A novel defrosting method in gasoline vapor recovery application

Accepted Manuscript A Novel Defrosting Method in Gasoline Vapor Recovery Application

Jierong Liang, Li Sun, Tingxun Li PII:

S0360-5442(18)31708-0

DOI:

10.1016/j.energy.2018.08.172

Reference:

EGY 13648

To appear in:

Energy

Received Date:

08 February 2018

Accepted Date:

22 August 2018

Please cite this article as: Jierong Liang, Li Sun, Tingxun Li, A Novel Defrosting Method in Gasoline Vapor Recovery Application, Energy (2018), doi: 10.1016/j.energy.2018.08.172

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.

ACCEPTED MANUSCRIPT

A Novel Defrosting Method in Gasoline Vapor Recovery Application Jierong Lianga,b , Li Sunc, Tingxun Lia,* , a

School of Engineering, Sun Yat-sen University, West XINGANG Road 135,Guangzhou, China, 510275 b Guangdong Shenling Environmental System Co., Ltd., XINGLONG 10th Road 8, Foshan, China, 528313 c Department of Chemical and Biochemical Engineering, Technical University of Denmark, 2800 Kongens Lyngby, Denmark *Corresponding author. Tel.: +86 020 3933 1115. E-mail address: [email protected] (Tingxun Li), [email protected]. Postal address: No. 135, Xingang West Rd., Guangzhou City 510725, Guangdong PR China. Abstract: Condensation method is comprehensively applied for gasoline vapor recovery (GVR), of which frosts in the heat exchanger is the greatest challenge, especially for the continuous long running cases. A novel dual channel GVR cascade refrigeration system with shell-tube heat exchanger was presented and tested in this paper. With onework-one-standby evaporator settings, combined with refrigerant evacuation and delay switching strategies, the defrosting of low temperature shell-tube heat exchanger was analyzed and solved. Also multi-stage cycle was introduced to supply three cooling stage, which cooled the gasoline vapor from ordinary temperature to about 70°C. By the means of industrial application validation and process calculation, the ability of the non-stop cooling during defrosting was verified. The refrigerant evacuation was proposed to prevent high pressure drop caused by frost accumulation, which also improved the cooling capacity by 28.2% and approached the defrost efficiency of 55.4%. In addition, it was found that delay switching can effectively reduce the capacity fluctuation. Based on sensitivity studies, 20 minutes delay was identified as the best switching timing for this device. The capacity of this system performed lower reduction, higher duty ratio and defrost efficiency. Keywords：Gasoline vapor, Dual channel, Defrost, Cascade cycle, Shell-tube

1. Introduction Ministry of Environmental Protection of China had stepped up efforts to control gaseous nonmethane hydrocarbons (NMHC) emitted from petrochemical industry, chemical terminals, and oil depots. Gasoline vapor containing volatile organic compounds (VOCs) was rooted in loading, unloading and storage operations (Fig.1), which not only resulted in air pollution and potential fire risks, but also brought about huge economic losses [1]. Many of separation methods have been developed for GVR, which include adsorption [2-3], absorption [4], condensation [5], and membrane [6]. Each of these methods was used in a certain range of flow rate and NMHC concentrations for 1

ACCEPTED MANUSCRIPT

economical consideration. The high concentration gasoline vapor was normally recovered by the condensation separation method for the advantage of convenient installation and direct resources recovery [5].

Fig.1 Schematic of VOCs emitting and its recovery

Currently low temperature gas cooling process mainly focused on liquefied natural gas (LNG, [7]) and cryogenic air separation process [8]. A limited number of studies have provided insight into the special GVR cooling process. Shie et al. [9] discussed the relationship of recovery efficiency and refrigeration temperature and concluded that treated gas should be refrigerated to about -73°C to meet the vapor recovery efficiency of 90%. Shi and Huang [5] presented a three-stage condensation system for industrial GVR, and compared the total cooling duties between single-stage and three-stage cooling processes. These studies prepared the ground for cooling temperature and stage settings of GVR. But the detailed understanding of the corresponding refrigeration cycle performance efficiency and stability is important for GVR applications. A three-stage refrigeration cycle for pure propane liquefaction was analyzed by Kalantar-Neyestanaki et al. [10], but not related to defrosting issues. To the best of our knowledge, no publication presented experimental or industrial investigation on the cooling cycle behaviors and especially defrosting control strategy for GVR systems. A three-stage cooling process of GVR (Fig.2) was designed as pre-cooling (1st stage), middlecooling (2nd stage) and super-cooling (3rd stage) with multi-temperature cycles. Cascade cycle was adopted to meet the cooling requirement of lower than -70°C, which was classified as classic and auto cascade. A classical cascade included two or more separate cycles with different refrigerant, while auto-cascade system was driven by a single compressor, using zeotropic mixture as refrigerant. Since auto-cascade was more sensitive against the classical cascade system [11], classical cascade

2

ACCEPTED MANUSCRIPT

system [12-13] was chosen in this work. Shell-tube heat exchanger was chosen as the evaporator for the highly required safety in explosive environment.

Fig.2 Temperature profile of cooling process

In contrast to LNG process, GVR cooling process has no pre-process such as dehydration [14-15], the vapor might contain multiple heavy ingredients such as benzene and water. These heavy components might freeze or frost during cooling process. According to demand of long continuous time working in shipping or tank farm GVR application [1], there is no interruption opportunity to remove the frost regularly. Frosting was the greatest obstacle which seriously penalized the device efficiency and functionality [16]. Therefore, defrosting method is essential for uninterrupted cooling of GVR. All defrosting methods have the common disadvantage of reducing the overall cooling efficiency because the energy is used for defrosting but not cooling. For the purpose of uninterrupted cooling, Jang et al. [17] designed a non-stop defrosting cycle with dual hot gas spray to dual outdoor heat exchanger. But different from normal heat pump systems that adopt commonly tube-fin heat exchangers, the GVR cooling system needs to solve the defrosting problems in shell-tube heat exchanger. In addition, GVR process might have much shorter frosting period and more frequently defrosting intervals due to its lower cooling temperature. A novel dual, parallel channel defrosting method was proposed and researched in this study. Distinct from conventional multi-evaporator system relating to variable refrigerant flow (VRF) systems [18-20], this method designed the switchable one-work-one-standby evaporator mechanism. 3

ACCEPTED MANUSCRIPT

The defrosting techniques used in the HVAC&R industry may be adopted in GVR cooling system. Amer and Wang [21] and Hu et al. [22] summarized defrosting methods as follows: (1) on-off cycling [23], (2) hot gas bypass [24-26], (3) reverse cycle [27-28], (4) electric resistive heating [29-30], (5) desiccants as dehumidifiers or (6) ultrasonic wave [31-32] and their combined methods [33-35]. Hot gas bypass (HGB) was employed here as the key candidate defrosting method instead of reverse cycle for simple structure [36]. Comparing with the defrosting of microchannel heat exchangers [37-38] and fin-tube exchangers [39], the defrosting in low temperature shell-tube heat exchanger shell side was seldom mentioned in previous studies, and was more complex (Fig.3). In conventional HGB mechanism, the residual refrigerant inside evaporator was removed by evaporation during defrosting [17, 30], which consumed the energy of hot gas [23]. Especially in low evaporating temperature conditions, this additional energy consumption will become more and even excessive. Thus, a new defrosting energy recovery strategy was provided, which contained evacuating the refrigerant inside the defrosting evaporators and the channel switching delay. Most cooling potential of the frost and cold refrigerant have been recovered. From the existing study [40-42], the defrosting process could be divided into: (ⅰ) defrosting initiation (pre-heating) stage, (ⅱ) melting and draining (vaporizing) stage and (ⅲ) defrost termination and recovery (dry heating) stage. In this defrosting strategy, on-off cycling and refrigerant evacuation methods were set in motion for stage (ⅰ). And different HGB mode and switching delay was adopted for stage (ii) and (iii). Also the electric distributed heater with conductive heat transfer was frequently in use for heating the shell and preventing re-frosting at the internal wall of the shell.

Fig.3 Schematic of shell-tube heat exchanger defrosting

4

ACCEPTED MANUSCRIPT

Table 1 - Technologies summation in this system Aspects

References

Descriptions

Differentiations in this system

Multi-evaporator

Cho et al. [33]

Evaporators for parallel use or

Switchable one-work-one-standby

Arpagaus et al. [42]

different modes

evaporator for continuous cooling

Multiple temperature levels Defrost

Amer and Wang [21]

Defrosting methods for

Combined defrosting method for

Song et al. [36]

microchannel and fin-tube HX

shell-tube heat exchanger

Jang et al. [17]

Evacuating refrigerant before defrosting Channel switching delay strategy

Multi-stage

Granwehr and Bertsch [44]

Two-stage cycle with closed

Adding dual, parallel evaporators for

cooling and

Mathison et al. [45]

economizer and second heat source

each stage

cascades

Shen et al. [46]

Cascade cycle with dual mode

Capacity control

Qureshi and Tassou [47]

HGB control

Slider for coarse modulation

Chen et al. [48]

On/off control

Dual HGB for fine modulation

Wang et al. [49]

Slider control

Liquid injection and on/off control

Xu et al. [50]

Liquid injection

for extreme conditions

Lin and Hrnjak, [52]

Two-phase thermosiphon loop for

Refrigerant thermosiphon loop

He et al. [53]

refrigerator and gas turbine

between liquid line and oil cooler

Dutta et al. [51] Oil cooling

Abu-Mulaweh, [54]

In this paper, a novel dual channel defrosting method for GVR was firstly studied to meet the long continues time working requirements, especially in shipping and tank farm GVR application. The cooling system consisted of switchable one-work-one-standby evaporators, multi-stage cooling and

5

ACCEPTED MANUSCRIPT

cascade cycle. The defrosting method was efficient for shell-tube heat exchanger and firstly adopted the refrigerant evacuation and switching delay. Other relevant technologies of this cooling system such as capacity control were summarized in Table 1. The ability of continuous cooling was validated by the industrial application. The method is also applicable to the cases of non-stop long-running gas cooling process, which contains heavy components. Nomenclature A1 regression coefficients h specific enthalpy, (kJ·kg-1) K Loss coefficient, , 2∆P/ρv2 L liquid level, (m) m mass, (kg·s-1) m mass rate, (kg) P pressure, (Pa); Power, (kw) Q cooling capacity, (kw) Q capacity rate, (kw·s-1) Re Reynolds number T temperature, (K) t time, (s) 𝑉 volumetric flow rate, (m3·s-1) Greek ρ density, (kg· m-3) Δ difference, (-) Subscripts 0 initial state dis discharge dry dry condition e end el electric evap evaporator f frost gl gasoline liquid gv gasoline vapor hot hot gas

hx heat exchanger i initiate in inlet j stage k component min minimum out outlet period operation period ref refrigerant s solid sf solid to fluid state steady steady state Abbreviation CDR Capacity duty ratio DSP Delay switching phase GVR Gasoline vapor recovery HGB Hot gas bypass HGBP Hot gas bypass phase HTC High-temperature cycle IHX Internal heat exchanger LNG Liquefied natural gas LTC Low-temperature cycle NMHC gaseous non-methane hydrocarbons REP Refrigerant evacuating phase TCHGB Temperature control HGB VRF Variable refrigerant flow

2. Process description As shown in Fig.4, the whole system can be divided into the follow processes: Gasoline vapor recovery: The gasoline vapor of ordinary temperature passed through the

6

ACCEPTED MANUSCRIPT

recuperator (REC1) for cooling down some degrees and recovering cooling energy of the outlet gasoline vapor (path 001-002). Then, the gasoline vapor followed three stage cooling with temperature lower than -70°C, each cooling stage consisted of dual switchable one-work-one-standby evaporators (path 013-014-015-016 for channel A and path 023-024-025-026 for channel B). After flowed out the recuperator, its temperature came to 10~20°C and emitted (path 007-008). Refrigeration cycle: two classical cycles were employed with different refrigerants. As a preliminary system, the criteria of refrigerant selection was first the safety, nonflammable refrigerant was preferred. And combined with the operation temperature, R404A was the refrigerant in HTC, while R23 acted as the low temperature cycle (LTC) refrigerant. The three groups of evaporators (Evap.1A~Evap.3A, Evap.1B~Evap.3B) operated at different evaporating pressures matching with the three cooling stages. Two separate heat exchangers in HTC was designed as the economizer, where the first one was for economizing function (ECO1), and the second one was for pre-cooling of gasoline vapor (path 115-116A-117A-118 for channel A and path 115-116B-117B-118 for channel B). Two-stage screw compressor with port (COMP1 and COMP2) for refrigerant injection was chosen in this study, which was manufactured by Fusheng Corporation and explosion-proof certificated. An internal heat exchanger (IHX) was installed in LTC to extend operation range and improve energy efficiency. To regulate the intermediate heat exchanger pressure of HTC during refrigerant evacuating phase (refer to section 2.2), the automatic control valve (ACK1) was installed. Drainage accumulation: gasoline condensate from all evaporators and recuperator is draining to the gasoline storage tank (V5), and condensing heat was recovered by liquid line (path 104-105). Oil cooling loop: an external oil separator was arranged for each compressor to remove oil from the high pressure discharged refrigerant gas. The separated oil was cooled in the thermosiphon oilcooler and then returned into the compressor (path 301-302 and path 303-304). Other bypasses: they included hot gas bypass (red lines) and liquid injection (carmine lines). The HGB provoked through both working and standby evaporator respectively in terms of defrosting and capacity control, which constituted the dual HGB mechanism and discussed in section 2.2. Liquid injection was added in each cycle (path 401, 402) to prevent excessively high discharge temperature when operating at extreme high compression ratios.

7

ACCEPTED MANUSCRIPT

Fig.4 Schematic of cooling system

2.1 Refrigerant circuits P-H plots for the refrigerant circuits were depicted in Fig. 5. In the diagrams corresponding to the position numbers in Fig.4, phases 101-102-103-104 indicated that the HTC refrigerant was cooled by the airflow and gasoline condensate and heated by the oil circuits. Phases 104-115-116-117 accomplished the economizing and pre-cooling functions, while phases 125-126-127 indicated the middle-cooling process. The cascade process within the intermediate heat exchanger was approaching the middle-cooling phases, which was however not depicted in the diagrams. For LTC in normal refrigerating operations, phases 202-203 and 205-206 represented the heat transferring in IHX. Phases 204-205 depicted the super-cooling process.

8

ACCEPTED MANUSCRIPT

(1)

(2) Fig.5 P-H diagrams of (1) HTC and (2) LTC

2.2 Channel switching and defrosting methodology The defrosting control methodology was illustrated in Fig.6. The two evaporators of each cooling stage were operating in one-work-one-standby procedure. The existing techniques to initiate and terminate defrost cycles are focused on time-temperature controlled, pressure differential, temperature difference [55] and heat transfer balance [56]. In this device, pressure drop of gasoline vapor was observed more notable response than temperature difference. The frosting condition was monitored by the differential pressure transmitter and flowmeter. The criterion for switching was that the pressure drop was twice of one in dry conditions (∆𝑃𝑑𝑟𝑦). ∆𝑃𝑑𝑟𝑦 adopted the loss coefficient equation of Reuter and Anderson [57] (Eq.[1]), which was suitable for bare tube bundle in dry mode. The calculation of ∆𝑃𝑑𝑟𝑦 was fitted by the tested airflow in this device according to Eq.[2]. 𝐾 = 14.5049𝑅𝑒

‒ 0.04678

[1]

∆𝑃𝑑𝑟𝑦 = 𝐴1𝑉

1.95322

[2]

Where K was Loss coefficient, Re was Reynolds number, A1 was regression coefficient and 𝑉 was volumetric flow rate. The defrost process is initiated when the pressure drop has reached a calculated value correspond to the current flow rate (i.e. 2×∆𝑃𝑑𝑟𝑦), and terminated when the evaporator temperature has reached a predetermined value. The procedure was divided into the following phases:

9

ACCEPTED MANUSCRIPT

Fig.6 Switching control conditions

Refrigerant evacuating phase (REP): this phase was to remove the cold liquid refrigerant [58] from the defrosting evaporator before the hot gas supply was initiated. Step1.1: closed all the liquid line solenoid valves (SV1-SV6), the refrigerant was evacuated to the reservoir. Step1.2: control valve (ACK1) was regulated for minor flow to enhance the vacuum of the suction line for refrigerant evacuation. Step1.3: the gasoline vapor still flowed through the defrosting channel for a period as a manner of on-off cycling defrosting.

10

ACCEPTED MANUSCRIPT

Step1.4: the outlet of Evap.1A and Evap.1B was reconnected to the suction line by electrical tee valve (TV1) for more thorough evacuation of refrigerant. Delay switching phase (DSP): this phase distributed the gasoline vapor between the dual channels for balancing the cooling and defrosting effect. Step2.1: opened the liquid line solenoid valves (SV1/3/5 or SV2/4/6) in the cooling channel. The refrigerant was migrated to the working evaporators, which were prepared to be placed back into low temperature service. Step2.2: constant frequency hot gas bypass (CFHGB) mechanism was invoked in defrosting evaporators to enhance the defrosting effect. The definition of CFHGB was described that HGB operated with fixed period and pulse width. As a lot of heat was needed for defrosting in this period, HGB spraying for too long would result potential risks to the system performance and the compressor safety. Step2.3: opened all the channel valves (MOV1-MOV4); both working and standby evaporators shared responsibility for the gasoline vapor cooling. Hot gas bypass phase (HGBP): the defrosting was dominated by HGB in this phase. Step3.1: closed the defrosting channel valves (MOV1/3 or MOV2/4). The gasoline vapor was switched to the cooling channel completely. Step3.2: the electrical heat-tracing system in defrosting evaporators started to heat the shell. Step3.3: the HGB mechanism was changed from CFHGB to temperature control hot gas bypass (TCHGB). TCHGB mode reduced the HGB frequency for energy saving, the PLC judged whether opened or stopped the HGB by the evaporator temperature signal.

3. Results and discussions The industrial application for the system proposed by this study was shown in Fig.7, which is located at Port of Lianyungang, China. Since there were no test standards or regulations for this type system of until now, the test condition was set as that the inlet gasoline vapor temperature was 25°C. The system performance was tested as the following four cases for 72 hours. (1) Case #1: twenty minutes delay switching. Refer to section 2.2, the switching delay was set to 20 minutes. The timing was approximately that the frosts at the defrosting evaporator of 1st 11

ACCEPTED MANUSCRIPT

stage have just been melted. (2) Case #2: forty five minutes delay switching. The procedure was the same as case #1, except that the delay was set to 45 minutes. The timing was approximately that frosts at the defrosting evaporator of 2nd stage have just been melted. (3) Case #3: no refrigerant evacuation. Refer to section 2.2, the process went through only DSP and HGBP. (4) Case #4: no delay switching. Refer to section 2.2, the process went through only REP and HGBP.

Fig.7 100~600 m3/h gasoline vapor cooling device for ship loading, serving 4 marine loading arms

3.1 Operation characteristics Stable pressure drop of the channel was an essential prerequisite of a successful continuous running, because the ship tanker should work in the design pressure range. Fig.8 compared the pressure drop of the gasoline vapor channel between case #1 and case #3. For case #3, the pressure difference increased continuously and could not return to normal range. The main reason was inefficient defrosting, which due to hot gas energy self-consumption by cold residual refrigerant in the defrosting evaporator. Therefore, the frosts accumulated all the time and resulted in the high pressure difference and operation failure after about 24 hours running. For case #1, despite the maximum and minimum values of the pressure difference in each period were not the same completely because of unstable working conditions, the value of the pressure difference was still able to return to normal range. This device was believed to be suitable for the long time non-stop operations.

12

ACCEPTED MANUSCRIPT

Fig.8 Pressure difference in case #1 and case #3

3.2 Capacity comparison With the components concentrations of inlet gasoline vapor were measured in advance, the cooling capacity was calculated by the enthalpy difference method on the gasoline vapor side in Eq.[3]. 3

𝑄𝑔𝑣 = ∑𝑗 = 1{(𝜌𝑔𝑣,𝑗 ‒ 1𝑉𝑔𝑣,𝑗 ‒ 1ℎ𝑔𝑣,𝑗 ‒ 1 ‒ 𝜌𝑔𝑣,𝑗𝑉𝑔𝑣,𝑗ℎ𝑔𝑣,𝑗) ‒ ∑𝑘[(𝑚𝑔𝑣,𝑘,𝑗 ‒ 1 ‒ 𝑚𝑔𝑣,𝑘,𝑗)ℎ𝑔𝑙,𝑘]}

[3]

The subscripts gv, gl and j represented gasoline vapor, gasoline liquid (condensate) and the state of after jth cooling stage (0 stage meant inlet state), respectively. 𝑉𝑔𝑣,0 represented the inlet volumetric flow rate and was measured by inlet flowmeter; 𝑚𝑔𝑣,𝑘,𝑗 represented the mass content of k component in gasoline vapor after jth cooling stage; ℎ𝑔𝑙,𝑘 represented the specific enthalpy of k component in gasoline liquid. 𝜌𝑔𝑣,𝑗, ℎ𝑔𝑣,𝑗, 𝑉𝑔𝑣,𝑗 and 𝑚𝑔𝑣,𝑘,𝑗 were calculated from Table. A.1.

13

ACCEPTED MANUSCRIPT

Fig.9 Cooling capacity comparison at one period

Fig.9 illustrated the cooling capacity values in different cases at one period. The corresponding temperature differences (ΔT) between the evaporator (T01-T06) and gasoline vapor channel (T11, T12, T21, T22 and T30), as well as the pressure drop (ΔP) were also recorded. In case #1, the cooling capacity was increased by 28.2% than case #3 without refrigerant evacuation. There are two main reasons for the capacity reduction of case #3: (1) Frost accumulation cause by defrosting failure (section 3.1) which decreased the heat transfer. It’s also checked by the temperature difference shown in Fig. 9, the ΔT of case #3 was 13°C on

14

ACCEPTED MANUSCRIPT

average, while ΔT of case #1 was only 5°C. (2) Residual refrigerant in defrosting evaporator was difficult to pump out only by the way of hot gas heating, which resulted in insufficiency of cycling refrigerant. Observing the region at initial relative time in Fig.9, larger ΔT and lower ΔP was measured. It indicated high superheat and lack of cycling refrigerant. The performance characteristics of case #2 was approaching to case #1. The only difference was the capacity decline in DSP, as the heat sink of defrosting channel was not enough to cool down the gasoline vapor after long switching delay. In case #4 without delay switching, the cooling capacity was worse than case #1 and 2#, which represented the delay switching cases. It also declined dramatically during channel switching, because the working evaporator did not reach steady state when just switching.

3.3 Stability and defrost efficiency analysis Cooling duty ratio (CDR) and cooling reduction ratio (CRR) were defined artificially as Eq.[4] and [5]. These two parameters were used to indicate the fluctuating period and range during defrosting. The lower CDR value was, the larger fluctuating period existed. While the higher CRR value meant the larger fluctuating range. 𝐶𝐷𝑅 = t𝑠𝑡𝑒𝑎𝑑𝑦 t𝑝𝑒𝑟𝑖𝑜𝑑

[4]

𝐶R𝑅 = (𝑄𝑠𝑡𝑒𝑎𝑑𝑦 ‒ 𝑄𝑚𝑖𝑛) 𝑄𝑠𝑡𝑒𝑎𝑑𝑦

[5]

t𝑠𝑡𝑒𝑎𝑑𝑦 and t𝑝𝑒𝑟𝑖𝑜𝑑 represented the times of the steady state and the whole operation period, respectively. 𝑄𝑚𝑖𝑛 represented the minimum cooling performance value during defrosting. Furthermore, these parameters were identified in Fig. 10. The cooling capacity in case #1 and case #3 was shown in Fig.10; as mentioned in section 3.1, the capacity measurement of case #3 was terminated in near 24 hours because of the high pressure drop. The values of CDR and CRR were summarized in Table. 2, which also compared to published heat pump defrosting studies. The fluctuating period and range in case #1 of this study were all smaller than the references’ studies. The situation of case #2 was similar to case #1 except about 10% reduction of both steady state time and minimum cooling performance values when channel switching. In case #3 without refrigerant evacuation, the value of CDR was much lower. That meant fluctuating 15

ACCEPTED MANUSCRIPT

period prolonged significantly, because the cycling refrigerant was insufficient for much time. Furthermore, the residual refrigerant inside the defrosting evaporator consumed the heat energy of hot gas, which increased the defrosting load. In case #4 without switching delays, the cooling capacity reduction became larger. Because the cooling potential of frost was wasted, and there is not enough time for the working evaporator to pre-cool. It was drawn that refrigerant evacuation and switching delay are effective methods for improving the energy conversion efficiency. Based on this, the system performed successfully in the non-stop cooling application with little fluctuations and reductions.

Fig.10 Cooling capacity in case #1 and case #3

Defrost efficiency [58-59] was defined as the ratio of the minimum energy required to melt the frost to the energy applied to actually melt the frost. Base on Hoffenbecker [59], the distribution of defrost energy was tracked in Appendix B. Since the defrost energy contribution of gasoline vapor Q𝑔𝑣 was the necessary heat source in cooling process, it was treated as no power consumption to defrost and not including in the applied energy term of defrost efficiency. η𝑑 = Q

Q𝑚𝑒𝑙𝑡

[6]

ℎ𝑔 + Q𝑒𝑙

Cycle-average defrost efficiency and energy contribution of case #1 and case #3 in different stages was evaluated in Fig.11. It’s found that most heat load appeared at the 2nd stage, because of large defrost temperature span and melting over half the frost at this stage. The phenomenon of frost accumulating at 2nd stage was also captured in Fig.9: the fast increment of ΔT at 2nd stage with the ascending of ΔP. On the other hand, gasoline vapor energy made an important positive effect. As a free heating factor, it even led to over 100% defrost efficiency at the 1st stage. Heat exchanger term

16

ACCEPTED MANUSCRIPT

dominated the parasitic heat load, because the heavy shell-tube should met the requirements of Chinese standard GB150 in such inflammable and explosive condition. Another inefficient factor was the refrigerant heating, which also played major role in case #3, and was eliminated by refrigerant evacuating in case #1. The values of overall defrost efficiency were also summarized in Table. 2. Compared to other published studies, the heavy evaporator structure and low temperature made this system a lower defrost efficiency, but the free heating of gasoline vapor compensated the short slab.

Fig.11 Defrost efficiency and energy consumption in case #1 and case #3 Table 2 - Performance and defrost efficiency comparison of some defrosting systems References Case #1

Key descriptions Switchable dual evaporator; non-stop defrosting; switching delay; refrigerant evacuation; cascade; evaporating temperature of 0/-35/-70°C

CDR

CRR

η𝑑

75%

10%

55.4%

Case #2

Over delay switching, evaporating temperature of 0/-35/-70°C

70%

20%

53.0%

Case #3

No refrigerant evacuation, evaporating temperature of 0/-35/-70°C

21%

27%

40.6%

Case #4

No switching delay, evaporating temperature of 0/-35/-70°C

55%

47%

46.6%

Kim et al. [34], Fig.7

Periodic reverse cycle defrosting

-

100%

-

Kim et al. [34], Fig.8

Dual hot gas bypass defrosting

-

53.8%

-

Kim et al. [34], Fig.9

Dual hot gas bypass with inverter heater

-

57.1%

-

25%

35.3%

-

Zhang et al. [62], Fig.12

Reverse cycle defrosting with using heat energy dissipated by the compressor

Jang et al. [17], Fig.9

Periodic reverse cycle defrosting

50%

100%

-

Jang et al. [17], Fig.9

Dual spray hot gas defrosting

75%

57.1%

-

-

73%

-

Qu et al. [63], Fig.11

Cascade air source heat pump, thermal energy storage based reverse cycle defrosting

Cho et al. [33], Fig.4

Multi-evaporator refrigeration; on-off cycle

60%

100%

-

Cho et al. [33], Fig.7

Multi-evaporator refrigeration; hot gas bypass

-

25.5%

-

Dong [28] Section 4

Reverse cycle defrost experiment, evaporating temperature of 0°C

-

-

Melo et al. [30] Table 1

Defrost experiments of electric heaters, freezer temperature of -23°C

Hoffenbecker [64] Table 3

Coil defrost model of hot gas, freezer temperature of -23°C

-

-

43.7%

Song [65] Table 6

Hot gas defrost experiment, evaporating temperature of 0°C

-

-

47.1%/58.8%

17

60.1% 48.0%

ACCEPTED MANUSCRIPT

From pressure control standpoint in case #1, Fig.12 illustrated the gauge pressure values at different positions of low pressure side (P01-P04). Each curves appeared periodic spike and sustained fluctuation. The pressure spike was synchronizing with the refrigerant switching to the warm evaporator after defrosting (step 2.1 of section 2.2); while the sustained fluctuation was synchronizing with the hot gas spraying solenoid valve action. Violent pressure spike would result in the mechanical damage to the compressor and sustained pressure fluctuation would affect the evaporating pressure of the simultaneous cooling. The pressure spike ranges were less than 0.2 MPa and safe for the compressor. The sustained fluctuations were less than 0.05 MPa, the corresponding effects on the evaporating temperature of the cooling side were less than ±2.5°C, which was important for the stable operation.

Fig.12 Low pressure in different positions for case #1

3.4 Sensitivity of the switching delay Sensitivity studies of different switching timing were carried out by analyzing the gasoline vapor 18

ACCEPTED MANUSCRIPT

temperatures (T30) in Fig.13. From the perspectives of the cooling temperature at steady state, case #1 and case #2 performed at the same level, and both better than case 4. As a failure case, the temperatures of case #3 cannot maintain a stable level and ascend dramatically with the existing switching strategy. Then we can summarize that inappropriate switching strategy will lead to performance reduction and even failure operation. The minimum temperature fluctuations were located in case #1, which were within 10°C. For case #2 of over delay switching, the fluctuations were about 5oC higher than case #1. The temperature fluctuations in case #4 appeared significantly higher than other normal cases, which was up to about 45oC. Therefore, the timing of switching was essential for the temperature fluctuations. If under-delay switching, the precooling at working evaporators before switched was not enough, the working evaporators were not ready for cooling just after defrosting. If over-delay switching, heavy components in the gasoline vapor can lead to frost at 2nd stage and even at 3rd stage when they flowed through the defrosting channel, which resulted in redundant defrosting and energy consumptions. The optimal switching delay was about 20 minutes for this device. At this time, the gasoline vapor temperature for defrosting at the outflow of 1st stage began to increase, and it’s supposed that the frosts at the defrosting evaporator of 1st stage have just been melted. With this prerequisite, the frost cooling potential was best recovered by gasoline vapor.

Fig.13 Comparison of the gasoline vapor cooling temperatures

19

ACCEPTED MANUSCRIPT

4. Conclusions A novel dual channel cooling system for GVR was developed in this study. Compared to conventional VOCs condensation method, the switchable one-work-one-standby evaporators and its control strategies were adaptive to non-stop running and high energy efficiency. According to industrial testing results, the following conclusions can be summarized: (1) Refrigerant evacuation ensured the amount of cycling refrigerant and therefore improved the cooling capacity by 28.2%. Otherwise, it might fail to defrost completely after long time running. (2) No switching delay didn’t result in notable capacity reduction of steady state, but affected its fluctuations. (3) Switching delay can bring free energy of gasoline vapor heating, refrigerant evacuation can eliminate the parasitic load of refrigerant heating, which were all profitable to defrost efficiency. (4) The best case of defrost in this study was 10% the performance reduction ratio, 75% the duty ratio, and 55.4% the defrost efficiency, which was competitive compared to the references’ studies. (5) Appropriate switching delay can optimize the performance and fluctuation by recovery of the frost cooling potential, which was an effective method of energy saving. The best timing of switching delay was about 20 minutes for this device. The main achievement of this study was the discovery of the dual switchable channel method to supply continuous long-running cooling in the multi-stage and cascade cycles. By adjusting the defrosting control strategy, solved the problem of frost interruption, which has been the greatest shortcoming of non-stop cooling process with heavy components. But the channel switching timing and criteria still needs to be optimized. We suggest developing system-level modeling integrated with a detailed frost growth model for future exploration, which is beneficial to energy performance analysis and model-based control design. Acknowledgement This work was supported by Guangdong Shenling Environmental System Co., Ltd. (China)

20

ACCEPTED MANUSCRIPT

Appendix A Table A.1 - Data sheet of gasoline vapor condensation process calculation results T oC

V

h

ρ

m3· h-

kJ·kg-

kg· m-

1

1

3

C2H6

C3H6

C4H10

2C4H10

1C4H8

2C4H8

C5H12

C6H14

Air etc

1.859

3.042

4.679

55.460

79.064

37.305

28.918

286.904

175.864

444.342

-

Components mass flow rate of gasoline vapor (kg/hr)

25

600.0

5

499.9

-854.1

1.824

2.977

4.586

51.971

73.122

33.479

26.162

206.029

71.333

442.135

-28

347.6

-402.1

1.688

2.630

4.074

34.003

43.594

16.452

13.419

37.132

3.017

432.479

-55

279.8

-224.2

1.734

2.246

3.535

18.807

20.724

5.778

4.991

3.259

0.042

425.717

-56

277.5

-218.9

1.736

2.216

3.494

18.006

19.668

5.405

4.678

2.942

0.037

425.437

-57

275.3

-213.8

1.739

2.185

3.451

17.213

18.638

5.049

4.380

2.656

0.033

425.166

-58

273.1

-208.9

1.742

2.153

3.407

16.431

17.637

4.712

4.095

2.398

0.029

424.905

-59

270.9

-204.1

1.745

2.120

3.361

15.662

16.667

4.392

3.824

2.167

0.026

424.655

-60

268.8

-199.4

1.749

2.085

3.313

14.908

15.729

4.090

3.567

1.958

0.023

424.416

-61

266.7

-194.9

1.752

2.050

3.264

14.171

14.825

3.805

3.324

1.769

0.020

424.188

-62

266.7

-194.9

1.752

2.050

3.264

14.171

14.825

3.805

3.324

1.769

0.020

424.188

-63

262.7

-186.3

1.760

1.975

3.160

12.753

13.120

3.284

2.878

1.446

0.016

423.763

-64

260.8

-182.3

1.764

1.937

3.106

12.075

12.321

3.047

2.674

1.308

0.014

423.567

-65

258.8

-178.5

1.769

1.897

3.050

11.418

11.557

2.825

2.482

1.182

0.013

423.380

-66

257.0

-174.8

1.773

1.856

2.992

10.783

10.829

2.617

2.302

1.069

0.011

423.204

-67

255.1

-171.3

1.778

1.815

2.934

10.171

10.135

2.422

2.133

0.967

0.010

423.037

-68

253.3

-167.9

1.783

1.772

2.873

9.582

9.476

2.240

1.975

0.874

0.009

422.880

-69

251.5

-164.7

1.789

1.729

2.811

9.016

8.850

2.070

1.827

0.790

0.008

422.731

-70

249.7

-161.7

1.794

1.685

2.748

8.474

8.256

1.912

1.689

0.714

0.007

422.592

-71

248.0

-158.9

1.800

1.640

2.683

7.955

7.695

1.764

1.560

0.645

0.006

422.460

-72

246.3

-156.2

1.806

1.595

2.618

7.459

7.164

1.627

1.439

0.583

0.006

422.336

-73

244.7

-153.7

1.812

1.549

2.551

6.986

6.663

1.499

1.327

0.526

0.005

422.220

-74

243.0

-151.3

1.818

1.502

2.483

6.535

6.191

1.380

1.223

0.475

0.004

422.110

-75

241.4

-149.1

1.825

1.455

2.414

6.106

5.747

1.269

1.125

0.429

0.004

422.008

1020.7

Appendix B The defrosting energy flows was quantified as follow: (1) Qℎ𝑔 - the amount of energy supplied to defrosting by the hot gas, which was calculated according to: 𝑡

Qℎ𝑔 = ∫𝑡𝑒𝑚ℎ𝑔(ℎℎ𝑔,𝑑𝑖𝑠 ‒ ℎℎ𝑔,𝑒𝑣𝑎𝑝,o𝑢𝑡) ∙ 𝑑𝑡𝑠𝑝𝑟𝑎𝑦 𝑖

21

[B.1]

ACCEPTED MANUSCRIPT

Where 𝑚ℎ𝑔 was the hot gas flowrate, ℎ𝑑𝑖𝑠 was hot gas enthalpy at discharge point; ℎℎ𝑔,𝑒𝑣𝑎𝑝,o𝑢𝑡 was hot gas enthalpy at the outlet of defrosting evaporator, 𝑡𝑠𝑝𝑟𝑎𝑦 was the hot gas spraying time, which is captured by the pulsing signal of solenoid valve; 𝑡𝑒 and 𝑡𝑖 were the timers which initiated and terminated the defrost period, respectively. In addition, 𝑚ℎ𝑔 was determined using the collected experimental data and the Fusheng Selection Software from the compressor manufacturer. (2) Q𝑒𝑙 - the amount of energy supplied to defrosting by the electric heater, which was calculated according to: 𝑡

[B.2]

Q𝑒𝑙 = ∫𝑡𝑒𝑃𝑒𝑙𝑑𝑡𝑒𝑙 𝑖

Where 𝑃𝑒𝑙 and 𝑡𝑒𝑙 were the power and heating time of the electric heater, the 𝑃𝑒𝑙 was held constant at 1.0 kw per evaporator. (3) Q𝑔𝑣 - the amount of energy supplied to defrosting by the gasoline vapor during the step 1.3 refer to section 2.2, which was derived by: 𝑡

[B.3]

Q𝑔𝑣 = ∫𝑡𝑒𝑄𝑔𝑣 ∙ 𝑑𝑡𝑔𝑣 𝑖

Where 𝑄𝑔𝑣 was calculated by Eq[4] using the local temperatures of defrosting channel. Since gasoline vapor duct is small-size radial and circular, its radial temperature distribution is assumed as uniform. 𝑡𝑔𝑣 is the time of gasoline vapor flowing through the defrosting channel. To prevent the confusion of melt and condensate drainage, dry nitrogen was chosen instead of gasoline vapor when estimating the frost mass. Therefore, Q𝑔𝑣 was certain underestimated as neglecting the latent heat of gasoline vapor condensation. (4) Q𝑚𝑒𝑙𝑡 - the amount of energy it took for the mass of frost to melting point and change phase from solid to liquid. Since the gasoline vapor was from refined oil, the frost component was assumed as only water. Then, we had: 3

Q𝑚𝑒𝑙𝑡 = ∑𝑗 = 1𝑚𝑓,𝑗 ∙ [∆ℎ𝑠𝑓 + 𝑐𝑝,𝑓,𝑠(𝑇𝑚𝑒𝑙𝑡 ‒ 𝑇𝑓,𝑖,𝑗)]

22

[B.4]

ACCEPTED MANUSCRIPT

Where ∆ℎ𝑠𝑓 was the enthalpy change from solid to liquid, 𝑐𝑝,𝑓,𝑠 was the specific heat of frost in solid phase, 𝑇𝑚𝑒𝑙𝑡 was the melting temperature, 𝑇𝑓,𝑖 and 𝑚𝑓,𝑖 were the frost temperature and mass at jth stage. 𝑚𝑓,𝑗 was evaluated by the liquid level of gasoline storage tank, and was estimated by: 𝑚𝑓,𝑗 = 𝑚𝑓(𝐿𝑗) ‒ 𝑚𝑓(𝐿𝑗 ‒ 1), 𝑗 = 1,2,3

[B.5]

Where 𝐿𝑖 was the liquid level of gasoline storage tank when the jth stage condensate began to drain and 𝐿0 was the liquid level before defrosting, as it’s observed that the melt drainage of different stage evaporator occurred in different time. (5) Q𝑒𝑥𝑐𝑒𝑠𝑠 - the energy provided beyond that needed to heat given frost past the melting point. 3

Q𝑒𝑥𝑐𝑒𝑠𝑠 = ∑𝑗 = 1𝑐𝑝,𝑓,𝑙 ∙ 𝑚𝑓,𝑗 ∙ (𝑇𝑒 ‒ 𝑇𝑚𝑒𝑙𝑡)

[B.6]

Where 𝑐𝑝,𝑓,𝑙 was the specific heat of frost in liquid phase, 𝑇𝑒 was the predetermined evaporator temperature when the defrost period was terminated. Q𝑒𝑥𝑐𝑒𝑠𝑠 was overestimated by using 𝑇𝑒 instead of the drainage temperature. (6) Q𝑟𝑒𝑓 - the energy stored in the refrigerant inside evaporator during a defrost period, which was calculated according to: 3

Q𝑟𝑒𝑓 = ∑𝑗 = 1m𝑟𝑒𝑓,i,𝑗(ℎ𝑟𝑒𝑓,𝑒,𝑗 ‒ ℎ𝑟𝑒𝑓,𝑖,𝑗)

[B.7]

Where ℎ𝑟𝑒𝑓,𝑖,𝑗 and ℎ𝑟𝑒𝑓,𝑒,𝑗 are the enthalpies of the refrigerant at jth stage evaporator when the defrosting initiating and terminating; 𝑚𝑟𝑒𝑓,𝑖,𝑗 is the mass of the refrigerant inside jth stage evaporator when the defrosting initiating, which was evaluated by the refrigerant recovery machine. (7) Qℎ𝑥 - the total thermal energy stored within the heat exchanger shell and tube after a defrost period was complete, which was calculated according to: 3

Qℎ𝑥 = ∑𝑗 = 1𝑚ℎ𝑥,𝑗 ∙ 𝑐

𝑝,ℎ𝑥

(𝑇𝑒 ‒ 𝑇𝑓,𝑖,𝑗)

[B.8]

23

ACCEPTED MANUSCRIPT

Where 𝑚ℎ𝑥,𝑗 was the mass of evaporator at jth stage and 𝑐𝑝,ℎ𝑥 was the specific heat of evaporator, as all shells and tubes of evaporators were all made of stainless steel. (8) Q𝑙𝑜𝑠𝑠 - the sensible energy convected to the ambient, also the latent energy caused by sublimation and re-evaporation from the tube surface, which was calculated by energy conservation: Q𝑙𝑜𝑠𝑠 = Qℎ𝑔 + Q𝑒𝑙 + Q𝑔𝑣 ‒ Q𝑚𝑒𝑙𝑡 ‒ Q𝑒𝑥𝑐𝑒𝑠𝑠 ‒ Qℎ𝑥

[B.9]

References [1] Huang, W., Bai, J., Zhao, S., Lv, A., Investigation of oil vapor emission and its evaluation methods. Journal of Loss Prevention in the Process Industries, 2011; 24(2): 178-186. [2] Liu, Y., Ritter, J.A., Kaul, B.K., Simulation of gasoline vapor recovery by pressure swing adsorption. Separation and Purification Technology. 2000; 20: 111–127. [3] Zhang, X., Gao, B., Creamer, A.E., Cao, C., Li, Y., Adsorption of VOCs onto engineered carbon materials: A review. Journal of Hazardous Materials 2017; 338: 102–123. [4] Roizard, D., Lapicque, F., Favre, E., Roizard, C., Potentials of pervaporation to assist VOCs' recovery by liquid absorption. Chemical Engineering Science, Chemical Engineering Science 2009; 64: 1927-1935. [5] Shi, L., Huang, W., Sensitivity analysis and optimization for gasoline vapor condensation recovery, Process Safety and Environmental Protection, 2014; 92: 807-814. [6] Yi, S., Wan, Y., Volatile organic compounds (VOCs) recovery from aqueous solutions via pervaporation with vinyltriethoxysilanegrafted- silicalite-1/polydimethylsiloxane mixed matrix membrane. Chemical Engineering Journal. 2017; 313: 1639–1646. [7] Tesch S, Morosuk T, Tsatsaronis G. Advanced exergy analysis applied to the process of regasification of LNG (liquefied natural gas) integrated into an air separation process. Energy, 2016: 117: 550-561. [8] Mehrpooya M, Sharifzadeh MMM, Rosen MA., Optimum design and exergy analysis of a novel cryogenic air separation process with LNG (liquefied natural gas) cold energy utilization. Energy. 2015; 90: 2047-2069. [9] Shie, J.L., Lu, C.Y., Chang, C.Y., Chiu, C.Y., Lee, D.J., Liu, S.P., Chang, C.T., Recovery of gasoline vapor by a combined process of two-stage dehumidification and condensation, J. Chin. Inst. Chem. Eng. 2003; 34 (6): 605–616. [10] Kalantar-Neyestanaki, H., Mafi, M., Ashrafizadeh, A., A novel approach for operational optimization of multi-stage refrigeration cycles in gas refineries. Int. J. Refrigeration 2017; 80: 169-181. [11] Ghorbani, B., Hamedi, M.H., Amidpour, M., Mehrpooya, M., Cascade refrigeration systems in integrated cryogenic natural gas process (natural gas liquids (NGL), liquefied natural gas (LNG) and nitrogen rejection unit (NRU)). Energy 2016; 115: 88-106. [12] Sun, Z., Liang, Y., Liu, S., Ji, W., Zang, R., Liang, R., Guo, Z., Comparative analysis of thermodynamic performance of a cascade refrigeration system for refrigerant couples R41/R404A and R23/R404A. Appl Energy 2016; 184: 19-25. [13] Kilicarslan, A., Hosoz, M., Energy and irreversibility analysis of a cascade refrigeration system for various refrigerant couples. Energy Convers Manage 2010; 51(12): 2947-2954. [14] Neagu, M., Cursaru, D. L., Technical and economic evaluations of the triethylene glycol regeneration processes in natural gas

24

ACCEPTED MANUSCRIPT

dehydration plants. Journal of Natural Gas Science and Engineering, 2017; 37: 327-340. [15] Santos, M. G. R. S., Correia, L. M. S., Medeiros, J. L. D., Araújo, O. D. Q. F., Natural gas dehydration by molecular sieve in offshore plants: impact of increasing carbon dioxide content. Energy Conversion and Management. 2017; 149: 760-773. [16] Li, C., Li, J., Wang, B., Xie, W., Zhongyi, Beijing, Theoretical Investigation on Condensing Characteristics of Air and Oil Vapor Mixtures. International Refrigeration and Air Conditioning Conference, 2008; Paper 907. http://docs.lib.purdue.edu/iracc/907. [17] Jang, J.Y., Bae, H. H., Lee, S. J., Ha, M. Y., Continuous heating of an air-source heat pump during defrosting and improvement of energy efficiency. Appl Energy 2013; 110: 9-16. [18] Matthew S. E., Bryan P.R., Decentralized model predictive control of a multi-evaporator air conditioning system. Control Engineering Practice 2013; 21: 1665–1677. [19] Zhu, Y., Jin, X., Du, Z., Fan B., Fu, S., Generic simulation model of multi-evaporator variable refrigerant flow air conditioning system for control analysis, Int. J. Refrigeration 2013; 36: 1602-1615. [20] Visek, M., Joppolo, C.M., Molinaroli, L., Olivani, A., Advanced sequential dual evaporator domestic refrigerator/freezer: System energy optimization. Int. J. Refrigeration 2014; 43: 71-79. [21] Amer, M., Wang, C. C., Review of defrosting methods, Renewable and Sustainable Energy Reviews, 2017; 73: 53-74. [22] Hu, B., Yang, D., Cao F., Xing Z., Fei, J., Hot gas defrosting method for air-source transcritical CO2 heat pump systems, Energy and Buildings 2015; 86: 864–872. [23] Mader, G., Thybo, C., A new method of defrosting evaporator coils. Appl. Therm. Eng. 2012; 39: 78–85. [24] Choi H. J., Kim B. S., Kim K. C., Defrosting method adopting dual hot gas bypass for an air-to-air heat pump. Appl Energy 2011; 88: 4544- 4555 [25] Yaqub M, Zubair SM, Khan J. Performance evaluation of hot-gas bypass capacity control schemes for refrigeration and airconditioning systems. Energy, 2000; 25:543–561. [26] Hu, B., Wang, X., Cao F., He Z., Xing Z., Experimental analysis of an air-source transcritical CO2 heat pump water heater using the hot gas bypass defrosting method, Appl. Therm. Eng. 2014; 71: 528-535. [27] Qu, M., Deng, S., Jiang, Y., An experimental study on the defrosting performance of a PCM-based reverse-cycle defrosting method for air source heat pumps. Int. J. Air Cond. Refrigeration, 2010; 4:327-337. [28] Dong, J.K., Deng, S.M., Jiang, Y.G., Xia, L., Yao, Y., An experimental study on defrosting heat supplies and energy consumptions during a reverse cycle defrost operation for an air source heat pump, Appl. Therm. Eng. 2012; 37: 380–387. [29] Tudor, V., Ohadi, M., Salehi, M.A., Lawler, J.V., Advances in control of frost on evaporator coils with an applied electric field, International Journal of Heat and Mass Transfer, 2005; 48 (21-22): 4428–4434. [30] Melo, C., Knabben, F.T., Pereira, P.V., An experimental study on defrost heaters applied to frost-free household refrigerators, Appl. Therm. Eng. 2013; 51: 239–245. [31] Zhang, L., Fujinawa, T., Saikawa, M., A new method for preventing air-source heat pump water heaters from frosting. Int. J. Refrigeration 2012; 35: 1327–1334. [32] Wang, D.Y., Tao, T.F., Xu, G.H., Luo, A.L., Kang, S.Y., Experimental study on frosting suppression for a finned-tube evaporator using ultrasonic vibration, Experimental Thermal and Fluid Science 2012; 36: 1–11. [33] Cho, H., Kim, Y., Jang, I., Performance of a showcase refrigeration system with multi-evaporator during on-off cycling and hot gas bypass defrost, Energy 2005; 30: 1915-1930. [34] Kim, J., Choi, H.J., Kim, K. C., A combined Dual Hot-Gas Bypass Defrosting method with accumulator heater for an air-to-air heat pump in cold region, Appl Energy 2015; 147: 344–352. [35] Tang, J., Gong, G., Su, H., Wu, F., & Herman, C., Performance evaluation of a novel method of frost prevention and retardation for air source heat pumps using the orthogonal experiment design method. Appl Energy, 2016; 169: 696-708. [36] Huang, D., Li, Q., Yuan, X., Comparison between hot-gas bypass defrosting and reverse-cycle defrosting methods on an air-towater heat pump, Appl Energy 2009; 86: 1697–1703.

25

ACCEPTED MANUSCRIPT

[37] Shao LL., Yang L, Zhang CL., Comparison of heat pump performance using finand-tube and microchannel heat exchangers under frost conditions. Appl Energy 2010; 87(4): 1187-1197. [38] Moallem E., Hong T., Cremaschi L., Fisher DE., Experimental investigation of adverse effect of frost formation on microchannel evaporators, Part 1: Effect of fin geometry and environmental effects. Int. J. Refrigeration, 2013; 36: 1762–1775. [39] Song M., Deng S., Dang C., Mao N., Wang Z., Review on improvement for air source heat pump units during frosting and defrosting. Appl Energy 2018; 211: 1150-1170 [40] Krakow KI, Yan L, Lin S., A model of hot gas defrosting of evaporators, Part 1: heat and mass transfer theory. ASHRAE Trans. 1992; 98(1): 451-461. [41] Krakow KI, Yan L, Lin S., A model of hot gas defrosting of evaporators, Part 2: experimental analysis. ASHRAE Trans. 1992; 98(1): 462–474. [42] Song, M., Deng, S., Xia, L., A semi-empirical modeling study on the defrosting performance for an air source heat pump unit with local drainage of melted frost from its three-circuit outdoor coil. Appl Energy. 2014; 136(C): 537-547. [43] Arpagaus, C., Bless, F., Schiffmann, J., & Bertsch, S. S., Multi-temperature heat pumps - a literature review. Int. J. Refrigeration, 2016; 69; 437-465. [44] Granwehr, E.V., Bertsch, S., Heat pump. Patent 2012; WO 2012/040864 A1. [45] Mathison, M.M., Braun, J.E., Groll, E.A., Performance limit for economized cycles with continuous refrigerant injection. Int. J. Refrigeration 2010; 34: 234-242. [46] Shen, J., Guo, T., Tian, Y., & Xing, Z., Design and experimental study of an air source heat pump for drying with dual modes of single stage and cascade cycle. Appl. Therm. Eng. 2018; 129: 280-289. [47] Qureshi, T., Tassou, S., Variable-speed capacity control in refrigeration systems, Appl. Therm. Eng. 1996; 16 (2): 103-113. [48] Chen, W., Xing, Z., Tang, H., Wu, H., Theoretical and experimental investigation on the performance of screw refrigeration compressor under part-load conditions, Int. J. Refrigeration 2011; 34: 1141-1150. [49] Wang, Z., Wang, Z., Wang, J., Jiang, W., Feng, Q., Research of thermal dynamic characteristics for variable load single screw refrigeration compressor with different capacity control mechanism. Appl. Therm. Eng. 2017; 110: 1172-1182. [50] Xu, X., Hwang, Y., Radermacher, R., Refrigerant injection for heat pumping/air conditioning systems: literature review and challenges discussions. International Journal of Refrigeration, 2011; 34(2): 402-415. [51] Dutta, A., Yanagisawa, T., Fukuta, M., An investigation of the performance of a scroll compressor under liquid refrigerant injection. Int. J. of Refrigeration 2001; 24: 577-587. [52] Lin, Z., Hrnjak, P., Thermosiphon with vapor separation: experimental comparison to conventional type. Appl. Therm. Eng. 2017; 121: 879-886. [53] He, T., Mei, C., Longtin, J. P., Thermosyphon-assisted cooling system for refrigeration applications. Int. J. Refrigeration, 2017; 74: 165-176. [54] Abu-Mulaweh, H. I., Design and performance of a thermosiphon heat recovery system. Appl. Therm. Eng. 2006; 26(5–6): 471477. [55] Borton, D.N., and D.H. Walker. Demand Defrost Controller for Refrigerated Display Cases. United States Patent 4993233. February 19, 1991. [56] Thybo, C., B.D. Rasmussen, and R. Izadi-Zamanabadi. Detecting Air Circulation Faults in Refrigerated Display Cabinets. New Technologies in Commercial Refrigeration. Urbana, IL: International Institute of Refrigeration, 2002. 211-217. [57] Reuter, H., Anderson, N., Performance evaluation of a bare tube air-cooled heat exchanger bundle in wet and dry mode. Appl. Therm. Eng. 2016; 105: 1030-1040. [58] Aljuwayhel, N. F., Numerical and Experimental Study of the Influence of Frost Formation and Defrosting on the Performance of Industrial Evaporator Coils (Ph.D. thesis). University of Wisconsin-Madison. 2006. [59] Kerschbaumer, H.G., Analysis of the Influence of Frost Formation on Evaporators and of the Defrost Cycles on Performance and

26

ACCEPTED MANUSCRIPT

Power Consumption of Refrigeration Systems. XIII International Congress of Refrigeration, 1971, Washington, D.C., Paper # 3.11 [60] Mohs, W. F., Kulacki, F. A. Heat and mass transfer in the melting of frost. Springer briefs in Applied Sciences & Technology; 2015, p. 6-8. [61] Hoffenbecker, N.., Investigation of Alternative Defrost Strategies (MS thesis). University of Wisconsin-Madison. 2004. [62] Zhang, L., Dong, J., Jiang Y., Yao Y., A novel defrosting method using heat energy dissipated by the compressor of an air source heat pump. Appl Energy. 2014; 133: 101-111 [63] Qu, M., Li, T., Deng S., Fan Y., Li Z., Improving defrosting performance of cascade air source heat pump using thermal energy storage based reverse cycle defrosting method. Appl. Therm. Eng. 2017; 121: 728-736 [64] Hoffenbecker, N., Klein, S. A., & Reindl, D. T. Hot gas defrost model development and validation. Int. J. Refrigeration 2005; 28(4): 605-615. [65] Song, M., Dang C., Mao N., Deng S., Energy transfer procession in an air source heat pump unit during defrosting with melted frost locally drainage in its multi-circuit outdoor coil 2018; 164: 109-120.

27

ACCEPTED MANUSCRIPT

Highlights 1. A novel switchable dual channel cooling system was proposed. 2. Combined defrosting method for low temperature shell-tube heat exchanger was presented. 3. Uninterrupted and lower reduction performance was validated by industrial application. 4. Sensitivity characteristics of defrosting control strategy were analyzed.

1