Applied Thermal Engineering 194 (2021) 117060

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Thermal-hydraulic performance investigation of an aluminium plate heat

exchanger and a 3D-printed polymer plate heat exchanger
S. Lowrey *, C. Hughes, Z. Sun
Department of Physics, University of Otago, PO Box 56, Dunedin, New Zealand

A R T I C L E I N F O A B S T R A C T

Keywords: Polymer heat exchangers have gained interest due to their favourable qualities over metal heat exchangers,
Heat exchanger including being lightweight, low cost, and having reduced fouling and corrosion. In this work, two identical
3D-printing prototype, counter-flow, air-to-air, plate heat exchangers were fabricated, one from aluminium and the other
Condensation
from thermoplastic using a fused filament fabrication 3D printer. The results have shown that under dry oper­
Dehumidification
Heat recovery
ating conditions, typical of certain domestic appliances, both heat exchangers demonstrate similar thermal
characteristics where the thermal effectiveness is greater than or equal to 0.50. Under wet conditions, the
aluminium heat exchanger out performs the polymer heat exchanger, and this effect increases with an increasing
temperature difference between the heat exchanger’s hot-side inlet and cold-side inlet. In terms of both thermal
and dehumidification capacity, the aluminium heat exchanger can outperform the polymer heat exchanger by as
much as 22% and 38%, respectively. Despite the lower performance of the polymer heat exchanger, the absolute
capacity and dehumidification rate are still reasonable and show promise for the use of polymer heat exchangers
as a good alternative to metal heat exchangers in heat recovery applications in domestic appliances, for example,
in clothes dryers or dehumidifiers, where operation takes place under both dry and wet operating conditions.

1. Introduction polymers, long-term seawater exposure may result in only minor

moisture-induced damage [8]. In addition to their favourable operating
Plate heat exchangers (PHEs) are widely used in heating, ventilation, characteristics, polymer heat exchangers also offer environmental ben­
air-conditioning and refrigeration (HVAC&R) technology. One applica­ efits over metal heat exchangers where the former requires two times
tion that has been receiving attention is in domestic appliances where less energy to produce a unit mass of material [9], and most polymer
they have been performance tested in an air-side energy recovery materials are recyclable, and can be reprocessed into other products
configuration in clothes dryers [1] and domestic scale refrigerative de­ [10]. Polymers are also easy to mold and the energy required to process
humidifiers [2,3]. The moist air that enters the hot ducts of a PHE is a specific shape is low [11]. Unlike metal units, plastic heat exchangers
cooled by the adjacent cold air-stream ducts. As a result, the cool sur­ can be easily contoured to fit available space, which is particularly
faces condense water vapour out of the warm air-stream. This can pro­ favourable for domestic appliance fabrication. Another major advantage
mote improved energy efficiency, by reducing the sensible cooling load is polymer’s comparatively low weight reduces the handling and
of the evaporator coil [2,3]. Recent work has shown that when testing a transportation emissions compared with metallic heat exchangers.
compact PHE in a domestic refrigerative dehumidifier, high rates of Other benefits of polymer heat exchangers include that they are less
condensation can occur in the PHE, which can account for ~ 25% of the influenced by thermal expansion and creep [7].
total dehumidification rate [4,5]. Polymer heat exchangers have been investigated in a wide number of
Polymer heat exchangers hold great potential as an alternative to applications and using a number of working fluid configurations
metal heat exchangers in a wide range of industrial applications including liquid-to-liquid, liquid-to-gas and gas-to-gas configurations
currently dominated by metal heat exchangers. Most polymer materials [12]. Waste heat recovery however is the primary application of poly­
are resistant to chemical acids, solvents and corrosive fluids, whereas mer plate heat exchangers due to their corrosion resistance in the
metals are susceptible to direct chemical dissolution [6,7]. For presence of condensation [13]. In this work, we focus on an air-to-air

* Corresponding author.
E-mail address: sam.lowrey@otago.ac.nz (S. Lowrey).

https://doi.org/10.1016/j.applthermaleng.2021.117060
Received 16 October 2020; Received in revised form 4 April 2021; Accepted 29 April 2021
Available online 14 May 2021
1359-4311/© 2021 Elsevier Ltd. All rights reserved.
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

plate heat exchanger. An experimental study was carried out using a demonstrating the benefits and feasibility for using polymer energy re­
polymer shell-and-tube heat recovery unit for greenhouses [14]. A five- covery heat exchangers in place of identical metal ones in domestic
tube bundle, with each tube having a wall thickness of 1 mm, was placed appliance applications.
within a single shell. Experimental data was in good agreement with the The specific objectives of this study are to compare the thermal
proposed model and the heat exchanger demonstrated a short payback characteristics of both prototype, plate heat exchangers operating under
period of three years, was easy to assemble, repair, maintain and operate dry and wet conditions to see how the overall cooling capacity and total
and was corrosion resistant. An 84% thermal effectiveness was thermal effectiveness compare, and most importantly, to quantify and
measured. A polymer PHE was experimentally investigated as a flue gas compare the moisture extraction rates. This investigation also addresses
heat recovery unit [15]. The channels were 1.5 mm thick polymer sheets the performance of a polymer heat exchanger in terms of fabrication via
spaced 1 cm apart. Experimental data was provided for dry and wet a simple fused filament 3D printing process. This paper addresses the
conditions. A numerical and experimental study was carried out on a following research questions for domestic scale, air-to-air, polymer plate
polymer PHE for the purpose of dehumidification and cooling [16]. A heat exchangers operating under psychrometric and airflow conditions
liquid desiccant was injected into one airstream to achieve dehumidi­ typical of domestic dehumidification:
fication while water was injected into the secondary airstream,
providing evaporative cooling. Streams were separated by 0.2 mm thick • What are the general thermal performance characteristics of a
polymer sheet. Numerical and measured data showed good agreement polymer plate heat exchanger, fabricated using fused deposition
[16]. method 3D-printing?
A disadvantage of polymer heat exchangers is their low thermal • How does a domestic scale polymer plate heat exchanger compare
conductivity compared to metal heat exchangers where polymer ther­ with an aluminium control heat exchanger under conditions of
mal conductivity is typically less than 1 W/m-K, which is two orders of relatively high-water vapour condensation?
magnitude lower than many metals [17]. In 2016, Trojanowski et al.
[18,20] investigated a heat exchanger fabricated from a thermally To address these questions, we use various methods in this work to
conductive polymer. Their results indicated that thermal conductivity assess condensation on the heat exchanger plates, including goniometry
values of stainless steel do not need to be achieved for similar sensible to obtain the static and dynamic contact angles, liquid displacement
heat recovery performance in a polymer heat recovery heat exchanger. testing, water retention testing and condensation performance testing to
However, during condensing operation, they noted that the lower obtain the moisture extraction rate, thermal capacity and thermal
thermal conductivity of the polymer heat exchanger would lead to effectiveness. Based on these measurements, the feasibility of using
higher surface temperatures and lower water condensation rates, but the polymer PHEs in place of metal PHEs in domestic appliances is assessed.
authors did not characterise how much the water condensation rate is
reduced. In 2020, Arie et al. [19] developed a novel water-to-air metal 2. Experimental methods
fiber composite heat exchanger produced by a 3D-printing method. The
composite heat exchanger demonstrated a thermal conductivity of 130 2.1. Plate heat exchanger fabrication
W/m-K [19].
In 2019, Amer et al. [7], presented a paper that investigated the Fig. 1 shows the plate heat exchanger design. The PHE was designed
suitability of plastic heat exchangers for dehumidification applications with a heat transfer surface area similar to that used in reference [21],
where their specific focus was hot and humid weather conditions. They which was also sized for domestic appliance applications. Two pro­
investigated three polymer heat exchangers, testing them at an ambient totypes were fabricated based on this design, one from aluminium, and
condition of 27 ◦ C for the dry-bulb temperature and 21.2 ◦ C for the wet- the other from polylactic acid (PLA). PLA has a glass transition tem­
bulb temperature. A polymer plate heat exchanger was used as the perature between 60 and 65 ◦ C and therefore was considered to be a
control against two polymer heat exchangers which used arrays of tubes, good material for testing in moist air conditions typical of domestic
one having an asymmetric array and the other having a uniform array. dehumidifiers where an energy recovery heat exchanger’s hot-side inlet
The control PHE consisted of PP plastic and a wall thickness of 0.13 mm. can be exposed to a temperature range of 5–15 ◦ C [2–5]. The weight of
They found that the novel polymer heat exchangers showed improve­ the aluminium heat exchanger was 7.68 kg and the polymer plate heat
ment in heat flux over the control polymer PHE, and this was in part exchanger weighed 2.51 kg, giving a polymer to aluminium heat
attributed to the condensate splashing mode that occurred on the tubes, exchanger weight ratio of ~33%.
which aided in reducing condensate retention [7]. The PHE inlets and outlets each have an identical face area. The
In 2020, Chu et al. tested an “assisted evaporator” in a refrigerative rectangular duct aspect ratio (height/width) and hydraulic diameter
dehumidifier with the aim of improving the dehumidifier’s moisture (DH = 4A/P) are 42.4 and 0.004 m, respectively. There were a total of 33
extraction rate (MER) and the energy efficiency [17]. It was found that hot-side ducts and 33 cold-side ducts, for a total of 66 ducts overall, or
an array of plastic straws placed between the evaporator and condenser 33 heat transfer units. Ducts were separated by a 1 mm thick plate,
could condense moisture from the entrained airflow where cold air off which can be seen in Fig. 1-C. The reason for the relatively high wall
the evaporator would cool the plastic tubes below the dew-point tem­ thickness was due to the 3D printer’s limitations where a thinner wall
perature. Under certain test conditions, it was found that the MER could could result in warping of the wall during the print process. Fig. 1-D
be increased by around 7–9%. However, it was found that the im­ shows the two inner duct strips used to prevent warping of the plates
provements could be lost if the straw array pressure drop became too during printing. These were used in both the aluminium and polymer
high as this reduced the system airflow [17]. heat exchangers to keep them identical. The heat exchanger duct and
Although many studies have been carried out with polymer heat face geometry is provided in Table 1.
exchangers, we are not aware of an investigation into small air-to-air,
counter-flow heat plate exchangers sized for domestic appliances, and 2.2. Liquid displacement measurement
their comparative thermal performance against a control metal heat
exchanger under dry and wet moist air conditions specific to the oper­ We note that the filament itself is not expected to be porous. How­
ation of appliances, e.g. dehumidifiers. Such a study is important due to ever, in anticipation of print imperfections that could lead to air gaps
increased interest in the adoption of energy recovery polymer heat ex­ within the polymer duct walls, liquid displacement measurements were
changers in domestic appliances [1,7,18,20] that can help reduce their conducted. The liquid displacement technique is described in detail in
energy consumption, promote reduced biological activity and lower reference [22]. The liquid displacement method was used to measure
cost. Therefore, this study addresses an important gap in the literature whether significant air space exists between the polymer plate walls as it

2
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

Fig. 1. (A) An image showing the heat exchanger geometry. (B) an image showing the geometry of a plate heat exchanger inlet (or outlet). (C) an expanded image of
the dashed circle shown in (B) which is a close up of the edge of the heat exchanger inlet/outlet showing the plate and duct thicknesses. (D) the internal layout of the
ducts showing the two internal duct supports. All images shown here are to scale.

when pumping-in. The pump-rate used for these measurements was

Table 1 0.05 µL/s. The contact angle hysteresis was then determined by taking
Summary of the heat exchanger geometric details.
the difference between the advancing and receding angles.
Duct-width (w) 2 mm Heat exchanger wall area 0.03 m2

Duct-height (h) 85 mm Heat exchanger face width 92 mm 2.4. Water retention measurements
(W)
Duct area (wxh) 170 Heat exchanger face height 230
mm2 (H) mm Dip testing measurements were performed by immersing the heat
Heat exchanger wall 1 mm Heat exchanger face area 0.02 m2 exchanger prototypes in a bath of room temperature water and then
thickness (WxH) rapidly lowering the water level below the heat exchanger via a hy­
draulic jack. During the lowering of the water level, the mass of the heat
exchanger is recorded continuously with a measurement being acquired
allows the void space volume ratio to be determined. To do this, a
once every 1 s using digital scales (Acculab, model number SVI-20B)
segment of 1 mm thick PLA was extracted from a heat exchanger plate.
having a resolution of 10 g and a 20 kg capacity. These dip testing ex­
Three mass measurements were then made using precision scales (Adam
periments were conducted five times per test condition which is rec­
Equipment PW124 Analytical Balance): (1) absolute ethanol is placed in
ommended to ensure the same dropping velocity [25], and the data was
a beaker and the mass is measured (m1); (2) then the PLA sample is
then averaged to determine the final steady water retention. The
placed in the beaker and the combined mass measured (m2); (3) the
average speed at which the water bath is lowered was ~0.09 ± 0.01 m/s.
sample is removed and the mass of ethanol is re-measured (m3). The void
Dip testing data for both the prototype heat exchangers (aluminium and
space volume ratio is then calculated as, (m1-m2)/(m2-m3).
polymer) was measured. The heat exchangers were kept in the same
orientation for these tests which is the same as their orientation during
2.3. Static and dynamic contact angle measurements wind tunnel testing (the orientation is shown in Fig. 1D).
To carry out this test, the dry heat exchanger mass is recorded and
Static and dynamic contact angles were measured on the aluminium the scales are then zeroed. The heat exchanger is immersed in the water
and PLA plate surfaces using an FTA200 goniometer. For static contact by raising the water bath and then given a resting time of 5 min to let air
angle (SCA) measurements, 6 µL volume drops of deionised water were bubbles escape from the ducts. We note that a visual inspection was
deposited onto the surface using a stepper-motor controlled syringe. made prior to lowering the water level to confirm that bubbles were no
This droplet volume corresponds to a droplet diameter that is less than longer present and 5 minutes was found to be enough time for this to
the capillary length, which is around 2.7 mm for water [23]. When this is always be the case. The water bath is then lowered. A total of five dip
satisfied, surface tension forces dominate the droplet shape rather than testing tests were carried out for each PHE so an average curve could be
gravitational forces [23]. In this condition, the measurements are car­ produced and the uncertainty for the measured data is taken as the
ried out by capturing profile images using a high-resolution CCD cam­ standard error in the mean.
era. The goniometer’s resolution is less than 1◦ and the system is Due to exterior surface imperfections on the two heat exchangers
calibrated against a standard (having a 90◦ contact angle) each time during the application of adhesive to hold the heat exchanger plates
before use. together (four duct segments in the case of the 3D-printed heat
Dynamic contact angles are measured to obtain the contact angle exchanger), aluminium tape was applied identically to the exterior
hysteresis (CAH) which can provide an indication of the degree of water surfaces of both heat exchangers except on the inlet and outlet faces. The
retention in a heat exchanger [24]. A surface having a low CAH will objective here was to make the effects of water retention identical on the
generally retain less water compared with a surface having a higher CAH exterior surfaces of both heat exchangers.
[24]. For CAH measurements, the advancing and receding angles were
measured by increasing or decreasing the water droplet volume on a 2.5. Test rig and instrumentation
surface until the drop starts to slip at each edge while maintaining a
fixed advancing angle, when pumping-out, or a fixed receding angle, The purpose of these experiments was to measure the steady-state

3
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

heat transfer processes under low relative humidity conditions (dry Sensor locations are shown in green in Fig. 2.
conditions) and the steady-state, simultaneous heat and condensation Air temperatures were monitored using Type-T thermocouples
processes under high relative humidity conditions (wet conditions) of housed in cylindrical stainless steel metal tubes. Type-T temperature
the two prototype heat exchangers. This was achieved by measuring sensor arrays were set up at the inlet and outlet of the cold-side of the
inlet and outlet dry-bulb temperatures, hot-side (HS) and cold-side (CS) PHE and at the hot-side outlet of the PHE to measure dry-bulb temper­
pressure drops, HS and CS air volume flow rates and the moisture ature. Outlet arrays consisted of six thermocouples evenly spaced within
extraction rate (condensation rate). The main comparative performance the rectangular cross-section and the cold-side inlet consisted of three
indicators used here are the overall heat transfer rate, moisture extrac­ thermocouples evenly spaced within the rectangular face. The thermo­
tion rate (MER) and thermal effectiveness. couples were calibrated against a platinum resistance thermometer with
The testing rig is shown schematically in Fig. 2. The rig is housed in a a measurement uncertainty of ±0.013 ◦ C over a temperature range of
psychrometric chamber that allows the hot-side moist air state to be 0–60 ◦ C, conforming with ASHRAE [26].
fixed, where the cold-side of the rig is coupled via ducting to an adjacent The thermocouples and pressure transducers were connected to a
psychrometric chamber that allows the cold-side moist air state to be dataTaker (model: DT800) data logger. The uncertainty in a thermo­
fixed. The hot- and cold-sides of the rig’s ducting consist of variable couple measurement is given by an error propagation associated with
speed controlled axial fans and temperature sensor arrays. The central the calibration against a platinum resistance thermometer, which
part of the rig is designed to house the test heat exchangers and separates complies with the ASHRAE standards of temperature measurement
the two air-streams. The entire rig is internally thermally insulated with [26]. The uncertainty in hot-side or cold-side temperature differences
a 25 mm foam insulation. System energy balance checks were performed was taken as the addition of the absolute errors associated with the two
to make sure system heat loss was low and the system energy balance temperatures.
was <5% comparing the total heat transfer rates inferred from the hot-
side data and the cold-side data. 2.6. System measurements
Airflow conditioning devices were constructed using arrays of
identical tubes (item 7, Fig. 2) and were placed downstream of both the The dry- and wet-bulb temperatures were continuously monitored in
PHE inlets and before thermocouple arrays. The test system was each chamber with ventilated psychrometers (item 2) [26] and the
equipped with Sensiron digital pressure transducers, with a measure­ relative humidity was calculated from these temperatures [27]. For a
ment uncertainty of ±1 Pa, to monitor the HS and CS pressure drop. steady-state temperature at a given location, e.g., the cold-side outlet of

Fig. 2. Schematic of the plate heat exchanger test rig.

4
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

a heat exchanger, the temperature is determined by summing over each 3. Experimental results
of the individual thermocouple measurements in an array, and this is
summed over to obtain the average temperature, that is, T = 3.1. Static and dynamic contact angle data
1 1
∑ N1 N2
N2 N1 i Tij , where N1 and N2 are the number of data points for a
Static contact angle measurements were made on the aluminium and
single thermocouple and number of thermocouples at the given location,
PLA surfaces which were the plate materials in the prototype heat ex­
respectively.
changers. Fig. 3A shows a top-view photograph of a 8 µL water droplet
The speed of the centrifugal fans (item 1) were controlled using
on an aluminium plate and Fig. 3B shows a cross-sectional CCD-image of
variacs (item 3). The air volume flow rate for the hot-side and cold-side
a ~6 µL water droplet on the aluminium plate. The top-view image
of the heat exchanger rig was measured by timing the inflation of a 100
shows that the water droplet has an isotropic in-plane wetting state, that
μm thick, cylindrical polythene bag, with a volume of 10.86 m3, using a
is, the surface tension forces between the droplet and solid surface are
digital stopwatch [28]. The end-point time of inflation is monitored
the same in all directions at the water/solid interface. The scanning
visually. The measurement is repeated three times and the variation of
electron microscope (SEM) images of the aluminium surface (Fig. 3C and
the end-point time is no larger than 3 s. Before each test, a fan is used to
3D) further reveal its relative smoothness. Fig. 4A shows a top-view
deflate the bag and a roller is used to flatten the bag before a mea­
image of a 8 µL water droplet on a PLA surface and Fig. 4B and 4C
surement is initiated. The bag inflation method has been shown to give
show CCD-images of a ~6 µL drop looking from either above and below
satisfactory agreement compared with using a vane anemometer or a
relative to Fig. A (Fig. 4B) and from either left or right relative to Fig. A
hot-wire anemometer to measure volume flow rate where the agreement
(Fig. 4C). Comparing the aluminium and PLA surfaces, we see from the
is within 2% [28].
photograph in Fig. 4A that the periodic microscale lines of the PLA
The PHE’s moisture extraction rate was determined by weighing the
surface promote an anisotropic wetting state. The SEM image of the PLA
mass of water in receptacle R1 (Fig. 2) and dividing the condensate mass
(Fig. 4D) shows that the pitch of the microscale lines is around 218 µm,
(mw) by the test run time (Δt, in hours), i.e. ṁw = mw /Δt. The conden­
and Fig. 4E shows that these lines have a height of around 51 µm. These
sate mass was measured using digital scales having a resolution of
lines arise from the 3D-printing process, where lines of filament are
±0.01 g and a 2000 g capacity. MER measurements were taken at least
deposited on the surface, the extruder is then lifted to deposit a subse­
three times for each condition with at least a one-hour duration between
quent layer, and so on. The drop clearly displays an anisotropic in-plane
each MER measurement. The MER measurement uncertainties reported
wetting state on this surface. It was not further investigated if these lines
here are the standard error in the mean.
could be further modified. Fig. 3C and 3D show the aluminium surface
The measured data is used to calculate the total thermal capacity of
also has a microstructure but the drop photograph in Fig. 3A shows an
the heat exchangers. If we consider a control volume between the inlet
isotropic wetting state. This is attributed to the micro-roughness being
and exit of the hot-stream area shown in Fig. 1D, at steady-state, the heat
much more variable compared with the PLA surface and the height of
transfer rate through the wall is given by
these microlines is tiny in comparison to the PLA microlines, moreover,
Q̇T = ṁa (ha1 − ha2 ) + ṁa (w1 hv1 − w2 hv2 ) − ṁw hw (1.1) there is an overlay of random roughness which will make for a more
heterogeneous surface roughness.
where ha , hv and hw are specific enthalpies of dry-air, water vapour and Fig. 3B shows the side-view CCD-image of a ~6 µL drop on the
saturated liquid water, respectively, w is the humidity ratio and ṁa is the aluminium surface. The static contact angle was found to be around
air mass flow rate. If sensible and latent cooling occur, as is the case in 91.0◦ ±1.2◦ , showing the aluminium is at the interface of the hydro­
the hot-side stream under certain conditions, the full expression allows philic/hydrophobic wetting regimes, where the boundary is typically
the heat transfer rate for the hot air-stream to be calculated. For the cold- defined as 90◦ . The anisotropic PLA surface has two distinctly different
side stream (behind the heat transfer wall shown in Fig. 1D), only sen­ static contact angles where observing the droplet from above or below
sible heating occurs, and as a result, the expression is reduced to relative to Fig. 3D (shown in Fig. 3E), the drop’s static contact angle is
101.20◦ ±0.70◦ , placing it well within the hydrophobic region. This is
Q̇T = ṁa (ha1 − ha2 ) (1.2)
attributed to the drop’s reduced ability to spread due to the higher
where w1 = w2 = w, hv1 = hv2 = hv and ṁw = 0. contact angle hysteresis (CAH) that is attributed to the microscale
The dry-side thermal effectiveness is calculated using the expression grooves. When viewing the droplet from the left or right relative to
ε = (TCSO − TCSI )/(THSI − TCSI ). Fig. 3D (shown in Fig. 3C), the droplet can more readily spread along the
microgrooves due to the lower contact angle hysteresis where the CAH

Fig. 3. (A) top-view photograph of droplet on aluminium. (B) side-view CCD-image of the drop on aluminium. (C) and (D) top-view SEM images of a droplet on
aluminium, where (D) gives a greater magnification of the surface.

5
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

Fig. 4. (A) top-view photograph of a droplet on PLA. (B) and (C) side-view CCD-images of the drop on PLA where image (B) is looking relative to the upper or lower
edge of (A) and (C) is looking relative to the left or right edge of (A).

along the lines is 49.19◦ ±8.00◦ and that against the lines is The main reason for the larger uncertainty measurements for the
70.42◦ ±6.10◦ . All static and dynamic contact angles and calculated dynamic contact angles with the PLA surface is due to the microgrooves.
contact angle hysteresis for the PLA and aluminium surfaces is sum­ The microgrooves create pinning of the droplet which can lead to the
marised in Table 2. drop moving at different times on the left-side and right-side of the drop
In order to compare the anisotropic PLA surface’s static and dynamic during the dynamic pump-in and pump-out process.
contact angles with the aluminium surface, here we compare the It is noted that the Wenzel micro-wetting state was observed for the
average in-plane static contact angle and CAH of the PLA surface with droplets on the grooved PLA surface. The Wenzel state is when a droplet
the static contact angle and CAH of the aluminium surface. The average fully wets the surface, as opposed to a state where the droplet sits upon
in-plane static contact angle of the PLA surface is 88.55◦ ±3.00◦ and the air cushions and the top of microscale grooves.
average static contact angle of the aluminium surface is 91.00◦ ±1.20◦ .
Therefore, the static CA for both surfaces are very similar within 3.2. Liquid displacement measurements
experimental error. Comparing the CAH for each surface, the average in-
plane CAH of the PLA surface is 59.81◦ ±14.10◦ and the average CAH for Void space for air to be trapped within the PLA walls was measured
the aluminium surface is 62.82◦ ±3.52◦ . These are once again very using the liquid displacement method [22] and was found to be around
similar. We note that the uncertainty is relatively large for the average 18% (volume ratio). This shows that there is opportunity for moisture to
in-plane CAH, and this is due to the variation in the receding contact be stored in the PLA wall and transferred from the hot air-stream to the
angle measurement. This measurement had a wide variation (while al­ cold air-stream.
ways being less than the static contact angle) and for this reason, the Imaging was conducted on a cross-section of PLA duct wall (that
dynamic angle measurements were carried out more times than the separates the hot and cold air-streams) to analyse the internal structure.
static contact angle measurements to improve the measurement. Fig. 5A shows an optical image of the 1 mm thick PLA wall and Fig. 5B
Because the aluminium’s average CA and CAH is very similar to the PLA shows an SEM image of the same cross-section but magnified. In Fig. 5A,
surface’s average in-plane static CA and CAH, the PLA’s microstructure it is clear that two columns of filament are used in preparing the wall and
is not expected to promote different condensation behaviour compared these are separated by open space. Fig. 5B shows a close-up SEM image
with the aluminium surface. where it appears that the two columns are connected via melted PLA,
which likely bridges the two filament columns during the print process.
Table 2 Due to imperfections in the PLA surface, we believe that during the
Static and dynamic contact angle measurements and calculated contact angle liquid displacement measurement, ethanol will penetrate within the two
hysteresis for PLA surface and aluminium surface. View A corresponds to Fig. 3B filament columns, explaining the void space volume ratio of 18%. We
and view B corresponds to Fig. 3C.
believe that these inner space regions may impact the polymer heat
Average Contact Angles PLA – View PLA – View Aluminium exchangers operation under wet conditions where moist air could
A B
penetrate this space within the filament layers leading to condensation
Static [-] 101.20 ± 75.90 ± 91.00 ± within the wall.
0.70 2.30 1.20
Advancing [-] 117.84 ± 80.89 ± 92.31 ±
1.18 3.82 0.72 3.3. Water retention measurements
Receding [-] 47.42 ± 31.70 ± 29.49 ±
4.92 4.18 2.80
The dip testing method was used to measure the water retention in
Contact angle hysteresis [-] 70.42 ± 49.19 ± 62.82 ±
6.10 8.00 3.52 the polymer PHE (P-PHE) and aluminium PHE (A-PHE). Fig. 6 shows the
Polymer surface, average in-plane 88.55 ± 3.00 NA water retention of the P-PHE (black curve) and the A-PHE (blue curve)
static CA [-] versus time. It was found for both heat exchangers that steady-state
Polymer surface, average in-plane 59.81 ± 14.10 NA water retention occurs after around 400 s, when the rate of water
CAH [-]
leaving the heat exchanger is 0.15 g/minute. This rate was considered as

6
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

Fig. 5. Images of the PLA wall’s cross-section. Left, optical image. Right, SEM image.

steady-state in dip-testing measurements carried out by Sommers et al.

[24]. Subsequent dip testing measurements were run for no less than
500 s (~8 min). These tests were repeated for each heat exchanger six
times to obtain an average water retention curve.
For both heat exchangers, the bulk of the water is removed in the first
50 s of being removed from the water bath. However, the water leaves
each heat exchanger at different rates in these two PHEs in the first 50 s.
The water leaves the A-PHE much faster than compared with the P-PHE,
but overall the water retention in each heat exchanger is the same.
Figs. 7 and 8 show photos of the inlet and outlet faces of the aluminium
and polymer heat exchangers, respectively, after a water retention test.
In Fig. 7, the upper images show that the aluminium inlet has no
condensate bridge formation. Whereas the lower images in Fig. 7 show
that a number of condensate bridges have formed at the aluminium heat
exchanger’s outlet face. In Fig. 8, there are a number of condensate
bridges formed within the P-PHE ducts at both the inlet and outlet faces.
Both inlet and outlet faces show that thin threads of filament have been
formed within the ducts. Furthermore, condensate bridges, showing a
range of sizes, have formed around these threads (Fig. 8, left image) and
Fig. 6. Heat exchanger mass as a function of time during the water retention
measurements. The blue curve is for the aluminium heat exchanger and the we note that these threads are also located well within the ducting and
black curve is for the polymer (PLA) heat exchanger. are likely throughout each duct in the polymer PHE. We suspect that
when the water leaves the polymer heat exchanger, the small threads set
up an additional resistance to the flow of free water as well as the surface

Fig. 7. Inlet and outlet faces on the aluminium plate heat exchanger for investigating retention behaviour after water retention testing. Upper images (A & B), inlet
face. Lower images (C & D), outlet face. Red arrows indicate locations of condensate bridging.

7
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

Fig. 8. Inlet face (left) and outlet face (right) on the polymer plate heat exchanger to investigate retention behaviour after water retention testing. Red arrows
indicate locations of condensate bridging.

bound water, resulting in a slower rate of water leaving the P-PHE temperature was less than the HS inlet dew-point temperature by at least
compared to the A-PHE. This is further supported by the static and dy­ 1 ◦ C. Therefore, while some latent cooling will have occurred in the
namic wetting results. These showed that the average in-plane static ducts at some of the 20 ◦ C ITD conditions, the condensation rate was not
contact angle and CAH of the polymer heat exchanger are very similar to measured as these tests were focussed on dry operating conditions. We
the quantities obtained for the aluminium heat exchanger, and as a note that the HS inlet dew-point temperature was less than the HS outlet
result, the water retained in both heat exchangers should be similar. dry-bulb temperature for all tests at an ITD of 5 ◦ C, 10 ◦ C and 15 ◦ C.
At steady-state, the water retained in the polymer heat exchanger is Fig. 10 compares the two heat exchangers for their HS heat transfer
630 g and in the aluminium heat exchanger it is 620 g. This difference is rate (A) and CS heat transfer rate (B). For the HS heat transfer rate, this is
small and primarily attributed to the filament bridging within the higher in the A-PHE compared with the P-PHE for an ITD of 20 ◦ C. This is
polymer heat exchanger ducts. These results reflect the fact that the in- in-line with the results above where we note that there is possibly
plane static and dynamic contact angles of the polymer PHE are very condensation when the ITD is 20 ◦ C given that the HS inlet moist air
similar to those in the aluminium heat exchanger. dew-point temperature is higher than the HS outlet dry-bulb tempera­
ture. However this is true for both heat exchangers at an ITD of 20 ◦ C at
the highest airflow. We note that latent cooling may be slightly higher in
3.4. Heat exchanger performance testing the P-PHE at this condition as the difference in the P-PHE’s HS and CS
heat transfer rate is wider at this condition (see Fig. 9) compared with
3.4.1. Dry conditions that for the A-PHE at the same condition. Comparing the CS heat transfer
Dry-tests were conducted to investigate the performance of the P- rates of both heat exchangers, overall the trends are similar, and their
PHE and A-PHE as well as compare the performance of both PHEs. A dry- disagreement is small, being less than or equal to 5% for all data points
test was considered as a test where the hot-side outlet temperature was except that at an ITD of 15 ◦ C for the mid-range AVFR. Here the relative
greater than the dew-point temperature based on the hot-side inlet moist difference in the heat transfer rate is ~ 12% but we note that the ab­
air conditions. solute difference is small, being around 37 W.
Fig. 9 shows the hot-side (HS) and cold-side (CS) heat transfer rates Fig. 11 shows the pressure drop versus the AVFR for the P-PHE (A)
versus the AVFR under dry operating conditions for the P-PHE (A) and and the A-PHE (B). For the P-PHE, the HS pressure drop is slightly higher
A-PHE (B). For both PHEs, the agreement between HS heat transfer rate than all other pressure drops. This is attributed to an oversight during
and CS heat transfer rate are in very good agreement, for inlet temper­ the experiment where the pressure drop was not properly corrected once
ature differences (ITD) of 5, 10 and 15 ◦ C. This confirms that the heat steady-state condensation was present and as a result, the airflow may
exchanger performance test rig has minimal heat and air leakage and be slightly higher to what was used in the HS heat transfer rate calcu­
minimal heat exchanger by-pass under dry operating conditions. For an lation but we emphasise that this is only the case for the P-PHE tests at
ITD of 20 ◦ C, a noticeable trend occurs for the P-PHE where at the two an ITD of 20 ◦ C.
higher AVFRs, the HS and CS heat transfer rates diverge. It is noted that For the other conditions, the pressure drop is relatively similar where
for both heat exchangers, for all the tests at an ITD of 20 ◦ C, the HS outlet

Fig. 9. Thermal capacity versus air volume flow rate for the polymer heat exchanger (A) and aluminium heat exchanger (B). The tests were carried at inlet tem­
perature difference conditions including 5 ◦ C, 10 ◦ C, 15 ◦ C and 20 ◦ C. The dashed and solid lines correspond to the hot-side and cold-side heat transfer rates,
respectively.

8
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

Fig. 10. Heat transfer rate versus air volume flow rate for the polymer and aluminium heat exchangers. Graph (A) compares the hot-side heat transfer rate and graph
(B) compares the cold-side heat transfer rate between both heat exchangers. The tests were carried out at inlet temperature difference conditions including 5 ◦ C,
10 ◦ C, 15 ◦ C and 20 ◦ C.

Fig. 11. Pressure drop versus air volume flow rate for the polymer (A) and aluminium (B) heat exchangers. Tests were carried at inlet temperature difference
conditions including 5 ◦ C, 10 ◦ C, 15 ◦ C and 20 ◦ C. Dashed and solid lines correspond to the hot-side and cold-side pressure drop, respectively.

a small divergence appears at the highest AVFR of around 32 Pa. For the conditions. For all of the testing presented in this section, the HS inlet
A-PHE, the pressure drops all show the same trend, however, they show relative humidity was 80% and the HS outlet temperature was always
a spread that remains between ~ 20 Pa, which is considered small less than the dew-point temperature. The condensation rate at the ITD of
enough to ignore for the sake of comparing the A-PHE and the P-PHE. 5 ◦ C resulted in a tiny MER (≪ 0.1 kg/hour) for both heat exchangers,
Comparing the P-PHE and A-PHE pressure drop, overall the P-PHE’s which is difficult to measure accurately and therefore these tests were
pressure drop is slightly higher than compared with the A-PHE’s pres­ not conducted.
sure drop where at the lower AVFR, the P-PHE’s pressure drop range is Fig. 12 shows the moisture extraction rate (MER) for the P-PHE and
106 Pa - 122 Pa and the A-PHE’s range is 88 Pa - 109 Pa. At the higher A-PHE. With the increase in the ITD, the MER goes up for both PHEs. In
AVFR, the P-PHE’s range is 169 Pa - 211 Pa and the A-PHE’s range is addition, as the ITD goes up, the MER difference between the two PHEs
160 Pa - 180 Pa. At both ends of the AVFR range, the pressure drop increases where the MER in the A-PHE is always greater. Considering the
ranges overlap, showing that the pressure drops are similar. If we ITD of 20 ◦ C, the relative improvement between the A-PHE’s and P-
compare just the dry tests, the P-PHE pressure drop range is still higher PHE’s MER is 22% at the lowest airflow, and 27% at the highest airflow.
than that of the A-PHE but is less when factoring wet test data. Fig. 13 shows the HS and CS heat transfer rates versus the AVFR for
The main reason for the higher pressure drop range in the P-PHE is the P-PHE (A) and the A-PHE (B). For the P-PHE, the agreement between
attributed to stray filament in the P-PHE ducts. This was discussed the HS heat transfer rate and CS heat transfer rate for each ITD is good at
earlier with regard to promoting condensate bridging during water higher airflows over the AVFR range. At the low airflows, the relative
retention tests. The stray filament was highly variable across P-PHE error is slightly higher, and this is attributed to the possibility that more
ducts with some ducts having a very smooth finish and others having a water is retained or transferred within or through the polymer walls, and
lot of stray filament. This will likely promote the higher pressure drop in therefore does not contribute to the calculation of HS heat transfer rate.
the P-PHE compared to the A-PHE for the same air volume flow rate. For the A-PHE, the agreement between the HS and CS heat transfer rates
for most data are in reasonably good agreement. For the ITD of 20 ◦ C at
3.4.2. Wet conditions the low AVFR, we attribute the lower HS heat transfer rate compared
Wet-tests were conducted to investigate the performance of the P- with the CS heat transfer rate to small water leakage points. Comparing
PHE and A-PHE as well as compare the performance of both PHEs under both PHEs, it can be seen that as the ITD increases, and at the two higher
conditions of latent cooling within the hot-side ducts. A wet-test was AVFRs, the heat transfer rates in the A-PHE are higher than the P-PHE,
considered as a test where the hot-side outlet temperature was lower showing that the A-PHE outperforms the P-PHE for the same moist air
than the dew-point temperature based on the hot-side inlet moist air conditions. The variability of the MER could be up 30% from hour to

9
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

outcome has occurred at the highest AVFR where the pressure drop
range here is 176 Pa–207 Pa, whereas for the same condition for the dry
testing, pressure drop range was 169 Pa–211 Pa. A similar outcome is
shown for the A-PHE where the pressure drop range across the AVFR is
small in absolute terms and similar to the ranges obtained for the dry
tests. This leads us to conclude that the pressure drop is not considerably
affected by condensation in the hot-side ducts of either heat exchanger.
We note that because the surfaces are on the hydrophobic/hydrophilic
wetting interface, we anticipate after prolonged operation that the
filmwise condensation mode prevails in both the aluminium and poly­
mer PHEs.
Fig. 16 shows the thermal effectiveness versus the AVFR for the P-
PHE (A) and the A-PHE (B) for both dry and wet conditions. The
effectiveness in both heat exchangers clearly improves significantly
between dry and wet conditions. For the P-PHE, the effectiveness range
for dry and wet conditions is 0.52–0.58 and 0.60–0.71. For the A-PHE,
these ranges are 0.50–0.57 and 0.60–0.73. So the effectiveness ranges
under both dry and wet conditions for both heat exchangers are similar
but the A-PHE does show slightly better performance overall as would be
expected from the higher heat transfer rates in the A-PHE under wet
operating conditions.
Fig. 12. Moisture extraction rate (MER) versus air volume flow rate for the
polymer (solid lines) and aluminium (dashed lines) heat exchangers. Tests were
4. Discussion
carried at inlet temperature difference conditions including 10 ◦ C, 15 ◦ C and
20 ◦ C and the hot-side and cold-side inlet relative humidities were fixed at 80%
and ≥90%, respectively.
Optical and SEM Imaging of the PLA duct walls (Fig. 5) showed that
the walls consisted of two columns of PLA filament that were separated
by an open gap which would be occupied by air. There was some PLA
hour for the same test conditions, and in some cases, a six hour test was
material that appeared to had melted and formed bridges between the
carried out to obtain an average MER (a minimum of 3 hours was used
two columns. Liquid displacement measurements of the PLA duct wall
for wet tests to obtain the average MER and this was used when vari­
found that the wall had an open-space of 18%. As discussed in Section
ability between tests was below 10%). Therefore, it is likely that for the
3.4.2, further heat exchanger performance testing did find that the P-
wet tests, the CS heat transfer rate provides the most accurate mea­
PHE had a lower dehumidification performance compared with the A-
surement of the heat exchanger capacity.
PHE where condensate could potentially be trapped in these gaps within
Fig. 14 shows the HS heat transfer rate (A) and CS heat transfer rate
the PLA duct walls, which can influence heat transfer.
(B) versus the AVFR for both heat exchangers. For all three ITDs, the A-
The static and dynamic contact angles were measured on the PLA
PHE shows higher capacity than the P-PHE. We consider the perfor­
and aluminium surfaces to see how they may respond to condensation
mance of each PHE by comparing the CS heat transfer rates as we believe
conditions and to see if their surface wetting properties were similar or
this provides the most accurate performance of capacity under wet
not as a difference may result in different condensation behaviour.
conditions, as described above. As the ITD increases, the divergence of
Photographic and SEM imaging showed that both surfaces have a
the two CS heat transfer rate curves for each PHE increases and this is
microscale roughness. This results in an anisotropic wetting state on PLA
again attributed to the higher degree of latent cooling in the A-PHE.
but only an isotropic wetting state occurs on the aluminium. SEM images
Fig. 15 shows the pressure drop versus the air volume flow rate for
of the aluminium show that the microscale roughness is not ordered like
the P-PHE (A) and the A-PHE (B). Both PHEs show that the pressure drop
that of the PLA which is uniform, periodic and much more comparable
is very similar for each test condition of the AVFR range. For the P-PHE,
with the size of the 8 µL droplet shown in the photographs. Despite the
at the lowest airflow, the pressure drop range is 112 Pa–127 Pa, whereas
different wetting states, comparing the average in-plane static contact
under dry conditions this variation was 106 Pa–122 Pa. A similar
angle and CAH of the PLA surface with that of the aluminium surface

Fig. 13. Thermal capacity versus air volume flow rate for the polymer (A) and aluminium (B) heat exchanger. The tests were carried at inlet temperature difference
conditions of 5 ◦ C, 10 ◦ C, 15 ◦ C and 20 ◦ C and the hot- and cold-side inlet relative humidities were fixed at 80% and ≥90%, respectively. Dashed and solid lines
correspond to the hot-side and cold-side heat transfer rates, respectively.

10
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

Fig. 14. Thermal capacity versus air volume flow rate for the polymer and aluminium heat exchangers. Graphs (A) and (B) compare the hot-side and the cold-side
heat transfer rates, respectively, between each heat exchanger. Tests were carried out at inlet temperature difference conditions of 5 ◦ C, 10 ◦ C, 15 ◦ C and 20 ◦ C and
the hot-side and cold-side inlet relative humidities were fixed at 80% and ≥90%, respectively. Dashed and solid lines correspond to the hot-side and cold-side heat
transfer rate, respectively.

Fig. 15. Pressure drop versus air volume flow rate for the polymer (A) and aluminium (B) heat exchanger. Tests were carried at inlet temperature difference
conditions of 5 ◦ C, 10 ◦ C, 15 ◦ C and 20 ◦ C and the hot- and cold-side inlet relative humidities were fixed at 80% and ≥90%, respectively. Dashed and solid lines
correspond to the hot-side and cold-side pressure drop, respectively.

Fig. 16. Thermal effectiveness versus air volume flow rate for the polymer (A) and aluminium (B) heat exchangers. The tests were carried out at inlet temperature
difference conditions of 5 ◦ C, 10 ◦ C, 15 ◦ C and 20 ◦ C. Dashed and solid lines correspond to tests carried out under wet and dry conditions, respectively.

shows that their wetting properties are very similar. As a result, we found that overall, the water retention behaviour is very similar in each
expect that the P-PHE and A-PHE behave similarly under condensing heat exchanger and these results are aligned with the contact angle
operating conditions. measurements which showed that overall, both heat exchanger mate­
Dip testing measurements were carried out to see if the water rials have very similar wetting properties. It is noted that while the water
retention properties of each heat exchanger are different. These tests retention in each heat exchanger is similar, their water retention curves

11
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

were different. This was attributed to the loose filament that would small as the MER itself is a small number. If chamber fluctuation
bridge between two duct walls. Some ducts had more of this print appeared to affect an MER result, a single steady-state test would run for
imperfection than others. This did not appear to significantly effect the up to six hours to obtain the average MER. Despite this attempt to
overall water retention but we believe that this may have resulted in the improve the MER measurement, we believe that this would result in
slightly higher pressure drop in the P-PHE compared with the A-PHE. some of the variability between the HS heat transfer rate and CS heat
This is discussed in more detail below. transfer rate. As a result, the CS heat transfer rate likely provides the
For dry operating conditions, both the polymer and aluminium heat most accurate measurement of the heat exchanger capacity under wet
exchangers show very similar thermal performance. The pressure drop conditions although we note that the HS heat transfer rate and CS heat
trend between both heat exchangers was found to be very similar. transfer rate remained in agreement to within 20% for both heat ex­
However, the polymer PHE did show a slightly higher overall trend in changers at the higher airflow rates. Relative differences larger than
pressure drop for the AVFR range. We attribute this to filament that was 20% occurred at low airflows, and this is attributed to small water leaks
present in the polymer ducts (see Fig. 8) which may have contributed to in both heat exchangers, and wall condensate retention and/or
this slightly higher pressure drop in the P-PHE. Despite the slightly condensate transfer across the polymer walls.
higher pressure drop, airflow calibration meant that the AVFR was kept For an ITD of 20 ◦ C and the lowest airflow, the A-PHE CS heat
the same in both heat exchangers for comparative testing. It is not transfer rate is only 5.6% greater than the P-PHE’s CS heat transfer rate
considered that the P-PHE’s pressure drop range is attributed to the and at the highest airflow, this relative increase rises to 13%. These
higher degree of latent cooling occurring in the P-PHE (which would differences in capacity at the low and high airflows for an ITD of 20 ◦ C
result in condensate within the hot-side ducts) as the pressure drops correspond to an MER of 0.07 kg/hour and 0.12 kg/hour respectively. It
remained very similar for the wet tests. was described earlier that the PLA duct walls have voids that could be
To assess the influence of the wall resistance on the overall heat penetrated with water and this could be accompanied by leaking into the
transfer coefficient, we take the wall resistance to be the wall thickness cold-side ducts. A small difference of 0.07 kg/hour to 0.12 kg/hour can
divided by the wall thermal conductivity. The aluminium and polymer potentially be explained by the possibility of condensate becoming
thermal conductivities are taken as 237 W/m-K [29] and 0.13 W/m-K trapped within the duct walls, possibly even leaking into the cold ducts,
[30], respectively. Given the wall thickness of 1 mm and the wall area of and then leaking back from the wall to the hot-side duct, and this could
around 0.031 m2, the wall thermal resistance for the A-PHE is 1.36 × explain the MER measurement variability.
10− 4 K/W and for the P-PHE is 4.2 × 10− 3 K/W. Now we consider the The change in pressure drop from dry to wet conditions for both heat
air-side thermal resistance and use the Nusselt number for a rectangular exchangers was found to be very small. This is attributed to both heat
duct as 7.54 which is valid when the duct aspect ratio (h/w) is larger exchangers having very similar surface wetting and water retention
than 8, which is the case here [31]. The air-side heat transfer coefficient properties. Because the water retention is relatively low, this suggests
is then given by h = Nu k
Dh [31], where Dh is the hydraulic radius, which for
that condensate does not build up in the form of a thick film or promote
a rectangular duct is given by Dh = 4AD/P [31], AD is the duct face area the formation of many condensate bridging sites, resulting in only a very
and P is the duct perimeter which is equal to 0.17 m, giving a hydraulic small change in pressure drop from dry to wet conditions.
radius of 3.9 × 10− 3 m. Taking the thermal conductivity of air as 0.026 The combined thermal effectiveness range for dry and wet conditions
W/m-K [32], the air-side heat transfer coefficient is ~ 50 W/m2-K, in the P-PHE was 0.52–0.71 and for the A-PHE it was 0.50–0.73.
giving a thermal resistance of 0.7 K/W. This air-side resistance is several Therefore, there is scope to make further improvements to this small
orders of magnitude greater than the wall resistance for both heat ex­ polymer plate heat exchanger in terms of thermal performance and
changers, and therefore helps explain why the dry performance of the dehumidification performance. As an example, the heat exchanger walls
heat exchangers is similar. were relatively thick (1 mm) in order to maintain strength of the poly­
A key question of this work was how a domestic scale polymer PHE’s mer heat exchanger, and also partly due to the low resolution of the
dehumidification capacity compares with an aluminium control PHE inexpensive 3D-printer, and thinner walls are known to reduce thermal
under conditions of relatively high-water vapour condensation. resistance [33]. A stronger polymer heat exchanger with thinner walls
Comparing the A-PHE and P-PHE in terms of dehumidification capacity could yield a higher thermal and dehumidification capacity, and thus
over the range of test conditions, the A-PHE outperformed the P-PHE at thermal effectiveness. This would also help to eliminate the possibility of
all airflow and ITD conditions. For the three ITDs tested, the average condensate being potentially trapped within the walls. This may also
relative increase in dehumidification capacity of the control metal PHE help to close the performance gap between the polymer and aluminium
compared with the polymer PHE was 24%, 29% and 23% for ITDs of PHE’s.
20 ◦ C, 15 ◦ C and 10 ◦ C, respectively. Determining the reasons for this are For the ITD test condition of 10 ◦ C the MER in the P-PHE is similar to
complex. For example, the low-resolution 3D-printing process would what is achieved in domestic refrigerative dehumidifiers that use energy
produce polymer walls that are semi-permeable compared with the recovery heat exchangers [21]. The MER for both the A-PHE and P-PHE
impermeable walls of the aluminium heat exchanger. Condensate may are close to 0.1 kg/hour, and this is similar to the MERs measured at the
become embedded within the heat exchanger walls. This then would energy recovery heat exchanger in such dehumidifiers [2,3]. Therefore,
result in a complicated wall thermal conductivity that would be a despite the reduced dehumidification capacity of the P-PHE compared to
weighted combination of thermal conductivities of water, air and PLA. the A-PHE, this suggests that the P-PHE is a good candidate for use in
Either way, this combined thermal conductivity would still be less than such an application and should be considered with other trade-offs
that of the aluminium heat exchanger. We note that reference [18,20] against a metal PHE, including weight and cost.
has indicated that the lower thermal conductivity of the polymer heat
exchanger’s walls would mean the hot-side surface would be higher than 5. Conclusions
compared with the metal heat exchanger’s hot-side wall. While it is
difficult to quantify what the actual reduction of the hot-side wall sur­ This study reports an investigation of two identically designed pro­
face temperature is in the polymer PHE compared with the aluminium totype counter-flow, plate heat exchangers sized for domestic appliance
PHE, it is likely that this is the case in the polymer heat exchanger tested applications, with one being constructed from aluminium and the other
here and the primary reason for the reduced moisture extraction rate. 3D-printed using the thermoplastic polylactic acid. The main aims of this
The MER could show variability between individual measurements study were to see how well the 3D printed PLA based heat exchanger
due to small psychrometric chamber fluctuations. While MER relative performed under dry and, especially, under wet operating conditions
error could be 30% at some test conditions, the absolute error is very compared with an aluminium control heat exchanger. Past research has
indicated that while similar sensible heat recovery is possible when

12
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

comparing polymer and metal heat exchangers, there is less work Declaration of Competing Interest
available on quantifying the difference in their dehumidification ca­
pacity. Furthermore, these heat exchangers tested here were sized such The authors declare that they have no known competing financial
they could be adopted in domestic appliances, such as domestic de­ interests or personal relationships that could have appeared to influence
humidifiers, where latent cooling can be relatively high [5]. In order to the work reported in this paper.
understand how each heat exchanger would perform in dry and wet
conditions, liquid displacement measurements were carried out to test Acknowledgements
how much water is retained within a segment of 3D-printed polymer
wall and goniometry measurements were carried out to quantify the The authors would like to acknowledge Professor Richard Blaikie for
heat exchanger wall wetting behaviour. In addition, both heat ex­ useful discussions. The authors also gratefully acknowledge financial
changers were tested for their water retention behaviour using a dip support from the Otago Energy Research Centre.
testing system and their condensing behaviour in a bespoke performance
testing rig. References
The surface wetting properties of each material were relatively
similar but the polymer heat exchanger’s duct walls were found to have [1] A.H. Nasution, P.G. Sembiring, H. Ambarita, Effectiveness of a heat exchanger in a
open-space within (18%). Furthermore, the 3D-printed material did heat pump clothes dryer, in: IOP Conference Series: Materials Science and
Engineering, Vol. 308, No. 1. IOP Publishing, 2018, pp. 012027.
have imperfections in some places and while the outside of the surface of [2] S. Lowrey, G. Carrington, Z. Sun, M. Cunningham, Experimental investigation of
the P-PHE could be improved for potential air and condensate leak sites, geared domestic refrigerative dehumidifier performance in New Zealand
this could not be fixed internally. household climates, Int. J. Refrig 35 (4) (2012) 750–756.
[3] S. Lowrey, G. Carrington, Z. Sun, Adapting a geared domestic refrigerative
Both the aluminium control and polymer heat exchangers were dehumidifier for low-temperature operation, Int. J. Refrig 41 (2014) 137–146.
found to have very similar thermal capacities when operating in dry [4] S. Lowrey, Z. Sun, A numerical model for a wet air-side economiser, Int. J. Refrig
conditions where their thermal effectiveness ranges were 0.50–0.57 and 60 (2015) 38–53.
[5] S. Lowrey, G. Le Bonniec, Z. Sun, Fluid flow modulation in a domestic refrigerative
0.52–0.58, respectively. Under wet operating conditions, the aluminium
dehumidifier with air-side gearing, Int. J. Refrig 106 (2019) 258–265.
control heat exchanger outperformed the polymer heat exchanger under [6] X. Chen, Y. Su, D. Reay, S. Riffat, Recent research developments in polymer heat
all wet test conditions (except two low airflow conditions where their exchangers–A review, Renew. Sustain. Energy Rev. 60 (2016) 1367–1386.
[7] M. Amer, M.R. Chen, U. Sajjad, H.M. Ali, N. Abbas, M.C. Lu, C.C. Wang,
heat transfer rates were the same) where the effectiveness ranges were
Experiments for suitability of plastic heat exchangers for dehumidification
0.60–0.73 and 0.60–0.71, respectively. The main reason for the capacity applications, Appl. Therm. Eng. 158 (2019), 113827.
being better in the aluminium control heat exchanger was the increased [8] F.E. Sloan, J.B. Talbot, Corrosion of graphite-fiber-reinforced composites
latent cooling capacity, which raised the dehumidification performance I—galvanic coupling damage, Corrosion 48 (10) (1992) 830–838.
[9] H.T. El-Dessouky, H.M. Ettouney, Plastic/compact heat exchangers for single-effect
compared to the P-PHE, however, the possibility of condensate retention desalination systems, Desalination 122 (2–3) (1999) 271–289.
within polymer PHE walls (and transfer from the hot ducts to cold ducts) [10] D.A. Reay, The use of polymers in heat exchangers, Heat Recovery Syst. CHP 9 (3)
may have made a small contribution to a reduced MER measurement. (1989) 209–216.
[11] P. Luckow, A. Bar-Cohen, P. Rodgers, J. Cevallos, Energy efficient polymers for gas-
The average relative dehumidification capacity improvement over all liquid heat exchangers, J. Energy Res. Technol. 132 (2) (2010).
test conditions was 25%, and while this is significant, we consider the [12] C. T’Joen, Y. Park, Q. Wang, A. Sommers, X. Han, A. Jacobi, A review on polymer
polymer PHEs dehumidification capacity high enough to warrant heat exchangers for HVAC&R applications, Int. J. Refrig 32 (5) (2009) 763–779.
[13] J.G. Cevallos, A.E. Bergles, A. Bar-Cohen, P. Rodgers, S.K. Gupta, Polymer heat
consideration of such heat exchangers in place of metal heat recovery exchangers—history, opportunities, and challenges, Heat Transfer Eng. 33 (13)
devices in domestic appliance applications by factoring the advantages (2012) 1075–1093.
of reduced bio-fouling, reduced weight and reduced cost. Moreover, a [14] D.R. Rousse, D.Y. Martin, R. Thériault, F. Léveillée, R. Boily, Heat recovery in
greenhouses: a practical solution, Appl. Therm. Eng. 20 (8) (2000) 687–706.
cheap off-the-shelf 3D-printer was used in this work, which resulted in
[15] L. Jia, X.F. Peng, J.D. Sun, T.B. Chen, An experimental study on vapor
polymer walls that were semi-permeable. Therefore, the dehumidifica­ condensation of wet flue gas in a plastic heat exchanger, Heat Transfer—Asian
tion capacity in such a polymer heat exchanger can likely be further Research: Co-sponsored by the Society of Chemical Engineers of Japan and the
Heat Transfer Division of ASME 30(7) (2001) 571–580.
improved by producing impermeable polymer walls.
[16] W.Y. Saman, S. Alizadeh, Modelling and performance analysis of a cross-flow type
The polymer PHE was also found to promote a higher pressure drop plate heat exchanger for dehumidification/cooling, Sol. Energy 70 (4) (2001)
compared with the aluminium heat exchanger, where for the highest 361–372.
airflow tests, the P-PHE’s pressure drop range was 176–207 Pa and for [17] W.X. Chu, C.H. Chiu, C.C. Wang, Improvement on dehumidifier performance using
a plastic assisted condenser, Appl. Therm. Eng. 167 (2020), 114797.
the aluminium control heat exchanger it was 153–176 Pa. [18] R. Trojanowski, T. Butcher, M. Worek, G. Wei, Polymer heat exchanger design for
Despite these differences between the two heat exchangers under wet condensing boiler applications, Appl. Therm. Eng. 103 (2016) 150–158.
operating conditions, they are considered relatively small and further­ [19] M.A. Arie, D.M. Hymas, F. Singer, A.H. Shooshtari, M. Ohadi, An additively
manufactured novel polymer composite heat exchanger for dry cooling
more, we expect that these differences can be reduced with improved 3D applications, Int. J. Heat Mass Transf. 147 (2020), 118889.
print settings which will be considered in a future study. Furthermore, [20] US Department of Energy, Energy conservation Program: energy conservation
the low thermal effectiveness of both heat exchangers under dry oper­ standards for residential dehumidifiers, 2016. [Online] https://www.regulations.
gov/document?D=EERE-2012-BT-STD-0027-0045 (accessed October 2020).
ating conditions shows that there is much scope for improving the per­ [21] S. Lowrey, Z. Sun, Experimental investigation and numerical modelling of a
formance of this small heat exchanger. One easy way to improve the compact wet air-to-air plate heat exchanger, Appl. Therm. Eng. 131 (2018)
effectiveness would be to use smaller duct walls which were limited to 1 89–101.
[22] A. Sionkowska, J. Kozłowska, Properties and modification of porous 3-D collagen/
mm in this study with the current 3D printer’s resolution.
hydroxyapatite composites, Int. J. Biol. Macromol. 52 (2013) 250–259.
This study has shown that under certain moist air conditions 3D- [23] D. Quéré, Non-sticking drops, Rep. Prog. Phys. 68 (11) (2005) 2495.
printed polymer plate heat exchangers show very good thermal char­ [24] A.D. Sommers, R. Yu, N.C. Okamoto, K. Upadhyayula, Condensate drainage
performance of a plain fin-and-tube heat exchanger constructed from anisotropic
acteristics compared to a metal control heat exchanger under dry and
micro-grooved fins, Int. J. Refrig 35 (6) (2012) 1766–1778.
wet operating conditions and could be considered for retrofitting ap­ [25] L. Liu, A.M. Jacobi, Issues affecting the reliability of dynamic dip testing as a
plications in domestic appliances such as dehumidifiers and could even method to assess the condensate drainage behavior from the air-side surface of
be trialled for home ventilation systems, especially given their benefits dehumidifying heat exchangers, Exp. Therm Fluid Sci. 32 (8) (2008) 15.
[26] American Society for Heating and Refrigeration Engineers ASHRAE, Standard
compared with metal plate heat exchangers such as reduced weight and Method for Temperature Measurement. ASHRAE, Atlanta, USA, 1986.
cost. [27] A. Wexler, R.W. Hyland, R.B. Stewart, Thermodynamic properties of dry-air, moist-
air and water and SI psychrometric charts, Reports from ASHRAE research
projects. 216-RP and 257-RP, ASHRAE J. (1984).
[28] C.G. Carrington, A. Marcinowski, W.J. Sandle, A simple volumetric method for
measuring airflow, J. Phys. E: Sci. Instrum. 15 (1982) 275–276.

13
S. Lowrey et al. Applied Thermal Engineering 194 (2021) 117060

[29] Engineering ToolBox, Thermal Conductivity of Metals, Metallic Elements and [31] Y.A. Cengal, A.J. Ghajar, Heat and Mass Transfer: Fundamentals and Applications,
Alloys, 2021. Accessed online March 2021. <https://www.engineeringtoolbox. fourth ed., McGraw-Hill Company, New York, 2011.
com/thermal-conductivity-metals-d_858.html>. [32] F.J. McQuillan, J.R. Culham, M.M. Yovanovich, Properties of Dry Air as One
[30] SD3D Printing, 3D Print with PLA Filament | 3D Printed PLA Material Data | SD3D Atmosphere. Rept. UW/M HTL, 8406. Microelectronics Heat Transfer Lab,
Printing, 2021. Accessed March 2021. <https://www.sd3d.com/wp-content/ University of Waterloo, Waterloo, Ontario, 1984.
uploads/2017/06/MaterialTDS-PLA_01.pdf>. [33] S. Pugh, G.F. Hewitt, H. Müller-Steinhagen, Fouling during the use of seawater as
coolant, Heat Transfer Eng. 26 (2005) 35–43.

14