Measurement of crispness in food products using

acoustic-mechanical techniques: a literature review

Sophia Davies

MSci Physics Project - ‘Physics of Cooking’

Léon van Riesen-Haupt, Prof. Peter Török & Dr. Carl Paterson

Department of Physics, Imperial College London

15 October 2014

1 Introduction
There are many interesting research topics in the physics
of cooking. Research on crispness has increased over the past
years due to rising consumer demand for crisp food prod-
ucts. Mechanical tests (Vincent, 1998; Ross, 1999; Ross and
Scanlon, 2004) have failed to produce significant correlation
with sensory crispness and are limited by the geometry and
composition of the food sample (Saeleaw and Schleining,
2011). Recently, there has been a particular focus on the
measurement of crispness using acoustic-mechanical param-
eters. Studies on crusted food products with a high-moisture
core have proved particularly inconclusive. Developing both
a better understanding and acoustic-mechanical measure of
crispness across all food types is critical to improving the
correlation of instrumental measurement with the sensory
perception of food texture. This review assesses studies that
have contributed to the development thus far; the first part Fig. 1. Instrumental setup of the Kramer shear-compression cell and
summarizes research on crispness of solid, brittle food prod- data acquisition system used by Chaunier et al. (2005).
ucts and the second focuses on that of crusted food products
with a high-moisture core. (R2 = 0.4) was found to be lower than both that with the
2 Acoustic-mechanical measurements average amplitude of acoustic signals and high-amplitude
fraction of acoustic signals (R2 = 0.63 and R2 = 0.65 re-
2.1 Crispness of solid, brittle food products
spectively).
Mohamed et al. (1982) stated that measurement of both Chen et al. (2005) investigated crispness in biscuits using
mechanical and acoustic properties could better predict sen- a three-point bending device and were the first to analyse
sory crispness than the measurement of either of these pa- mechanical and acoustic data using the Acoustic Envelope
rameters alone, and was confirmed using solid, brittle food Detector (AED) (Fig. 2). The study established the link
products by the work of Chaunier et al. (2005), Chen et al. between mechanical and acoustic properties and is thus in-
(2005) and Varela et al. (2006). This integrated measure- trinsic to all work that has since followed.
ment approach has since become the dominant method used It was found that there was an almost one-to-one cor-
to investigate crispness of food products and underpins the respondence between force drops in the force-deformation
majority of the developments made in acoustic measure- curve and acoustic signals in the sound pressure level (SPL)-
ment. deformation curve (Fig. 3). It was stated that a one-to-one
correspondence should not otherwise be expected if such
2.1.1 The relationship between acoustic and mechanical pa- drops in force do not occur over very short time periods
rameters (as they did here); the mechanism behind sound dissipa-
Chaunier et al. (2005) investigated crispness in cornflakes tion works very differently to that responsible for drops in
using a Kramer shear-compression cell, acquiring the me- force in the force-deformation curve. However, it was found
chanical and acoustic data simultaneously using Labview that the second derivative of the force-deformation curve
software (Fig. 1). Spectral analysis was performed to pro- linked well with the acoustic signals, indicating the release
duce normalised power spectra in the frequency domain. of energy through the air of fracture events. It was further
The correlation of sensory crispness with maximum force hypothesised that if the fraction of total dissipated energy

1
in the form of detectable SPL was common to all fracture
events, then there should also be a link between the ampli-
tude of the gradient of the force-deformation curve and the
amplitude of the acoustic signals. However, this was not ev-
idenced by the data and the reason given for this was related
to problems with sound propagation paths in the biscuits.
The work of Varela et al. (2006) on the crispness of al-
monds, using a compression cylinder and AED (Fig. 4),
supported the findings of Chen et al. (2005) through both
acoustic-mechanical and microstructural measurement. The
work was significant in showing how a chemometric ap-
proach to the analysis of acoustic and mechanical data pro- Fig. 4. Instrumental setup of the compression cylinder and data ac-
quisition system used by Varela et al. (2006).
vides a good objective measure of sensory crispness. Fur-
thermore, using a principal component analysis (PCA) (Fig.
5) proved particularly successful in correlating combined
sensory, instrumental and compositional measurements. The
first PC was found to contain both acoustic and mechanical
measures and it was therefore concluded that both of these
were required for complete evaluation of sensory crispness.
PCA and one/two/three-way analysis of variance are be-
coming increasingly popular techniques in the analysis of
instrumental measurements of sensory food properties.

Fig. 5. Example PCA 3-D loading plot of almonds roasted for different
amounts of time (Varela et al., 2006).

The measurements of Salvador et al. (2009) on potato

chips indicated a good correlation between sensory crispness
and the number of force and sound peaks about the point of
maximum force. This conclusion is supported by the more
recent study of Arimi et al. (2010) of the same parameters in
biscuits, where a correlation coefficient exceeding 0.77 was
Fig. 2. Instrumental setup of the three-point bending probe and data found.
acquisition system used by Chen et al. (2005).

2.1.2 Measurement techniques

One of the major limitations of acoustic-mechanical mea-
surement is the dependence of the results on the method
used and type of food tested. The result of this is thorough
preliminary testing (Chaunier et al., 2005; Chen et al., 2005;
Varela et al., 2006).
Varela et al. (2006) found that different types of compres-
sion cause changes in the fracture mechanics and, in turn,
changes in the acoustics. Castro-Prada et al. (2007) found
the amplitude of the recorded acoustics to depend on the
fracture event, available propagation paths and the position
of the microphone. Chen et al. (2005) found changes in the
acute angle positioning of the microphone to have little to no
impact on measurement, agreeing with the statement made
Fig. 3. Typical force- and SPL-deformation curves (in the frequency
domain) produced by the three-point bend test of Chen et al. (2005) by Castro-Prada et al. (2007) that free-field microphones
on a certain brand of biscuit. have the least directional dependence of existing models.

2
However, alterations in the distance from the fracture event of Visser et al. (2008) was used to measure the number of
were seen to cause more significant changes in measurement. peaks in the region up until the crust penetration point in
Castro-Prada et al. (2007) discussed how free-field condi- both the force- and SPL-time curves as a function of the size
tions need to be created when using these microphones and of the peak (threshold). Preliminary tests indicated a suit-
that these can be achieved by using an acoustically isolated able minimum threshold value above which the contribution
chamber, which also acts to reduce the signal-to-noise ratio, of noise to the measured number of peaks was reduced. It
and positioning the microphone a specific distance (based was found that multiplying the number of peaks by each
on its geometry) from the source. threshold ‘enhanced clarity’, particularly for higher thresh-
Acoustic artefacts have also been found to have a signif- olds. There was a greater distinction between samples with
icant impact on measurements. Castro-Prada et al. (2007) different pre-frying times across all thresholds for the acous-
found an interesting contribution from mechanical appara- tic data than there was for the mechanical data, indicating
tus. Aluminium probes in both three-point bending and acoustic measurements were more sensitive to differences in
wedge compression devices caused resonance. A Perspex fracture behaviour. The study also investigated the relation-
blade was thus used as the preferred probe. ship between moisture levels and crispness, considering both
Additionally, the increased use of AED demands careful the moisture content of the entire French fry and that solely
testing of integration times (Chen et al., 2005). The use of of the crust. It was concluded that differences in crispness
acoustic transducers is also particularly problematic since due to different pre-frying times could only be understood
they are limited to detecting sounds conducted through the in terms the latter.
air; those conducted by the jawbone have been proven to be
particularly important in the perception of sensory crispness 2.2.2 Analysis of the entire crust-core structure
(Saeleaw and Schleining, 2011). There are some studies in the literature that analyse
2.2 Crispness of crusted food products with a high-moisture properties of the entire crust-core structure of breaded fried
core chicken nuggets. Antonova et al. (2003) investigated the
correlation between instrumental parameters and sensory
Whilst there are numerous studies in the literature on crispness of such nuggets. Measurements of both ultra-
crispness of solid, brittle food products (Chen et al., 2005; sonic and mechanical properties were made using 250-kHz
Chaunier et al., 2005; Varela et al., 2006; Castro-Prada et al., dry-coupling ultrasonic transducers and a Kramer shear-
2007; Salvador et al., 2009), there remains very little conclu- compression cell, respectively. Ultrasonic velocity was found
sive work pertaining to that of crusted food products with to have a higher correlation (R2 = 0.83) with sensory crisp-
a high-moisture core (Antonova et al., 2003). ness than mechanical peak force (R2 = 0.64). Sahin et al.
(2005) and Firdevs Dogan et al. (2005) both analysed the
2.2.1 Individual analysis of the crust texture of nuggets using a conical probe to find the maxi-
Studies have mostly focused on the crusts of battered or mum peak force required for 25% penetration, the former
breadcrumb-coated food products, analysing these by sep- study proving to be particularly inconclusive. Varela et al.
arating them from the core either before (Fan et al., 1997; (2008) suggest that the limitation of such methods is that
Mohamed et al., 1998) or after the frying process (Baix- the mechanical parameters they measure relate to hardness
auli et al., 2003; Ling et al., 1998; Maskat and Kerr, 2002; and toughness, which are not direct measurements of crisp-
Salvador et al., 2005). Ross and Scanlon (2004) investi- ness in crusted food products.
gated the fracture mechanics of fried potato crust using the Varela et al. (2008) developed a new method to assess the
former crust-core separation method for tensile tests. How- crispness of crusted food products with a high-moisture core.
ever, such analysis methods are limited in that they account The study involved acoustic-mechanical analysis of chicken
solely for the outermost layer of the composite food product nuggets that were cooked in various different ways, combin-
(Varela et al., 2008). Indeed the deformation of the exterior ing parameters derived from the force-deformation curves of
definitely depends on the mechanics of the crust, but the the entire sample with those of the simultaneously emitted
mechanics of the core also has a contribution (Varela et al., sound. Preliminary mechanical tests (including penetration,
2008). The contribution of each layer to such behaviours compression and cutting) identified that cutting with a Per-
remains difficult to quantify (Luyten et al., 2004). spex blade was the best method to achieve good discrimi-
To reduce the complication of the mechanics of the core, nation and reproducibility of force curves and sound peaks.
where both shear and compression forces are present (Varela The AED used in previous studies (Chen et al., 2005; Varela
et al., 2006), fracturing data is often only analysed up un- et al., 2006) recorded the SPL. Both force-deformation and
til the point at which the crust is penetrated (Sanz et al., SPL-deformation curves were plotted together using syn-
2007). Van Loon (2005) performed mechanical tests on chronised signals for real-time analysis.
French fries, analysing the region in the force-deformation Fig. 6 shows the texture profiles for the four different
curve up until penetration of the crust, using the method cooking methods studied. Force curves DF and CO had
of Visser et al. (2008). This method counts the number of similar texture profiles and comprised two distinct regions.
peaks above a set threshold. The number of peaks during The first was an increase in force with very few significant
fracture was found to correlate with sensory crispness. Sanz drops in force, corresponding to deformation of the sample
et al. (2007) were the first to measure crispness of French without any cutting. The second was an increase in the
fries using an acoustic-mechanical analysis. The method number of significant drops in force, corresponding to frac-

3
ture caused by cutting. The pick-up in force preceding this and sound events. Chen et al. (2005) also predicted that
marked the termination of the propagation of this fracture one-to-one correlation would unlikely be found for soft food
by heterogeneities and the beginning of further deformation. materials. Using these studies, it was suggested that a sig-
Both the jaggedness and high number of force drops in these nificant amount of the energy applied to the food product
two profiles was indicative of crispness. was likely stored elastically in the soft core.
Analysis of mechanical parameters included both the
area under the force-deformation curve and the total number
of force peaks. It was found that the latter better discrim-
inated between the crispness of the samples, agreeing with
earlier studies done on solid, brittle foods (Chen et al., 2005;
Varela et al., 2006; Vincent, 1998). Similar analysis was per-
formed for the acoustic parameters and the number of sound
peaks was again found to be the best discriminator between
the crispness of the samples. Particular emphasis was placed
on the fact that this preferred parameter related directly to
crispness since it measures only the fracture events that emit
sound.
Complementing the work of Varela et al. (2008), Primo-
Martı́n et al. (2008b) used an acoustic-mechanical method
to study the fracture behaviour of bread crust and what af-
fect certain ingredients had on this. The jagged structure of
the force-deformation curve seen to arise from the fracture
events was attributed to the gradual breakdown of the crusts
cellular structure. The distinction between the texture pro-
Fig. 6. Force-deformation curves of a certain sample of chicken nuggets file of crusted foods and solid, brittle ones was exemplified
subject to four different cooking methods: conventional oven (CO) using a separate mechanical analysis of a baked crispy roll
(thin black line), microwave oven and susceptor (MW + S) (thick black
line), deep-frying (DF) (thin grey line), and microwave oven (MW)
with the crumb removed. As expected, the jaggedness was
(thick grey line) (Varela et al., 2008). common to both samples, but there was a continuous de-
cline in force after the first force maximum for the crispy
Fig. 7 shows an overlay of both the force-deformation roll. Therefore, this study supported conclusions that the
and SPL-deformation curves for the DF method. As ex- soft core of crusted foods allows for further deformation and
pected, peaks in SPL accompanied all significant drops in fracture events after the first major fracture event.
force. However, caution was given to the assumption that
such drops in force arise solely from fractures in the crust.
It was highlighted that many of these could also arise from
other heterogeneities in either the crust-core interface (e.g.
air spaces) or in the core itself (e.g. bubbles). It was noted
that some small drops in force did not appear to correspond
to any peaks in SPL and it was suggested that these could
indicate deformations of the core producing no sound and
thus no contribution to sensory crispness. Discussion on the
non-stringent requirement for one-to-one correlation of the
peaks was also given based on the distinction between the
origin of sound and force signals given by Chen et al. (2005).
A suggested extension was quantification of the correla-
tion of the peaks by measuring the fraction of applied en-
ergy released as detectable sound. Reference was made to
the findings of Chen et al. (2005) concerning the time corre-
lation of fracture events with energy dissipation. However,
calculation of the second derivative of the force-deformation
curves did not improve correlation. The reason given for Fig. 7. Force- (black line) and SPL-deformation (grey line) curves for a
certain sample of chicken nuggets cooked using the DF method (Varela
this (and the challenges this problem poses in general) was et al., 2008).
explained using the work of Luyten et al. (2003) on how
the contribution of each form of deformation energy (stored Primo-Martı́n et al. (2008b) added to the acoustic anal-
elastic energy, dissipated fracture/other energy) to the re- ysis of Varela et al. (2008) by recognising a reduction in
leased energy depends on the material. Luyten et al. (2003) acoustic emission once passing the first force maximum. A
showed that dissipative energy processes other than frac- suggested reason for this, other than just a deeper penetra-
turing have an impact on the available energy for fracture tion into the crumb, was that fracturing involves the opening

4
of pre-existing cracks on the lateral part of the bread that Luyten, H., J Plijter, J. and Van Vliet, T. (2004), ‘Crispy/crunchy
form on cooling. This could perhaps suggest that the frac- crusts of cellular solid foods: a literature review with discussion’,
Journal of Texture Studies 35(5), 445–492.
ture mechanics of crusted foods with high-moisture cores is
inherently different to that of solid, brittle foods and de- Luyten, H., Plijter, J. and Van Vliet, T. (2003), Understanding the
pends strongly on the cooling process. sensory attributes crispy and crunchy: an integrated approach, in
‘Proceedings of the 3 rd International Symposium on Food Rheology
In a further study of the fracture behaviour of bread and Structure, Zurich, Switzerland’, pp. 379–384.
crust, Primo-Martı́n et al. (2008a) focused on the effect of
Maskat, M. Y. and Kerr, W. L. (2002), ‘Coating characteristics of
air and vacuum cooling for different storage times. A greater
fried chicken breasts prepared with different particle size breading’,
number of large fracture events were measured for the vac- Journal of food processing and preservation 26(1), 27–38.
uum cooling process. Force and sound events were found
Mohamed, A., Jowitt, R. and Brennan, J. (1982), ‘Instrumental and
to decrease with increasing storage time, correlating to an sensory evaluation of crispness: I—in friable foods’, Journal of Food
increase in the measured moisture content of the crust. An Engineering 1(1), 55–75.
extension of this work would be to focus on why this in-
Mohamed, S., Hamid, N. A. and Hamid, M. A. (1998), ‘Food com-
creased moisture content reduces the number of measured ponents affecting the oil absorption and crispness of fried batter’,
fracture events. Journal of the Science of Food and Agriculture 78(1), 39–45.
3 Conclusion Primo-Martı́n, C., De Beukelaer, H., Hamer, R. and Van Vliet, T.
(2008a), ‘Fracture behaviour of bread crust: Effect of bread cooling
The majority of acoustic-mechanical research on crisp- conditions’, Journal of Food Engineering 89(3), 285–290.
ness is focused on only one of what is an amalgamation of
Primo-Martı́n, C., De Beukelaer, H., Hamer, R. and Van Vliet, T.
complex behaviours in food products. The literature per- (2008b), ‘Fracture behaviour of bread crust: effect of ingredient
taining to measurement in crusted food products with a modification’, Journal of cereal science 48(3), 604–612.
high-moisture core is particularly inconclusive and research
Ross, K. A. (1999), A fundamental mechanical analysis of the texture
into possible differences in the underlying physical mecha- of french fries, Master’s thesis.
nisms is required. Indeed current studies have already sug-
Ross, K. A. and Scanlon, M. G. (2004), ‘A fracture mechanics analysis
gested a number of important extensions.
of the texture of fried potato crust’, Journal of food engineering
Conclusively, development of both the understanding and 62(4), 417–423.
instrumental measure of crispness across all food types will
Saeleaw, M. and Schleining, G. (2011), ‘A review: Crispness in dry
significantly improve the correlation of instrumental mea- foods and quality measurements based on acoustic–mechanical de-
surement with sensory perception of food texture. structive techniques’, Journal of Food Engineering 105(3), 387–399.
References Sahin, S., Sumnu, G. and Altunakar, B. (2005), ‘Effects of batters
Antonova, I., Mallikarjunan, P. and Duncan, S. (2003), ‘Correlating containing different gum types on the quality of deep-fat fried
objective measurements of crispness in breaded fried chicken nuggets chicken nuggets’, Journal of the Science of Food and Agriculture
with sensory crispness’, Journal of food science 68(4), 1308–1315. 85(14), 2375–2379.

Salvador, A., Sanz, T. and Fiszman, S. (2005), ‘Effect of the addition

Arimi, J., Duggan, E., O’sullivan, M., Lyng, J. and O’Riordan, E.
of different ingredients on the characteristics of a batter coating for
(2010), ‘Development of an acoustic measurement system for ana-
fried seafood prepared without a pre-frying step’, Food Hydrocolloids
lyzing crispiness during mechanical and sensory testing’, Journal of
19(4), 703–708.
Texture studies 41(3), 320–340.
Salvador, A., Varela, P., Sanz, T. and Fiszman, S. (2009), ‘Under-
Baixauli, R., Sanz, T., Salvador, A. and Fiszman, S. (2003), ‘Effect of standing potato chips crispy texture by simultaneous fracture and
the addition of dextrin or dried egg on the rheological and textural acoustic measurements, and sensory analysis’, LWT-Food Science
properties of batters for fried foods’, Food Hydrocolloids 17(3), 305– and Technology 42(3), 763–767.
310.
Sanz, T., Primo-Martı́n, C. and Van Vliet, T. (2007), ‘Characterization
Castro-Prada, E. M., Luyten, H., Lichtendonk, W., Hamer, R. J. and of crispness of french fries by fracture and acoustic measurements,
Van Vliet, T. (2007), ‘An improved instrumental characterization effect of pre-frying and final frying times’, Food Research Interna-
of mechanical and acoustic properties of crispy cellular solid food’, tional 40(1), 63–70.
Journal of Texture Studies 38(6), 698–724.
Van Loon, W. A. M. (2005), Process innovation and quality aspects of
Chaunier, L., Courcoux, P., Della Valle, G. and Lourdin, D. (2005), French fries, PhD thesis.
‘Physical and sensory evaluation of cornflakes crispness’, Journal of
texture studies 36(1), 93–118. Varela, P., Chen, J., Fiszman, S. and Povey, M. (2006), ‘Crispness
assessment of roasted almonds by an integrated approach to texture
Chen, J., Karlsson, C. and Povey, M. (2005), ‘Acoustic envelope detec- description: Texture, acoustics, sensory and structure’, Journal of
tor for crispness assessment of biscuits’, Journal of Texture Studies Chemometrics 20(6-7), 311–320.
36(2), 139–156.
Varela, P., Salvador, A. and Fiszman, S. M. (2008), ‘Methodological
Fan, J., Singh, R. and Pinthus, E. J. (1997), ‘Physicochemical changes developments in crispness assessment: Effects of cooking method on
in starch during deep-fat frying of a molded corn starch patty’, Jour- the crispness of crusted foods’, LWT-Food Science and Technology
nal of food processing and preservation 21(6), 443–460. 41(7), 1252–1259.

Firdevs Dogan, S., Sahin, S. and Sumnu, G. (2005), ‘Effects of soy and Vincent, J. F. (1998), ‘The quantification of crispness’, Journal of the
rice flour addition on batter rheology and quality of deep-fat fried Science of Food and Agriculture 78(2), 162–168.
chicken nuggets’, Journal of Food Engineering 71(1), 127–132.
Visser, J. E., Luyten, H., Hamer, R. J. and van Vliet, T. (2008),
Ling, D., Gennadios, A., Hanna, M. A. and Cuppett, S. L. (1998), A mechanical–acoustical test for the determination of crispness, a
‘Quality evaluation of deep-fat fried onion rings’, Journal of food study on cracotte crackers, Master’s thesis. (submitted for publica-
quality 21(2), 95–105. tion).