HAL Id: hal-02739625
https://hal.inrae.fr/hal-02739625
Submitted on 2 Jun 2020
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Interest of the modelling approach to predict the quality of processed foods - the example of salted meat products
Alain Kondjoyan, Pierre-Sylvain Mirade, Stéphane Portanguen, Jean-Dominique Daudin
To cite this version:
Alain Kondjoyan, Pierre-Sylvain Mirade, Stéphane Portanguen, Jean-Dominique Daudin. Interest of the modelling approach to predict the quality of processed foods - the example of salted meat products.
Food Tech Congress, 2014, Novi Sad, Serbia. �hal-02739625�
INTEREST OF THE MODELLING APPROACH TO PREDICT THE QUALITY OF PROCESSED FOODS - THE EXAMPLE OF SALTED MEAT PRODUCTS
Alain Kondjoyan*, Pierre-Sylvain Mirade, Stéphane Portanguen, Jean-Dominique Daudin
INRA, QuaPA-Meat Product Quality Unit, Centre of Clermont-Ferrand/Theix, 63122 Saint-Genès- Champanelle, France
*Corresponding author:
E-mail address: alain.kondjoyan@inra.clermont.fr
ABSTRACT
Experimental results from literature on the evolution of food quality during meat treatments are very difficult to use in practice to improve food quality since they often correspond to a specific equipment and a few process conditions. This leads to contradictions between studies and makes difficult to transpose results from one case to another. Mathematical modelling which combines transfers phenomena to chemical reactions is appropriate to respond to this situation. This paper illustrates how this has been applied to predict the final quality of cooked meat and dry-cured ham. The approach presented in this paper can be generalized to other foods and other processes to find out the best scenarios to obtain a targeted quality by taking into account both initial food characteristics and types of industrial equipment.
Keywords: Model, transfer, reactions, quality, meat
INTRODUCTION
Improving quality and safety of processed foods has been the objective of industrials and scientists for many years. However, technologies equipment and know-how vary a lot from one company to another. Initial quality of the agricultural and fish products is also variable depending on the animal genotype and the way they have been fed and bred. Moreover, quality is most of the time analyzed averagely while physicochemical, structural and biochemical changes are far to be uniform in solid foods. This can explain why quantitative literature results are often contradictory and always difficult to transpose from one case to another. This leads scientists and engineers to repeat experiments as soon as the type or size of the products, the types of equipment or the processing conditions are changed.
Mathematical modelling which couples transfers phenomena to chemical reactions is appropriate to respond to this situation. This paper illustrates how this approach was used in the case of meat products. It is divided into three parts, dealing successively with the modelling of heat and mass transfers, then with the modelling of chemical reactions and finally with the predictions issued from their combination.
TRANSFER MODELLING
Food processes involve exchanges of energy, water and chemical compounds. Temperature and concentration of chemical compounds are not uniform in solid foods. The local temperature and concentration values determine the rate of the reactions responsible for the development of food quality. Thus, the development of quality is different from one place to another in the food. Transfer modelling aims at calculating the gradients of temperature and concentrations in the food all along the processing. In our examples, heat and mass transfers are modelled during two processes: (1) the cooking of meat products, (2) the processing of dry-cured ham.
Heat and mass transfer during meat cooking.
During cooking, the temperature increase in the meat piece leads to water debinding from the myofibrillar proteins and water migration under pressure in channels of different dimensions formed by the contraction of the complex muscle structure (Bouharara, 2011, 2012). Mechanical forces involved in meat contraction are generally not taken into account and water debinding and migration are determined by mathematical relations based on the fact that: (1) at a given temperature the water concentration tends to an equilibrium value for long cooking time, and (2) the dynamic of the phenomena is somehow linked to the difference of water concentration at a given time and at equilibrium. A simple approach has been described by Kondjoyan et al. (2013) where water debinding and migration is assimilated to a first order chemical reaction. Under this assumption, the modelling can be divided into two stages. In the first stage the temperature kinetics are calculated inside the sample. In-product heat transfer can be assumed as purely conductive and energy exchanges at the boundaries can be determined by a Newtonian law. The reliability of the simulated temperatures is then essentially dependent on how accurately sample dimensions are known and on how accurate of the parameter values which are introduced into the model are (Kondjoyan et al., 2013; Oillic et al., 2011). In the second stage, the water concentration is calculated using the first-order kinetic models. The dependence of the reaction rate on temperature can be determined by an Arrhenius law, the local temperature being that issued from the heat transfer model. The rate constant of the kinetic models has also to be dependent on the distance between the local point and the solid surface to mimic juice migration. The value of the water content in the meat at equilibrium has to be determined experimentally at different temperature. Experiments have been conducted on beef, lamb and horse (Oillic et al., 2011) and more recently on pork salted and unsalted (Bombrun, 2014). The shape of the curve which gives the equilibrium water content as a function of temperature is always the same, expect for very special muscles. This transfer modelling approach was validated to predict the water concentration in beef roasts of parallelepiped shapes subjected to oven cooking conditions (Kondjoyan et al., 2013). The geometrical domain was meshed and finite element methods were used under COMSOL® Multiphysics 3.4 to simulate the evolution of temperature, water concentration and weight loss during meat cooking. The rectangular cuboid geometry of the sample was meshed using between 3600 and 170,000 tetrahedrons depending on the sample's dimensions. The solver used a conjugate gradient method and an algebraic multigrid preconditioning to solve the equations.
The solution took from 2 to 20 min depending on processing conditions, using a PC with a 2.33 GHz Intel Core2Duo Processor and 32 Gb RAM. The results of Bombrun et al. (2014) leads to the conclusion that the results on non-salted meat can be transposed to slightly salted meat under laboratory conditions.
Transfers during the dry-cured ham process.
Dry-cured ham elaboration is a very long process separated in 4 different stages: salting (duration of about 15 days), post-salting (8 weeks), pre-drying (1 week) and drying (5 to more than 18 months). During the process, heat transfer in the product can considered as purely conductive. Salt diffusion and water migration can be modelled using classical Fick equation, considering that the diffusivity of water depends on water concentration. Exchanges of energy at the surface shall include the effect of water evaporation and mass transfer coefficient can be determined from the heat transfer coefficient using the Lewis relation.
During post-salting, salt still goes on diffusing from the surface inside the ham while water migrates from inside the ham to the outer zones where evaporation took place. Specific mass flow rate boundary conditions have then to be imposed at the ham-air interface. Heat and mass transfers are coupled through the value of the water activity at the surface of the product which takes into account the local values of the water and of the salt concentration.
The quality of the dry-cured ham depends on the type of muscle involved in the biochemical reactions. Thus, muscles internal distribution has to be taken into account in the representation of the product as well as the complex shape of the surface. The process of constructing an accurate 3D ham representation required a series of 2D slices of a fresh ham
that can be typically provided by Computed Tomography (CT), which is a rapid and non-destructive X-ray imaging technique. Building the geometry is a complex procedure which shall include: noise reduction, smoothing, re-sampling and rescaling. Pertinent thresholding is required to separate the rind, the different muscles, the bone and the thin film of fat which lies between the SM and BF muscles. During our data treatment 181 X-ray CT images of a fresh ham were segmented using a specific software called Mimics® (Materialise, Leuven, Belgium).The segmented objects were simultaneously connected. The obtained 3D ham geometry was then smoothed and meshed many times in order to obtain a high quality surface mesh. After that, a volumetric tetrahedral mesh with extra smooth boundary was obtained. Finally, the 3D ham model consisting in 202,000 tetrahedral meshes and containing 5 different groups of muscle was imported into Comsol® Multiphysics software.
Once implemented solving all these equations lasted about 3.5 h on a 3 GHz Xeon 8-processors PC with 48 Go of RAM to model what happened during the salting and post- salting stages in terms of water and salt transfers, with a time step of 0.1 day.
Results of the 3D simulated mass transfer prove that after one week of process, the salt penetration was only visible in the few first centimeters from the salting surface of the ham, leading to a salt concentration of 10% of the total matter (TM) in this area, whereas it was 2- 3% TM in the middle part of the ham and negligible near to the rind side of the ham opposite the salting surface (Fig. 1). At the end of salting, salt has more diffused inside in the ham and salt concentration reached 5% TM in the middle part of the ham while it remained relatively low in the deeper zones of the ham (Fig. 2). During the post-salting stage, salt content became more uniform in the product. Also after six weeks, the simulated value of the salt concentration reached 4%-5% TM and 2-3% TM in the middle of the ham and its rind opposite salting zones, respectively (Fig. 2). All along the process water concentration decreased due to evaporation. This effect was all the more important that the ham muscle was close to the salting surface.
Figure 1: Distribution of proteolysis index (PI), salt and water contents predicted during the salting stage, at mid-salting and at the end of salting.
Figure 2: Distribution of PI, salt and water contents predicted during the post-salting stage, at mid- period and at the end of post-salting.
REACTION MODELLING
Two different qualities were analyzed depending on the type of process. In the example on meat cooking, the studied quality was the content in micronutrients while it was the texture of the product which was analyzed in the case of the dry-cured ham.
Loss of micronutrient during meat cooking.
Meats are rich in micronutrients (B vitamins, iron, zinc and selenium) which can be easily assimilated. The example is focused here on the water soluble vitamins B3 and B6, but the same approach can also be extended to vitamin B12 or to Heme iron. During cooking, water soluble micronutrients can be lost both because of thermal denaturing and because of juice expelling. The loss by the juice can be calculated from the previous mass transfer model as soon as the concentration of micronutrient in the juice is known, while thermal degradation can be described using a first order chemical reaction. Concentrations in the juice and parameters of the degradation reaction have been determined using separate experiments and data treatments.
Proteolysis and texture during dry-cured ham processing.
A major sensory characteristic to be followed during dry-cured ham processing is texture evolution which can lead to problems, such as softness, pastiness, crusting and impede slicing, or give a mouth-coating sensation. Texture evolution can be linked to the variation of the proteolysis index (PI) which can be rapidly measured by fluorescence using a fluorescamine-specific labelling method (Harkouss et al., 2012). Time evolution of proteolysis index has been measured and modelled on five different types of pork muscle under different accurate temperature and water activity (water and salt contents) laboratory conditions.
Statistical analysis of the experimental results showed that the PI was correlated positively with temperature and water content, but negatively with salt content. Applying response surface methodology and multiple linear regressions enabled us to build phenomenological models relating PI to water and salt content, and to temperature (Harkouss et al., 2014).
COMBINED TRANSFER-REACTION MODELS
The quality of the product can be predicted thanks to the combination of the transfer and of the reaction models.
During meat cooking, experiments on the thermal degradation of the B vitamins confirm the literature knowledge that B3 Vitamin was thermally resistant while B6 vitamin can be degraded by temperature. Thus, the loss of vitamin B3 was predicted afterwards directly from the calculated quantity of expelled juice while the thermal degradation of the vitamin B6 was added to the quantity of B6 expelled in the juice to determine the total loss of this vitamin.
The combined model was validated during the oven-roasting of meat cuts of different sizes heated under different conditions. Then, the validated model was used to predict the loss of vitamins B3 and B6 under different heating conditions. Simulated results lead to the conclusion that during grilling or pan frying of steaks, the loss in vitamin B, was only due to juice expelling, and was ranging from 4 to 23% depending on the degree of doneness.
During roasting, the loss of vitamin B3 in beef meat was in between 25 and 32% mainly depending on the final core temperature of the meat (50–70 °C), while it was in between 30 and 41% during simmering. The combined model was used to calculate the loss of B vitamins under a lot of practical cooking conditions. The calculated data were gathered in table sheet to be used by industrials and nutritionists (Table 1).
Vitamin B3 experiment B3 losses
Estimated quantity
of vit. B3 remaining (exp-cal)/exp
Muscles
Cooking time (min)
Temp.
(°C) MS (%)
B3 (µg/g)
B3 (µg/gMS)
Standard deviation (µg/gMS)
% µg/gMS %
Masseter 0 / 26.1 32.3 124.0 7.7 / / /
Masseter 60 / 38.8 34.7 89.8 13.2 40.0 74.4 17.1
Steak LT 0 / 23.3 60.6 260.1 6.3 / / /
Steak LT 4.37 55 28.1 66.7 236.6 10.8 17.2 215.4 9.0
Roast 0 / 22.5 61.1 271.7 4.2 / / /
Roast
slice 76 55 34.1 56.1 164.7 23.5 29.4 151.8 7.8
Table 1: Example of calculations of the loss of B3 vitamin for different muscles cooked under different conditions. The actual Tables gathered much more different cooking conditions and also the loss of B6 vitamin and, heme iron.
The quality of the dry-cured ham was determined from the proteolysis indexes (PI) which were compared at the end of post-salting stage with the experimental PI values measured in samples extracted from industrial Bayonne hams. In-ham PI evolution was determined using the local calculated values of the water and salt concentrations. According to the phenomenological model of proteolysis, the predicted PI values were logically lower in areas highly dried and highly salted, as close to the salting surface, than in the wet-low-salted areas. During the two first weeks of the process, the in-ham PI values i increased up to 3-4%
(Fig. 1). Proteolysis went on during the post-salting stage leading to local PI values ranging from 7 to 9%. At the end of the post-salting stage, i.e. after 11 weeks of process, the lower mean values of PI were still predicted in the most dried and salted group of muscles, i.e. the
“grosse noix” (Fig. 2). The difference in mean PI values exceeded 3% between the “sous- noix” (12.5%) and the “grosse noix” (9%); the mean PI value of the entire ham was intermediate (10.7%). In fact, these calculated mean PI values globally agreed with those
measured on samples extracted from two types of muscle (BF and SM) of industrial Bayonne dry-cured hams, at the end of the post-salting stage. Accurate analysis of results reveals that the difference in terms of time course of PI values between the 3 groups of muscles appeared early in the second or third day of salting.
CONCLUSIONS
The way how transfer-reaction models can be built and used has been illustrated in two cases: either cooked meat or dry-cured ham. In the case of cooked meat, the models were simple and rapid and were used to simulate the cooking yield and cooking loss of water soluble micronutrients under a lot of different conditions. The calculated data were gathered in tables to be distributed to industrials and nutritionists. In the case of the dry-cured ham, a complex 3D geometry of the product including an accurate location of the different muscles has been built and meshed. Transfers of water and salt have been calculated numerically and used to predict the evolution of proteolysis in the product during the different stages of the processing. The dry-cured ham process last several months. Thus, improving the quality of this product by testing experimentally other process conditions is very limited and very costly while the “3D numerical ham” model can help to asses rapidly various scenarios and select the best one. Up to now this approach was used to find how to reduce the salt content in the dried products without modifying their quality. Further simulations will take into account product safety by introducing the prediction of microbial growth or inactivation. More generally, the approach presented in this paper can be generalized to other foods and processes to find out the best scenarios to obtain a targeted quality; both the initial food material characteristics and types of industrial equipment can be considered. In this paper, the chemical reactions were considered as simplistic apparent reactions while oxidation or Maillard reactions in foods involve hundreds of elementary reactions. Research is ongoing to reliably predict the kinetic of compounds formation in these complex reactions systems.
ACKNOWLEDGEMENTS
Works on the salted meat products were funded by the Na- integrated program (ANR-09- ALIA-013-01) financed by the French National Research Agency. Authors thank IFIP for providing CT images during the study on the dry-cured ham.
REFERENCES
Bombrun, L., Gatellier, P., Portanguen, S., Kondjoyan, A. Analysis of the juice and water losses in salted and unsalted pork samples heated in water bath. Consequences for the prediction of weight loss by transfer models. Meat Science, DOI: 10.1016/j.meatsci.2014.07.033.
Bouhrara, M., Clerjon, S., Damez, J. L., Chevarin, C., Portanguen, S., Kondjoyan, A., Bonny, J. M. (2011).
Dynamic MRI and thermal simulation to interpret deformation and water transfer in meat during heating.
Journal of Agricultural and Food Chemistry, 59(4), 1229-1235.
Bouhrara, M., Clerjon, S., Damez, J.-L., Kondjoyan, A., Bonny, J.-M. (2012). In Situ Imaging Highlights Local Structural Changes during Heating: The Case of Meat. Journal of Agricultural and Food Chemistry, 60(18), 4678-4687.
Harkouss, R., Mirade, P. S., Gatellier, P. (2012). Development of a rapid, specific and efficient procedure for the determination of proteolytic activity in dry-cured ham: definition of a new proteolysis index. Meat Science, 92(2), 84-88.
Harkouss, R., Safa, H., Gatellier, P., Lebert, A., Mirade, P. S. (2014). Building phenomenological models that relate proteolysis in pork muscles to temperature, water and salt content. Food Chemistry, 151, 7-14.
Kondjoyan, A., Oillic, S., Portanguen, S., Gros, J. B. (2013). Combined heat transfer and kinetic models to predict cooking loss during heat treatment of beef meat. Meat Science, 95(2), 336-344.
Oillic, S., Lemoine, E., Gros, J.-B., Kondjoyan, A. (2011). Kinetic analysis of cooking losses from beef and other animal muscles heated in a water bath — Effect of sample dimensions and prior freezing and ageing. Meat Science, 88(3), 338-346.
