at SciVerse ScienceDirect

Applied Thermal Engineering 51 (2013) 239e245

Contents lists available

Applied Thermal Engineering

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

An experimental study on defrost heaters applied to frost-freehousehold refrigerators

Cláudio Melo*, Fernando T. Knabben, Paula V. PereiraPOLO Research Laboratories for Emerging Technologies in Cooling and Thermophysics, Department of Mechanical Engineering,Federal University of Santa Catarina, Campus Universitario, Trindade, 88040970 Florianópolis, SC, Brazil

h i g h l i g h t s

< A purposed designed testing apparatus was constructed and used during the experiments.< The defrost efficiencies of three distinct heaters were measured and compared.< The calrod heater was found to be the best option for the refrigerator under study.

a r t i c l e i n f o

Article history:Received 6 June 2012Accepted 25 August 2012Available online 10 September 2012

Keywords:DefrostHeaterRefrigeratorEfficiencyEvaporator

* Corresponding author. Tel.:þ55 48 3234 5691; faE-mail address: melo@polo.ufsc.br (C. Melo).

1359-4311/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.applthermaleng.2012.08.04

a b s t r a c t

In this study the performance of defrost systems applied to household refrigerators was experimentallyevaluated through a purpose-built testing apparatus. The test bench is comprised of a calorimeter,a refrigerated cabinet and a humidifying system. Three distinct types of electrical heaters (distributed,calrod and glass tube) and three actuation modes (integral power, power steps and pulsating power)were investigated. The experiments were carried out under controlled conditions in order to ensure thesame frost accumulation pattern over the evaporator. It was found that the defrost efficiency of the threetypes of heaters is practically the same for each operating mode. The highest efficiency of approximately48% was obtained with the glass tube heater operating in power steps. However, this heater reached thehighest temperature levels. The calrod heater seems to be the most appropriate not only because of itsefficiency, which is compatible with the other options, but also due to its low cost and easy installation.However, standardized energy consumption tests need to be carried out before this finding can begeneralized.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Frost is essentially a porous medium composed of humid air andice crystals, formed from the desublimation of the water vaporcontained in the air stream. In a typical frost-free refrigerator frostshould only form over the evaporator surface, which, unfortunately,rarely happens [1]. Frost formation is a common phenomenonobserved in household refrigerators, being strongly dependent onthe infiltration of warm and humid air due to periodic door open-ings and through the door gaskets.

The accumulation of frost over the evaporator surface reducesthe cooling capacity and consequently the overall performance ofthe refrigerator. Such a loss occurs because the frost layer not onlyincreases the thermal resistance between air and evaporator butalso the air side pressure drop, decreasing the fan-supplied airflow

x: þ55 48 3234 5166.

All rights reserved.4

rate. To avoid the evaporator blockage and the capacity reduction,defrosts must be periodically carried out. During defrost thecompressor and fan remain off and part of the heat provided by theelectrical heater is transferred to the refrigerated compartments.Therefore, the compressor on-time will be longer after defrost inorder to compensate for this extra thermal load. Additionally, thedefrost system increases the air side pressure drop, which reflectsnegatively on the product performance. Then, although needed, thedefrost system increases the energy consumption of the householdrefrigerators.

Hot-gas, reverse-cycle and electrical heaters are the mostcommon defrost techniques. The first two consume less energy butrequire some system modifications, which makes them unfeasiblefor household refrigerators. The electrical heaters consume moreenergy, increase the air side pressure drop and are subjected tocorrosion problems, but are much cheaper and therefore are themost typical method for frost removal in low capacity evaporatorsof frost-free refrigerators [2], which are the scope of this study.

Nomenclature

Romanc specific heat, kJ/kg KE energy, kJh water fusion latent heat, kJ/kgk student factor, dimensionlessm frost mass, g or kgt time, s or minT temperature, �CU expanded propagated uncertaintyu combined uncertaintyW defrost power, W

Greekh defrost efficiency, dimensionlesss standard deviation

Subscriptsd defrostfz freezerh heateri iceid idealmelt meltp constant pressurer realsl solid to liquidw evaporator wall

C. Melo et al. / Applied Thermal Engineering 51 (2013) 239e245240

The most common types of heaters are: (i) aluminum tube(distributed), (ii) glass tube, and (iii) tubular metal sheathed (cal-rod). The aluminum heaters, shown in Fig. 1a, are widespread usedin household frost-free refrigerators. In this case conduction heattransfer is dominant, since the heater is in direct contact with theevaporator fins. Such heaters, besides providing a uniform heatdistribution, do not increase the tube temperature substantially. Forthis reason they are recommended for application with the refrig-erant HC-600a, especially when the refrigerators are testedaccording to the ISO 8561 [3] standard, which uses the temperatureof the warmest package contained in the freezer compartment asthe reference for the energy consumption measurement.

In addition, the distributed heater rarely damages the internalplastic and polystyrene parts near the evaporator. On the otherhand, it considerably increases the air side pressure drop, maypresent corrosion problems [2] and its manufacturing and instal-lation procedures are more complex.

The glass tube heaters have quite a simple design, as shown inFig. 1b, do not present corrosion problems but increase the air sidepressure drop because they are installed at the evaporator leadingedge. In addition they require a protection cover to avoid theshattering of the glass when hit by melted water droplets, whichincreases further the pressure drop. The calrod heaters are alsoinstalled at the evaporator leading edge, with a negative impact onthe air side pressure drop (see Fig. 1c). Such heaters might overheatthe plastic parts if the heat flux exceeds a certain limit. On the otherhand they are extremely simple and easy to install. Both calrod andglass tube heaters dissipate heat mostly by radiation and convec-tion and may reach temperatures of over 300 �C whereas thedistributed models rarely exceed the 100 �C barrier.

A large number of publications have focused on the defrostprocess in evaporators for commercial applications [4e9]. Stoecker[4] conducted one of the first studies on hot-gas defrosting,providing recommendations for the appropriate design of the by-pass lines. Niederer [5] observed that only 15e25% of thereleased heat is effectively used during the defrosting. Cho et al.[6] compared two defrost strategies in a three-evaporator system:hot-gas by-pass and cycle defrost. The authors reported that thetemperature fluctuations within the refrigerated compartmentswere lower when the hot-gas technique was adopted. Byun et al.[7] studied the effect of the refrigerant by-pass mass flow rate onthe performance of hot-gas defrost systems. They concluded thatthe fluctuations in the operating conditions of the refrigerationsystem increase with the by-pass mass flow rate. Minglu et al. [8]installed a heat exchanger with PCM (phase change material) atthe internal coil of a heat pump whose defrost was performed byreversing the cycle. They verified that the prototype developed

attenuated the temperature variations inside the occupied envi-ronment, improved the thermal comfort and reduced the defrosttime. Dong et al. [9] carried out a theoretical and experimentalstudy to investigate the time and efficiency of a heat pump reverse-cycle defrost process. The authors observed that when the internalcoil fan was kept on during the defrost efficiencies up to 60% wereobtained. They emphasized, however, that removing more energyfrom the internal environment to improve the defrost process couldaffect the thermal comfort, which would require the use of alter-native sources of thermal energy, such as PCMs.

The heat exchangers used in the above-mentioned studiesdiffer from those employed in domestic refrigerators. The face areais considerably larger, the evaporation temperature is higher andthe temperature, humidity and airflow profiles at the evaporatorinlet are uniform. In duplex frost-free refrigerators the evaporatoris submitted to two air streams, one from the freezer compart-ment (colder, dry and with a higher airflow) and another from thefresh-food compartment (hotter, humid and with a lower airflow),which causes non-uniform frost accumulation over the evaporator.In spite of the large amount of publications related to defrost incommercial tube-and-fin heat exchangers, when it comes todomestic appliances the literature is scarce. Notable publicationsinclude those by Kim et al. [2], Bansal et al. [10] and Ozkan et al.[11]. Kim et al. [2] conducted a comparative study of differenttypes of defrost heaters applied to a side-by-side refrigerator. Theauthors provided no information on the operating conditions oron how these conditions were controlled. It is worth mentioningthat for an appropriate comparison it is essential that the frostmass is the same for all tests and this can only be ensured by anappropriate control system. Bansal et al. [10] studied the heatdistribution from a calrod heater within a vertical freezer. Theyestimated the distribution through a mathematical model basedon the association of convection and radiation thermal resistancesand noted that only 32% of the total energy is effectively absorbedby the frost layer. Ozkan et al. [11] experimentally analyzed thedefrost process in a two-compartment refrigerator with analuminum distributed heater. The authors, however, removed thewall that separates the compartments (mullion), turning a dual-refrigerator into an all-refrigerator. With the aid of an endo-scopic camera they observed that the frost formation is moreintense on the first rows of the evaporator, which indicates thatmost of the defrost power should be dissipated in this region. Thisfinding corroborates the observations presented by Knabben et al.[12] where the same type of heater was tested in a similarrefrigerator. Knabben et al. [12] also noted that the gradualreduction of the heater power during the defrost processincreased the efficiency by 118%.

Fig. 1. Defrost heaters: (a) distributed, (b) glass tube, (c) calrod.

C. Melo et al. / Applied Thermal Engineering 51 (2013) 239e245 241

In this context, the present study aims to experimentally inves-tigate the defrost efficiency of three distinct types of electric heaters(distributed, calrod and glass tube), submitted to operating condi-tions typically observed in real applications. Additionally, the effectof time and heater power on the defrost efficiency was explored.

2. Experimental work

2.1. Setup

A purpose-built experimental facility was constructed in orderto reproduce the operating conditions of a typical dual-compartment frost-free refrigerator. The apparatus is basicallycomprised of a calorimeter (see Fig. 2), a 263-l frost-free refriger-ator cabinet, a humidifying system and a climate-controlledchamber. The refrigerator is a top-mount type, with a 60-l freezerat the top and a 203-l fresh-food compartment at the bottom. Theoriginal cooling system was removed and replaced by the calo-rimeter. The calorimeter allows the evaporating temperature to becontrolled by varying the condenser fan speed. The speed iscontrolled by a PID controller which acts based on the refrigerantpressure signal at the evaporator inlet. During all tests the evapo-rator was kept flooded with HFC-134a (i.e., no superheating) byadjusting the opening of the metering valve mounted in series withthe capillary tube.

The air temperaturewithin the refrigerator was controlled by PID-driven electrical heaters, strategically placed in the freezer and fresh-food compartments. In order to facilitate the internal temperaturecontrol the damper that controls the air distribution between thecompartmentswas held in a constant position. The evaporator surfacetemperature was measured by 10 T-type thermocouples, withamaximumuncertaintyof�0.2 �C, attachedalong the coil. Thedefrostheater temperatures were monitored by K-type thermocouples.

To accelerate the frost formation, water vapor was releasedinside the fresh-food compartment by a 400 W PID-driven elec-trical heater immersed in a 700 ml water reservoir. The PID signalwas provided by a relative humidity transducer (TESTO transmitter6681, probe 6614, with an absolute uncertainty of 1%) positioned at

Fig. 2. Schematic view of the experimental apparatus.

the center of the fresh-food compartment. The freezer air relativehumidity was not controlled because its values are intrinsicallyhigh, above 70% [12].

In order to visualize the frost formation process and evaluate thedefrost time, a 40 mm-thick insulating glass window was installedin the freezer rear wall, as illustrated in Fig. 3. The defrost powerwas varied by a 1.8 kVA varivolt and monitored by a power trans-ducer (Yokogawa 2385, with an uncertainty of �1 W).

Eleven tests were carried out to evaluate the efficiency of threedistinct defrost heaters: distributed, calrod and glass tube, withnominal powers of 160, 300 and 160W, respectively. The distributedheaterhas four rowsof tubes (5mmdiameterand380mmlength) atboth sides of theevaporator. The calrod type is comprisedof onlyonetube (5mmdiameter and380mmlength). Theglass tubeheater alsoconsists of a single tube (10 mm diameter and 310 mm length).Additionally, three operating modes were tested: integral power,power steps and pulsating power. The purpose was to analyze theeffect of different operating times and power combinations on thedefrost efficiency. Thefirstmode (integralpower) is the simplest andthemost commonly used in real applications. In this case the heateris kept on during the whole defrost process. The second mode(power steps) consists of gradually reducing the heater powerduring the defrosting period. As the frost mass decreases along thetime it is expected that less energy is needed to complete theprocess, thus increasing the defrost efficiency. The third mode(pulsating power) follows the same idea as the previous one: powermodulation during the defrosting. In this case the heater is activatedin an on-off cyclic pattern, always with nominal power.

To standardize the tests and provide a fair comparison betweenthe different options, the test conditions were kept constant toensure that not only the frost mass but also its distribution andstructure were approximately the same at the beginning of defrost.Therefore, the ambient temperature was kept at 25.0 � 0.7 �C andthe air temperature inside the freezer and fresh-food compartmentswere controlled at�18.0� 0.2 �C and 5.0� 0.4 �C, respectively. Theevaporator surface temperaturewas held at�23.0� 0.5 �C,with theevaporator fully activated, i.e., no superheating. As soon as thesteady-state conditionwas achieved the humidifying tray was filledwithwater through a special tube to avoid door openings. After that,the water heater was activated and controlled to generate a relativehumidity of 90.0 � 1.2% inside the fresh-food compartment. Afterapproximately 9 h the calorimeter, the humidifying system and theevaporator fanwereswitchedoff and thedefrostheater switchedon.The water mass from the defrost process was collected by a watertray placed near the refrigerator drain and weighted with a scale(0.1 g accuracy). The defrosting process was endedwith basis on thevisual inspection of the evaporator and also on the collected watermass, which was kept constant in all tests.

2.2. Data reduction

The different heater options and actuation modes werecompared through a parameter known as the defrost efficiency.This parameter is defined as the ratio between the ideal energy, i.e.,

Fig. 3. Test section: (a) evaporator and defrost heaters, (b) glass window behind the evaporator.

C. Melo et al. / Applied Thermal Engineering 51 (2013) 239e245242

that required to melt the frost layer, and the energy that is actuallyreleased by the electrical heater, expressed as follows:

hd ¼ Eid=Er (1)

The ideal energy, Eid, is composed of a sensible part, responsible forthe temperature rise to 0 �C, and a latent part, responsible for thephase change from solid to liquid. The real energy, Er, is the timeintegration of the power dissipated by the defrost system,

Eid ¼ mcp;iðTmelt � TwÞ þmhsl (2)

Er ¼Zt

0

Wddt (3)

where cp,i is the ice specific heat (w1.9 kJ/kg K) and Tmelt and hsl are,respectively, the water fusion temperature (0 �C) and latent heat(w333.6 kJ/kg). The parameter Tw stands for the average temper-ature of the evaporator tubes, while m is the frost mass and Wd isthe defrost power.

The defrost efficiency uncertainty was assessed based on themethodology proposed by INMETRO [13]. As the defrost efficiencyis not directly measured, but is calculated from other measuredvariables, the expanded propagated uncertainty can be calculatedby the following equation,

UðhdÞ ¼ f ðx1; x2.xnÞ ¼ k

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiXni¼1

�vhdvxi

uðxiÞ�2

vuut (4)

where k stands for the Student factor (w2), x1 to xn for themeasured variables (such as the defrost mass and evaporatorsurface temperature), and u(xi) for the combined uncertainty ofeach of these variables, calculated based on the standard deviation,s, and the accuracy of the transducer, uf,

Fig. 4. Visualization o

uðxiÞ ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiu2f þ s2

q(5)

In addition to the defrost efficiency other parameters wereanalyzed, such as the average air temperature of the freezercompartment after defrost and the maximum temperature of theheater during defrost.

3. Results

Fig. 4 illustrates the frost formation process at different timesalong the tests. It can be noted that the frost accumulation is moreintense in the central region where the supercooling degree e

defined as the difference between the air dew point temperatureand the evaporation temperature [14] e is much higher than in thelateral parts of the evaporator, where only a modest frost accu-mulation can be seen, in spite of the higher airflow rate of theseregions. It can also be seen that the spatial frost distribution isapproximately parabolic: more mass in the center and on the firstrows and less mass at the sides and on the final rows. As shown inFig. 4, after 4 h the central region is already partially clogged on thefirst rows. As a consequence, part of the air stream from the fresh-food return duct, which is warmer and more humid, is directedtoward the sides, accelerating the frost accumulation in theseregions.

Table 1 lists the eleven tests that were carried out and theirrespective time and power combinations. Other parameters arealso presented: maximum temperature of the defrost heater, Th,freezer air temperature after defrost, Tfz, evaporator surface averagetemperature after defrost, Tw, frost mass, m, and defrost efficiency,hd. The latter was calculated with an average uncertainty of �2.5%.The distributed heater temperature was measured at two points,one at the center and another at the bottom, while the calrod andglass tube temperatures were measured only at the center. Due tosome experimental problems these measurements were not per-formed for the first three tests.

f frost formation.

Table 1Experimental results.

Test Heater Mode t [min] Wd [W] Th [�C] Tfz [�C] Tw [�C] m [g] hd [%]

1 Distributed Integral 19.6 160 e �9.2 23.1 155.2 31.12 Distributed Integral 15.2 160 e �13.6 12.9 149.0 38.63 Calrod Integral 11.9 300 e �14.8 22.3 153.2 27.54 Calrod Integral 19.3 160 343.4 �11.8 16.4 155.1 34.05 Glass tube Integral 16.8 160 354.1 �13.3 9.5 152.0 36.06 Distributed Steps 28.5 160 to 20 73.8 �5.8 4.3 157.1 45.77 Calrod Steps 22.3 160 to 20 288.9 �10.0 6.8 149.9 43.48 Glass tube Steps 21.9 160 to 20 340.5 �11.2 2.2 147.4 48.09 Distributed Pulsating 22.9 160 or 0 70.4 �8.9 9.3 154.8 39.910 Calrod Pulsating 23.2 160 or 0 278.6 �10.0 9.6 155.7 42.711 Glass tube Pulsating 22.6 160 or 0 340.8 �10.5 5.0 146.2 39.8

C. Melo et al. / Applied Thermal Engineering 51 (2013) 239e245 243

In frost-free refrigerators defrost usually takes from 10 to 30minand is commonly performed with a constant power. The purpose oftest 1 was to assess the defrost efficiency using the original heaterof the product (a 160 W distributed heater) for approximately20 min. As indicated in Table 1, a low efficiency was obtained whichmeans that only a small part of the heat was actually used to meltthe frost while the rest was transferred to the compartments. It canalso be concluded that themass of water to be collected in the othertests should be close to 155 g. This valuewas thus considered as oneof the criteria for ending the defrosting process in the further tests.

Fig. 5 shows the temperature transients at the evaporator inlet,center and outlet tubes during the defrost process of test 1. It can benoted that the process occurs in three distinct stages: (i) sensibleheat transfer until the coil reaches 0 �C, (ii) latent heat transfer at0 �C, and (iii) extra sensible heat transfer until the end of theprocess. Ideally, as observed in Fig. 5, the defrosting should endafter approximately 9 min. However, this does not happen becausethe frost formation over the evaporator is not uniform. It can beseen, for instance, that the outlet tube temperature remained at0 �C for approximately 13 min, which probably means that up tothis moment frost was still melting in this region.

Considering the first test results a second test with a freeoperating time was conducted, i.e., defrost would end after therewas no frost over the evaporator and the mass of collected waterwas the same as in test 1. It was noted that after 15 min all frost wasremoved and the defrosting was thus ended. As a net effect theefficiency increased from 31.1% to 38.6%. It was also observed thatthe defrost time of test 2 was very close to the time required for theoutlet tube temperature to exceed 0 �C in test 1 (see Fig. 5), whichindicates a possible correlation between the outlet tube tempera-ture and the defrost end. This evidence was observed for all eleventests, as shown in Fig. 6 for tests 2, 5 and 10. Coincidently, in real

Fig. 5. Temperature distribution along the evaporator: test 1.

applications, this temperature, measured by a bimetallic or elec-tronic sensor (usually called the defrost terminator), is used asa parameter to end the defrosting process. According to themanufacturers, the defrost terminator, when placed at the positionwhere the evaporator takes longer to be warmed, guarantees thecomplete elimination of the frost after the temperature exceeds10 �C. The defrosting of tests 3 and 4 was performed by the calrodheater, firstly with the nominal power (300 W) and then with thesame power used for the distributed heater. It was observed thatthe efficiency of test 3 was only 27.5%, which is 11.1 percentagepoints lower than test 2. The defrost time was reduced, but thepower practically doubled, most of the energy being transferred tothe area surrounding the evaporator. Although the power was thesame, the defrost efficiency of test 4 was slightly lower than that oftest 2. In comparisonwith test 3, the power was practically half andthe defrost timewas almost double, which had a positive impact onthe process efficiency. The test that was carried out with the glasstube heater operating in the integral mode revealed a defrost effi-ciency similar to those of the same capacity. The heater tempera-ture, however, reached considerably high levels, which indicatesthe need for special safety protections. As stated in Fig. 7, the calrodand glass tube heaters reach temperatures above 300 �C (tests 4and 5) and the glass tube heater heats up more rapidly during theinitial transient period.

As the defrost energy is a function of the accumulated frostmass, tests 6, 7 and 8 were carried out. Defrost was conducted bygradually reducing the heater power in four 40 W steps, as illus-trated in Fig. 8a. The steps lengths were defined based on the testswith the integral power operating mode. In this mode and with thedistributed heater (test 2) defrost lasted 15.2 min and the averageevaporator temperature reached 0 �C after 8 min. The first steplength was then defined as 8 min, remaining further 7.2 min to

Fig. 6. Evaporator outlet temperature.

Fig. 7. Heater temperature during defrost: integral mode.

Fig. 9. (a) Pulsating mode: test 11, (b) heater temperature during the defrost.

C. Melo et al. / Applied Thermal Engineering 51 (2013) 239e245244

conclude the defrosting, assuming that the defrost times for theintegral and steps mode are identical. As the use of 40-W steps wasselected, 2 min steps were adopted. An analogous procedure wasadopted for the calrod and glass tube heaters where step lengths of3 and 2minwere obtained, respectively. Unfortunately the adoptedsteps were not effective at completely eliminating the frost layer,which is evident considering the last step length. Nevertheless, asexpected, the defrost efficiency increased in all cases. The defrosttime, however, particularly for the distributed heater, increased insuch way that it led to a considerable temperature rise within thefreezer compartment after defrosting. Regardless of the result ob-tained with the steps operating mode, it is evident that throughproper refinements substantial efficiency improvements can beachieved.

It was also noted that the calrod and glass tube heaters onceagain reached high temperature levels, while the distributed heater

Fig. 8. (a) Steps operating mode, (b) heater temperature during defrost.

remained below 100 �C (see Fig. 8b). One of the advantages of thesteps mode is that the heater temperature decreases with time,thus avoiding superheating of the parts surrounding theevaporator.

The purpose of tests 9, 10 and 11 was to explore the effect of thepulsating operating mode, as illustrated in Fig. 9a. The length ofeach pulse was defined applying the same procedure used for thestep length. It can be observed that the pulsating mode providesdefrost efficiencies which are higher than the integral mode butlower than the steps mode. It can also be seen that with thepulsating mode the defrost time with the distributed heaterbecame lower than that obtained with the steps mode, which wasreflected in the freezer air temperature. Fig. 9b shows the heatertemperatures during defrost. Once again it can be observed that thecalrod and glass tube heaters reached considerably high tempera-tures during the initial transient period. However, during the

Fig. 10. Comparative: defrost efficiency.

C. Melo et al. / Applied Thermal Engineering 51 (2013) 239e245 245

cycling period the average temperature was substantially reduced,evidencing an advantage of this mode compared to the integralpower mode.

4. Conclusions

Considering the three operating modes investigated in thisstudy, the power steps mode was found to be the most efficient, asillustrated in Fig. 10 (tests 6, 7 and 8). In general it was concludedthat all three types of heaters presented practically the samedefrost efficiency in each operating mode. When operating in thesteps mode, the glass tube heater provided the highest efficiency ofapproximately 48%, although the efficiency values obtained withthe other heaters were very close. However, both defrost time andfreezer air temperature after defrost were higher. The calrod andglass tube heaters did not warm the freezer as the distributed onedid, but reached considerably higher temperatures, which couldcontribute towarming up the freezerwhen the fan restarts, in a realapplication. Considering that the defrost efficiency of the calrodheater is very close to the values obtained with the other types andthat it is less expensive and easier to install, one can conclude thatthis type can be recommended for installation in a real application.It can also be concluded that it is possible to achieve significantimprovements in the defrost efficiency by testing differentcombinations of time and power. It should be noted however thatthis analysis is limited, as previously stated. The gradual powerreduction requires an operating time increase, which cannot belong, since the compressor should not remain off for extendedperiods. If this happens, the air temperature within the compart-ments rises and the compressor run time increases after defrost,thus increasing the energy consumption. Therefore, it is recom-mended that standardized energy consumption tests are carriedout in order to identify the impact of the defrost efficiency on therefrigerator performance.

Acknowledgements

This study was carried out at the POLO Laboratories underNational Grant No. 573581/2008-8 (National Institute of Science

and Technology in Cooling and Thermophysics) funded by theBrazilian Government Agency CNPq. The authors thank WhirlpoolS.A for the financial support. The authors also thank Rafael Gõesand Rodrigo Freitas for their help with the experiments.

References

[1] C.J.L. Hermes, R.O. Piucco, J.R. Barbosa, C. Melo, A study of frost growth anddensification on flat surfaces, Experimental Thermal and Fluid Science 33(2009) 371e379.

[2] Y. Kim, A. Tikhonov, Y. Shin, J. Lee, Experimental study on high performancedefrosting heater for household refrigerator, 13th International Heat Confer-ence, Sydney, Australia, 2006.

[3] ISO 8561, Household Frost-free Refrigerating Appliances e Refrigerators,Refrigerator-freezers, Frozen Food Storage Cabinets and Food Freezers Cooledby Internal Forced Air Circulation e Characteristics and Test Methods, Inter-national Organization for Standardization, Geneva, Switzerland, 1995.

[4] W.F. Stoecker, Selecting the size of pipes carrying hot gas to defrost evapo-rators, International Journal of Refrigeration 7 (4) (1984) 225e228.

[5] D.H. Niederer, Frosting and defrosting effects on coil heat transfer, ASHRAETransactions 82 (1) (1986) 467e473.

[6] H. Cho, Y. Kim, I. Jang, Performance of a showcase refrigeration system withmulti-evaporator during on-off cycling and hot-gas bypass defrost, Energy 30(10) (2005) 1915e1930.

[7] J.S. Byun, J. Lee, C.D. Jeon, Frost retardation of an air-source heat pump by thehot gas bypass method, International Journal of Refrigeration 31 (2) (2008)328e334.

[8] Q. Minglu, X. Liang, S. Deng, J. Yiqiang, Improved indoor thermal comfortduring defrost with a novel reverse-cycle defrosting method for air sourceheat pumps, Building and Environment 45 (11) (2010) 2354e2361.

[9] J. Dong, S. Deng, J. Yiqiang, X. Liang, Y. Yao, An experimental study ondefrosting heat supplies and energy consumptions during a reverse cycledefrost operation for an air source heat pump, Applied Thermal Engineering37 (2012) 380e387.

[10] P. Bansal, D. Fothergill, R. Fernandes, Thermal analysis of the defrost cyclein a domestic freezer, International Journal of Refrigeration 33 (3) (2010)589e599.

[11] D.B. Ozkan, P. Ozil, C. Inan, Experimental investigation of the defrostingprocess on domestic refrigerator finned tube evaporators, Heat TransferEngineering 33 (6) (2012) 548e557.

[12] F.T. Knabben, C.J.L. Hermes, C. Melo, Numerical and experimental investiga-tion of the frosting and defrosting processes in no-frost evaporators, XICongress of Refrigeration and Air Conditioning e CIAR, Mexico, 2011.

[13] INMETRO, Guide to Express Measurement Uncertainties, third ed., ABNT,INMETRO, Rio de Janeiro, Brasil, 2003.

[14] R.O. Piucco, C.J.L. Hermes, J.R. Barbosa, C. Melo, A study of frost nucleationon flat surfaces, Experimental Thermal and Fluid Science 32 (8) (2008)1710e1715.