NAUKA TECHNIKA
5Nr I-IV/2015 Polska Energetyka Słoneczna 47
AMMONIA – CARBON ADSORPTION CYCLE RESEARCH AT THE
UNIVERSITY OF WARWICK
A.M. Rivero-Pacho1, R.E. Critoph
1, S.J. Metcalf
1, M. van der Pal
2
1 University of Warwick, Coventry, CV4 7AL, United Kingdom,
2 ECN, Petten, 1755 ZG, The Netherlands,
ABSTRACT
Active carbon-ammonia cycles have been developed from the
1980’s in the USA and are still a major research interest at
the University of Warwick, where systems have been built
for car air conditioning, solar refrigeration and gas-fired heat
pumps. The basic cycles are introduced and a brief
description of the historical development is presented. The
paper then presents work to date on domestic gas fired heat
pumps with preliminary results from the latest system under
test, and describes plans and the prospects for future
products. The possibility of using carbon-ammonia in thermal
transformers for industrial use is modelled theoretically and
the prospects discussed and compared with using ammonia
and chemical salts.
INTRODUCTION
The most basic adsorption refrigeration cycle is
shown in Fig. 1 where two linked vessels are presented.
The vessel on the left hand side acts as the generator,
containing the solid adsorbent and adsorbed refrigerant,
whilst the vessel on the right hand side acts as a
receiver, condenser and evaporator during the cycle,
containing refrigerant gas. In the Stage (a) of the cycle
the system is at low pressure and ambient temperature.
The adsorbent contains a high concentration of
adsorbed refrigerant whilst the right hand vessel
contains refrigerant gas. When the vessel on the left is
heated, the adsorbed refrigerant gets driven out and the
pressure of the system increases.
In Stage (b) the pressure of the system is high
enough so that the refrigerant gas starts to condense in
the right hand vessel whilst rejecting heat. After that
the generator is cooled back to the initial low
temperature readsorbing the refrigerant in the adsorbent
material which creates a drop in the pressure of the
system. When the pressure drops, the liquid refrigerant
boils and produces the cooling effect, absorbing heat.
In Stage (c) the system returns back to the initial state
and the cycle is completed. The cycle provides
heating/cooling discontinuously for approximately half
of the cycle time. If two or more generators are
operated out of phase the heating/cooling could be
produced continuously.
When continuous heating/cooling is produced the
right hand vessel mentioned above is usually split into
two separate components: a condenser and evaporator.
Each generator of the system is connected to the
condenser and to the evaporator via separate automatic
check valves as shown in Fig. 2a). Once the refrigerant
is desorbed from the generator, it is condensed in the
condenser and it passes through an expansion valve to
reduce its pressure. After that the refrigerant evaporates
in the evaporator and finally it can be adsorbed in the
generator completing a cycle. The ideal adsorption
cycle assumes that the condensation and evaporation
occur at constant pressures and that the void volume in
the bed is zero so that the pressurisation and
depressurisation are isosteric processes.
Fig. 1 – Simple adsorption cycle schematic (Metcalf et al., 2012)
NAUKA TECHNIKA
48 Polska Energetyka Słoneczna Nr I-IV/2015
Fig. 2 – a) Refrigerant circuit for one sorption generator b) p-T-x (Clapeyron) diagram for a single adsorption cycle
Figure 2b) presents the adsorption cycle and it can
be described in four stages:
Stage 1 à 2 – isosteric pressurisation: The
adsorbent in the bed is heated and its pressure rises
from evaporating to condensing pressure.
Stage 2 à 3 – isobaric desorption: The adsorbent
continues to be heated and starts desorbing refrigerant
at the condensing pressure while the refrigerant
condenses in the condenser.
Stage 3 à 4 – isosteric depressurisation: The
adsorbent in the bed is cooled and its pressure
decreases from condensing to evaporating pressure.
Stage 4 à 1 – isobaric desorption: The adsorbent
continues to be cooled and starts adsorbing refrigerant
at the evaporating pressure while the refrigerant
evaporates in the evaporator.
The COP (Coefficient of Performance) of the basic
adsorption cycle presented above is quite low. In order
to increase it, heat regeneration techniques that allow
efficient heat transfer from one adsorption generator to
another should be used. Multiple bed cycles are
a simple but way of achieving heat recovery between
the generators. In addition a good heat transfer design
in the generators is needed to achieve a good
performance.
At the University of Warwick heat driven
adsorption cycles are chosen to be developed over
mechanically driven vapour compression system as
they use primary energy sources more efficiently and
can use energy from renewable sources such as solar
collectors or waste heat from other sources. Adsorption
systems are preferred over absorption ones because of
their low operational cost and maintenance, higher
reliability, simple operations and no crystallisation,
corrosion or chemical disposal issues.
REFRIGERATION
EU-TOPMACS project. Introduction The developed cooling system in this project was
designed to simulate a mobile air conditioning system
(MACS) for a Class C passenger vehicle (such as
a Ford Focus or Fiat Bravo) with a 1.9 l turbo diesel
engine. This choice was very challenging due to the
high engine efficiency and therefore low waste heat
availability.
The engine coolant was to be used to provide the heat
input to the cooling system at a temperature of 90 °C
and a nominal flow rate of 24 l/min. The cooling power
required was determined to be 2 kW and, although it is
highly variable during the driving cycle, the nominal
heat input available is 5 kW – thereby necessitating a
nominal COP 0.4.
Generator design and sorption material The sorption generator used was a nickel-brazed
stainless steel design with 29 layers of active carbon
adsorbent each 4 mm thick. By incorporating the
carbon adsorbent in thin layers, conduction path length
through the material is reduced and the area for fluid
heat transfer is increased which enables rapid
temperature cycling and thus a high SCP (Specific
Cooling Power). The separating stainless steel plates
are constructed from chemically etched shims with
0.5 mm square water flow channels on a 1 mm pitch.
These channels give a high heat transfer coefficient and
a large heat transfer area, further improving heat
transfer performance. The square design ensures equal
flow path lengths in every channel and therefore even
heating and cooling of the adsorbent. The internal
pressure (up to 20 bar when condensing at 50°C) is
withheld by the stainless steel shims which act as
supporting webs to the outer wall, which only needs to
be 3 mm thick despite being straight. The open end of
the front face as shown in Figure 3a) is used to insert
and remove the carbon in order that a range of
adsorbents can be tested. Once the generator is filled
with carbon water manifolds and pressure flanges are
put into place prior to testing.
The sorption material used in the generators is
active carbon type SRD1352/3, especially good for
refrigeration applications, based on coconut shell
precursor manufactured by Chemviron Carbon Ltd.
1 kg of grains of carbon were mechanically compressed
between the shims of each generator reaching a bulk
density of 435 kg/m3 and yield a thermal conductivity
of 0.42 W/(mK). The heat transfer coefficient between
the water channels and the shims was 4150 W/K and
between the shims and the carbon 420 W/K. The
overall heat transfer coefficient of each generator was
380 W/K (Tamainot-Telto et al., 2009).
NAUKA TECHNIKA
Nr I-IV/2015 Polska Energetyka Słoneczna 49
Fig. 3 – a) Plate heat exchanger generator design, b) Schematic diagram of laboratory MACS
System description A schematic diagram of the system is shown in
Figure 3b). The engine coolant is alternately passed
through the two generator beds in order to heat them.
A second pumped coolant loop is used to recover heat
between the two beds. An air to water heat exchanger
placed in front of the vehicle radiator (labelled
adsorption heat exchanger) is used to cool the
generator beds to ambient temperature.
An interconnecting pipe with a valve is also
incorporated which enables the ammonia side of the
two generators to be connected for mass recovery
purposes. In this process, the heated high-pressure bed
is connected to the cooled low-pressure bed and
ammonia is transferred from the high pressure to the
low. This increases the concentration change in the
adsorbent during the cycle, thereby increasing SCP and
COP. Check valves are used to control the flow of
ammonia between the generators and the condenser
and evaporator, which are as per a conventional
system. One key difference however is the use of an
indirect evaporator with an intermediate chilled water
glycol loop – this prevents leakage of toxic ammonia
into the cabin, which could occur with a direct
evaporator. For the purposes of controlling both the
condensing and evaporating temperatures during the
experimental tests, an indirect condenser
(ammonia/water) and a temperature controlled bath
linked to the evaporator (ammonia/water) are utilised.
Experimental results and analysis Tests were carried out under the normal European
summer conditions, condenser temperature of 32°C and
evaporating temperature of 20°C. With an operating
driving temperature of 80°C and 1.3 m3/h flow on each
coolant (source and sink) and with mass recovery, the
system met the cooling power target: 1.3 kW vs.
1.2 kW target. However the COP was below the target:
0.23 vs. 0.52. The cooling power achieved corresponds
to a power density (the cooling power per unit volume)
of 93 W/l based on generator volume and 62 W/l based
on the total system volume. The specific cooling power
(the cooling power per unit mass of carbon) SCP is
about 0.650 kW/kg. With increased driving
temperature up to 90°C, the cooling power increased to
1.6 kW, exceeding the 1.2 kW target by 33%. The COP
was 0.22, which is close to the target value of 0.24. The
decreased COP obtained with higher driving
temperature is due to the fact that the cycle time was
not optimised for each condition. The power density
was 114 W/l based on generator volume and 77 W/l
based on total system volume. The SCP reaches
0.8 kW/kg. The effect of coolant flow rates through the
generator from 0.46 to 1.25 m3/h over the range tested
is minimal. The system should therefore not be
significantly affected by the variation in the coolant
water flow rate from the engine during the driving
cycle. The current preliminary performance could be
improved when operating the system with both mass
and heat recovery and with an optimised control
strategy (cycle time).
Quadcon refrigerated container with solar
adsorption refrigeration system
Introduction The aim of this project was to develop and built
a stand-alone solar powered adsorption refrigeration
device. The refrigerator would be integrated with
a standard US Army QuadCon insulated container with
energy requirement based primarily on solar thermal
and solar photovoltaic power. In order to achieve this,
the development of a high adsorbent capacity, high-
density and high thermal conductance carbon along
with a compact plate heat-exchanger could lead to
a truly grid-free/fuel-free refrigeration system. The
goal of the project was to achieve in a QuadCon
a consistent 5ºC air temperature during both day and
night time operations and during conditions which
necessitated door openings during the day.
Generator design and sorption material The generators used in this project were very
similar to the ones used in the EU-TOPMACS project
but in this case the number of carbon slots was
decreased to 10 and their thickness increased to 12 mm
as a carbon with higher thermal properties was used.
The sorption material used in the generators is
a polymer mix of active carbon SDS3 based,
manufactured by Entegris Inc. (previously ATMI Inc.).
2 kg of this carbon mixture were compressed into
smooth-sided monolith blocks and placed between the
generator shims using a high thermal conductivity
epoxy coating to improve heat transfer between carbon
and shims. The bulk density of the carbon mixture was
895 kg/m3 and yielded a thermal conductivity of
2.04 W/(mK) (Carruthers, 2011).
NAUKA TECHNIKA
50 Polska Energetyka Słoneczna Nr I-IV/2015
Fig. 4 – a) Solar Thermal Panels at QuadCon Adsorption Refrigeration System, b) System schematics
System description A schematic diagram and photograph of the
system is shown in Figure 4b). The system consists of
two beds and has mass and heat recovery, very similar
to EU-TOPMACS project. The evaporator is immersed
in an ice bank of nearly 200 l within the QuadCon.
Experimental results and analysis While the machine was tested (March in Arizona,
US) the average solar radiation was 650 W/m2
(peaking at 750 – 800 W/m2), which meant that the
target water (heat transfer fluid) temperatures of 140ºC
needed to yield a good COP was reached for at least
3 h of the 8 h of sun contact. Low cooling powers of
around 400 W were registered during testing. The COP
reached between 0.3 and 0.5 corresponding to cooling
cycles between 250 and 600 s. An average cycle had
180 s of heat/cooling period, 10 s mass recovery and
20 s heat recovery. If the solar radiation was very high
the heat recovery was eliminated to ensure a good
temperature control of the generators.
It is believed that due to the low solar radiation and
due to the lower than expected heat transfer properties
in the generators the system did not achieve the target
level of refrigeration.
HEAT PUMPING: 2-BED HEAT DRIVEN HEAT
PUMP
Introduction The heat pump developed in this project was
intended to be used in a domestic environment (space
heating), replacing a gas condensing boiler. For space
heating of a typical family home in the UK, a three
bedroom semi-detached house, which is required to be
maintained at an internal temperature of 18°C, the heat
pump should supply a heating power of 7 kW (Metcalf,
2009). The 2-bed machine would be driven by the heat
supplied by a gas burner and used pressurised water as
heat transfer fluid. The water used in the heating
system of the house is passed through the ammonia
condenser and then through the cooler (fluid to fluid
heat exchanger) where it increases its temperature.
Generators design and sorption material
The type of heat exchanger used in this project was
shell and tube. The core of generators was made of
stainless steel 304 and their tubes had an outer diameter
of 1.2 mm and an inner diameter of 0.8 mm. The end
plates of the generators had a diameter of 144.5 mm
and 1777 tubes that were nickel brazed together
creating the core of the generator as it can be observed
in Figure 5a). After the core was filled with 3 kg of
a mixture of carbon grains and powder, it was slid
inside the two generator shells. Finally, two flanges
were attached at the ends of the shell completing the
generator assembly.
The generator is effectively a thermally driven
compressor and is the most critical part of the design.
Originally, with the old shell design (one shell and two
end flanges), it was discovered that the generator
flanges contained 10 kg of stainless steel which
reduced significantly the COP. In order to reduce the
thermal mass of the generator shell new domed end
flanges that reduce the mass of steel of the generator
from 10 kg to 2 kg were designed and manufactured
and will be soon tested in the machine. A comparison
of both shell versions can be observed in Figure 5b)
and the new shell version of the generators already
installed in the heat pump can be seen in Figure 5c).
The sorption material used in the generators is
active carbon type 208C, especially good for heat
pumping applications, based on coconut shell precursor
manufactured by Chemviron Carbon Ltd. 3 kg of
a mixture 2/3 grains and 1/3 powder of carbon was
vibrated in the shell and tube heat exchanger reaching
a bulk density of 640 kg/m3 and yield a thermal
conductivity of 0.3 W/(mK). The heat transfer
coefficient between the water channels and the shims
was 5500 W/K and between the shims and the carbon
1300 W/K. The overall heat transfer coefficient of each
generator was 1000 W/K (Rivero Pacho, 2014).
So
lar
coll
ecto
rs a
nd
ho
t o
il p
um
p
Sorption
Generator
G1
Sorption
Generator
G2
Evaporator / ice store
Coole
rC
onden
ser
Ammonia
Solar
collector oil
loop
Adsorber Oil
Cooling loop
Ambient air
Intercooling
loop
Check valve
Cold space
FAN
Two-way valve Solenoid valve
NAUKA TECHNIKA
Nr I-IV/2015 Polska Energetyka Słoneczna 51
Fig. 5 – a) Generator core, b) Comparison of old and new version of generator shell,
c) Generators installed in the machine
System description
The heat pump prototype was constructed to test the
performance of the generators along with the carbon
mixture and to validate the computational model
developed. The water schematics and the ammonia
pipework of the heat pump are shown in Figure 6a), b).
The components used in the prototype were:
· Generators: as previously described.
· Condenser and cooler: plate heat exchangers.
· Evaporators: Even though the heat pump was
design to be air sourced, during the testing the
type of evaporator used was flooded
evaporator (water source) since it was more
convenient and controllable.
· Receiver: cylindrical stainless steel vessel
made by the University of Warwick
workshops.
· Expansion valves: electronic controlled
especially design for ammonia.
· Check valves: poppet type check valves later
modified and changed for a ball and conical
seat design.
· Gas burner / Heater: even though the heat
pump was design to be heat-driven by gas,
during the testing the hot water was supplied
by an electric heater since it was more
convenient and controllable.
· Water pumps: fitted in the generators water
circuit, in the load circuit and in the heat
recovery circuit.
· Expansion vessel: used to pressurise the water
circuit as it operated at high temperatures (120
to 170°C).
Fig. 6 – a) Water and b) Ammonia pipework circuits
NAUKA TECHNIKA
52 Polska Energetyka Słoneczna Nr I-IV/2015
Experimental results and analysis
The machine, with the old version of the generators,
was tested with a driving temperature of 150°C and
with an evaporating temperature of 5 °C (temperatures
for normal winter conditions in the UK vary between
0 – 7°C). The machine was tested for two different
cases of delivery temperature:
· Underfloor heating: 36°C flow delivery– 26°C
return.
· Low temperature radiators: 50°C flow
delivery– 40°C return.
The results of the machine testing at the different
delivery temperatures can be seen in Figure 7a). When
compared to the simulation prediction it was observed
that the experimental performances were lower than the
model prediction. The experiment circled, was taken as
an example to be analysed. The COP obtained during
testing was 1.3 and as it was said before, it was lower
than the modelled predicted one that was 1.38. If in the
model, 10 kg of stainless steel (shell flanges and end
plates) were added to the thermal mass of the generator
the model predicted a COP of 1.29, very similar to the
one obtained during testing. This indicated that the
thermal mass of the generator was too high and a new
shell design was needed in order to increase the
performance of the machine. As described in the
previous section, ‘3.2 Generators design and sorption
material’, new domed end flanges and shells that
weight only 2 kg were design and manufactured and
according to the simulation model they would help
increase the COP of the machine from 1.29 to 1.35. At
the time of writing the generators with their new shells
are being installed in the machine in order to be tested
and matched against the model.
In Figure 7b) the experimental heating powers of
the condenser and cooler of the circled case in Figure
7a) are plotted and compared with the prediction of the
correspondent model and taking into account the 10 kg
of extra stainless steel for the old shell design. It is
possible to observe that the experiment and the
simulation are very similar.
Fig. 7 – a) COP vs Heating Power for underfloor heating and low temperature radiators (circled example case), b)
Heating Power vs time of the circled underfloor heating case
THERMAL TRANSFORMERS
Introduction
Where thermally-driven heat pumps pump heat
from low temperature to a medium temperature level
using the work potential of heat flowing from high to
medium temperature, thermal transformers use the
work potential of heat flowing from a medium
temperature level to a low temperature level, allowing
upgrade of heat from medium temperature to high
temperature. Figure 8 gives a schematic representation
of a thermal transformer based on the chemisorption of
ammonia by two (different) salts. This batch process
consists of a ‘charge’ or regeneration phase in which
the ammonia is desorbed from the high temperature
salt (HTS) and adsorbed by the low temperature salt
(LTS). This is achieved by cooling the LTS to ambient
temperature, where the equilibrium pressure of the
reaction between ammonia and LTS is lower than that
of ammonia and HTS at a medium temperature level.
At the end of this phase all ammonia is adsorbed to the
LTS and no ammonia remains in the HTS. By heating
the LTS with ammonia to the medium temperature
level, the equilibrium pressure of the sorption reaction
rises until it exceeds the equilibrium pressure of
reaction of ammonia with the HTS. As ammonia starts
reacting with the HTS, heat is released causing the
HTS to rise in temperature. Once the desired, high
temperature is reached, the remaining heat of reaction
can be extracted and used.
Thermal transformers can be applied when
sufficient heat at medium temperature is available and
high temperature heat is required. This typically occurs
in industrial processes where waste heat is released at
temperatures of up to 150°C (Spoelstra, 2002) where
simultaneously there is a demand for medium pressure
steam (180 - 200°C).
The first attempt by the Energy Research Centre of
the Netherlands (ECN) to build a thermal transformer
consisted of a shell and tube type reactor in which salt
1,00
1,05
1,10
1,15
1,20
1,25
1,30
1,35
5 6 7 8 9 10 11
CO
P (
hea
tin
g)
Heating Power (kW)
Underfloor heatingLow temperature radiators
0
10
20
30
40
50
60
0 120 240 360 480
Hea
tin
g P
ow
er (
kW
)
Time (s)
Experiment
CondenserExperiment Cooler
NAUKA TECHNIKA
Nr I-IV/2015 Polska Energetyka Słoneczna 53
was deposited without any support material or matrix.
Due to the poor thermal conductivity of the salts as
well as the agglomeration of salt after several cycles of
expansion and shrinking, the salt became a compact,
solid block that was no longer capable of adsorbing
ammonia. Later concepts used nickel foam and
aluminium foam to enhance heat transfer and to
stabilize the salt matrix. Although the results improved,
there were still issues with salt escaping from its matrix
and also for full scale CAPEX were too high and
sensible heat losses significant. This led to the
conclusion that the performance of the salt-matrix
should be studied in more detail before constructing
another reactor or thermal transformer system. This
was done in a setup where a single tube with salt in
a matrix could be repeatedly exposed to any desired
ammonia pressure and temperature (van der Pal, 2013).
By measuring heat fluxes and mass flows, the
performance of the salt in the matrix could be
monitored over many cycles. A matrix of ENG
(Expanded Natural Graphite) and salt proved to be the
most promising combination with stable performance
with MgCl2, CaCl2 and LiCl in ENG-matrices over
1000+ sorption cycles.
At the University of Warwick, in collaboration with
ECN, the ENG-salt matrix option has been further
developed. In the INTERACT project, part of H2020
Marie Curie programme, research is done to achieve
a sorbent reactor with a power density of 1 MW/m3. It
is built by SpiraxSarco and consists of 10 tubes and has
a length of 1200 mm. The inner diameter of the tubes is
1 inch and contains the ENG with a porosity of 90%
which has subsequently been filled with CaCl2 using
a salt solution, yielding a CaCl2 to salt ratio of
approximately 2:1. Pressurised water flows around the
outside of the tubes to provide/extract heat to/from the
sorbent material. Figure 9a) shows the reactor design
made by SpiraxSarco.
Fig. 8 - Schematic representation of a chemisorption-based thermal transformer cycle using two salts
Fig. 9 – a) Schematic drawing of reactor for a 10 kW chemisorption-based thermal transformer, b) Model
calculation of the advancement of the sorption reaction CaCl2.(4-8)NH3 in CaCl2-ENG matrix as a function of time at
five different distances from the tube-wall
NAUKA TECHNIKA
54 Polska Energetyka Słoneczna Nr I-IV/2015
Initial model calculations, that include heat transfer
and kinetics, show rapid adsorption and desorption of
NH3 from CaCl2 for driving temperatures of 10 K as
can be seen in Figure 9b). The kinetic parameters and
thermal resistance and conductivity used in this
simulation have been validated in separate
experiments. Future experiments will focus on
validating the performance in the reactor and the
translation of these results into an economical design of
a heat-driven heat transformer for industrial
application.
REFERENCES
Carruthers, J. D., 2011, Development of a QuadCon
refrigerated container with first generation
prototype solar adsorption refrigeration system,
Technical report NATICK/TR-11/016.
Metcalf, S. J., 2009, Compact, efficient carbon-
ammonia adsorption heat pump, Ph.D. Thesis,
University of Warwick, UK.
Metcalf, S. J., Critoph, R. E., Tamainot-Telto, Z., 2012,
Optimal cycle selection in carbon-ammonia
adsorption cycles, International journal of
refrigeration, 35, pp. 571-580.
Rivero Pacho, A. M., 2014, Thermodynamic and heat
transfer analysis of a carbon – ammonia adsorption
heat pump, Ph.D. Thesis, University of Warwick,
UK.