Influence of temperature on heat transfer coefficient during moderate vacuum deep-fat frying

Journal of Food Engineering 113 (2012) 167–176

Contents lists available at SciVerse ScienceDirect

Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng

Influence of temperature on heat transfer coefficient during moderate vacuum deep-fat frying Jorge Mir-Bel, Rosa Oria, María L. Salvador ⇑ Plant Foods Research Group, University of Zaragoza, Miguel Servet 177, 50013 Zaragoza, Spain

a r t i c l e

i n f o

Article history: Received 9 April 2012 Received in revised form 1 June 2012 Accepted 18 June 2012 Available online 26 June 2012 Keywords: ‘‘Churros’’ Heat transfer Moisture loss Potato Vacuum frying

a b s t r a c t The objective of this study is to analyze the influence of temperature and reduced pressure on the convective heat transfer coefficient, h, during frying of products with different area/volume ratio. h was determined from surface temperature and moisture loss experimental data during frying of potato cylinders and ‘‘churros’’, at different oil temperatures (100, 120 and 140 °C) and moderate vacuum (19.5– 25.9 kPa). The results obtained during vacuum frying were compared with those obtained at atmospheric pressure, both for the same oil temperature (140 °C) and for the same thermal gradient (40 °C). During frying, h changes considerably, reaching a maximum between 700–1600 Wm2 K1 in vacuum frying and 800–2000 Wm2 K1 in atmospheric frying. To quantify the effect of oil temperature, pressure and size of the product on h, a parameter called ‘‘bubbling efficiency’’, BE, was defined. BE relates the bubble departure radius and the area/volume ratio of the product. An equation (the derivative of the Gompertz function) was proposed to estimate the mean heat convective coefficients for each frying condition as a function of BE (R2 = 0.957). The relation between h and BE shows a maximum corresponding to an optimal bubbling pattern. Most of the vacuum frying settings are outside this optimum, being affected by the insulation effect of bubbles covering the surface. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction Deep-fat frying is a thermal food processing method. It basically consists of cooking food by immersion in oil or fat at a high temperature, 150–200 °C. Frying is considered a complex process of food dehydration in which many factors contribute, including temperature, warm up time, oil type, size and nature of the food product, and in which many phenomena occur, such as water loss, oil uptake, crust formation, gelatinisation of starch, colour changes, shrinkage and expansion, hydrolysis of frying fats and interaction between food constituents and oxidised lipids. In recent decades much effort has been made to identify how these factors influence oil uptake on the part of the product. Reducing the high fat content in final products which sometimes reaches a third of their total weight (USDA, 2008) is an objective for meeting the demand for healthier products. Frying under reduced pressure (vacuum frying) is one way of reducing oil uptake in fried products (Shyu and Hwang, 2001; Garayo and Moreira, 2002; Granda et al., 2004; Fan et al., 2005; Shyu et al., 2005; Da Silva and Moreira, 2008; Mariscal and Bouchon, 2008; Pérez-Tinoco et al., 2008; Mir-Bel et al., 2009; Troncoso and Pedreschi, 2009; Troncoso et al., 2009), ⇑ Corresponding author. Tel.: +34 976 762739; fax: +34 976 761612. E-mail address: [email protected] (M.L. Salvador). 0260-8774/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jfoodeng.2012.06.009

and this has encouraged the appearance on the market of vacuum fried products as well as studies into how vacuum frying affects the organoleptic and nutritional properties of fried products. Simultaneous heat and mass transfer occurs during frying. Heat is transferred from the oil to the product surface initially via free convection and subsequently, during the surface boiling and falling rate stages, by turbulent convection. This is because when boiling point is reached, moisture is evaporated from the product and the vapour generation results in an ‘‘explosion’’ of bubbles (Farkas et al., 1996a,b). In vacuum frying, it would be expected that the heat transfer would be at the boiling state practically from the start of the frying since the sensible heat required to heat the surface of the product until boiling point is much lower. Furthermore, it may be supposed that there will be differences in heat transfer compared to atmospheric pressure frying given that the vapour bubbles formed will be of a different size and will be produced in different quantities and at different velocities. The heat transfer coefficient during frying is an important parameter in the modelling and calculation of fryer systems (Tangduangdee et al., 2003; Datta, 2007; Halder et al., 2007). The heat flux has a relevant role in the formation and quality of the crunch layer and on the development of characteristic properties of the end product such as the colour (browning), texture or flavour (Sosa-Morales et al., 2006). To quantify the heat transfer

168

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

coefficient, it must be taken into account that this will depend on the specific set-up of the system as there is no one standard method for its determination. There are in fact three methods to measure the convective heat transfer coefficient (Alvis et al., 2009): steady-state measurement of surface temperature, transient measurement of temperature, and heat flux measurement at the surface. A frequently applied method consists of measuring the surface temperature of the product over time and the corresponding water loss (Costa et al., 1999; Hubbard and Farkas, 1999; Budzaki and Seruga, 2005a,b; Erdogdu and Dejmek, 2010), assuming that all the heat received by the product is used in evaporating its water content, calculating previously the sensible heat fraction (Costa et al., 1999) or estimating by numerical integration the temperature data at different locations across the sample thickness (Farinu and Baik, 2007, 2008). A different method based on experimental time–moisture content and time–temperature data at the geometric centre of the product has been proposed by Yildiz et al. (2007). Another alternative is to take the values of saturation temperatures as the surface temperatures values. This approach was used by Erdogdu and Dejmek (2010) for the determination of the heat transfer coefficient during high pressure frying of potatoes. There are few studies in the literature dealing with heat transfer outside atmospheric pressure conditions. Erdogdu and Dejmek (2010), frying potato slabs under super-atmospheric pressures, found h values higher than those obtained at atmospheric conditions, but they did not study the effect of the oil temperature. In fact, working under atmospheric pressure the influence of vacuum and oil temperature on the convective heat transfer coefficient is remains unclear. Yamsaengsung et al. (2008) proposed a model to predict the heat and mass transfer during vacuum frying of potato chips, considering a constant heat transfer coefficient of 250 Wm2 K1 which was taken directly from atmospheric frying studies. More recently, Pandey and Moreira (2011), and Yagua and Moreira (2011) calculated the convective heat coefficient for the vacuum frying of potato chips. They found that h changed considerably as frying progressed, reaching a maximum between 2200 and 2650 Wm2 K1; moreover, it increased with temperature during the first moments of the frying process and diminished for extended times. They did not compare the obtained h values with atmospheric data so it is difficult to establish any conclusion about the effect of vacuum on convective heat transfer during frying. The different conditions and products used in h determination during frying make it very difficult to reach overall conclusions valid for all frying cases. For example, most published works have found that h increases with the temperature, while others have concluded that it decreases (Alvis et al., 2009). This situation could be related to the different vapour bubble patterns developed during each frying process. Although many theoretical and experimental studies have been conducted to understand the mechanisms of heat transfer during nucleate pool boiling, neither of the proposed models has been successfully applied to explain the mechanisms of heat transfer during deep-fat frying. Erdogdu and Dejmek (2010) made a first attempt to link frying heat transfer with nucleate boiling correlations but they did not find any clear relationship. In conclusion, there is no global explanation of the effect of pressure and temperature over heat transfer in frying processes. The aim of the present work is to analyze the influence of temperature and reduced pressure on the convective heat transfer coefficient during frying. Products with different area/volume ratios were fried, at temperatures between 100 and 140 °C in moderate vacuum and atmospheric conditions, and the convective heat transfer coefficients were obtained and compared.

2. Materials and methods 2.1. Sample preparation The fried products were potato cylinders and ‘‘churros’’. The latter are sometimes known as Spanish doughnuts. They are cylindrical in shape with a ridged surface, made with a fried dough of wheat flour, water and salt. The potatoes (Solanum tuberosum) of the ‘‘Agria’’ cultivar were obtained from local distributors and stored at 7 °C and 90% relative humidity until subsequent use. After 24 h at room temperature cylinders of different radii (1 and 0.5 cm respectively) and 5 cm in length were extracted from the middle of the potato, using a metal punch. The area/volume ratio of the potato cylinders were 4.43 and 2.40 cm1 for the thin and thick size, respectively. The initial moisture content was 0.8096 ± 0.0144 (g water/g of potato). Frozen commercial readyto use ‘‘churros’’ were obtained from a local retailer (La Cocinera, Spain) and stored at 20 °C until use. They were 11 cm in length with an area/volume ratio of 7.20 cm1. Their initial moisture content was 0.6050 ± 0.0114 (g water/g of dough). Sunflower oil with a high oleic acid content specially prepared for frying was used (Titan, Koipe, Spain); the oil/product proportion was 3.5 L of oil for every 100 g of product. After all the experiments, residual oil was handled by an authorized manager for recycling. 2.2. Experimental system Frying was carried out in an electric (2000 W) vacuum cooker (Gastrovac ICC, Spain) with a capacity of 8 L. The vacuum frying system is illustrated in Fig. 1.This was thermostatically controlled to maintain the set frying temperature within ±1 °C. The cooker was equipped with a vacuum pump (model 24207, JP Selecta, Spain) allowing a minimum absolute pressure of 21 kPa, a vacuum measuring gauge (model 16012, JP Selecta, Spain) and a valve (Legris, France) to control vacuum break velocity (Mir-Bel et al., 2009). The product to be fried was placed in a wire basket to prevent it from floating in the oil. The basket can be in two positions: raised or submerged in the oil. Each frying operation consisted of an initial depressurization step with the product outside the oil (approximately 1 min), immersion of the product once the vacuum reached the target value, the frying period, subsequent removal of the product above the oil level and a cooling stage inside the cooker. This post-frying stage included gravity draining during 1 min and a subsequent pressurization stage. In the vacuum experiments, the product was fried at 27.5 kPa. This working pressure allowed water elimination

Fig. 1. Schematic of the vacuum frying system used.

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

at less than 100 °C, but above the gelatinisation temperature of potato starch, 65 °C (Karlsoon and Eliasson, 2003). The changes in pressure and temperature were recorded with a piezoresistive pressure transducer (Picovacq PT, Digiterm, France) placed inside the frying oil. Before each experiment, the oil was kept at the working temperature at the maximum vacuum level for one hour, and discarded after a maximum of five hours of use. Three frying temperatures of 100, 120 and 140 °C were used in this study. Also, experiments at atmospheric pressure were carried out at 140 °C using the same fryer but with all the valves opened. This value used at atmospheric pressure may appear somewhat lower than is usual for this type of process, but it allows a double comparison. It coincides with one of the temperatures tested under vacuum and it involves a thermal driving force, 40 °C, similar to the other (40.5 °C, in the case of 100 °C and vacuum frying). This parameter is calculated as the difference between the oil temperature and the boiling point of water at the working pressure. This has been used by other authors as a comparative value for this type of process (Mariscal and Bouchon, 2008). The frying was done in batches, corresponding to the different temperature, pressure, frying time or product. In each batch, four thin cylinders, or three thick cylinders, or three ‘‘churros’’ were fried. 2.3. Direct method used to determine h To study the temperature history during frying, three (0.5 mm diameter) K-type thermocouples (Pico technology, United Kingdom) were connected across the products slightly below their surface. Great care was taken to avoid separation of the thermocouple on extraction, as described by other authors (Costa et al., 1999). The highest temperature value was taken as the surface temperature. A data logger (Testo 177-T4, United Kingdom) and a portable computer were used to acquire real time sample temperature data. The monitoring of the temperature evolution was carried out in triplicate (three batches) for each frying temperature, pressure and product type. The temperature data shown are the average values of these three sets of measurements. Moisture content, M, was determined at different frying times between 5 s and 18 min. After each frying, the product was then removed from the oil, and the surface oil was removed with paper towels. The moisture content of the raw and fried product was determined by weight loss after drying of 3 g of ground sample in a forced convection oven (AACC, 1986). The moisture contents were divided by the initial values, Mo, to obtain the adimensional moisture. The test was performed in eight replicas. The total heat transferred by convection from oil to product is equal to the sum of energy spent on heating the sample and energy spent on water evaporation. As an approximation to the calculation of h, it is assumed that all the heat received by the product is used only for evaporating water (an approximation that has greater validity the lower the Stefan numbers, in our conditions their values are between 0.06 and 0.12). In this way the convection coefficient, h, can be calculated from the expression:

h¼

dm k dt AðT 1  T s Þ

169

2.4. Statistical analysis The data were analysed using an EXCEL 2007 spreadsheet (Microsoft, USA). All fittings were solved using the SOLVER function, which employs an iterative least squares fitting routine to produce the optimal goodness of fit between data and function. 2.5. Theoretical approach As an approach to dealing with the complexity of heat transfer during frying, certain parallels can be drawn with the mechanisms through which a liquid at saturation temperature may be converted to vapour by the addition of heat from a hot surface in a container in which the liquid is confined (‘‘pool boiling’’). Vapour bubbles are also formed during frying, although not from the boiling of the oil. The heat transfer occurs between the solid surface and the liquid surrounding it, but in the opposite direction. The aim of this approximation is to evaluate and quantify the influence of the vapour bubbles on the heat transfer. Heat transfer for pool boiling fluids in contact with a hot surface is complex. It depends of the interaction between generated bubbles and the heating surface. The heat transfer in these systems could be summarized by the classic curve of heat flux versus temperature difference between surface, Ts, and liquid saturation, Tsat (Bell and Mueller, 2001), Fig. 2. As the surface temperature increases, the liquid is superheated by direct contact with the solid surface and vapour bubbles are formed at preferred nucleation sites. The bubbles grow until buoyant forces pull them free from the surface. In this regime of ‘‘nucleate boiling’’, the heat flux increases as the temperature difference increases. If the release of vapour is very vigorous (at higher temperature differences) a maximum heat flux is reached, and the flow of liquid to the surface is just sufficient to supply the vapour. The vapour generated is unable to escape, the bubbles cover the surface and the heat flux falls. As a consequence of this, the convective heat transfer coefficient during pool boiling also shows a maximum (Fig. 2). The shape of the convective heat coefficient versus temperature difference curve in pool boiling is very close to those obtained during frying (Hubbard and Farkas, 1999; Costa et al., 1999; Farinu and Baik, 2007; Yagua and Moreira, 2011). Although in frying the heat flow comes from the oil instead of the solid, the

ð1Þ

where dm/dt is the rate of the water loss from the product (kg s1), k is the latent heat of the water evaporation (J kg1), A is the external area of the product (m2), T1 is the oil temperature (K) and Ts is the surface temperature of the product (K). Similar expressions were used for other authors to determine convective coefficients during frying (Budzaki and Seruga, 2005a,b; Yagua and Moreira, 2011).

Fig. 2. Typical saturated pool boiling curve. Heat flux, Q (solid line); heat transfer coefficient, h (dash line).

170

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

evolution of the bubbling pattern is equivalent. The high rate of the water loss originates a screening effect in such a way that the solid surface is practically inaccessible for the oil. From this point, the heat flux diminishes. One of the most important features of pool boiling processes that affects the heat transfer is the departure radius of the bubbles generated, Rd (Kim and Kim, 2006). This parameter has been the subject of numerous investigations. Several correlation equations that may be used to predict the departure radius of bubbles were proposed and compiled in the work of Kim and Kim (2006), based

on experimental studies involving high-speed movies of the boiling process. Many of the correlations included the Bond number, Bo, defined as:

Bo ¼

gðqoil  qv Þð2Rd Þ2

r

ð2Þ

where g is gravity, qoil is the oil density, qv is the vapour density, and r is the oil surface tension. The frying conditions (temperature, pressure, and oil properties) could modify the departure radius of the bubbles. The temperature

Fig. 3. Surface temperature during frying at different conditions (experimental and calculated data): 140 °C and atmospheric pressure (d, continuous line), 140 °C and vacuum (s, dashed dotted line), 120 °C and vacuum (h, dashed line), and 100 °C and vacuum (D, dotted line). (a) Thick potato cylinders, (b) Thin potato cylinders, and (c) ‘‘Churros’’.

171

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

dependence could be set introducing expressions for the oil surface tension and density as a linear function of temperature (e.g. such expressions for the surface tension can be found in Bouchon and Pyle, 2005, for palm olein; and for the density the function can be calculated from experimental sunflower oil data, Moreira et al. (1999). The pressure dependence of this factor could be introduced using the correlation proposed by Cole and Shulman (1966), obtained working at subatmospheric pressures:

Bo1=2 ¼

1000 P

ð3Þ

where P is the pressure in mmHg. The success of this expression is apparently a result of the fact that 1000/P approximates the combined pressure dependence of the physical properties that appear in more complex relations. For similar oil properties, Eq. (3) shows that bubbles generated in vacuum frying are bigger than in atmospheric frying. This indicates that the thermal driving force is not sufficient for comparing the heat transfer of frying processes at different pressures. Unlike the case of pool boiling, the bubbling pattern and its screening effect in frying will also be affected by the shape and size of the product. To take these factors into consideration, a characteristic parameter related to the shape and size of the product has to be included. As a first simple approach, the volume/area ratio is proposed. The quotient between the departure radius of the bubbles and the volume/area ratio, defined as ‘‘bubbling efficiency’’ (BE), is suggested as a means of quantifying the effect of the bubbles generated from the product with the available surface for the heat transfer.

BE ¼

20 s for the thin ones, and 10 s for the ‘‘churros), the temperature starts increasing again until it comes close to that of the oil. This behaviour is similar to that described by other authors for atmospheric pressure frying (Budzaki and Seruga, 2005a; Farinu and Baik, 2008). In the case of vacuum frying, the boiling temperature at the working pressure is around 60 °C. At this point the heating slope is so fast that the slowing down in the surface temperature is too short to be visible, except for the thick potato cylinders fried at 100 °C (coinciding with the case where the thermal driving force and the area/volume ratio are the lowest). Despite this effect, the surface temperature trends are similar for the different products under the same pressure and temperature frying conditions. The surface temperature values, Ts (°C), were adjusted to the following sigmoid function:

T s ¼ T f  expðA  expðB  tÞÞ

ð5Þ

where t(s) is the time, and A, B (s1) and Tf are the fitting parameters. Tf is related to the surface temperature at t = 1. The fitting parameter values and the coefficients of determination, R2, are shown in Table 1. Some of the coefficients of determination are not very close to 1, but their use simplifies the subsequent calculation of the convective heat transfer coefficients. Tf decreases for each one of the frying conditions as the volume/area of the product decreases; that is, as the ‘‘bubbling efficiency’’, BE, increases. This shows that the screening effect is greater for higher BE values. 3.2. Moisture content

ARd V

ð4Þ

3. Results and discussion 3.1. Temperature at the surface of the products The surface temperatures at different frying oil temperatures are shown in Fig. 3 for the different products: thick (a) and thin (b) potato cylinders, and ‘‘churros’’ (c). The products fried at atmospheric pressure have an initial temperature increase almost linear with time until the boiling point of water is reached (100 °C). The temperature then remains almost stable for a short period of time since the heat supplied is spent on moisture evaporation at this stage. After some time (around 30 s for the thick potato cylinders,

The moisture contents (expressed as remaining moisture related to initial moisture, M/Mo) at different frying times and oil temperatures are shown in Fig. 4 for the different products: thick (a) and thin (b) potato cylinders, and ‘‘churros’’ (c). The evolution of the water loss is similar to that which can be obtained in a drying process in which three stages can be distinguished: heating without evaporation, surface evaporation at a constant rate and a period of decreasing rate of water loss. In the cases studied, the first step can only be seen at 100 °C for the thin potato cylinders and for the ‘‘churros’’, and at the three temperatures when working in vacuum for the cylinders with the largest diameter; in no case does the stage last for more than 30 s. These data are confirmed by the experimental observations. When the oil temperature was 100 °C, a period of time passed when the cylinders were already

Table 1 Fitting parameters for Eq. (5). Frying conditions

Parameters

Product Thick potato cylinder

Thin potato cylinder

‘‘Churros’’

100 °C Vacuum

Tf (°C) A B (s1) R2

95.9 ± 0.1 0.83 ± 0.01 0.015 ± 0.001 0.990

95.1 ± 0.1 0.83 ± 0.01 0.027 ± 0.001 0.974

94.5 ± 0.1 0.86 ± 0.01 0.041 ± 0.001 0.970

120 °C Vacuum

Tf (°C) A B (s1) R2

114.9 ± 0.2 0.59 ± 0.02 0.013 ± 0.001 0.900

112.3 ± 0.1 0.99 ± 0.02 0.037 ± 0.001 0.904

112.2 ± 0.1 1.16 ± 0.01 0.061 ± 0.001 0.917

140 °C Vacuum

Tf (°C) A B (s1) R2

137.2 ± 0.3 0.61 ± 0.02 0.011 ± 0.001 0.909

135.9 ± 0.1 0.64 ± 0.01 0.023 ± 0.001 0.917

135.2 ± 0.1 0.69 ± 0.02 0.046 ± 0.001 0.916

140 °C Atmospheric

Tf (°C) A B (s1) R2

136.7 ± 0.1 0.88 ± 0.01 0.011 ± 0.001 0.977

134.6 ± 0.1 0.89 ± 0.01 0.015 ± 0.001 0.954

134.4 ± 0.1 1.17 ± 0.01 0.027 ± 0.001 0.976

172

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

Fig. 4. Evolution of moisture during frying at different conditions (experimental and calculated data): 140 °C and atmospheric pressure (d, continuous line), 140 °C and vacuum (s, dashed dotted line), 120 °C and vacuum (h, dashed line), and 100 °C and vacuum (D, dotted line). (a) Thick potato cylinders, (b) Thin potato cylinders, and (c) ‘‘Churros’’.

immersed in the oil without any bubbles being produced, while for the other conditions the bubbles began immediately after the product was placed in the oil. In this period the heat transfer between the frying medium and the product is controlled by natural convection. The period of time in which the product is confined in the cooker nearly the hot oil, until the depressurization takes, favours the heating of the product, despite the cooling effect that the reduce pressure could cause, shortening the duration of this stage compared to atmospheric pressure frying. Thus, at 30 s the M/Mo ratio is 0.990 for the larger size potato cylinders when frying

at 140 °C at atmospheric pressure, and 0.916 at the same temperature but in vacuum conditions. Other authors have found similar results. Garayo and Moreira (2002) have shown in the case of fried chips that this stage is especially short and difficult to quantify. The second stage is characterised by the surface becoming saturated because the movement of the water from the interior to the surface occurs at the same rate as the surface evaporation. Comparing the vacuum frying cases, it can be seen that the higher the temperature of the oil, the shorter the duration of this stage and the greater the rate of water loss. For the same oil temperature

173

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

the water loss occurs at a faster rate under vacuum conditions. This second stage in all the cases studied is of short duration (less than 90 s). The studies reported in the literature that identify this stage have shown that the period is so short (Gamble et al., 1987) that it can be considered negligible, especially for vacuum processes (Garayo and Moreira, 2002). When the moisture level in the product is so low that its surface is no longer wet, the third stage of a decreasing drying rate begins. In this last stage two phases can be distinguished in the product, a nucleus that retains the greater part of the remaining moisture and a semi-permeable crust that acts as a barrier to the water loss. The drying rate is controlled by moisture diffusion mechanisms. For the three samples studied, this period continues until the frying is finished. A constant moisture value is not reached at the given frying time, unlike in other types of chip product (Garayo and Moreira, 2002). Similar to what occurs in atmospheric pressure frying (Krokida et al., 2000; Farinu and Baik, 2008), as the oil temperature increases, the sample moisture content for the same frying period decreases. Most authors have also observed this effect when studying various vacuum fried products such as apple slices (Shyu and Hwang, 2001), Sepat Siam fish chips (Suwanchongsatit et al., 2004) or carrot chips (Fan et al., 2005). Only at high temperatures (165–180 °C) has it been observed that the effect of the temperature is negligible (Tan and Mittal, 2006). The size and shape of the product also significantly affects the moisture content during frying. The moisture content of products is lower for higher area/volume ratios and for the same frying time (Fig. 4) following the same tendency as in atmospheric pressure frying (Krokida et al., 2000; Farinu and Baik, 2008). The drying rate depends noticeably on the working pressure. There is a negative correlation between pressure and moisture loss rate for the same oil temperature, as can be seen for an oil temperature of 140 °C. This is because in vacuum frying, the boiling point of the water is reduced and therefore the water in the potato cylinders begins to vaporise more quickly. If we compare the rates of moisture loss in experiments with the same thermal gradient (the same difference between the oil temperature and the boiling temperature of the water at the working pressure), we can see that the moisture loss rate is greater under atmospheric pressure at 140 °C than at 100 °C under vacuum. Mariscal and Bouchon (2008) made the same comparison with apple slices and obtained the same results. According to these authors, the differences can be partly due to microstructural changes that might occur during the initial depressurization step and to the water vapour accumulation in the head space of the fryer. In this case the pressure of the system was monitored constantly and no change was registered, suggesting that the cause may be the accumulation of vapour in the interior of the products that would increase the saturation temperature, making

the effective thermal gradient lower than the nominal value. As the frying progresses, the overpressure reduces because less vapour is produced or because the degradation of the crust creates preferential paths or cracks enabling the retained vapour to escape. This would explain why at the end of frying the water loss rate under atmospheric pressure is similar to that of vacuum frying at 100 °C. The moisture loss data, during the period of the decreasing rate of water loss, were fitted to the following exponential equation as a function of frying time:

M=Mo ¼ C  expðD  tÞ

ð6Þ 1

where C (g of water/g of initial water) and D (s ) are the fitting parameters. These values together with the coefficients of determination are shown in Table 2. In all cases, R2 > 0.960, so the fit may be considered good. The derivate of Eq. (6) is used to determine the velocity of the water loss, and then the convective heat transfer coefficients by means of Eq. (1). 3.3. Convective heat transfer coefficients With the surface temperature, Ts, and the velocity of moisture loss data, dm/dt, the convective heat transfer coefficients, h, were determined using Eq. (1). Fig. 5 shows the coefficients for all the conditions studied. In all cases it can be observed that h reaches a maximum after which it decreases due to the diminishing rate of the moisture loss. The results obtained are in agreement with those of Hubbard and Farkas (1999, 2000), Budzaki and Seruga (2005a,b) and Farinu and Baik (2008), who determined the h values for atmospheric pressure frying of potato dough, sweet potato discs and potato cylinders, respectively. In these studies, the h values reached a maximum and later diminished in the falling rate stage until becoming fairly constant as frying continued. In addition, these authors concluded that the maximum convective heat transfer coefficient reached during frying increases as the oil temperature increases, even suggesting an exponential trend (Baik and Mittal, 2002). The maximum h values are around 1600 Wm2 K1 in the case of thick potato cylinders, between 1300 and 1600 Wm2 K1 for thin potato cylinders and ranging from 700 to 1100 Wm2 K1 for ‘‘churros’’. These maximum values are lower than those obtained by Yagua and Moreira (2011) in the vacuum frying of potato chips. The reason could be that chips have a higher external surface than the products used in this study; which involves a very different bubbling pattern. The values of h at atmospheric pressure ranged from 800 to 2000 Wm2 K1, and increase as the area/volume ratio increases. Comparing the h values under vacuum conditions for each product at different temperatures, it can be observed that, after

Table 2 Fitting parameters for Eq. (6). Frying conditions

Parameters

Product Thick potato cylinder

Thin potato cylinder

‘‘Churros’’

100 °C Vacuum

C (g/g) D (s1) R2

0.994 ± 0.011 0.0006 ± 0.00002 0.979

0.989 ± 0.007 0.0013 ± 0.00003 0.998

0.975 ± 0.019 0.0021 ± 0.0001 0.985

120 °C Vacuum

C (g/g) D (s1) R2

0.956 ± 0.017 0.0007 ± 0.00005 0.960

0.878 ± 0.029 0.0015 ± 0.0001 0.979

0.938 ± 0.024 0.0028 ± 0.0002 0.985

140 °C Vacuum

C (g/g) D (s1) R2

0.935 ± 0.016 0.0007 ± 0.00005 0.974

0.851 ± 0.029 0.0019 ± 0.0002 0.986

0.912 ± 0.026 0.0040 ± 0.0003 0.986

140 °C Atmospheric

C (g/g) D (s1) R2

0.959 ± 0.014 0.0006 ± 0.00003 0.962

0.902 ± 0.023 0.0014 ± 0.0001 0.979

1.032 ± 0.023 0.0037 ± 0.0002 0.985

174

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

Fig. 5. Convective heat transfer coefficients during frying: 140 °C and atmospheric pressure (continuous line), 140 °C and vacuum (dashed dotted line), 120 °C and vacuum (dashed line), and 100 °C and vacuum (dotted line). (a) Thick potato cylinders, (b) Thin potato cylinders, and (c) ‘‘Churros’’.

the maximum, the coefficients tend to approach one to each other, being at the last frying stage higher as the temperature decreases. A similar phenomenon was also described in studies at atmospheric (Hubbard and Farkas, 1999) and vacuum frying (Yagua and Moreira, 2011), and it was related to the different evolution of water loss for each temperature. This would explain the inverse

relation obtained by Yildiz et al. (2007) between oil temperature and the convective heat transfer coefficient, since they used a numerical model that discards the initial period of time and estimates an average h value. Although the qualitative behaviour is similar, the numerical data show important differences comparing h values at vacuum

175

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

and atmospheric conditions for the three products. In the case of thick potato cylinders, the h values are higher for vacuum than for atmospheric frying for all temperatures and during all the frying time. For the thin cylinders, atmospheric h values are lower during the first 7 min; after that, they approach vacuum h values, even exceeding them during the fall period. The frying of ‘‘churros’’ follows a similar pattern to that of the thin cylinders. The atmospheric h values are lower for the first 3 min, increasing sharply and markedly surpassing the vacuum h values during the remaining frying time. To explain the influence of pressure, oil temperature and area/ volume ratio of the product on the convective heat transfer coefficients, the ‘‘bubbling efficiency’’, BE, was calculated for the twelve conditions studied by means of Eq. (4). Rd values were estimated from Eq. (2) once Bo was calculated from Eq. (3). The results of these calculations are shown in Table 3. Furthermore, the average convective heat transfer coefficients were determined for each condition, until the moisture reached 1/3 of the initial value. The h values increase with the BE parameter up to values of around 1.5, from which point they begin to decrease (Fig. 6). The relation between the ‘‘bubbling efficiency’’, BE, and the covered surface, CS, could be expressed with a Gompertz function. This sigmoid expression should reflect that the covered surface is low for low BE values, sharply increases for medium values as the bubbling increases enhancing the convective heat transfer coefficient, and tending to an asymptotic final value corresponding to a full screening effect at high BE values. The convective heat transfer coefficient could be calculated as the derivative of the Gompertz function, Eq. (7):

h¼

dCS d½P1  expðP2  expðP3  BEÞ ¼ dBE dBE

ð7Þ

where P1, P2 and P3 are the fitting parameters. The proposed model gives a good estimation for the relation between h and BE, with a coefficient of determination of R2 = 0.957, being P1 = 3808 ± 99 Wm2 K1, P2 = 3.28 ± 0.11 and P3 = 0.70 ± 0.02. Fig. 6 can help to understand the results of the convective heat transfer coefficients obtained (Table 3). In the case of thick potato cylinders, all the BE values are under 1.5 so that the bubble screening is low and the use of vacuum enhances the heat transference. The vacuum h values are thus higher than those obtained under atmospheric frying as the Bo number in vacuum conditions is higher (Eq. (3)), and as a consequence Rd (Eq. (2)) and BE (Eq. (4)) are also higher. On the other hand, before the maximum h is reached a decrease in BE causes a drop in the h values, and this occurs when the temperature increases. In the case of thin potato cylinders, the BE values corresponding to vacuum frying are higher than 1.5. This indicates that bubble screening is important in these working conditions, with two important consequences: the differences between atmospheric and vacuum frying h values are lower than for thick cylinders, and the temperature improves the heat transfer because a decrease in BE values in this part of the curve leads to an increase in h. For the ‘‘churros’’, the bubble efficiencies in the vacuum cases are the highest of the three products, therefore the h values are the lowest. The convective heat transfer coefficient at atmospheric frying is higher than for the potato cylinders because the BE is closer to the optimum value. To attain the better frying conditions, the ‘‘bubbling efficiency’’ parameter can help to determine how far the system is from the optimal bubbling pattern. Its inclusion in mathematical models involving heat transfer during frying can be an adequate option to quantify the effect of the oil temperature, pressure and

Table 3 Departure radius of the bubbles, Rd, bubbling efficiency, BE, and mean convective heat transfer coefficient, h, for different frying conditions. T (°C)

P (kPa)

Rd (m)

100 120 140 140

19.5 22.4 25.9 101.3

0.00588 0.00497 0.00433 0.00106

h (Wm2 K1)

BE Thick cylinders

Thin cylinders

‘‘Churros’’

Thick cylinders

Thin cylinders

‘‘Churros’’

1.41 1.19 1.04 0.26

2.58 2.19 1.90 0.47

4.22 3.57 3.11 0.76

944.41 954.12 891.04 495.00

856.52 923.35 984.09 693.02

417.65 527.00 650.44 732.04

Fig. 6. Relationship between the mean convective heat transfer coefficient, h, and the bubbling efficiency, BE: thick potato cylinders (D), thin potato cylinders (h), and ‘‘churros’’ (), h calculated as the derivative of the Gompertz function (solid line).

176

J. Mir-Bel et al. / Journal of Food Engineering 113 (2012) 167–176

volume/area ratio of the product on the mean convective heat transfer. The fluid dynamics of the system for frying chips is different, so this model could not be applied in that case. In fact, estimations with the available data for vacuum frying (1.33 kPa) of potato chips have shown poor results (Yagua and Moreira, 2011). Chips lost moisture at such a fast velocity that the time to reach 1/3 of the initial moisture was too short (<52 s) for the bubbles to have time to develop the pattern described in the present work. 4. Conclusions During frying of potato cylinders and ‘‘churros’’, the convective heat transfer coefficient changes considerably reaching a maximum. The influence of temperature (140–100 °C), pressure (19.5– 101.3 kPa) and size of the products (area/volume ratios between 2.4 and 7.2 cm1) may be quantified with the use of a parameter called the ‘‘bubbling efficiency’’, BE, which relates the bubble departure radius with the area/volume ratio of the product. As this parameter increases, the convective heat transfer coefficient increases up to a limit at which point the values begin to decrease. This inflection point can be related to the optimal bubbling pattern, as it reflects an agreement between the efficient renewal of the oil boundary layer surrounding the fried product and the insulation effect of bubbles covering the surface. Acknowledgements The authors express their gratitude to the ‘Ministerio de Educación y Ciencia’ (Spain) (Project: AGL2007-64252/ALI) and to the ‘Diputación General de Aragón’ (Project: ALCOTEC 2009/0196) for providing financial support for the study. References AACC, 1986. Approved Methods of the American Association of Cereal Chemists, AACC, Minneapolis, MN. Alvis, A., Vélez, C., Rada-Mendoza, M., Villamiel, M., Villada, H.S., 2009. Heat transfer coefficient during deep-fat frying. Food Control 20, 321–325. Baik, O., Mittal, G.S., 2002. Heat transfer coefficients during deep fat frying of tofu disc. Transactions of the American Society of Agricultural Engineer 45, 1493– 1499. Bell, K.J., Mueller, A.C., 2001. Wolverine Tube Heat Transfer Data Book, Wolverine Tube Inc., available from: . Bouchon, P., Pyle, D.L., 2005. Modelling oil absorption during post-frying cooling II: solution of the mathematical model, model testing and simulations. Food and Bioproducts Processing 83 (C4), 261–272. Budzaki, S., Seruga, B., 2005a. Determination of thermal conductivity and convective heat transfer coefficient during deep fat frying of ‘‘Krostula’’ dough. European Food Research Technology 221 (3–4), 351–356. Budzaki, S., Seruga, B., 2005b. Determination of convective heat transfer coefficient during frying of potato dough. Journal of Food Engineering 66 (3), 307–314. Cole, R., Shulman, H.L., 1966. Bubble departure diameters at subatmospheric pressures. Chemical Engineering Progress Symposium Series 62, 6–16. Costa, R.M., Oliveira, F.A.R., Delaney, O., Gekas, V., 1999. Analysis of the heat transfer coefficient during potato frying. Journal of Food Engineering 39 (3), 293–299. Da Silva, P.F., Moreira, R.G., 2008. Vacuum frying of high-quality fruit and vegetablebased snacks. Lebensmittel-Wissenschaft und-Technologie-Food Science and Technology 41 (10), 1758–1767. Datta, A.K., 2007. Porous media approaches to studying simultaneous heat and mass transfer in food processes. Part II: property data and representative results. Journal of Food Engineering 80 (1), 96–110. Erdogdu, F., Dejmek, P., 2010. Determination of heat transfer coefficient during high pressure frying of potatoes. Journal of Food Engineering 96 (4), 528–532. Fan, L.P., Zhang, M., Xiao, G.N., Sun, J.C., Tao, Q., 2005. The optimization of vacuum frying to dehydrate carrot chips. International Journal of Food Science and Technology 40 (9), 911–919. Farinu, A., Baik, O.-D., 2007. Heat transfer coefficients during deep fat frying of sweetpotato: effect of product size and oil temperature. Food Research International 40 (8), 989–994.

Farinu, A., Baik, O.-D., 2008. Convective mass transfer coefficients in finite element simulations of deep fat frying of sweetpotato. Journal of Food Engineering 89 (2), 187–194. Farkas, B.E., Sing, R.P., Rumsey, T.R., 1996a. Modeling heat and mass transfer in immersion frying. Part I model development. Journal of Food Engineering 29 (2), 211–226. Farkas, B.E., Sing, R.P., Rumsey, T.R., 1996b. Modeling heat and mass transfer in immersion frying. Part II model solution and verification. Journal of Food Engineering 29 (2), 227–248. Gamble, M.H., Rice, P., Selman, J.D., 1987. Relationship between oil uptake and moisture loss during frying of potato slices from c.v. record UK tubers. International Journal of Food Science and Technology 22, 233–241. Garayo, J., Moreira, R.G., 2002. Vacuum frying of potato chips. Journal of Food Engineering 55 (2), 181–191. Granda, C., Moreira, R.G., Tichy, S.E., 2004. Reduction of acrylamide formation in potato chips by low-temperature vacuum frying. Journal of Food Science 69 (8), E405–E411. Halder, A., Dhall, A., Datta, A., 2007. An improved, easily implementable, porous media based model for deep-fat frying. Trans IChemeE, Part C, Food and Bioproducts Processing 85 (C3), 209–219. Hubbard, L.J., Farkas, B.E., 1999. A method for determining the convective heat transfer coefficient during immersion frying. Journal of Food Engineering 22 (3), 201–214. Hubbard, L.J., Farkas, B.E., 2000. Influence of oil temperature on convective heat transfer during immersion frying. Journal of Food Processing and Preservation 24 (2), 143–162. Karlsoon, M.E., Eliasson, A.Ch., 2003. Gelatinization and retrogradation of potato (Solanum tuberosum) starch in situ as assessed by differential scanning calorimetry (DSC). Lebensmittel-Wissenschaft und-Technologie-Food Science and Technology 36, 735–741. Kim, J., Kim, M.H., 2006. On the departure behaviors of bubble at nucleate pool boiling. International Journal of Multiphase Flow 32, 1269–1286. Krokida, M.K., Oreopoulou, V., Maroulis, Z.B., 2000. Water loss and oil uptake as a function of frying time. Journal of Food Engineering 44 (1), 39–46. Mariscal, M., Bouchon, P., 2008. Comparison between atmospheric and vacuum frying of apple slices. Food Chemistry 107 (4), 1561–1569. Mir-Bel, J., Oria, R., Salvador, M.L., 2009. Influence of the vacuum break conditions on oil uptake during potato post-frying cooling. Journal of Food Engineering 95 (3), 416–422. Moreira, R.G., Castell-Perez, M.E., Barrufet, M.A., 1999. Deep-Fat Frying: Fundamental and Applications. Aspen Publishers, Gaithersburg, MD. Pandey, A., Moreira, R.G., 2011. Batch vacuum frying system analysis for potato chips. Chemical Engineering Progress Symposium Series. http://dx.doi.org/ 10.1111/j.1745-4530.2011.00635.x. Pérez-Tinoco, M.R., Pérez, A., Salgado-Cervantes, M., Reynes, M., Vaillant, F., 2008. Effect of vacumm frying on quality of pineapple chips. Journal of the Science of Food and Agriculture 88 (6), 945–953. Shyu, S.L., Hwang, L.S., 2001. Effects of processing conditions on the quality of vacuum fried apple chips. Food Research International 34, 133–142. Shyu, S.L., Hau, L.B., Hwang, L.S., 2005. Effects of processing conditions on the quality of vacuum-fried carrot chips. Journal of the Science of Food and Agriculture 85 (11), 1903–1908. Sosa-Morales, M.E., Orzuna-Espíritu, R.O., Vélez-Ruiz, J.F., 2006. Mass, thermal and quality aspects of deep-fat frying of pork meat. Journal of Food Engineering 77 (3), 731–738. Suwanchongsatit, W., Oupadissakoon, Ch., Yamprayoon, J., Jangchud, K., 2004. Frying process improvement and shelf life studies of fried boneless salted Sepat Siam. Kasetsart Journal-Natural Science 38 (6), 142–149. Tan, K.J., Mittal, G.S., 2006. Physicochemical properties changes of donuts during vacuum frying. International Journal of Food Properties 9 (1), 85–98. Tangduangdee, C., Bhumiratana, C.S., Tia, S., 2003. Heat and mass transfer during deep-fat frying of frozen composite foods with thermal protein denaturation as quality index. Science Asia 29, 355–364. Troncoso, E., Pedreschi, F., 2009. Modeling water loss and oil uptake during vacuum frying of pre-treated potato slices. Lebensmittel-Wissenschaft und-Technologie – Food Science and Technology 42 (6), 1164–1173. Troncoso, E., Pedreschi, F., Zuñiga, R.N., 2009. Comparative study of physical and sensory properties of pre-treated potato slices during vacuum and atmospheric frying. Lebensmittel-Wissenschaft und-Technologie – Food Science and Technology 42 (1), 187–195. USDA, U.S. Department of Agriculture, Agricultural Research Service, 2008. USDA national nutrient database for standard reference. Release 21. Nutrient data laboratory home. Available from: . Yagua, C.V., Moreira, R.G., 2011. Physical and thermal properties of potato chips during vacuum frying. Journal of Food Engineering 104, 272–283. Yamsaengsung, R., Rungsee, Ch., Prasertsit, K., 2008. Simulation of the heat and mass transfer processes during the vacuum frying of potato chips. Songklanakarin Journal of Science and Technology 30 (1), 109–115. Yildiz, A., Palazoglu, T.K., Erdogdu, F., 2007. Determination of heat and mass transfer parameters during frying of potato slices. Journal of Food Engineering 79 (1), 11–17.