-phase reactions considered for thermochemical energy
storage, requires a catalyst. An essential tool for the design of reactors is a numerical model capable of
predicting the conditions at all points within such catalytic reactors. The usual approach for a packed-bed
reactor is to use a ‘pesudo-homogeneous’ model. This means treating the catalyst bed as a continuum
and representing the conditions at each point by local averages of the various quantities. This is done by
defining an ‘effective conductivity’ (Keff) and an ‘effective diffusivity’ (Deff), both of which will vary with
local conditions. The general three-dimensional energy conservation equation with heat conduction as
the only heat transfer process is:
pu)
p v - v e - v.(KCffv T ) - R A H - -v*v P= - 7. (2)

Similarly, the three-dimensional mass-conservation equation in terms of reaction extent is:

The reaction extent (or conversion) is the mass fraction of the primary reactant 6.e. ammonia) consumed.
Richardson et al. (19881, have developed a two-dimensional pseudo-homogeneous model for packed-bed,
isothermal-walled, steam methane reforming reactors. The model covers steady-state behaviour in
radially symmetric constant diameter annular cylindrical beds. The Fortran implementation of this model
has been adapted by this author to cover generalised reaction systems and arbitrary flux or temperature
profiles at the wall. In particular, it has been applied to the modelling of ammonia dissociation and
synthesis reactors.
The model uses the simplifying assumptions that average longitudinal flow dominates longitudinal
diffusion, that viscous forces and pressure drops are insignificant, that the average velocity and density
are functions of longitudinal position only and that the gradients of specific heat, effective conductivity
and effective diffusivity are much smaller than the temperature and reaction-extent gradients. The energy
conservation equation becomes:

The boundary conditions are:

T = Ti, at z =0
 HIGH PRESSURE NH, DISSOCIATION 967

The mass conservation equation becomes:

(6)

The boundary conditions for mass conservation are:

Richardson et al. surveyed the literature to find appropriate correlation expressions for calculation of the
effective conductivity, the effective diffusivity and the wall heat transfer coefficient. Although the
pressure-dependent term has been dropped from the energy equation, the model does evaluate pressure
drops using the Ergun equation - see for example Hill (1977). For the ammonia reaction the Temkin
Pyzhev intrinsic rate expression has been used (Vancini 1971):

Where a is an empirical parameter with a value of 0.5.

In addition to thermodynamic data for all the reactants, input data comprises a number of parameters
relating to the catalyst’s particle properties and intrinsic kinetics, plus inlet mass flow and temperature
and some specification of the conditions that apply at the reactor walls. The program allows the
specification of multiple reactor sections, this is accomplished by calling the reactor simulation block of
the program once for each section. The average exit reaction extent and pressure from each section are
used as the inlet values for the subsequent section in conjunction with the specified inlet temperature.
For the work reported here, the intrinsic rate parameters, (namely the activation energy and
pre-exponential factor) were initially taken from the work of Nandy and Lenz (19841, but were treated as
the main variables for fitting the model to the experimental results.

3. EXPERIMENTAL APPARATUS
The open-loop energy transfer experimental arrangement used is shown in Figure 1. Ammonia is
periodically drawn from a supply vessel, pressurized and passed to the accumulator. It is subsequently
displaced by constant pressure nitrogen gas and passes via a metering value to the electrically heated
dissociation reactor. This dissociator and its associated counterflow heat exchanger are illustrated in
Figure 2. It is based on two parallel sections of 21.3 mm 0.d. 15.8 mm i.d. ‘Inconel’ tubing. For the work
reported here these two sections were filled with ICI 47-1 nickel-on-alumina ammonia-dissociation
catalyst. The catalyst consists of 4.5 mm long 4.5 mm diameter cylindrical pellets of gamma alumina with
an impregnation of nickel of approximately 10% by weight. The pellets were crushed to approximately
25% of their original size resulting in the bed parameter values shown in Table 1.
Heat at a rate of up to 2.3 kW is provided to each branch by stainless steel sheathed heating elements
wound around the outside. Reactor temperature is controlled by two (PID) proportional-plus-integral-
plus-differential feedback temperature controllers that determine the power levels for the two heat
elements. Small (5 mm o.d.1 tubes run up the centre of each branch to measure internal temperature
profiles. Stainless steel sheathed thermocouple probes can be driven along their lengths to measure
internal temperature profiles. Incoming ammonia passes through one passage of the heat exchanger, to
the top of the first reactor column. The reactants leaving the bottom of this column are transferred to the
bottom of the second and then travel back to the top where they pass into the exit passage of the heat
exchanger. The open-loop arrangement incorporates a synthesis reactor housed inside a cold-walled
pressure vessel filled with nitrogen. For the dissociator experiments presented here, the dissociation
968 K. M. LOVEGROVE

r e ~ l a t . 4N,

t
h
controlled pressure I
ratridion
Pump * I L D
I
disposal
I
I

Figure 1. The open-loop energy-transfer experimental arrangement

products flowed via the synthesizer bypass path (shown in Figure 1) directly to the balancing separator. At
the system exit, a combination of the reaction products plus nitrogen from a pressure regulator, pass
through a restriction value to a disposal system. The nitrogen pressure regulator compensates for
variations in volumetric flow rate from the balancing separator, keeping the exit pressure constant. The
ammonia accumulator is kept at a sufficiently high pressure so that the pressure drop between this and
the exit pressure occurs largely across the metering valve. Inlet mass flow is determined by the setting of
this metering value.
The balancing separator is a 360 ml vessel mounted on an electronic beam balance. Measurement of
the rate of liquid accumulation during reaction allows the extent of reaction to be calculated from the
inlet ammonia mass flow. This calculation takes account of the composition of the gas phase via
knowledge of the ammonia vapour pressures. After including the contribution of the uncertainty
associated with linear fits to the balancing separator mass rate of change, the net uncertainty in the value
for the mass flow of ammonia in exit reactant streams was between 3 and 4%. The uncertainty in outlet
ammonia mass flow rate generally dominates the uncertainty in inlet mass flow, which is measured with a
capacitative level sensor on the ammonia accumulator. Reaction extent values obtained have uncertain-
* *
ties in the range 15% at reaction extents around 0.2 to 15% at reaction extents around 0.75.
In addition to the variables already mentioned, thermocouples are also used to measure temperatures
on the outside of the reaction chambers, heat exchanger, and interconnections. There are also internal
thermocouples measuring reactant inlet and exit temperature from the heat exchanger. Pressures are
measured by transducers immediately ahead of the metering valve, at the heat exchanger inlet and at the
system exit, together with a differential pressure transducer that measures the pressure drop across the
reactor plus heat exchanger. These variables are recorded every 20 s by a computer-based data
acquisition and control system.

4. COMPARISON OF MODEL WITH EXPERIMENT

Figures 3 and 4 show representative modelled and experimentally measured temperature profiles and
modelled reaction extent profiles together with the measured exit values.
 HIGH PRESSURE NH, DISSOCIATION 969
*-------------------------- 1

I
I
I
II i
I II
II 1I
I
I
I

Ractorutarul
I
rkrwampla I
(K-tTpc);
column 1 at 7.4, 30,
52, k 77.5 an,
cd- 2 at 5.5, 33,

Heat bsch.nser ther-

moewpla;
J-trp.at 10, 20, 30.
50.70 k 90 an,

1
K-type .t 110,130 k
110 cm.

Figure 2. The ammonia dissociator and counterflow heat exchanger (columns are vertically mounted in the laboratory)

The distance scale corresponds to position along the reactor measured from the inlet port of the first
chamber. The connecting tube between the two chambers has been ignored with the position of the inlet
of the second chamber set to equal the position of the exit of the first. The discontinuity between the
profiles from the two reaction chambers and the drop in temperature near their ends indicates that there
are considerable thermal losses from these regions. A short unheated section at the beginning of the first
chamber is reflected in the unchanging temperatures there.
970 K.M. LOVEGROVE

Table 1. Dissociator catalyst particle parameters

Parameter Value
~~~~ ~ ~ ~~

Average particle mass 00769 f 00002 g

Average particle volume 004457 f OoO04 cm'
Effective particle diameter 0223 f 0.002cm
Void fraction 0462 f 0005

0 lo00 I500 2000

Distance(mm)
Figure 3. External(---), average (-) and internal ( G - )modelled temperature profiles and internal (+) and external(.) experimen-
tal values for 0507 g s-', 11.4 MPa, and 650°C

Flux profiles for the reactor, which are required as input for the numerical model, have been
determined from measurements of the electrical heater power corrected for thermal losses. The thermal
losses have been quantified by determining the temperature dependence of the heat transfer coefficient
to the environment at each point along the reactor. This determination is based on measured tempera-
ture profiles obtained under conditions of no flow and with pure nitrogen flow. The results of a complete
analysis of the energy balance of the conditions from Figures 3 and 4 are presented in Figure 5.
With the input of the appropriate flux profile and after adjustment of the intrinsic rate parameters, the
model proved very effective in reproducing the longitudinal internal and external temperature profiles
observed experimentally. Experiments under no reaction conditions with pure nitrogen flow were also
carried out. These produced similar temperature profiles and confirmed the validity of the wall heat
transfer coefficient and effective conductivity correlations used. The irregularities in the measured
profiles, which are attributed to packing inhomogeneities in the relatively narrow catalyst bed, are of
course not predicted by the model. The model quite successfully predicts the magnitude of the
external-to-internal temperature drop along the length of both chambers. This provides support for its
ability to predict radial temperature distributions. In all cases, both measured and predicted tempera-
tures profiles for the first reactor chamber show a noticeable shoulder at around 550 "C.The shape of the
predicted reaction extent profiles confirms that this corresponds to the point at which the reaction rate
increases rapidly and thus reduces the rate of temperature increase produced by the heat flux.
 HIGH PRESSURE NH, DISSOCIATION 971

Figure 4. Equilibrium (---)and average (-) modelled reaction-extent profiles corresponding to Figure 3

Conduction
from end
5.7f3W
01 Conduc ti on
from end
21f10W
-1L

Loor through Loss through

insulation of ineulation of
end re ion end reglon
11.3*3# 64f15W

Loee through Heater Lose through

ineulation of input ineulation of
middle region 4
QOlf9W 01 middle region
67.5i3W 15 lf4W
k

e
l n
0, c7 Heater
P
k
Q)
e
a
bd
c
u c
u Total heater power 1414fllW
Total l o s s 4 13f20W
]Net i n m t 1001f23Wl

Loee through
ineulation of Loes through
end region ineulation of
47flOW end region
a
Conduction
from end
19*1 ow
01 I I lo
3
Conduction
from end
19*1 ow
I "
eaclant flow-
Figure 5. Energy balance corresponding to the conditions of Figure 3
972 K. M. LOVEGROVE

The model has been used to produce predicted relationships between the operating temperature,
pressure and massflow and the conditions at the reactor exit. The dependence of reactor temperatures
and reaction extents on operating conditions reflects the behaviour of the dissociator in conjunction with
the counterflow heat exchanger. The heat exchanger is essentially a positive feedback mechanism for the
reactor. Its behaviour can be characterized by two observations. If other factors are equal, the inlet
temperature seen by the first reaction chamber increases with the exit temperature of the second
reaction chamber. If the exit temperature of the second chamber remains unchanged, the inlet
temperature for the first chamber increases with flow rate, owing to the reduced significance of thermal
losses, and with exit reaction extent, due to the reduction in specific heat mismatch.
Whilst it would be possible to model the heat exchanger behaviour in conjunction with the reactor, it
is the behaviour of the reactor which is of primary interest. The approach taken has been to relate the
inlet temperatures for the two chambers to the varying operating parameter using a one-dimensional
curve fit to the experimental values in each case. Interpolated values from these fits have been used as
input data for the reactor model. Similarly, the experimentally observed enthalpy changes from each
chamber have been the subject of curve fitting, with interpolated enthalpy change values being used to
calculate scale factors applied to the net input flux profiles from three chosen baseline conditions.
Different baseline conditions were used for the investigation of the effect of temperature, pressure and
mass flow variation.

4.1 Variation of temperature

The results of the experimental investigation of the effect of operating temperature on exit reaction
extent are presented in Figure 6 together with the predictions of the model. The average flow rate for
these runs was 0.447 g s-' and the pressure was 11.4 M Pa, actual flow rates differed by up to 0-06 g s-'
and pressures by 0-7 M Pa. As expected, there is a dramatic increase in exit reaction extent as the
operating temperature is increased. The temperature versus reaction-extent relationship predicted by the
model is confirmed by the experimental points, close to within the bounds of their error bars. The curve
generated by the model indicates an asymptotic approach to equilibrium conditions that begins at
approximately 750°C. The curve generated by the model was obtained using intrinsic rate parameter
values selected to give the best overall prediction of reaction extents for all the conditions examined.
The error bars associated with the experimental values represent the uncertainty in the calculation of
exit reaction extent based on the determination of inlet and exit ammonia mass flow. Point-to-point
fluctuations are the result not only of random errors, which have been quantified by the error bars, but
also of variations in the operating conditions or of deviations from steady-state behaviour.
Although diffusion and heat transfer processes are affected by the operating temperature, the most
significant effect on reactor behaviour is via the temperature dependence of the intrinsic reaction rate.
Figure 7 illustrates the effect of varying the activation energy on the predictions of the model. These
curves were obtained by making compensating changes to the pre-exponential factor so that the intrinsic
rate at a temperature of 640°C would remain constant. The fact that the family of curves pivots around a
maximum external temperature value of approximately 660°C indicates that this is the value at which the
reaction-rate-weighted average temperature within the reactor is 640 "C.
Nandy and Lenz (1984) determined activation energies for high pressure ammonia dissociation for two
nickel-based catalysts operating at pressures up to 10 MPa. For 6.9 wt % nickel on alpha-alumina they
obtained values between 258 and 283 kJ mol-I, for 8.1 wt% nickel on gamma-alumina, which corre-
sponds closely to the composition of the ICI 47-1 material used for this project, they obtained a value of
192 kJ mol-I. An activation energy of 192 kJ mol-' produces a maximum temperature exit reaction
extent relationship which is in good agreement with the experimental points, although a slightly lower
value of 170 W mol-' produces the best fit; however, this figure has an estimated uncertainty of f30%.
The experiments performed by Nandy and Lenz were more appropriate for the direct determination of
intrinsic rate parameters than the ones reported here, hence the most appropriate conclusion is that the
experimental results reported here support their value of 192 kJ mol-'.
 HIGH PRESSURE NH, DISSOCIATION 973

1.0 ,

550 600 650 760 750

'0
Maximum temperature (Deg C)
Figure 7. The effect of variation of operating temperature on exit reaction extent modelled using different activation energies

On the other hand, it can be seen from Figure 7 that an activation energy of 349 kJ mol-' as
suggested by Williams and Carden (1979b), is not supported by these results. Although they used the
identical ICI 47-1 catalyst material, their determination of activation energy must be questioned since it
involved the use of Arrhenius plots of measurements of extrinsic rates made using large catalyst particles,
rather than the finely crushed catalyst necessary for direct measurement of intrinsic rates.
974 K.M.LOVEGROVE

Figure 8 illustrates the sensitivity of the predictions of the model to variation of the pre-exponential
factor using the 192 H mol-' activation energy. The value determined on the basis of all the
experimental results is 3.0 X mols-' cm-3 atm-'.

4.2 Variation of mass flow

The investigation of the effect of mass flow variation was carried out at an average pressure of 10.3 MPa,
(with variations up to 0.6 MPa) and with temperature setpoints of 650°C. Figure 9 shows the experimen-
tal points and the modelled variation of reaction extent with mass flow.
As expected, the exit reaction extent decreases as the residence time within the catalyst bed is
decreased by increasing the mass flow. The nonlinearity of the relationship results from the counter
effects of increases in the intrinsic reaction rate with larger departure from chemical equilibrium plus a
combination of the effect of increased mass flow on the performance of the heat exchanger (given a
higher inlet temperature), and in enhanced heat transfer and diffusion processes within the bed.
Using the rate parameters discussed above, the model has predicted a behaviour of the reaction-extent
versus mass-flow relationship which is in good agreement with that indicated by the experimental points.
The deviation of experimental values from the model curve is, in a number of cases, outside the range of
the uncertainty associated with the reaction exLentcalculation. No correlation between the point-to-point
variation and the pressure or chamber inlet temperatures has been identified: however, the measure-
ments were taken from experimental runs on different dates and some correlation with date has been
noted indicating the possible presence of other small systematic effects.

4.3 Variation of pressure

The results together with model predictions for the investigation of the effect of system pressure are
presented in Figure 10. The average mass flow used was 0.229 g s-', with variations up to 0.07 g s-'.
Setpoint temperature was maintained at 650" C. Both the experimental results and the predictions of the

500 550 600 650 700 7

Maximum temperature (Deg C)
Figure 8. The effect of variation of operation temperature on exit reaction extent modelled using different pre-expotential factors
(activation energy 1.92 KJ mol-').
 HIGH PRESSURE NH, DISSOCIATION 975

Figure 9. The effect of variation of mass flow on exit reaction extent, for 1 0 3 MPa and 650°C maximum temperature.

model show the exit reaction extent increasing at lower pressures: however, the model predicts exit
reaction extents just above the upper limit of the range of uncertainty associated with most of the
experimental points. Reducing pressure for a given massflow reduces the residence time within the bed,
this would work to decrease the exit reaction extent. However a lower pressure also shifts the chemical
equilibrium to increase the intrinsic reaction rate and this effect dominates that of the decreased
residence time.
Although the model predictions for exit reaction extent are higher than most of the experimental
values, an adjustment to the intrinsic rate pre-exponential factor would produce good agreement. The
model curve has been produced using the previously quoted value, which was chosen to give the best
overall agreement with all the experimental results. It can be noted that the measurement at 9.9MPa is
consistent with the majority of experimental points and is less than the modelled variation of flow rate
curve (it appears as the 0.24 g s-l measurement in Figure 9).
The point at 7 MPa is higher than the general trend, although only just beyond the range of the error
bars: it corresponds to the lowest inlet flowrate of the series, the variation of -0.05 g s-' from the
average is almost sufficient to account for the discrepancy.
The effect of pressure on the exit reaction extent is relatively minor: however, in terms of the primary
quantity measured under experimental conditions, ie. the liquid ammonia mass flow in the exit reactant
stream, the effect is dramatic. The lowest pressure measurement at 5.27 MPa resulted in no liquid
accumulation in the balancing separator. The extra contour plotted in Figure 10 represents the mass
fraction H2/N2 in the gas phase of a two-phase reactant stream entering the separator for the 5.27 MPa
measurement). This contour intercepts the experimental points at the pressure at which the ammonia
vapour pressure is a sufficiently large enough fraction to prevent the formation of a liquid phase.
When the exit liquid mass flow is small relative to the inlet mass flow, the calculation of reaction
extents from balancing separator measurements relies heavily on the validity of the gas phase composi-
tion data used, but becomes progressively less affected as the exit liquid mass flow increases. Thus the
calculation of exit reaction extent at 5.27 MPa is entirely based on the gas composition data used,
whereas the calculation for 16 MPa, is almost independent of the data. If the composition data were
erroneous the effect would be to vary the curvature of the experimental reaction extent pressure
976 K. M.LOVEGROVE

1.0 ,

."
Y

0
2 0.4
& -

o0 5
Pressure
10 (MPa) 15
Figure 10. The effect of variation of pressure on exit reaction extent, for 0.23 g s-' and 650°C maximum temperature (- model
predictions, --- mass fraction of H,/N, in the gas component at 290°C

relationship whilst leaving the high pressure values unaffected. The consistency between the shape of the
modelled and experimental reaction-extent versus pressure relationships suggests not only that the model
has successfully predicted the effect of pressure, but that the method and data used for reaction extent
calculations are sound.
The inlet temperatures for the two reactor chambers demonstrated no resolvable variation with
pressure. For modelling purposes, the average values of (211 f 5)" C for the first chamber, and
(546 f 10)"C for the second chamber were used. The model predictions indicate a variation in exit
temperatures of only f3" C, consistent with the apparently constant value indicated by the experimental
measurements, thus indicting that the variation in reaction extent has indeed resulted from pressure
related variations in the average reaction rate.

5. CONCLUSIONS
The work done by the author on the Fortran implemented reactor model of Richardson et af. involved
the generalization of the geometry, the reaction system and the wall boundary conditions and improve-
ments in the algorithm. The 'model' itself, as embodied in the physical assumptions made and the choice
of empirical correlations for the calculation of effective conductivity, effective diffusivity and other
properties, has not been changed. This model performed well in comparison with their experimental
results for steam reforming of methane. With the input of the appropriate thermodynamic properties for
the ammonia system reactants and parameters for the catalyst particles in the dissociator, a simple
adjustment of the pre-exponential factor for the intrinsic rate was all that was required to reproduce the
experimentally observed effects of the variation of temperature, pressure and flow rate. This degree of
success suggests that the model can now be used with reasonable confidence for simulation of reactor
configurations and operating conditions outside the range covered by these experiments.
The experimental investigation of the behaviour of the ammonia dissociator has shown the experimen-
tal arrangement, the instrumentation and the analysis techniques used all to be successful. For the
purposes of a closed-loop energy transfer demonstration, it is apparent from the experimental results
 HIGH PRESSURE NH, DISSOCIATION 977

obtained that operation at a nominal power level of 1 kW is possible using the present catalyst. This
could be achieved with an inlet flowrate of 0.41 g s-’ and with setpoint temperatures at 725” C to give an
exit reaction extent of 0.75. This performance would not be significantly affected by variation of the
pressure between 10 and 30 MPa.

NOMENCLATURE

= initial concentration (mol m-3)

= effective diffisivity (m2s-’)
= specific enthalpy (J kg-’)
= void fraction
= activation energy (kJ mol-’1
= enthalpy change (J mol-’1
= wall heat transfer coefficient (W m-2 K-’ 1
= effective conductivity (W m-’ K-’)
= equilibrium constant
= Boltzmann’s constant (J K-’)
= pre-exponential factor (mol s-’ cm-3 atm-’1
= average molecular weight (g mol-’1
= pressure (Pa)
= radial length variable (m)
= reaction rate (mol s-’ m-’1
= temperature (K)
= specific internal energy (J kg-’)
= average longitudinal gas velocity (m s-’1
= velocity (m s - ’ )
= longitudinal length variable (m)
= reaction extent (conversion)
= density (g m-3)
= initial density (g m-3)
= dissipation (W m-3)
= vector differentiation operator

REFERENCES
Carden, P.O. (1987). ‘Direct work output from thermochemical energy transfer systems’, Int. J. Hydrogen Energy 12, (1) 13-22.
Carden, P.O. and Williams, O.M. (1978). ‘The efficiencies of thermochemical energy transfer’, Int. J. Energy Research, 2 389-406.
Buck, R., Muir J., Hogan, R., and Scocypec, R. (1991). ‘Carbon dioxide reforming of methane in a solar volumetric receiver/reac-
tor: the CAESAR project’, Solar Energy Materials, 24,449-463.
Diver, R., Fish, J., Levitan, R., Levy, M., Meirovitch, M., Rosin, H., Parypatyadar, S. and Richardson, J. (1992) ‘Solar test of an
integrated sodium reflux heat pipe receiver/reactor for thermochemical energy transport’, Solar Energy, 48, (l), 21-30.
Hill, C.G. (1977). ‘Engineering kinetics and reactor design’, Wiley, New York.
Lenz, T., Wright, J. and Chubb, T. (1978). ‘Engineering design study of conversion of solar energy to chemical energy through
ammonia dissociation’, Technical Report, Department of Agricultural and Chemical Engineering, Colarado State University,
U.S.A.
Levitan, R., Rosin, H. and Levy, M. (1989). ‘Chemical reactions in a solar furnace - direct heating of the reactor in a tubular
receiver’ Solar Energy Materials, 24, 464-477.
Levy, M., Levitan, R., Rosin, H., and Rubin, R. (1993). ‘Solar energy storage via a closed loop chemical heat pipe’, Solar Energy, 50,
179.
Lovegrove, K. (1993a). ‘Thermodynamic limits on the performance of a thermochemical energy storage system’ Int. J. Energy
Research, 17, 817-829.
Lovegrove, K. (1993b). ‘Exergetic optimization of a thermochemical energy storage system subject to real constraints’, Inr. J. Energy
Research, 17, 831-845.
978 K. M. LOVEGROVE

Nandy, S. and Lenz, T. (1984). ‘Observations on the catalytic decomposition of ammonia at high temperatures and pressures’,
American Int. Chem. Eng. J. 30 (31, 504-507.
Richardson, J., Parypatyadar, S. and Shen, J. (1988). ‘Dynamicsof a sodium heat pipe reforming reactor’, American Inst. Chem. Eng.
I. W5) 743-752.
Vancini, C. (1971). ‘Synthesis of Ammonia’, Macmillan, London, U.K.
Williams, 0. and Carden, P. (1979a). ‘Ammonia dissociation for solar thermochemical absorbers’ Int. 1. energy Research, 3,129-142.
Willaims, 0. and Carden, P. (1979b). ‘Energy storage efficiency for the ammonia/hydrogen-nitrogen thermochemical energy
transfer system.’ Int. J. Energy Research, 3, 129-142.