PREDICTING SENSORY COHESIVENESS, HARDNESS AND SPRINGINESS OF SOLID FOODS FROM INSTRUMENTAL MEASUREMENTS

Transcription

1 Struttura e proprietà degli alimenti PREDICTING SENSORY COHESIVENESS, HARDNESS AND SPRINGINESS OF SOLID FOODS FROM INSTRUMENTAL MEASUREMENTS

2 Struttura e proprietà degli alimenti Texture is the sensory and functional manifestation of the structural, mechanical and surface properties of foods detected through the senses of sight, hearing, touch and kinesthetics (Szczesniak, 2002). It depends on product physical and physicochemical properties and on the unique, complex features of the human senses (Peleg, 1987). Texture is also widely recognized as an important quality attribute for product acceptability affecting consumer perception.

3 Struttura e proprietà degli alimenti According to an engineering approach of food processing, texture attributes are a measure of performance when food interacts with the consumer. They are the result of structure and composition which are obtained by submitting the ingredients to the sequence of operations which comprise a given food process. From this point of view, like the performance of any other engineering material the texture attributes of foods can be described by intrinsic material properties which, according to existing theories, can even be defined by adequately selecting ingredient compositions and food process parameters.

4 Struttura e proprietà degli alimenti Understanding the relationship between food texture perception and food structure is of increasing importance for companies wishing to produce texturally attractive food products (Wilkinson et al., 2000). This study is complicated by the dynamic nature of texture perception and by the presence of large individual differences in oral processes. Thus, a multidisciplinary approach is required, integrating three research areas: sensory science, physiology and food structure research, i.e. the study of rheological parameters, microstructure and other relevant food engineering properties.

5 Struttura e proprietà degli alimenti The stimuli of texture perception are primarily of a mechanical nature. Therefore to establish a relationship between perceived texture and food characteristics, it is essential to understand the mechanics or the rheology of food deformation (Peleg, 1987).

6 Struttura e proprietà degli alimenti The most recurrent parameters among solid food texture attributes are hardness, springiness, cohesiveness and friability. Hardness is a textural characteristic estimated during the first mastication: strength is applied on the food product in an approximately linear way and can be satisfactorily reproduced instrumentally through a uniaxial compression test (Shama and Sherman, 1973).

7 Struttura e proprietà degli alimenti The present work aims to investigate the relationships between mechanical textural sensory attributes (hardness, cohesiveness and springiness) and instrumental measurements from cyclic compression and stress relaxation tests, using 15 food samples.

8 MATERIALS AND METHODS Food samples Food products used for this study, consisting of well-known commercial brands, were chosen to represent a wide range of texture characteristics. Food Samples Big fruit candy Dofour Sliced white bread Mulino bianco Candy Fruit joy Jelly candy Gelly Firm egg white Ovito Galbanino cheese Galbani Little loaf Mulino bianco Madeleines Le Petit di Playtime Mini Babybel cheese Bel Angel cake Bisconova Sandwiches Mulino bianco Toffee candy Elah Trolly kiss Candy Trolly Würstel without skin Wuoi Citterio Würstel without skin Wuoi Citterio (raw)

9 MATERIALS AND METHODS Sample preparation In order to achieve comparable results with data related to different food categories, sample size was standardized for both sensory and instrumental analysis. For most samples analyzed, 15 mm diameter and 15 mm height cylinders were prepared with a special metal cable cylinder, while candy samples, due to their cylindrical shape and dimensions, were used in unmodified form. Rectangular toffee candy samples (19 mm wide, 15 mm thick and 10 mm high) were cut in order to obtain samples 15 mm wide and thick and 10 mm high. Würstel samples were boiled for 5 minutes and egg samples for 20 minutes.

10 Sensory analysis Analysis was performed by means of a panel composed by eight trained assessors in an ISO 8589:1998 compliant sensory analysis location. Food hardness, cohesiveness and springiness were evaluated by using a continuous structured scale (10 cm), without reference. Every sample was tested in a randomized design with 3 replications. Attribute definition and evaluation techniques used are described in table. Attributes Definitions Evaluation techniques Cohesiveness Hardness Springiness Mechanical textural attribute relating to the degree to which a food can be deformed before it breaks. Mechanical textural attribute relating to the force required to compress the sample. Mechanical textural attribute relating to the rapidity and degree of recovery from a deforming force. Compress the sample with molars and evaluate the amount of deformation before rupture Compress the sample with molars and evaluate the required force. Compress the sample partially with fingers and evaluate the degree and rapidity of recovery

11 Instrumental Analysis Cyclic compression and stress-relaxation tests were performed by means of an Instron Universal Testing Machine (Instron Ltd., mod High Wycombe, GB). Cyclic compression tests. Cyclic compression tests were performed using a 60 mm diameter piston and with load bearing cells (10 N, 1 kn) chosen according to the attainable maximum load on samples during test execution. Data were acquired through INSTRON Series IX software. Tests were performed with a crosshead speed of 50 mm/min, imposing 50% as maximum deformation for the first cycle, with preventive confirmation of breakup within such a range of the examined foods. Five compression/decompression cycles were performed. Stress-relaxation tests. Stress-relaxation tests were performed imposing a deformation from 2% to 16% so that deformation was included in the linear viscoelasticity region, achieved with a crosshead speed of 100mm/min. Stress was registered during 5 minutes. Every test was replicated 15 times for each different food sample.

12 DINAMOMETRO

13 E = modulo elastico σ = sforzo a rottura ε = deformazione a rottura W = Energia

14 RILASSAMENTO DELLO SFORZO Si applica al campione una deformazione costante e istantanea e si misura la variazione dello sforzo nel tempo.

15 RILASSAMENTO DELLO SFORZO Lo sforzo diminuisce nel tempo secondo la relazione di decadimento: t σ () t = σ exp 0 τ Tempo di rilassamento τ = η E

16 Data analysis From the force/deformation curves of the first two compression test cycles, TPA parameters related to cohesiveness (A2/A1), hardness (S1) and springiness (d2/d1) were calculated according to Bourne (1978). Young s Modulus (E), Strain (di) and Stress (Si) at peak as well as Irrecoverable Strain (ri) and Irrecoverable work (Li) were acquired during the five cycles of experimental cyclic compression tests. Irrecoverable work (Li) was obtained by the difference between compression work and recoverable work. From stress-relaxation response Peleg s linearization model parameters, K1, K2, were estimated by best fit regression (Peleg and Calzada, 1976).

17 Statistical analysis Sensory and instrumental data were submitted to analysis of variance and Duncan s test (p 0,05) to assess significant differences related to texture characteristics among different products. Sensory and instrumental data were correlated by Linear and non- Linear Regression, and by Partial Least Squares regression (PLS) analyses. Partial Least Squares regression is a statistical modeling technique used to compare two sets of data by seeking out latent variables common to both data sets (Martens and Martens, 1986). Instrumental parameters were used as predictors of all three sensory variables (PLS2 option). Relative Ability of Prediction (RAP) was used in order to estimate the model s predictive quality.

18 RESULTS AND DISCUSSION Sensory analysis Results of sensory analysis are shown in the figures (Duncan s Test, p 0.05). Cohesiveness, hardness and springiness showed significant variation among the food samples analyzed. Hardness and springiness of the examined foods are uniformly distributed on the evaluation scale, while for cohesiveness the opposite can be observed.

19 a 10 b b 8 c 6 cd 4 2 cd a 10 b RESULTS AND DISCUSSION 8 c cd d d Sensory analysis d d d e ef f fg gh h 6 e ef f Toffee candy Trolli kiss candy Fruit joy candy Galbanino cheese Cooked würstel Mini Babybel cheese Raw würstel Big fruit candy Jelly candy Firm egg white Little loaf Sliced white bread Sandwiches Madeleines Sponge cake Toffee candy cohesiveness scores Fruit joy candy Jelly candy Raw würstel Big fruit candy Cooked würstel Trolli kiss candy Mini Babybel cheese Galbanino cheese scores Madeleines Firm egg white Sponge cake Sandwiches Sliced white bread Little loaf a 10 b b 8 hardness d d 6 e e e g g h i il l Firm egg white Cooked würstel Raw würstel Trolli kiss candy Jelly candy Big fruit candy Sliced white bread Sandwiches Little loaf Mini Babybel cheese Galbanino cheese Madeleines Sponge cake Fruit joy candy scores Toffee candy c c c g h Springiness f i

20 RESULTS AND DISCUSSION Cyclic compression test In figure the sensory characteristics are measured against the TPA parameters. From the TPA parameters only sensory hardness was successfully correlated with TPA parameters (R2 = 0.875). Indeed, no correlation was found between cohesiveness and TPA-cohesiveness and between springiness and TPAspringiness. a Sensory Cohesiveness b Sensory Hardness c Sensory Springiness ,2 0,4 0,6 0,8 1 TPA Cohesiveness y = 1,98Ln(x) + 13,60 R 2 = 0,875 0,00 0,02 0,04 0,06 TPA Hardness ,0 0,2 0,4 0,6 0,8 TPA Springiness

21 RESULTS AND DISCUSSION Cyclic compression test From the force-deformation 700 s 1 curves Young s Modulus (E), 600 L 1 Strain (d i ) and Stress (S i ) at 500 peak as well as Irrecoverable Strain (r i ) and Irrecoverable 400 Work (L i ) were also monitored 300 during the five compression 200 test cycles. A typical cyclic compression/decompression 100 curve is shown for the Big Fruit 0 Candy sample (Figure 5), d 1 where all parameters r 1 monitored during the five cycles of compression test are represented. Stress (kpa) Big Fruit 0 0,5 1 1,5 2 2,5 3 3,5 4 Hencky's Strain

22 RESULTS AND DISCUSSION 0,06 0,05 R 2 = 0,99 big Big fruit mbb Mbb cheese wurstel Cooked cooked würstel Cyclic compression test A mathematical model was fitted in order to describe every parameter Y, with the exception of E, as a function of cycle number. logy=c i *n -Ai In the above equation C i is the parameter value for n=1 and A i is an index of parameter decay rate. The n value corresponds to the cycle number (n= 1, 2, 3, 4, 5). Both the observed and predicted values of stress at peak, irrecoverable strain and irrecoverable work for Big Fruit candy, Mini Baby Bel cheese and cooked Würstel samples are shown in figure. stress irrecoverable strain Irrecoverable work 0,04 0,03 0,02 0,01 0 0,6 0,5 0,4 0,3 0,2 0,1 0, ,012 0,009 0,006 0,003 R 2 = 0,98 R 2 = 0, number of cycle R 2 = 0,94 R 2 = 0,88 R 2 = 0,88 R 2 = 0,93 R 2 = 0,85 number of cycle y = C 2 n -A2 y = C 3 n -A3 y = C 4 n -A4 0,000 R 2 = 0, number of cycle

23 RESULTS AND DISCUSSION The same behavior was observed for all mechanical parameters and for all food samples; the results of parameters Ci and Ai, according to equation, are presented in table. From R2 values it is evident that the model was able to correctly describe the mechanical behavior of all samples. Food samples E *C 1 A 1 R 2 C 2 A 2 R 2 C 3 A 3 R 2 C 4 A 4 R 2 Big fruit candy d 0.5 a d c d fg c d cd 0.93 Sliced white bread a 0.5 a b a a e a a b 0.95 Firm egg white a 0.5 a a ab d a ab a b 0.91 Fruit joy candy e 0.5 a cd d a h b f b 0.90 Galbanino cheese b 0.5 a c ab b f b b b 0.94 Jelly candy a 0.5 a c b a fg b c c 0.90 Little loaf a 0.5 a a a a b a a b 0.92 Madeleines c 0.5 a b ab c e a b c 0.89 Mini Babybel cheese b 0.5 a c ab c fg b b c 0.89 Angel cake a 0.5 a a a b d a a b 0.95 Sandwiches a 0.5 a a a a b a a a 0.86 Toffee candy f 0.5 a e c d i d e d 0.89 Trolly candy a 0.5 a a ab a c a a b 0.96 Cooked Würstel without skin a 0.5 a a ab a b ab b b 0.85 Raw Würstel without skin b 0.5 a a b a a b b b 0.89

24 RESULTS AND DISCUSSION Stress-relaxation test. Stress-relaxation test results were normalized with respect to stress at initial time as σ(t) = σ(t)/σ(t0), in order to better compare the different samples analyzed. Figure shows an example of a stress-relaxation curve for three food samples. The curves show a slope decreasing as time increases. This means that the stressrelaxation speed is higher at the beginning, and decreases as time progresses. From this figure, one can observe that Big Fruit, cooked Würstel and Mini Babe Bel cheese samples behave as a typical viscoelastic solid, because stress for long testing times assumed non-zero values

25 Stress-relaxation test All food samples tested showed similar behavior. Curves obtained were fitted through Peleg s model. This is a simplified approach in representing relaxation curves of food materials, resulting in a linear regression RESULTS AND DISCUSSION t/y(t) = K 1 t+k 2 1/K 1 is related to the initial stress decay rate; 1/K 2 to a hypothetical asymptotic level non-relaxed at long times. Food samples K 1 K 2 R 2 Big fruit candy e e Sliced white bread cd c Firm egg white d e Fruit joy candy b a Galbanino cheese b d Jelly candy bc b Little loaf c d Madeleines c c Mini Babybel cheese a e Angel cake c c Sandwiches cd c Toffee candy b a Trolly candy c bc Cooked würstel without skin d d Raw würstel without skin e e 0.997

26 RESULTS AND DISCUSSION Prediction of sensory attributes from instrumental parameters 0.7 W Springiness X A 3 Hardness X 0.5 W K W 1 K 2 PLS Component A 2 W Cohesiveness A 4 W X E W C 2 W C 4 W A 1 W C PLS component 1 W

27 RESULTS AND DISCUSSION Prediction of sensory attributes from instrumental parameters The total percent variance explained by the first two components of PLS analysis was and the individual percent variances were for cohesiveness, for hardness and for springiness. The RAP indices were 0.78, 0.85 and 0.82 for cohesiveness, hardness and springiness, respectively. Thus, hardness was the best predicted sensory attribute, followed by springiness and cohesiveness. Indeed, cohesiveness showed the lowest RAP.

28 RESULTS AND DISCUSSION In this figure the graphs of observed variables versus predicted from instrumental measurements by the PLS regression analysis are shown. The position on the graphs of the different samples confirms a good relation between instrumental parameters and sensory hardness, even sensory springiness is predicted quite well by PLS model, whereas cohesiveness is not so accurately predicted. a observed cohesiveness b observed hardness c observed springiness predicted cohesiveness predicted hardness predicted springiness

29 CONCLUSION The major finding of this study is that both stress-relaxation and cyclic compression tests need to be performed for correct prediction of sensory responses. Texture Profile Analysis is unable to predict the examined sensory characteristics, except for hardness which showed a good nonlinear relation with the TPA parameter. As a sensory attribute, hardness as well as springiness is predicted well by means of the instrumental parameters used. Instead the instrumental prediction of cohesiveness is not so accurate. This may be due to the fact that analyzed samples are not uniformly distributed on a sensory cohesiveness scale