A dehumidification process with cascading desiccant wheels to produce air with dew point below 0 °C

A dehumidification process with cascading desiccant wheels to produce air with dew point below 0 °C

Accepted Manuscript A Dehumidification Process with Cascading Desiccant Wheels to Produce Air with Dew Point below 0°C Qun Chen, Jim R. Jones, Richard...

NAN Sizes 0 Downloads 34 Views

Accepted Manuscript A Dehumidification Process with Cascading Desiccant Wheels to Produce Air with Dew Point below 0°C Qun Chen, Jim R. Jones, Richard H. Archer PII: DOI: Reference:

S1359-4311(17)38198-X https://doi.org/10.1016/j.applthermaleng.2018.10.114 ATE 12858

To appear in:

Applied Thermal Engineering

Received Date: Revised Date: Accepted Date:

1 January 2018 18 October 2018 24 October 2018

Please cite this article as: Q. Chen, J.R. Jones, R.H. Archer, A Dehumidification Process with Cascading Desiccant Wheels to Produce Air with Dew Point below 0°C, Applied Thermal Engineering (2018), doi: https://doi.org/ 10.1016/j.applthermaleng.2018.10.114

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A Dehumidification Process with Cascading Desiccant Wheels to Produce Air with Dew Point below 0°C Qun Chena,*, Jim R. Jonesb, Richard H. Archera a, Massey Institute of Food Science and Technology, School of Food and Nutrition, Massey University, Private Bag 11-222 Palmerston North 4442, New Zealand b, School of Engineering and Advanced Technology, Massey University, Private Bag 11-222 Palmerston North 4442, New Zealand

* Corresponding author Qun Chen [email protected] Massey Institute of Food Science and Technology, School of Food and Nutrition, Massey University, Private Bag 11-222 Palmerston North 4442, New Zealand Tel: +64 6 951 9291

1

Abstract

Moisture can be removed from air either by desiccant dehumidification or by cooling air to its dew point by refrigeration. In this paper, a new dehumidification process consisting of cascading desiccant wheels is proposed for producing dry air with a dew point lower than 0°C (sub-zero dew point). Heat and mass transfer within the desiccant wheels is mathematically modelled and case study scenarios are used to explore the performance of the process. Energy consumption and cost are compared for dehumidification processes consisting of either a mechanical refrigeration unit, a single desiccant wheel, or of counter current sequential desiccant wheels. To dehumidify moist air (relative humidity 78%) from a dry bulb temperature of -12.5 ºC to a dew point of around -30ºC, the electricity demand for the mechanical refrigeration system (with COPre of 4~6.2) is 4.2~6.6 MJ/kg water removed. In contrast, when two cascading desiccant wheels are used, electricity demand dropped to 0.6 MJ/kg water removed when the second cascade utilised a recuperator and a heater, and 2.6 MJ/kg water removed when the second cascade utilised a heat pump. However, the low electricity demand in the first case is balanced by heat demand, requiring an extra 5MJ/kg water removed in addition to the electricity, but this becomes competitive when the heat/electricity cost ratio falls below 0.3. Two other conventional desiccant systems were compared, which exhibited similar electricity demand but far greater heat demand. In conclusion, cascading desiccant wheels are a cost effective method for advancing dehumidification processes.

Keywords: Desiccant wheel, Dehumidification, Sub-zero dew point, Energy consumption, Refrigeration and air conditioning

2

Highlights

1. An energy-cost effective desiccant wheel dehumidification process is proposed. 2. A case study is conducted to simulate sub-zero dew point air dehumidification. 3. Energy demands for four dehumidification processes are assessed in the case study. 4. The cascading desiccant wheel process features lowest electricity and heat inputs.

3

1 Introduction

Humidity in air is a key determinant of human comfort and health, as well as of equipment stability, industrial process performance and stability of material properties. Air conditioning is so important that HVAC systems reportedly consume up to 11% of the total electricity consumption in the EU [1,2]. In manufacturing facilities in the US, the energy consumption of HVAC systems amounts to 30-40% of total energy [3]. Hence, energy efficiency improvement in HVAC systems is pivotal for energy conservation [4,5].

A typical dehumidification process falls into one of the following two types. Moisture in humid air can be removed either by cooling to the dew point to condense, which is usually achieved by mechanical refrigeration. Or, moisture can be removed by adsorption onto desiccant materials. The moisture-laden desiccant materials then need to be regenerated before recycling back for dehumidification. Both dehumidification approaches require expenditures of energy. Latent heat of moisture condensation and/or heat of sorption is inescapable, and some sensible heat needs to be removed in all practical circumstances. With the growing concerns of energy security and environmental sustainability, more energyefficient dehumidification technologies are desirable.

Dehumidification through condensation provides both dehumidified air and cooling benefits for processes or enclosures. Condensing dehumidifiers are compact, ubiquitous and have been widely applied across industrial and commercial sectors successfully. However, this application of vapour-compression refrigeration has been facing economic and environmental regulatory issues, such as high electricity demand, wear in compressors, phase-out of ozonedepleting or potent greenhouse refrigerants under the Montreal Protocol and Paris Agreement

4

[6]. On the other hand, when the dehumidification requirement does not need the associated cooling, or when the required dew point is unusually low, desiccant dehumidification is the method of choice [7]. More specifically, there exist situations that demand very dry air with a low dew point. For instance, medical and pharmaceutical processes require compressed air to be highly dry as water vapour may be treated as a contaminant [8]. Air used in lithium-ion battery rooms, environmental test chambers, and some processes of the beverage, food, chemical and pharmaceutical industries usually requires sub-zero dew points with some below -40ºC [9,10]. Furthermore, food and pharmaceutical systems can be compromised by the presence of liquid water generated during defrosting coils.

The application of desiccant wheels for dehumidification in HVAC systems has been extensively investigated [11,12,13]. Various models have been developed for simulating the performance of a desiccant wheel [14,15,16,17]. These systems generally employ a single desiccant wheel to reduce the humidity of a moist air stream as it is blown through the adsorption sector of the wheel. As the desiccant wheel turns, it moves from the adsorption sector to the regeneration sector, through which hot air is blown to remove the moisture, then back to the adsorption sector. This inevitably causes the dehumidified air to gain a portion of heat released from moisture adsorption as well as extra sensible heat from the thermal mass of the wheel as it rotates from the hot regeneration sector to the cool dehumidification sector. This can result in an extra need for cooling the dehumidified air, which is often achieved by using a cooling unit or sensible heat recovery wheel [18]. More complex arrangements, i.e., multi-stage or multiple desiccant wheels in series, can achieve the same result with improved efficiency. Ando et al. [19] reported a double stage dehumidification where humid air goes successively through two desiccant wheels alternating with two sensible heat wheels so that dehumidified air leaving each desiccant wheel can be cooled. A similar two-wheel

5

configuration has been employed to exploit recovered waste heat in the shipping applications [20]. More recently, the performance of such multi-stage desiccant conditioning systems has been further studied under various conditions[21,22]. In all cases reported, the total moisture load is shared between multiple wheels arranged in series with respect to the conditioned air.

In this work, a different configuration of desiccant dehumidification process is proposed. The objective of this contribution is to consider a low energy cost dehumidification process using cascading desiccant wheels that can produce dehumidified air with a dew point of -40 ~ 0°C. In this cascading arrangement, one desiccant wheel (the first wheel) produces dry air for an external process or application while one or more other desiccant wheels are employed in cascade to prepare low humidity air for regenerating the previous desiccant wheel. In such a set-up, as distinct from multiple stages in series systems[19-22], it is the moisture which is conveyed from wheel to wheel so that the last, and highest temperature desiccant wheel in the cascade, can be regenerated at high thermal efficiency.

2 Material and methods

2.1 Process configuration

A dehumidification process with cascading desiccant wheels is proposed in this work, as presented in Fig. 1. The process mainly consists of a sequence of desiccant wheels (DW 1 to DW N), where the next wheel in the sequence provides regeneration air for that prior. Each subsequent wheel operates at higher temperature and transfers moisture into a progressively smaller flow of air. Moisture is lifted up and rejected to a final regeneration air loop which is either heated ambient air (Fig 1a and optionally incorporating a recuperator) or a heat pump

6

(Fig. 1b). In Fig. 1, heat pumps are shown simplistically and may require an additional dump condenser. Inevitably some excess sensible heat flows back down the desiccant wheel sequence and is scavenged by a refrigerant evaporator acting on the final conditioned process air.

Back to processes

To the air

Cooler or refrigerant evaporator

From industrial processes A DW 1 D

E

B

Heat exchanger

Heater

DW 2

From ambient air

F

DW 3

DW N

C

(a) Back to processes

Cooler or refrigerant evaporator

Condenser

Compressor

From industrial processes

DW 1

DW 2

DW 3

DW N

Expansion valve

Evaporator

(b) Fig. 1 Configuration of the proposed dehumidification process with cascading desiccant wheels (a) with a heat recuperator and (b) with a heat pump.

The simplest form of the proposed dehumidification process has two desiccant wheels as in Cases 1 and 2 as shown in Fig. 2(a) with a recuperator and Fig. 2(b) with a heat pump. The performance of this dehumidification process with two cascading desiccant wheels is analysed and compared with several other dehumidification processes also depicted in Fig. 2. Several conventional dehumidification methods are presented in Figs. 2(c), (e) and (f) (Cases

7

3, 5 and 6). In Fig. 2(e) and (f), the working air from an external industrial process is cooled to a dew point temperature by condensation onto the surface of one or two evaporator coils. The dehumidified air is then heated to a specified temperature as required in the external process in the refrigerant condenser, which expels the heat that is absorbed in the evaporator. The refrigeration system works as one or two heat pumps where extra heat from the compressor is rejected to ambient. Fig. 2(c) presents a commonly used dehumidification process with one single desiccant wheel. Moist air from an external process is dehumidified in the desiccant wheel. The temperature of the dehumidified air is then modulated according to the requirement of the external process before the air is recirculated back to the process. The desiccant wheel is regenerated by ambient air that first recuperates a portion of sensible heat from the regeneration air exiting the desiccant wheel then is preheated up to a high temperature by a heater.

These simple processes are compared to a dehumidification process composed of counter current sequential desiccant wheels, Case 4 in Fig. 2(d) [19-24]. As shown, the moist working air is first dehumidified through the first desiccant wheel (DW 1), cooled to increase the relative humidity and then further dehumidified through the second desiccant wheel (DW 2). The temperature of the dehumidified working air is conditioned again by a refrigeration evaporator before the air is recirculated back to the external processes. DW 2 and DW 1 are sequentially regenerated by a stream of hot air in a counter-current direction to the flow of the working air. These two desiccant wheels effectively operate as a single wheel of double depth with integral cooling.

8

To the air 9

Cooler or refrigerant evaporator

Back to processes

3

Heater

Heat exchanger 6 From ambient air

2 7

From industrial processes 1

Back to processes

High temperature desiccant wheel

5

Refrigerant condenser

2

Compressor

7

From industrial processes 1

8

Low temperature desiccant wheel

Cooler or refrigerant evaporator

3

8 High temperature desiccant wheel 5

Low temperature desiccant wheel

(a) Case 1

(b) Case 2

3

Back to processes

2

From industrial processes 1

Refrigeration evaporator

Desiccant wheel

Heater

Refrigerant evaporator

4

4

Back to processes

Expansion valve 6

Refrigerant evaporators

5

From industrial processes 1

DW #1

2

3

9

6

4 DW #2

8

7

5 Heater To the air

Heat exchanger

Heat exchanger 10

4

(c) Case 3 3

1

(d) Case 4 Back to processes 5

Refrigeration condenser

Compressor From industrial processes

Expansion valve

To the air

From 6 ambient air

From ambient air

Back to processes

Heater

7

Refrigeration condenser 1

Compressor 1 From industrial processes

2

1

Refrigeration evaporator

Refrigeration condenser 2

4

Expansion valve 1 Compressor 2

Refrigeration evaporator 1

2

3 Expansion valve 2

Refrigeration evaporator 2

(e) Case 5 (f) Case 6 Fig. 2 Configuration of the dehumidification processes, (a) cascading desiccant wheels with a heat recuperator, (b).cascading desiccant wheels with a heat pump, (c) single desiccant wheel, (d) counter current sequential desiccant wheels, (e) dehumidification by refrigeration (one heat pump), (f) dehumidification by refrigeration (two heat pumps)

2.2 Mathematical modelling

9

The performance of the dehumidification processes is explored through numerical analysis. Specifically for the desiccant wheels, differential governing equations of mass and energy conservation are solved, based on the following assumptions [16, 17]: a) The transport of moisture and energy only occurs at the interface between the air flow and the desiccant, i.e., only gas-side resistance is considered (GSR model [25]). As the size of desiccant material is very small, the resistance of heat and mass transfer within the solid desiccant particles could be neglected [26]. b) Due to the small size of the solid desiccant, the adsorbed moisture concentration could reasonably be assumed to be in local equilibrium with the air close to it [25,27]. c) The air flow in the wheel channels is assumed to be one-dimensional steady laminar flow (Re < 200, which is much less than the upper limit of Reynolds number, 2000~2300, for a laminar flow in a pipe [28]). d) Lewis number is assumed equal to unity (Le=1) for the air flow [26].

The mass balance of moisture in the air flow can be described as,

u a  a Ac

X a  , z   k m Pc  X d  X a  z

(1)

where Xa (kg/kg d.a.) is the humidity of air, ua (m/s) the air velocity, Ac (m2) the crosssectional area of a wheel channel, ρa (kg/m3) the density of air, km (kg/m2s) mass transfer coefficient, and Pc (m) the perimeter of a wheel channel. Xd is the humidity of air close to and in local equilibrium with the surface of the desiccant. For the mass balance of moisture in the desiccant material,

 d Ad 

Wd  , z   k m Pc  X d  X a  

(2)

where ρd (kg/m3) the density of desiccant material, Ad (m2) the area of desiccant material and ε the mass fraction of desiccant material. The moisture content Wd is determined by the 10

adsorption isotherm of the desiccant material. For silica gel in the high temperature desiccant wheel (HTDW), the moisture adsorption isotherm is described by the polynomial [16,29]:

Wd  0.0059  1.0377  5.0667 2  23.5 3  48.8 4  44.6 5  14.9 6

(3)

For 3A molecular sieve used in the low temperature desiccant wheel (LTDW), the isotherm is obtained from fitting the adsorption curve at 0°C [30]:

Wd  0.02 ln    0.23

(4)

where, the relative humidity  can be derived from the absolute humidity as,



Xd patm X d  M w M a ps

(5)

where patm (Pa) is the atmospheric pressure, ps (Pa) the saturation pressure of moisture, Mw and Ma (kg/mol) the molecular weight of moisture and air, respectively.

In general, the adsorption isotherm for a desiccant material is a function of relative humidity and temperature. Of these, relative humidity exerts much more significant effects on the moisture content of hygroscopic materials than temperature [31]. In the temperature and humidity ranges that this work concerns, it is reasonable to simplify the dependence of the adsorption isotherms merely on relative humidity [29,31].

Equations of energy conservation for both the air flow and the desiccant material can be expressed as, u a  a Ac

 d Ad



  k P T

 Ta 

(6)

  k P T

 Td   k m H ad Pc  X a  X d 

(7)

 c pa  X a c pv Ta  , z  z



h

c

d

 c pd  Wd c pw Td  , z  

h

c

a

where Ta and Td (K) are the temperature of air and desiccant material, kh (W/m2·K) the heat transfer coefficient, cpa, cpv, cpd, and cpw (kJ/kg·K) the specific heat of air, water vapour, 11

desiccant material and water, respectivelty. Had (kJ/kg) is the heat of moisture adsorption. In the model, its value is approximately equal to the latent heat of vapourisation of water [32].

The initial and boundary conditions for the cascading desiccant wheels are as follows. For the low temperature desiccant wheel,

X 1,ap  ,0  X 1,api ; T1,ap  ,0  T1,api ; W1,dp 0, z   W1,dr  r , Z  z  ; T1,dp 0, z   T1,dr  r , Z  z  For the high temperature desiccant wheel,







; T2,ar  ,0  T2,ari ; W2,dr 0, z   W2,dp  p , Z  z ; T2,dr 0, z   T2,dp  p , Z  z



The model has been validated for a single desiccant wheel and shows good agreement with experimental data [16,17]. The geometry of desiccant wheel channels and other properties are similar to those presented by Chung et al [16]. As depicted in Fig. 3, each sinusoidal channel in the desiccant wheels is 1.75 mm in height and 3.5 mm in width with the hydraulic diameter of 1.42 mm [16]. The desiccant wall thickness is 0.15 mm.

Fig. 3 Schematic of the desiccant channels (a) front view, (b) side view.

The pressure loss (ΔP) across the desiccant wheels consists of the pressure drop through the desiccant channels and pressure drops due to the channel entrance and exit, which can be calculated by [33],

12

2

  1 4c   f L   P   a u a2 1   1.5  2  Dh   Re Dh 

(8)

where the friction factor f for the fully developed laminar flow in the desiccant channels is approximately 44.7 [34], with L (m) the thickness of the desiccant wheel, Dh (m) the hydraulic diameter of a desiccant channel and c (m) the desiccant thickness.

2.3 Case studies

A case study is carried out for evaluating the cascading desiccant wheel dehumidification process in comparison with other configurations as shown in Figs. 2(b-d). Without losing generality, the process parameters are assumed according to a food processing facility [5], as listed in Table 1. For simplicity, the efficiency of the heat exchanger is assumed to be 80% in the case study as it is generally in the range of 50~80% [35]. The ambient temperature is set at 28°C with relative humidity of 65% [36].

Table 1 Parameters for the case study Air from an external industrial process Temperature Tapi -12.5 Humidity Xapi 1.0 Dew point Tdew -15.2 Air flow rate 10 ap Required dehumidified air back to the external process Temperature Tapo -10.0 Humidity Xapo 0.20 Dew point Tdew -31.4 Regeneration air from the ambient Air flow rate for Cases 1 and 2 1.11 ar Air flow rate for Cases 3 and 4 3.33 ar Ambient temperature Tari 28 Humidity Xari 15.6 13

°C g/kg d.a.* °C kg/s °C g/kg d.a. °C kg/s kg/s °C g/kg d.a.

Relative humidity

ari

65

%

Desiccant wheels P/R ratio(wt/wt)

3:1

Rotation speed

5

rev/hour

Superficial air velocity**

ua

1

m/s

Wheel depth

L

150

mm

Channel height

a

1.75 [16]

mm

Channel width

b

3.5 [16]

mm

Desiccant thickness

c

0.15 [16]

mm

HTDW in Cases 1 and 2

Silica gel

LTDW in Cases 1 and 2, Cases 3 and 4

3A molecular sieve ε

0.7 [16]

Heat capacity

cpd

0.921 [16]

Density

ρd

720 [16]

Desiccant fraction

kJ/kg·K kg/m3

*d.a.: dry air **Normalised to normal conditions (20°C, 1atm)

3 Results and discussion

This section presents the performance of the dehumidification with cascading desiccant wheels in the case study based on numerical simulation. The results are compared with those of a refrigeration dehumidification system, a single desiccant wheel and two counter current sequential desiccant wheels dehumidification systems.

3.1 Dehumidification process parameters

Although the mathematical model of the heat and mass transfer process in desiccant wheels has been validated by other researchers [16,17], the model is further verified in this work. The performance of a single desiccant wheel has been simulated to compare with data presented by a manufacturer[37] and Angrisani et al [11], as shown in Fig. 4. As can be seen, 14

the simulation results agreed very well (relative errors within 5%) with the data reported in the literature. This implies that the mathematical model is reasonable for analysing the performance of the dehumidification configurations.

(a)

(b)

Fig. 4 Comparison of simulation results for a single desiccant wheel with experimental data, (a) from a manufacturer brochure [37] and (b) from Figs. 4 and 5 in [11]

Table 2 lists the major stream properties in the six dehumidification configurations depicted in Fig. 2. The state of these streams for Cases 1-4 is shown in psychrometric charts (Fig. 5). Dehumidification through a refrigeration system (Cases 5 and 6) is straightforward. The moisture-laden working air is directly cooled to the dew point of -31.4°C through refrigerant evaporator coils where extra moisture turns to frost. The dehumidified working air is heated up to -10°C through refrigerant condenser coils where the air gains the heat that is absorbed and carried over by the refrigerant from the evaporator. Case 6 is an advance over Case 5 in that two heat pumps are used to share the load, each working over a reduced temperature range.

When desiccant wheels are used for dehumidification, as in Cases 1 – 4, the working air gains heat through the thermal mass of desiccant wheels carrying over both heat released from moisture adsorption and a part of the sensible heat from the regeneration air. In Cases 1 and 2 15

where cascading desiccant wheels are applied, the low temperature desiccant wheel is regenerated by the intermediate air on temperature below 18°C (Figs. 4a and 4b). In the low temperature desiccant wheel, nearly 60% of the heat released by the intermediate air is used to desorb moisture from the desiccant material, the balance being sensible heat carried over from the regeneration sector to the dehumidified working air. In Case 3, the single desiccant wheel that is used to dehumidify the working air is regenerated by the hot regeneration air of 125°C. The energy for moisture desorption employs only 30% of the heat extracted from the regeneration air. This corresponds to a significant amount of sensible heat carry-over from the regeneration side to the dehumidified working air. Consequently, the temperature of the dehumidified air rises to -6.3°C after it exits the desiccant wheel. In Case 4, the working air is dehumidified through DW1 and DW2 sequentially. Nearly 80% of the moisture removed is adsorbed through DW1. This desiccant wheel is reactivated by the regeneration air of approximately 68°C, which contributes 35% of the change in sensible heat for desorption of moisture from the desiccant material. The rest of the moisture is removed from the working air into the regeneration air by DW2, which uses only 17% of the sensible heat released by the regeneration air for desorption of moisture from the desiccant material. Comparing these show that desiccant wheel cascading systems avoid undue carryover of the sensible heat load.

Table 2 Steady state process variables of the dehumidification processes in the case study

L*

1 2 3 4 5 6

Case 1 X h

Case 2 T X h

Case 3 T X h

Case 4 T X h

Case 5 T X

°C

g/kg kJ/kg d.a. d.a.

°C

°C

°C

°C

-12.5 -9.1 -10.0 6.0 16.2 28.0

1.00 -9.98 0.20 -8.57 0.20 -9.44 3.36 14.6 0.97 18.9 15.6 68.1

-12.5 -9.5 -10.0 3.9 12.8 16.6

T

g/kg kJ/kg d.a. d.a. 1.00 0.29 0.29 3.18 1.03 10.0

-9.98 -8.78 -9.23 12.0 15.6 42.1

-12.5 -6.3 -10.0 28.0 125 106

g/kg kJ/kg d.a. d.a. 1.00 0.20 0.20 15.6 15.6 17.9

-9.98 -5.72 -9.41 68.1 169 156

16

-12.5 -8.2 -11.9 -9.6 -10.0 28.0

g/kg kJ/kg d.a. d.a. 1.00 0.41 0.41 0.25 0.25 15.6

-9.98 -7.17 -10.9 -8.89 -9.33 68.1

h

g/kg kJ/kg d.a. d.a.

T °C

-12.5 1.00 -9.98 -12.5 -31.4 0.20 -31.0 -21.8 -10.0 0.20 -9.44 -31.4 -22.3 -10.0

Case 6 X

h

g/kg kJ/kg d.a. d.a. 1.00 0.54 0.20 0.20 0.20

-9.98 -20.5 -31.0 -21.8 -9.44

7 75.0 15.6 117 60.0 10.0 86.6 8 45.0 22.7 104 33.4 16.4 75.8 9 33.0 22.7 91.5 10 *L is the stream number depicted in Fig. 2.

33.0 17.9 79.3 -

68.1 61.1 48.5 33.0

15.6 16.1 17.8 17.8

110 104 95.1 79.0

Humidity, g/kg air

(a)

Dry-bulb temperature, C

+ Humidity, g/kg air

(c)

Dry-bulb temperature, C

17

Fig. 5 Psychrometric charts of key process streams in (a) Case 1, (b) Case 2, (c) Case 3, and (d) Case 4.

3.2 Energy cost

Dehumidification is an energy intensive process. Electrical power is required for the operation of the refrigeration units shown in Fig 2 as well as to drive fans to overcome pressure loss through desiccant wheels, and a further smaller amount of electrical energy to turn the desiccant wheels. In addition, heat is required to regenerate the desiccant wheels, shown labelled as heaters in Figs. 2(a, c and d). Heat maybe supplied by electrical heating or an alternate heat source, e.g., combustion, which is why it is differentiated from electrical energy for this analysis. The total electricity and heat demands for all cases are given in Table 3.

Cases 1-4 with desiccant wheel dehumidification require fans to circulate the air through the drying chamber and across the desiccant wheels. According to Eq. (8), the pressure drop of air flow across a desiccant wheel is approximately 90 Pa, close to the lower limit of those

18

reported in the literature [33]. Assuming the fan efficiency to be 70% [38], the required electrical energy for air handling is able to be calculated, as shown in Table 3. The electricity for turning a desiccant wheel is only approximately 6 kJ/kg water removed and is neglected in this work. Electrical energy is also needed to run the refrigeration units in cases 1-4. In Case 1, the refrigerant evaporation temperature is assumed to be -15°C and the condensing temperature 33°C because the required dehumidified working air temperature is -10°C and the ambient temperature is 28°C. The COPre for the refrigeration unit is then estimated to be 2.7 [39]. For Case 1 with the proposed cascading desiccant wheels and a recuperator, the sensible heat to be removed from the dehumidified working air is only 0.85 kJ/kg d.a. Thus, the electricity required by the refrigeration unit is 0.40 MJ/kg water removed. In Case 2, while the electricity consumed by the refrigeration unit for cooling the working air is only 0.24 MJ/kg water removed, the heat pump has an additional requirement of 2.11 MJ/kg water removed. In the closed regeneration cycle, the heat pump can be reasonably assumed to work between 10°C (refrigerant evaporation temperature) and 65°C (refrigerant condensing temperature) with the COPre of 2.5[39]. In Case 3 where the dehumidification process consists of one single desiccant wheel, the sensible heat extracted from cooling the working air from -6.2°C down to -10°C is 3.77 kJ/kg d.a, and so the required electricity is 1.73 MJ/kg water removed for collecting this amount of heat through a refrigeration unit. In Case 4, the working air is first cooled from -7.5°C down to -12.4°C after being dehumidified by DW1 and then cooled from -9.6°C down to -10°C. The total sensible heat to be recovered through the refrigerant evaporator is 5.31 kJ/kg d.a, and so the electricity consumption for the refrigeration unit is 2.03 MJ/kg water removed.

For comparison, in the refrigeration system of Case 5, the coefficient of performance of refrigeration (COPre) is estimated to be 4.0 [39]. As a result, the electricity needed for

19

dehumidification by the refrigeration unit is 6.6 MJ/kg water removed. A two-stage heat pump or multiple heat pumps in an optimal configuration could offer higher COP re. In Case 6, for instance, the overall COPre of two single-stage heat pumps cycling between -35 and 18°C, and between -27 and -5°C could reach 6.2 (7.0 and 5.6 for each respectively, as shown in Table 2). The electricity consumption for dehumidification is 4.22MJ/kg water, lower than Case 5 but still higher than Case 2. In both refrigeration systems, the extra heat from the compressors needs to be rejected to ambient. A down side of Cases 5 and 6 is that the dehumidification process must be halted to defrost the evaporator coil. The heat required for defrosting is neglected in these two case.

Heat energy, as distinct from electrical energy, is required for preheating the regeneration air in three of the cases with desiccant wheel dehumidification, Cases 1, 3 and 4 (calculated from the data listed in Table 2 and presented in Fig. 5), noting that Case 2, which employs a heat pump, does not have a heat demand. Although some 80% of sensible heat is assumed to be recuperated from the regeneration air after it leaves the desiccant wheels, net heat demands for desiccant wheel dehumidification processes are significant, among which Case 1 is the lowest.

Table 3 Energy consumption for the dehumidification processes Case 1 Case 2 Case 3 Case 4 Case 5 COPre 2.7 2.7 and 2.5 2.7 2.7 4.0 Electricity for refrigeration 0.40 0.24+2.11 1.73 2.03 6.56 Electricity for fans 0.23 0.26 0.17 0.39 Total electricity demand 0.63 2.61 1.90 2.42 6.56 Heat consumption 5.39 16.6 12.6 All electricity demands and heat consuptions are expressed as MJ/kg water removed

Case 6 7.0 and 5.6 4.22 4.22 -

Heat and electricity demands given in Table 3 are used to estimate the energy costs for the five dehumidification processes shown in Fig. 2. The energy costs depend heavily on the 20

supply prices of heat and electricity. For example, heat supply from burning natural gas can be assumed to have an energy efficiency of 80%. Note that the EU-28 average price of natural gas for industrial consumers was €8.3/GJ (i.e. €30/MWh) and that of electricity was €31.7/GJ (or €114/MWh) in 2016 [40]. In the North Island of New Zealand, the annual average price of natural gas for industrial use in 2016 was NZ$6.1/GJ (NZ$22/MWh) and that of the electricity was NZ$31.4/GJ (NZ$113/MWh) [41]. Fig. 6 compares the energy costs among the cases of all the dehumidification processes on the basis of different scenarios. Dehumidification systems with desiccant wheels (Cases 1~4) have lower energy costs than refrigeration dehumidification (Case 5) in the North Island of New Zealand. However, in the European Union, the refrigeration dehumidification is more advantageous over the process with a single desiccant wheel because the price ratio of natural gas to electricity in EU is higher than that in the North Island of New Zealand. If the heat is assumed to be supplied in the form of steam, considering the steam cost of US$45/MWh and electricity cost of US$120/MWh[42], the energy costs for all the five cases are in increasing order , Case 2 < Case 1 < Case 6 < Case 5 < Case 4 < Case 3.

21

Fig. 6 Energy consumption and cost comparison of the dehumidification processes. The histograms are the energy consumption per kg water removed and the points are the costs in North Island New Zealand, the European Union and using steam as the heat source where steam is produced using coal.

Nonetheless, it is evident that the energy cost of the proposed dehumidification process of cascading desiccant wheels (Cases 1 and 2) is the lowest among all the cases of interest. In particular, the energy cost for the proposed dehumidification process of cascading desiccant wheel with a heat pump regeneration cycle (Case 2) is approximately 60% lower than the refrigeration dehumidification system (Case 5) due to the constantly lower electricity consumption. Although both Case 2 and Case 5 use heat pumps, they operate in different temperature ranges with different cooling loads. Because the heat pump in Case 5 works at lower temperature with higher cooling load for the evaporator coils, the volumetric flow rate of refrigerant vapour is about 20 times of that in Case 2 (using R134a as the refrigerant). Thus, much larger sizes of coils and compressor are required in Case 5 than in Case 2 despite the compression ratio for the latter being slightly higher (about 3.8 and 4.7, respectively [38]). As a result, the equipment cost of desiccant wheels in Case 2 is offset by the reduced cost of the heat pump compared with Case 5. Moreover, periodic defrosting is needed for the evaporator coils of the heat pump in Case 5.

As previously noted, Case 2 is a closed loop system while Case 1 uses ambient air for regeneration. The closed loop enables a non-air atmosphere for oxygen sensitive substrates. If such a closed loop is not necessary, Case 1 becomes cost competitive with Case 2 if the price ratio of heat to electricity is less than 0.3. Indeed, when waste heat from industrial processes is easily accessible at low cost, the energy costs of all the desiccant

22

dehumidification processes (Cases 1, 3 and 4) become significantly lower than the refrigeration dehumidification (Case 5). Because cascade desiccant dehumidification systems require both heat and electricity, they are amendable to be coupled to a small-scale combined heat and power (CHP) unit, for example, an internal combustion engine with a generator. Thus, the proposed dehumidification process becomes a stand-alone system when it is coupled with the CHP unit. It should be noted that the current work is based on mathematical model simulation. Experimental work is required in next stage to reveal the details of the process performance so that the operating conditions could be further optimised.

4 Conclusions A new configuration for desiccant dehumidification is reported, where all wheels subsequent to the first are a regeneration cascade that progressively concentrates the moisture inot hotter, more humid and smaller flow until exhaust. The last wheel is regenerated at high thermal efficiency. The simplest cascade, consisting of two wheels, were numerical simulated as Case 1 with a recuperator and heater on the second cascade, and Case 2 with a heat pump on the second cascade. They were compared to other conventional dehumidification approaches of a single desiccant wheel (Case 3), a pair of wheels with counter current flow (Case 4) and mechanical refrigeration (Case 5). The cascading wheels exhibited the lowest the electricity consumption and in heat demand. Case 2 did not require any additional heat. Its electricity demand is 43% of a standard mechanical refrigeration system (Case 5). When the costs of heat and electricity are differentiated, the effect of the cheaper cost of producing heat advantages Case 1 over Case 2 when the heat to electricity cost ratio falls below 0.3. Neither of the other conventional desiccant systems (Cases 3 and 4) was competitive with the cascade systems (Cases 1 and 2). Therefore, the proposed dehumidification process with cascading

23

desiccant wheels is promising, in terms of energy consumption and cost, for the production of dry air with dew point lower than 0°C.

Acknowledgements This work was supported by FIET programme from the Ministry of Business, Innovation and Employment (MBIE), New Zealand. The authors thank Prof D J Cleland for his comments on the manuscript.

Nomenclature

M patm ps Pc Re T u W X z Z

cross-sectional area, m2 desiccant channel wall thickness, m specific heat capacity, kJ/kg·K hydraulic diameter of desiccant channels, m friction factor of desiccant channels specific enthalpy, kJ/kg heat transfer coefficient, kW/m2·K mass transfer coefficient, kg/m2s desiccant wheel depth, m mass flow rate, kg/s molecular weight, kg/mol atmospheric pressure, Pa saturated pressure of moisture, Pa perimeter of the wheel channel, m Renolds number temperature, K air velocity, m/s moisture content in the desiccant material absolute humidity, kg/kg air axial coordinate along the wheel depth, m depth of the desiccant wheel, m

ε η

mass ratio of the desiccant material effectiveness parameter

A c cp Dh f h kh km L

24

ρ τ

density, kg/m3 time, s relative humidity

Subscripts 1 2 a c d i o p r v w

low temperature desiccant wheel High temperature desiccant wheel air channel of the desiccant wheel desiccant material inlet outlet process stream for a desiccant wheel regeneration air for a desiccant wheel water vapour water

References

[1] P. Bertoldi, B. Atanasiu. Electricity Consumption and Efficiency Trends in European Union: Status Report 2009. Luxembourg: Joint Research Centre, EC. [2] I. Knight. Assessing electrical energy use in HVAC systems. REHVA Journal. 2 (2012) 611. [3] K. Chinnakani, A. Krishnamurthy, J. Moyne, A. Arbor, F. Gu. Comparison of energy consumption in HVAC systems using simple ON-OFF, intelligent ON-OFF and optimal controllers. 2011 IEEE Power and Energy Society General Meeting. [4] S.Y. Ahmed, P. Gandhidasan, A. Al-Farayedhi. Pipeline drying using dehumidified air with low dew point temperature. Applied Thermal Engineering. 18 (1998) 231-244. [5] I.C. Claussen, T.S. Ustad, I. Strommen, P.M. Walde Atmospheric freeze drying - a review. Drying Technology. 25 (2007) 947-957.

25

[6] W. Hwang, S. Choi, D. Lee. In-depth analysis of the performance of hybrid desiccant cooling system incorprated with an electric heat pump. Energy. 118 (2017) 324-332. [7] N. Subramanyam, M.P. Maiya, S.S. Murthy. Application of desiccant wheel to control humidity in air-conditioning systems. Applied Thermal Engineering. 24 (2004) 2777-2788. [8] S.P. Denyer, R.M. Baird. Guide to Microbiological Control in Pharmaceuticals and Medical Devices. 2nd ed. Boca Raton: CRC Press, 2006. [9] H. Leuenberger. Spray freeze-drying - the process of choice for low water soluble drugs? Journal of Nanoparticle Research. 4 (2002) 111-119. [10] J. Zhao, Z. Lu, N. Liu, H.-W. Lee, M.T. McDowell, Y. Cui. Dry-air-stable lithium silicide-lithium oxide core-shell nanoparticles as high-capacity prelithiation reagents. Nature Communications. 5 (2014) 5088. [11] G. Angrisani, A. Capozzoli, F. Minichiello, C. Roselli, M. Sasso. Desiccant wheel regenerated by thermal energy from a microcogenerator: experimental assessment of the performances. Applied Energy. 88 (2011) 1354-1365. [12] K. Daou, R.Z. Wang, Z.Z. Xia. Desiccant cooling air conditioning: a review. Renewable and Sustainable Energy Reviews. 10 (2006) 55-77. [13] D. La, Y. Dai, Y. Li, R. Wang, T. Ge. Technical development of rotary desiccant dehumidification and air conditioning: a review. Renewable and Sustainable Energy Reviews. 14 (2010) 130-147. [14] L.A. Sphaier, W.M. Worek. Analysis of heat and mass transfer in porous sorbents used in rotary regenerators. International Journal of Heat and Mass Transfer. 47 (2004) 3415-3430. [15] T.S. Ge, Y. Li, R.Z. Wang, Y.J. Dai. A review of the mathematical models for predicting rotary desiccant wheel. Renewable and Sustainable Energy Reviews. 5 (2008) 1485-1528.

26

[16] J.D. Chung, D.-Y. Lee, S.M. Yoon. Optimization of desiccant wheel speed and area ratio of regeneration to dehumidification as a function of regeneration temperature. Solar Energy. 83 (2009) 625-635. [17] S. De Antonellis, C.M. Joppolo, L. Molinaroli. Simulation, performance analysis and optimization of deisccant of deisccant wheels. Energy and Buildings. 42 (2010) 1386-1393. [18] D.B. Jani, M. Mishra, P.K.Sahoo. Solid desiccant air conditioning – A state of the art revew. Renewable and Sustainable Energy Revews. 60 (2016) 1451-1469. [19] K. Ando, A. Kodama, T. Hirose, M. Goto, H. Okano. Experimental study on a process design for adsorption desiccant cooling driven with a low-temperature heat. Adsorption. 11 2005) 631-635. [20] G. Zheng, C. Zheng, G. Yang, W. Chen. Development of a new marine rotary desiccant air-conditioning system and its energy consumption analysis. Energy Procedia. 16 (2012) 1095-1101. [21] T.S. Ge, Y.J. Dai, R.Z. Wang, Y. Li. Performance of two-stage rotary desiccant cooling system with different regeneration temperatures. Energy. 80 (2015) 556-566. [22] M. Gadalla, M. Saghafifar. Performance assessment and transient optimisation of air precooling in multi-stage solid desiccant air conditioning systems. Energy Conversion and Management. Energy Convrsion and Management. 119 (2016) 187-202. [23] E. N. Jensen, T.R. Olesen. Patent No. US 2016/0250583 A1, 2016 [24] R. Tu, X. Liu, Y. Jiang. Performance analysis of two-stage desiccant cooling system. Applied Energy. 113 (2014) 1562-1574. [25] J.D. Chung. Modelling and analysis of desiccant wheel. in N. Enteria, H. Awbi, H. Yoshino. Desiccant Heating, Ventilating, and Air-conditioning Systems. Singapore: Springer. 2017

27

[26] R.S. Barlow. Analysis of the Adsorption Process and of Desiccant Cooling Systems – A Pseudo- Steday-State Model for Coupled Heat and Mass Transfer. Golden: Solar Energy Research Institute. 1982. [27] C.E.L. Nobrega, N.C.L. Brum. Desiccant-Assisted Cooling: Fundamental and Applications. London: Springer. 2014. [28] F.M. White. Fluid Mechanics. New York: McGraw-Hill, 2009. [29] A. Ramzy, R. Kadoli, A. Babu. Experimental procedure to develop the isotherm equation for moisture adsorption on silica gel particles. Journal of Multidisciplinary Engineering Science and Technology. 2 (2015) 71-76. [30] R. Lin, A. Ladshaw, Y. Nan, J. Liu, S. Yiacoumi, C. Tsouris, D.W. DePaoli, L.L. Tavlarides. Isotherms for water adsorption on molecular sieve 3A: inlfuence of cation composition. Industrial & Engineering Chemistry Research. 54 (2015) 10442-10448. [31] S. Weintraub. Demystifying silica gel. In Object Specialty Group Postprints 9, American Institute for Conservation, 2002. Available at http://www.apsnyc.com/uploads/demystifying%20silica%20gel%20-%20updated.pdf. [32]A. Kodama, T. Hirayama, M. Goto, T. Hirose, R.E. Critoph. The use of psychrometric charts for the optimisation of a thermal swing desiccant wheel. Applied Thermal Engineering. 21 (2001) 1657-1674. [33] M. Goldsworthy, S. White. The performance of desiccant wheels for desiccant airconditioning. In C. E. L. Nobrega, N. C. L. Brum ed. Desiccant-Assisted Cooling: Fundamentals and Applications. London: Springer, 2014. [34] L.Z. Zhang, J.L. Niu. Performance comparisons of desiccant wheels for air dehumidification and enthalpy recovery. Applied Thermal Engineering. 22 (2002) 13471367. [35] Carbon Trust. Heat recovery: A guide to key systems and applications. 2011.

28

[36] R. J. Love, D. J. Cleland, I. Merts. B. Eaton. What is the optimum compressor discharge pressure set point for condensers? EcoLibrium. 2005. [37] Seibu Giken Co., Ltd. Brochure of “DRY-SAVE”desiccant dehumidifier rotors (2000). https://www.sgamerica.com/product-docs. Last accessed in August 2018. [38] P.G. Schild, M. Mysen. Recommendations on Specific Fan Power and Fan System Efficiency. AIVC Technical Note 65. 2009. [39] IPU and Technical University of Denmark. CoolPack, Version 1.50. 2017. [40] Eurostat. Eurostat: Statistics Explained. Retrieved from http://ec.europa.eu/eurostat/statistics-explained/index.php/Main_Page. Last accessed in June 2017. [41] MBIE. Ministry of Business, Innovation and Employment: Prices. Retrieved from http://www.mbie.govt.nz/info-services/sectors-industries/energy/energy-datamodelling/statistics/prices. Last accessed in May 2017. [42] T.G. Walmsley, M.R.W. Walmsley, M.J. Atkins, J.R. Neale. Thermo-economic optimisation of industrial milk spray dryer exhaust to inlet air heat recovery. Energy. 90 (2015) 95-104.

29

Highlights

1. An energy-cost effective desiccant wheel dehumidification process is proposed. 2. A case study is conducted to simulate sub-zero dew point air dehumidification. 3. Energy demands for four dehumidification processes are assessed in the case study. 4. The cascading desiccant wheel process features lowest electricity and heat inputs.

30