JFS: Food Engineering and Physical Properties

Bigelows General Method Revisited: Development of a New Calculation Technique

R. SIMPSON, S. ALMONACID, AND A. TEIXEIRA ABSTRACT: This article describes the development of a comprehensive procedure for broader application of the General Method in calculating thermal processes. In addition to the capability of calculating lethality (or process time to achieve a specified lethality) for a given set of heat-penetration data, the procedure can also calculate process time or lethality for process conditions different from those of the original heat-penetration tests. The new procedure was experimentally tested against the Ball Formula Method and numerical finite difference methods. It often safely predicted shorter process times than the Ball Formula Method by accurately accounting for slow comeup and cool-down phases of the retort process. Keywords: general method, formula method, process evaluation techniques

Introduction

HERMAL PROCESSING IS AN IMPORTANT

method of food preservation in the manufacture of shelf-stable canned foods and has been the cornerstone of the foodprocessing industry for more than a century. Thermal process calculations, in which process times at specified retort temperatures are calculated to achieve safe levels of microbial inactivation (lethality), must be carried out carefully to ensure public health safety. However, overprocessing must be avoided because thermal processes also have a detrimental effect on the quality (nutritional and sensorial factors) of foods. Therefore, the accuracy of the methods used for this purpose is of importance to food science and engineering professionals working in this field. The first procedure to calculate thermal processes was developed by W.D. Bigelow in the early part of the 20th century and is usually known as the General Method (Bigelow and others 1920). The General Method makes direct use of the time-temperature history at the coldest point to obtain the lethality value of a process. The procedure was carried out graphically using a plot of lethal rate against time to produce a lethality curve, the area beneath which corresponded to the accumulated lethality delivered by the process. If more or less lethality were required, the procedure was repeated with an estimate of the cooling portion of the cold spot temperature (cooling profile) advanced or retarded on a trial-and-error basis until the desired lethality was achieved. This is the reason why this method was known as the graphical trial-and-error method (Stumbo 1973).

Bigelows procedure earned the name General method because it applies to any product/process situation. Since it relies solely on the measured cold spot temperature, it is blind to process conditions, mode of heat transfer, product properties, or container size and shape. This immunity to product/process conditions has always been the strength of the General Method, in addition to its unquestioned accuracy. For this same reason, the greatest limitation of the General Method was that it could be used only to calculate process times for the same retort temperature used in the heat penetration test from which the cold spot temperature profile was obtained. Thus, it has limited predictive power (Pham 1987). Over time, several improvements were introduced to the original General Method, such as those contributed early on by Ball (1928) and Schultz and Olson (1940), and then later by Patashnik (1953) and Hayakawa (1968). The lack of programmable calculators or personal computers until the latter part of the 20th century made this method very long, tedious, and impractical for most routine applications, and it soon gave way to formula methods offering shortcuts. In response to this need, a semi-analytic method for thermal process calculation was developed and proposed to the scientific community by Ball (1923). This is the wellknown Formula Method, and works in a different way from the General Method. It makes use of the fact that the difference between retort and cold spot temperature decays exponentially over process time after an initial lag period. Therefore, a semilogarithmic plot of this temperature difference

over time (beyond the initial lag) appears as a straight line that can be described mathematically by a simple formula, and related to lethality requirements by a set of tables that must be used in conjunction with the formula. However, there are several assumptions made that cause the method to lose accuracy in many situations. According to Holdsworth (1997), most Formula Methods have been applied to metallic cans or glass jars that can be processed in pure steam or water-cook retorts with rapid come-up-times. Recent developments with retortable flexible pouches and semirigid bowls and trays have made it necessary to reexamine process calculation methods. These packages are processed with steam-air mixtures in the system and often require relatively slow come-up times, which can introduce additional error with use of formula methods. Most workers in this field will agree that the General Method is more accurate than the Formula Method, but the popularity of the Ball Formula Method as a tradition throughout the food-canning industry continues to be overwhelming (Merson and others 1978). According to Teixeira (1992), the limiting factors that historically deterred the use of the Bigelow General Method have long since been overcome with the advent of programmable calculators and personal computers. The goal of this study was to reintroduce the General Method as a more accurate, powerful, and easy-to-use method of thermal-process calculation. Specific objectives were as follows: Develop a procedure that would integrate the lethality calculation by the General
2003 Institute of Food Technologists
Further reproduction prohibited without permission

Food Engineering and Physical Properties

1324

JOURNAL OF FOOD SCIENCEVol. 68, Nr. 4, 2003

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1324

7/23/2003, 12:02 PM

General method revisited . . .

Table 1Can dimensions and processing conditions for Figure 1 to 7 Figure 1 2 3 4 5 6 7
a TRT(t)

Can size (mm) 52 74 99 52 52 99 74 38 116 141 38 38 141 116

Can (common name) 70 g tomato paste Nr 1 tall Liter 70 g tomato paste 70 g tomato paste Liter Nr 1 tall

TRTa,b (C) 116 121 121 116 116 121 130

Pt (min)
50 90 95 28 13 70 40.1

Fo (min)

Valuec

Table 2Process lethality and process time calculated from experimental data presented in Figure 8 using RGM and FM Evaluation Technique FM RGM

9.8* 28.5* 8.5* 3.1** 0.2** 2.0** 6.0

Fo Value 1 (min)
3.59 6.31

Pt 2 (min)
111 102

1 F value calculated from experimental data. o 2 Calculated for a F = 6.3 (min). r

= a + bt, ( 0 < t < CUT ), where a = 40 (C) and b can be evaluated, in each case, considering the known value of a and that TRT(5) is the value reported as processing temperature per process. b TRT = Constant ( t > CUT ). cCalculated with General Method. *Fp > Fr ** F p < F r Fr = 6 min; CUT = 5 min; Conduction heated product , = 1.7 10 7 m/s 2 ; T w = 18 (C), IT = 70 (C)

5% Bentonite, 1-kg cans (98 110 mm) j h = 1.93, f h = 71.73 (min) IT = 17.8 C, TRT =121 C, T r = 121 C, T w = 24.2 C

Description and identification of the problem

A detailed procedure was developed to give practical use to the General Method proposed by Bigelow and others (1920) so that it would include the following capabilities: (1) calculation of lethality ( Fo value) for a given set of heat penetration data, (2) calculation of process time to achieve a specified lethality ( Fr value) from a given set of heat-penetration data, and (3) calculation of either process time or lethality for alternative process conditions different from those used during the original heatpenetration tests (even if there is slow come-up time) with no further experimental data. The first capability is a straightforward execution of the General Method with no involvement of heat transfer. The accumulated lethality is calculated by numerical integration of the lethal rate along the cold spot temperature profile measured in the heat penetration test (or provided by simulation of a test). The second capability requires execution of the first as a starting point. Integration must then be repeated with the cooling portion of the cold spot temperature profile advanced or retarded on a trial-and-error basis until the desired lethality is achieved.
JFS is available in searchable form at www.ift.org

Figure 1Simulated heat-penetration data for analysis (Fp > F r) for Case 1 Situation 1 Vol. 68, Nr. 4, 2003JOURNAL OF FOOD SCIENCE 1325

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1325

7/23/2003, 12:02 PM

Food Engineering and Physical Properties

Method with principles of the heattransfer theory. Demonstrate its ability to evaluate processes at different conditions from those used in heat-penetration tests (retort temperature, initial temperature, and so on). Demonstrate its ability to take into account slow come-up and cool-down phases. Demonstrate that the procedure performs with at least the same ease of use and reliability as the Formula Method but with better accuracy.

The true profile of these cooling curves will be altered on the basis of when the onset of cooling occurs because different internal temperature distributions change continuously during heating. At best, these cooling temperature profiles are only crudely and conservatively estimated in the traditional use of the General Method. Herein lies one of the existing weaknesses that was addressed by integrating heat-transfer concepts to more accurately predict the true alternate cooling temperature profiles. The third capability has not been possible with traditional use of the General Method. This application requires accurate prediction of the entire cold spot temperature profile under totally different conditions of retort and/or initial product temperatures, including retorts with unusually slow comeup times. These profiles can also be accu-

rately predicted by integrating the heattransfer concepts developed here. Much of the significance of the work reported here stems from the heat-transfer concepts developed in the following section.

Heat-transfer concepts
Most mathematical models for predicting time-temperatures histories in food products at a given point normally need to assume one of the basic modes of heat transfer. Two extreme cases have their own analytical solutions: (1) perfect mixing of a liquid (forced convection), and (2) homogeneous solids (pure conduction). Most foods are an intermediate case, and these extreme solutions would give a guideline for the usefulness of temperature-time histories (profiles) developed here. Heat-transfer model for perfect mixing. For forced convection (agitated liquids), it is possible to assume that temperature inside the can is uniformly distributed but time-

General method revisited . . .

Table 3Process lethality and process time calculated from experimental data of a broken-heating curve presented in Figure 9 using RGM and FM Evaluation technique FM RGM

must be the same at different TRT and/or IT: (7) (4) From Eq. 7, the dimensionless temperature ratio can also be expressed as:

Fo (min)
35 40

Value1

(min) 26 25

Pt2

Product: meat chunks, UT can 73 115 (mm) j h = 5.65, f h = 12.85 (min), f h2 = 47.96 (min) IT = 46.5 C, TRT =127.9 C, T r = 121 C, T w = 25.9 C

1 F value calculated from experimental data. o 2 Calculated for a F = 6.0 (min). r

Slow come-up time with perfect mixing. Eq. 5 was derived from Eq. 2, solving an ordinary differential equation and assuming a linear retort temperature profile (that is, simulating temperature profile during come-up time).

dependent. A transient energy balance, taking the container as a system, gives: (1) (5) where retort temperature is time-dependent and expressed as: TRT (t) = a + bt and Eq. 5 is valid for: 0 < t CUT. For t > CUT, temperature T (or TC.P.) can be expressed by Eq. 3 using an appropriate initial temperature (constant TRT). Provided that fh is defined as ln10 [MCp/ UA] (Merson and others 1978), Eq. 5 can be rearranged and expressed as:

(8)

Heat transfer model for pure conduction. Heat transfer for pure conduction is based on Fouriers equation and can be written as:

(2) Provided that the cans inside temperature is uniformly distributed, T also denotes the cold spot temperature (T = TC.P.). Using the initial condition as T = IT at t = 0, and T at time t > 0, the integration of Eq. 2 renders:

(9) If thermal conductivity ( k) is independent of temperature and the food material is assumed isotropic, as it is for most foods at the sterilization temperature range, then Eq. 1 becomes:

Food Engineering and Physical Properties

(3) (10) The dimensionless temperature ratio for forced convection (Eq. 3) is dependent on geometry, thermal properties, and time. Therefore, the liquids aforementioned ratio (6) Further working on Eq. 6 renders: Although solutions for different geometries are not necessarily straightforward, in general, for any geometry, the dimensionless temperature ratio for constant retort temperature can be expressed as (Carslaw and Jaeger 1959):

(11) If initial temperature distribution, geometry, product (thermal properties), and time are maintained constant (just changing TRT and/or IT ), then the dimensionless temperature ratio of the solid must be the same at different TRT and/or IT: (4) It is important to point out that Eq. 11 is valid for constant retort temperature (TRT); so is Eq. 4. A simplified analytical solution for homogeneous solids confined in a finite cylinder is presented in Eq. 12 (Merson and others 1978). This simplified solution is only
JFS is available in searchable form at www.ift.org

Figure 2Simulated heat-penetration data for analysis (Fp > F r) for Case 1 Situation 2 1326 JOURNAL OF FOOD SCIENCEVol. 68, Nr. 4, 2003

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1326

7/23/2003, 12:02 PM

General method revisited . . .

valid for long periods of time (after the initial lag period when the Fourier number is greater than 0.6), in addition to assuming a Biot number over 40 (meaning that the external heat resistance is negligible in comparison with the internal resistance).
Table 4Can dimensions and processing conditions for experimental data shown in Figure 10 Process 1 2 TRT (C) 114.5 117.6 IT (C) 17.3 63.1

F o1 (min) RGM
7.0 13.6

F o1 (min) FM
6.0 9.9

Pt2 (min) RGM

36 16

Pt2 (min) FM
40 23

Product: mussel ( Mytilus chilensis ) in brine, can format 100 22 (mm) f h = 9.11 (min) a; j h = 1.21a a Obtained from the process at TRT = 114.5 (C)

1 F value calculated from experimental data. o 2Calculated for a F = 6 (min) r

(12)

Slow come-up time with conduction heating. Gillespy (1953) and Hayakawa (1974) have developed methods to determine center temperature where the heating profile was time-dependent (for example, linear or exponential). According to Holdsworth (1997), the method is applicable to packs, which have a slow come-up, for example, conduction heating products in flexible pouches or plastic containers. Gillespy (1953) developed an equation for a slab of material being heated with a linear temperature gradient valid during come-up time. Hayakawa (1974) developed a similar expression for finite cylinders. Expressions for conduction heating products of other geometries (for example, parallelepiped) with a linear temperature gradient can be found in Carslaw and Jaeger (1959) and Luikov (1968). According to Carslaw and Jaeger (1959) and Luikov (1968) it is possible to find a dimensionless temperature ratio equation suitable for a linear heating profile during come-up time in conductive heating products. Heat-transfer model: a general approach. Although the heat-transfer mechanisms are rather dissimilar, both models (pure conduction and forced convection), within certain limitations, can be described by the same mathematical expression that was presented by Ball (1923):

Figure 3Simulated heat-penetration data for analysis (Fp > F r) for Case 1 Situation 3

(13)

Where:

As was shown by Datta (1990), the latter expression is valid not only for finite cylinders, but also for arbitrary shapes (rectangu-

Figure 4Simulated heat-penetration data for analysis (Fp < F r) for Case 2 Situation 1 Vol. 68, Nr. 4, 2003JOURNAL OF FOOD SCIENCE 1327

JFS is available in searchable form at www.ift.org

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1327

7/23/2003, 12:02 PM

Food Engineering and Physical Properties

General method revisited . . .

lar, oval shape, and so on). The main limitations are that for heat conduction foods, it is only valid for heating times beyond the initial lag period (when the Fourier number is greater than 0.6). An interesting, practical, and general conclusion that can be drawn from the heattransfer theory presented here is that Eq. 4 remains independent of container geometry and heat-transfer mode (conduction or forced convection) and requires only a constant retort temperature: (4) and for the cooling phase: ature ratio for conduction-heating products. These statements have been supported by computer-aided experiments that demonstrate that Eq. 8 is an accurate and secure way to transform data obtained for conduction-heating products subjected to a linear temperature profile during come-up time. The importance and relevance is that we will be able to transform the raw data from heat penetration tests and use the General Method, not only to directly evaluate the raw data, but also to evaluate processes at different conditions (retort temperatures, initial temperatures, longer or shorter process times) than those originally recorded. may be bigger, smaller, or equal to Fr. Therefore, 3 situations arise: Fp > Fr, Fp = Fr, and Fp < Fr. Situations such as Fp > Fr and Fp < Fr are of interest for further analysis and they will be called Case 1 (Fp > Fr) and Case 2 (Fp < Fr ). Case 1 are all those heat-penetration tests in which final lethality is bigger than the required lethality, so the processing time must be shortened, for example, to find a new processing time shorter than the real processing time, so that Fp Fr and Fp Fr is a minimum. Within this case, 3 different situations may arise. Figures 1, 2, and 3 were computer-generated to show and analyze the 3 different situations (specifications given in Table 1). In the first situation (Figure 1), shortening the process time was straightforward because with the new process time (for an F value equal to or bigger than Fr), the temperature inside the can is uniform and it can be assumed that the cooling temperature profile would be the same as the original. In the second and third situations, the problem is different and more complicated. In Figure 2, the temperature at the coldest point (for the adjusted process) would be lower than retort temperature, giving a non-uniform temperature distribution inside the can. In this situation, the temperature at the cold spot (referred to as the heating part) during the cooling phase will have inertia. To evaluate process lethality (for the adjusted process), it is necessary to generate data for the cooling phase. In this case, the use of Eq. 14 will generate data assuming no inertia. Although this is not completely accurate, the resulting error in predicted lethality coming from the cooling phase will be on the conservative (safe) side. In Figure 3, similar to Figure 2, the temperature at the cold spot for the process is lower than the retort temperature as well as for the adjusted process. Although, both curves have inertia, again the use of Eq. 14 will not be accurate but will safely predict lethality from the cooling phase for the adjusted process. In this application, the use of Eq. 14 will predict inertia but less pronounced than in the real curve. Case 2 includes all those heat-penetration tests in which final lethality is lower than required, so the processing time must be extended, that is, find a new processing time longer than the test processing time, so that Fp Fr and Fp Fr is a minimum. In this case, the cooling phase temperature profile must be displaced to the right in order to extend process time. Three situations need to be considered and analyzed; they are presented in Figure 4, 5, and 6 (specifications given in Table 1). In the first situation
JFS is available in searchable form at www.ift.org

Methods and Materials

(14)

Thermal process evaluation (to calculate lethality)

The method must allow for the calculation of the F value for a set of data obtained experimentally or by simulation. Given that the data are not continuous, the integration procedure should be done numerically (Gauss, Simpson, trapezoidal, and so on) or alternatively fit the data by interpolation method (such as cubic spline) and integrate the lethality analytically.

Although Eq. 4 is valid only for constant retort temperature profiles, Eq. 8 has shown that similar expressions for the dimensionless temperature ratio can be derived for the case of slow come-up time (for example, linear temperature rise during come-up time for forced convection heating products). Even though Eq. 8 was derived for situations in which forced convection is the dominant heating methodso as to use a single equation for data transformationthis one will also be used on products in which the ruling heating mechanism is conduction. As was previously mentioned by some authors (Carslaw and Jaeger 1959; Luikov 1968), it is feasible to derive a dimensionless temper-

Food Engineering and Physical Properties

Thermal time adjustment (to calculate process time)

To determine the processing time so that the F value obtained ( Fp) is equal or greater than the required lethality (Fr), the Fp value had to first be determined, with the original heat penetration data. This Fp value

Figure 5Simulated heat-penetration data for analysis (Fp < F r) for Case 2 Situation 2 1328 JOURNAL OF FOOD SCIENCEVol. 68, Nr. 4, 2003

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1328

7/23/2003, 12:02 PM

General method revisited . . .

(Figure 4), the process adjustment is straightforward. The two assumptions needed are that the coldest temperature could be maintained at the same level and that the new cooling temperature profile would be the same as the original one. In Figure 5 and 6, it was first necessary to generate more data for the heating process to be able to extend the process time using Eq. 13, followed by the need to generate the new cooling temperature profile. According to heat-transfer theory, the recorded cooling temperature profile will have more pronounced inertia than the new cooling temperature profile; therefore, in these 2 situations (Figure 5 and 6), the use of Eq. 14 would generate not only inaccurate but also unsafe data. To avoid this problem, Eq. 14 was applied, considering only the data starting at point A as seen in Figure 5 and 6 (A pinpoints the maximum temperature recorded in the process). Since these data lack inertia, they could generate the new cooling temperature profile on the safe side. case of Figure 7, two aspects should be carefully considered when changing processing conditions (TRT and/or IT): (1) maintain the same come-up time and (2) decide how to generate the new cooling temperature profile, specifically in the presence of different inertia. First, although it is a limitation, come-up time must be maintained. Second, the cool-down temperature data transformation could lead to a new cooling temperature profile with less inertia (for example, the new retort temperature is lower than the original). In this case, the transformed data would overestimate the F value that could result in an unsafe process. In this situation (which is very rare; Figure 7 presents an extreme situation), a safe procedure should be to follow the recommendation explained in Case 2, Figure 5 and 6. To manage the already transformed data (new processing conditions), it is necessary to follow the procedure explained in the section Thermal time adjustment (to calculate process time) to adjust the process.

Thermal-process evaluation at conditions other than those experimentally recorded (or generated by simulation)
Sometimes it is useful to obtain a process evaluation at different conditions other than those used for the original heat-penetration test and avoid or significantly reduce the number of new experiments. The new process conditions could be: initial food temperature, retort temperature, and/or cooling temperature. The new time-temperature data should be predicted using adequate mathematical models (if the type of food allows it) or using the dimensionless temperature ratio concept developed in this study that is applicable to any kind of food. The dimensionless temperature ratio concept could be used for any kind of geometry. However, in real process situations, the retort temperature is not always constant (for example, come-up time) and will impair the theoretical validity of the concept derived for dimensionless temperature ratio, as has been discussed in the literature (Shultz and Olson 1940). In the present work the retort temperature was divided into 3 parts: (1) come-up time (TRT(t) = a + bt), (2) process temperature (TRT = Constant), and (3) cooling temperature (Tw = Constant). Eq. 8 was used to transform the original data to the newest processing conditions (TRT and/or IT), assuming a linear temperature profile during come-up time; and Eq. 4 and 14 were used for the constant retort temperature (TRT) and cooling water temperature (Tw) conditions, respectively. In the
JFS is available in searchable form at www.ift.org

Figure 6Simulated heat-penetration data for analysis (Fp < F r) for Case 2 Situation 3

Figure 7Simulated heat-penetration data for analysis (changing processing conditions, TRT) Vol. 68, Nr. 4, 2003JOURNAL OF FOOD SCIENCE 1329

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1329

7/23/2003, 12:02 PM

Food Engineering and Physical Properties

General method revisited . . .

Table 5Comparison of operator process time (Pt) required for the same lethality at alternative retort temperatures calculated by RGM and FM with actual operator process time in data set generated by finite differences (F.D.) (conduction heated product) at each retort temperature. TRT (C) 110 115 125 130 F.D. (min) 436 341 245 217 RGM (min) 461 367 251 227 Error (%) 5.73 7.62 2.45 4.61 FM (min) 465 370 275 247 Error (%) 6.65 8.50 12.24 13.82

Validation
Thermal process evaluation and adjustment. To compare results from the Revisited General Method (RGM) developed in this study with those from the Formula Method (FM), sets of computer-simulated data as well as experimental data were analyzed. Experimental data. Figure 8 shows experimental data taken from Teixeira and others (1999) for a thermal process that was evaluated with the RGM as well as with the traditional FM. Figure 9 shows experimental data for a broken-heating curve (Tucker 2002). Results of process evaluations (Figure 8 and 9) by both procedures as well as adjusted processes for a specified F value are depicted in Table 2 and 3. Simulated data. To analyze extreme situations, heat penetration data (cold spot temperature profiles) were generated at different retort temperatures using a finite difference solution of the conduction heattransfer equation for a cylindrical can (Teixeira and others 1969) and for forced convection product using Eq. 3 and 5 that were developed in this study. The data set generated at 120 C was then used as a starting point (reference process) to calculate process times to achieve a specified lethality by both methods. Calculations with the FM were executed with computer software publicly available at the Purdue Univ. Food Science Dept. website <http://cifmc. foodsci.purdue.edu/ball/ball.cfm>. Calculations with the RGM (thermal process evaluation at different conditions and adjustments) were executed according to the procedure described previously in the Methodology section. Thermal process evaluation at different conditions other than Recorded and Adjustment for a specified Fo value. Figure 10 represents experimental data collected in a seafood-processing plant to test the developed procedure for changing retort temperature and/or initial temperature and then adjusting the process to a specified Fr value. Process specifications and calculations are given in Table 4.

F r = 6 (min); Conduction heated product, can (603 909); = 1.25 10 7 (m 2 /s); CUT = 30 (min); TRT = 120 (C), IT = 50 (C), Tw = 20 (C); f h = 289.2 (min); j ch = 1.8

Food Engineering and Physical Properties

Figure 8Experimental data for a heat-penetration test (taken from Teixeira and others (1999)

Results and Discussion

Thermal-process evaluation and adjustment
Experimental data. First, for the situation shown in Figure 8, application of the RGM is no different from the original General Method because no data transformation was involved, and the data were directly evaluated. Table 2 shows the results from evaluating experimental data using the RGM and the FM. The FM severely underestimated the Fo value whereas the RGM
JFS is available in searchable form at www.ift.org

Figure 9Experimental data for a heat-penetration test (broken heating curve). (Tucker 2002) 1330 JOURNAL OF FOOD SCIENCEVol. 68, Nr. 4, 2003

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1330

7/23/2003, 12:03 PM

General method revisited . . .

produced an estimate in very close agreement. In this example, the FM overestimated operator process time (Pt) by nearly 9% compared with the RGM. This is a reflection of the inherent weakness in the FM for evaluation of lethality at the onset of the cooling phase (Merson and others 1978; Spinak and Wiley 1982). This is particularly significant in the situation shown in Figure 8 because the accumulated lethality during the cooling phase is greater than that accumulated during the heating phase. Figure 9 represents experimental data for a broken heating curve. Again the FM underestimated the accumulated lethality compared with the RGM (Fo = 35 min compared with 40 min). When adjusting process time for an Fo value of 6 min, the RGM gave a slightly shorter operator process time (Pt = 25 min compared with 26 min). Simulated data. Computer-supported experiments were developed to analyze extreme situations. Extreme cases were selected as (1) being pure conduction and forced convection, (2) having extreme fh values, and (3) having a slow come-up time. Conduction-heated product. Table 5 compares process times for 4 alternative processes (reference process was developed at 120 C). Product (thermal diffusivity) and can dimensions were chosen to have a high fh value (289.2 min). In addition, the reference process considered a very high comeup time (30 min). In all 4 computer-simulated experiments, the RGM had less error than the FM. Note that the error in the RGM, in this example, was insensitive to retort temperature. On the other hand, in the case of the FM, the error was significantly higher as retort temperatures approached 130 C, as others have shown (Holdsworth 1997; Smith and Tung 1982). Figures 11 and 12 show graphically how the RGM was applied to generate and adjust the new process starting with the reference process at TRT = 120 C. Figure 11 depicted results from changing retort temperature from 120 to 130 C. First, the come-up cold spot temperature profile was transformed using Eq. 8. Second, for the period of constant retort temperature, Eq. 4 was used. Third, given that the reference process (TRT = 120 C) had less inertia at the cooling phase compared with the new process at 130 C, the dimensionless temperature ratio concept was directly applied to the entire cooling phase. Finally, the new process (TRT = 130 C) was adjusted for a Fo value of 6 min according to the procedure described in methodology. Figure 12 depicted results from changing retort temperature from 120 to 110 C. In this case, the data at 120 C (reference process) were insufficient to achieve
JFS is available in searchable form at www.ift.org

Table 6Comparison of operator process time (Pt) required for the same lethality at alternative retort temperatures calculated by RGM and FM with actual operator process time in data set generated by analytical solution (forced convection product) at each retort temperature. TRT (C) 110 115 125 130 Analytical solution (min) 130.9 75.9 40.1 32.0 RGM (min) 130.6 75.8 40.1 32.0 Error (%) 0.23 0.13 0.00 0.00 FM (min) 127.1 72.5 37.7 30.1 Error (%) 2.90 4.48 5.99 5.94

Convection heated product, can dimensions 0.1 m dia 0.1 m height Fr = 6 (min); U = 100 (W/m 2 C); TRT = 120 (C), IT = 50 (C), T w = 20 (C); CUT = 30 (min); fh = 45.3 (min); j ch = 1.2

Figure 10Experimental data obtained in seafood-processing plant for a heatpenetration test at 2 retort temperatures

Figure 11Thermal process at 130 C for a conduction-heated product obtained from a reference process at 120 C Vol. 68, Nr. 4, 2003JOURNAL OF FOOD SCIENCE 1331

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1331

7/23/2003, 12:03 PM

Food Engineering and Physical Properties

General method revisited . . .

a Fo value of 6 min at 110 C. Then, using Eq. 13, more heating data were generated at 120 C before proceeding with data transformation to 110 C. To transform thermal data from 120 to 110 C, Eq. 8 was used for comeup, Eq. 4 for holding time, and finally Eq. 14 for the cooling phase but starting from point A (Figure 12), avoiding inertia to safely predict the cooling phase at 110 C. Convection-heated product. Table 6 compares process times for 4 alternative processes (reference process was developed at 120 C). Given that Eq. 6 was derived for perfect mixing (linear temperature profile during come-up time), the dimensionless temperature ratio (Eq. 8) could accurately transform the original time-temperature data (for 0 < t < CUT). Equation 4 was used for t > CUT, and Eq. 14 for the cooling-down phase. As seen in Table 6, the error attributed to the RGM was approximately zero in all 4 cases, regardless of the new retort temperature. It is also important to note that in these 4 examples, the FM predictions resulted in an error found on the risk side.

Thermal process evaluation at conditions other than those recorded, and adjustment for a specified Fo value
Experimental data presented in Figure 10 were selected as being normal thermal processing data. First, come-up between both processes was similar but with slight differences. Second, initial temperatures (IT ) were different; and third, the retort temperature (TRT = 117.6 C) was constant but with slight variations during the process. Thermal process data were evaluated with the RGM and the FM (Fo values are depicted in Table 4). In both cases, the FM underestimated Fo value in relation to the RGM, revealing once again that the higher the retort temperature, the lower the prediction capacity of the FM. Taking Process 1 (Table 4) as a reference process, thermal process data were transformed from TRT = 114.5 C to TRT = 117.6 C according to the RGM procedure and then used for process evaluation to compare with results from the FM. When comparing operator process time (Fr = 6 min), FM overestimated both processes by approximately 10% and 44%, respectively, compared with RGM (TRT = 114.5 C and TRT = 117.6 C). According to Figure 13, the RGM has a very accurate prediction capacity and the predicted Fo value is on the safe side ( Fo value of 6.07 min compared with 6.14 min).

Food Engineering and Physical Properties

Figure 12Thermal process at 110 C for a conduction-heated product obtained from a reference process at 120 C

Conclusions
procedure for the General Method to give it the same ease of use as the Balls Formula Method. The RGM allows the same calculations as the FM, but with more accuracy. The proposed methodology was rather accurate when used for thermal process adjustment. The processing time was always estimated (overestimated) with an error less than 5% (in all cases under study). For the FM, it was common to find errors of 10% to 20% or more. The new procedure safely predicted shorter process times than those predicted by the FM. These shorter process times may have an important impact on product quality, energy consumption, plant production capacity, and adequate corrections for online control (process deviations). On the other hand, the FM predicted unsafe processes for forced convection products analyzed in this study. Further testing with experimental data must be done on this developed procedure
JFS is available in searchable form at www.ift.org

T WAS POSSIBLE TO DEVELOP A SYSTEMATIZED

Figure 13Use of dimensionless temperature ratio concept over experimental data to generate data at a different retort temperature 1332 JOURNAL OF FOOD SCIENCEVol. 68, Nr. 4, 2003

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1332

7/23/2003, 12:03 PM

General method revisited . . .

to check its performance and then it can be made available in computer userfriendly software. The developed procedure could be extended to pasteurization processes, UHT/ HTST processes, and so on. Future studies should consider the possibility of changing come-up time, the nonlinear temperature profile during come up time, and the capability to evaluate broken heating curves at different conditions than the originally recorded temperature (TRT and/or IT). = rate of heat transfer R = inside radius of can TA = extrapolated initial can temperature obtained by linearizing entire heating curve of a can T = temperature TC.P. = temperature in the coldest point TC.P. = new temperature in the coldest point IT = initial temperature ITW = initial temperature cooling phase IT = new initial temperature IT W = new initial temperature cooling phase TRT = retort temperature TRT : new retort temperature Tw = cooling temperature Tw = new cooling temperature Tr = reference temperature, 121.1 (C) t = time U = global heat transfer coefficient
sterilizing value of a thermal process. Food Technol 22(7):93-5. Hayakawa KI. 1974. Response charts for estimating temperatures in cylindrical cans of solid food subjected to time variable processing temperatures. J Food Sci 39(6):1090-8. Holdsworth SD. 1997. Thermal processing of packaged foods. London: Blackie Academic & Professional. p 146-61. Luikov AV. 1968. Analytical heat diffusion theory. New York: Academic Press, Inc. p 300-50. Merson RL, Singh RP, Carroad PA. 1978. An evaluation of Balls formula method of thermal process calculations. Food Technol 32(3):66-76 Patashnik M. 1953. A simplified procedure for thermal process evaluation. Food Technol 7(1):1-6 Pham QT. 1987. Calculation of thermal process lethality for conduction-heated canned foods. J Food Sci 52(4):967-74. Schultz OT, Olson FC. 1940. Thermal processing of canned foods in tin containers. III. Recent improvements in the General Method of thermal process calculation. A special coordinate paper and methods of converting initial retort temperature. Food Res 5:399. Smith T, Tung MA. 1982. Comparison of Formula Methods for calculating thermal process lethality. J Food Sci 47(2):626-30. Spinak SH, Wiley RC. 1982. Comparisons of the General and Ball Formula Methods for retort pouch process calculations. J Food Sci 47(3):880-4, 888. Stumbo CR. 1973. Thermobacteriology in food processing. 2nd ed. New York: Academic Press. p 14351. Teixeira AA. 1992. Thermal process calculations. In: Heldman DR, Lund DB, editors. Handbook of food engineering. New York: Marcel Dekker, Inc. p 563619. Teixeira AA, Balaban MO, Germer SPM, Sadahira MS, Teixeira-Neto RO, Vitali AA. 1999. Heat transfer model performance in simulation of process deviations. J Food Sci 64(3):488-93. Teixeira A, Dixon J, Zahradnik J, Zinsmeiter G. 1969. Computer optimization of nutrient retention in the thermal processing of conduction-heated foods. Food Technol 23(6):845-50. Tucker GS. 2002. Personal Communication. Chipping Campden, Gloucestershire, U.K.: Campden and Chorleywood Food Research Assn. MS 20020308 Submitted 5/20/02, Revised 12/8/02, Accepted 1/22/03, Received 1/27/03
We kindly appreciate the contribution made by Mr. Cristian Corts and particularly to Mrs. Paula Solari.

Nomenclature
A = area a and b = constant of linear equation describing retort temperature profile [TRT(t) = a + bt] b = new slope of the linear equation describing retort temperature profile [TRT(t) = a + bt] a = new constant of the linear equation describing retort temperature profile [TRT(t) = a + bt] Cp = heat capacity of food CUT = come-up time = energy per mass unit Fo = sterilizing value at 121.1 C Fp = process sterilizing value Fr = required sterilizing value f = rate factor (related to slope of semi-log heat-penetration curve) fh and fc = heating and cooling rate factors (related to slope of semi-log heat-penetration curve) j = dimensionless lag factor jh and jc = heating and cooling lag factors k = thermal conductivity of food l = height of canned content M = product mass Pt = operator process time (time that is measured from when the retort reaches processing temperature [TRT] until the steam is turned off ).

Greek letters
: = thermal diffusivity of food ( = k/Cp) : = density of food = differential or nabla operator ( = ) 2 = laplace operator ) ( =

References
Ball CO. 1923. Thermal processing time for canned foods. Bull. Nr 7-1 (37). Washington, D.C.: Natl. Res. Council. Ball CO. 1928. Mathematical solution of problems on thermal processing of canned food. Berkley, Calif.: Univ. Cal. Pub. In Pub. Health 1, N 2, 15-245. Bigelow WD, Bohart GS, Richardson AC, Ball CO. 1920. Heat penetration in processing canned foods. Bull. Nr 16-L Res. Washington, D.C.: Lab. Natl. Canners Assn. Carslaw HS, Jaeger JC. 1959. Conduction of heat in solids. London: Oxford Univ. Press. p 63-4. Datta AK. 1990. On the theoretical basis of the asymptotic semi logarithmic heat penetration curves used in food processing. J Food Eng 12:177-90. Gillespy TG. 1953. Estimation of sterilizing values of processes as applied to canned foods. II. Packs heating by conduction: complex processing conditions and value of coming-up time of retort. J Sci Food Agric 4:553-65. Hayakawa KI. 1968. A procedure for calculating the

Authors Simpson and Almonacid are with the Dept. de Procesos Qumicos, Biotecnolgicos, y Ambientales, Univ. Tcnica Federico Santa Mara; P .O. Box 110-V; Valparaso, Chile. Author Teixeira is with the Dept. of Agricultural and Biological Engineering, Frazier Rogers Hall, P. O. Box 110570, Univ. of Florida, Gainesville, FL 326110570. Direct inquiries to author Simpson (E-mail: ricardo.simpson@pqui.utfsm.cl).

JFS is available in searchable form at www.ift.org

Vol. 68, Nr. 4, 2003JOURNAL OF FOOD SCIENCE

1333

jfsv68n4p1324-1333ms20020308-TLS-3pgsColor.p65 1333

7/23/2003, 12:03 PM

Food Engineering and Physical Properties