Ultrasonics 42 (2004) 155–159

www.elsevier.com/locate/ultras

Ultrasonic inspection of batters for on-line process monitoring

Jordi Salazar *, Antoni Tur�o, Juan A. Ch�avez, Miguel J. Garc�ıa

Departament d’Enginyeria Electr�onica, Universitat Polit�ecnica de Catalunya, Campus Nord, M�odul C-4, C/ Jordi Girona 1–3, 08034 Barcelona, Spain

Abstract

A new method for on-line batter monitoring using ultrasound techniques is presented. Air or gas incorporation is done during the

beating process which produces bubbles in the mixture. The density and the compressibility of the batter vary as a function of

mixing time and are quality index of batter. Traditionally, a batter sample of a fixed volume is removed and weighted in order to

determine its density. This is a time consuming process.

Batters are air filled mixtures of high viscosity which do not support significant transmission of ultrasound. For this reason

conventional ultrasonic density sensors for liquids are not suitable for this application. Therefore, a special transducer has been

developed. The sensor was constructed using a piezoelectric ceramic at the fundamental frequency of 1 MHz. Instead of measuring

density, in this work, changes in compressibility in batters are monitored by measuring the acoustic impedance of the batter. Main

advantage of this novel approach is that changes in acoustic impedance are easier to detect than changes in density especially when

air incorporation is in small quantities. Experimental results on different liquids and batters with different gas contents are presented

and discussed.

� 2004 Elsevier B.V. All rights reserved.

Keywords: Ultrasound; Density; Acoustical impedance; Batter

1. Introduction

Automation of the food industry requires fast and

reliable measurements of the physical properties ofmaterials during processing. The mixture of wheat flour,

water, yeast and other ingredients produces either a

dough or a batter with specific visco-elastic character-

istics capable of retaining gas and producing aerated

goods. Within the baking industry, the control of batter

properties is required to achieve final product quality

and consistency [1–3].

With regard to batters, air or gas incorporation isdone during the beating process which produces bubbles

in the mixture. The density and the compressibility of

the batter vary as a function of mixing time and it is,

therefore, in this phase when the density changes should

be measured. The final product quality as well as the

cake volume is strongly related to the amount of gas

bubbles present in the batter. Traditional methods for

batter testing are slow and off-line and do not provide

*Corresponding author. Tel.: +34-93-401-56-74; fax: +34-93-401-

67-56.

E-mail address: jsalazar@eel.upc.es (J. Salazar).

0041-624X/$ - see front matter � 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.ultras.2004.02.017

fundamental rheological information. Normally, a bat-

ter sample of a fixed volume is removed and weighted in

order to determine its density. This is a time consuming

process. There is therefore a need for the development offast and on-line instruments capable of providing rele-

vant data for baking.

In this work, we propose that changes of density in

batters are measured by using ultrasonic techniques.

Ultrasonics provides a non-destructive, rapid and low

cost technique for the measurement of physical food

characteristics such as density, speed of sound in a

material, or acoustic impedance. As the presence ofbubbles produces changes in both density and com-

pressibility of the batter, a rapid way to monitor these

changes is using ultrasound by measuring changes in

acoustic impedance instead of density, thus giving very

high sensitivity measurements.

Benefits to the industry of this on-line measuring

technique include better control of product quality,

improving processing efficiencies and reduction inwastage. Also, this allows operators to adjust processing

parameters during rather than after production run.

For this purpose a transducer has been developed

especially thought for high-viscous materials or air filled

156 J. Salazar et al. / Ultrasonics 42 (2004) 155–159

mixtures. Despite based on conventional ultrasonic

density sensors for liquids, the transducer has been

adapted to properly work with air filled mixtures such as

batters. Section 2 describes the ultrasonic transducer

design, and measurements and results are discussed inSection 3. Finally, Section 4 concludes this paper and

outlines future work.

2. Transducer design

The density q of a liquid is related to its acoustic

parameters, speed of sound c and acoustic impedance Z,through q ¼ Z=c. The speed of sound c can be deter-

mined from the time of flight of a pulse along a known

path. The acoustic impedance Z can be calculated from

the measurement of the reflection coefficient C of a

sound at the interface of a material with a known

acoustic impedance (reference material) and the un-

known liquid (batter in our case). A brief description of

different sensors which use this acoustic principle can befound in [4].

However, it should be noted that when measuring

batters difficulties arise due to their high values of

attenuation to ultrasound. Values greater than 500 dB/

cm have been measured. Batters as well as doughs are

air filled mixtures of high viscosity which do not support

significant transmission [5]. Additionally, coupling be-

variation of den

0.00999%

0.00999%

0.01000%

0.01000%

0.01001%

0.01001%

0 250 500 750 1000 1250

gas content (ppm

variation of acoustic imped

0%5%

10%15%20%25%30%35%

0 250 500 750 1000 1250 1gas content (ppm)

Fig. 1. Variation in acoustic impedance and velocity (top), and density (

tween the sensor and the batter is low which prevents the

acoustic wave to be transmitted into the batter. There-

fore, conventional density sensors proposed in [4] could

not properly work with batters since the high attenua-

tion they exhibit would completely attenuate the ultra-sound wave when travelling through, avoiding in this

way the measurement of velocity, and consequently, of

density.

Therefore, in this paper it is proposed to measure

acoustic impedance changes in batters instead of density

to overcome these difficulties. The main advantage of

this approach is that changes in acoustic impedance are

much greater than changes in density and thereforeeasier to be measured and a better precision can be

achieved in the measurement. This behaviour can be

observed in Fig. 1, which depicts for a given gas–liquid

mixture the variation in acoustic impedance and velocity

(top), and density (bottom) as a function of ppm units of

gas. While density changes are almost negligible to the

incorporation of small quantities of gas, the acoustic

impedance and velocity values of the mixture experi-ment a huge and similar variation, and plots of both

parameters appear overlapped. The origin of this vari-

ation is not due to the effect on the density but rather

that on the compressibility of the mixture [6].

Construction of the sensor is depicted in Fig. 2 (top).

This sensor is based on the same working principle that

those described in [4] for the impedance measurement

sity

1500 1750 2000

)

density

ance and velocity

500 1750 2000

velocityacous. imped.

bottom) of a gas–liquid mixture as a function of ppm units of gas.

AmeasAref

amplitude

time

Referencematerial

Piezoceramic

ReferencematerialAir Batter

Za<<Z0 Z0 Z0 ZbA0 A0

Aref Ameas

Zc

l1 l2

AA AA

Air

Z <<Z0 Z0 Z0A0 A0

Aref Ameas

Zc

l1 l2

Air

Z <<Z0 Z0 Z0A0 A0

Aref Ameas

Zc

Air

Z <<Z0 Z0 Z0A0 A0

Aref Ameas

Zc

l1 l2

Fig. 2. Construction of the ultrasonic sensor and received signals.

J. Salazar et al. / Ultrasonics 42 (2004) 155–159 157

except for the housing that has been modified. This

housing is terminated in the form of a 45� conical tip

such that it enters the batter cleanly without trapping

any external air bubbles on the outer surface as it entersinto the batter sample. When working with air filled

mixtures of high viscosity such as batters, this approach

is also advantageous since the reflection coefficient at the

interface between the buffer rod and batter is close to )1and changes in the ultrasound wave echo due to varia-

tions of the batter would be difficult to detect. However,

after a double reflection in the conical tip, the amplitude

of this echo will be affected by the square of the reflec-tion coefficient which will improve sensor sensitivity.

Both the piezoelectric ceramic and the buffer rod of

reference material have cylindrical shapes. The piezo-

electric ceramic operates at the fundamental frequency

of 1 MHz. Dimensions of the sensor are: l1 ¼ 30 mm,

l2 ¼ 45 mm, and 40 mm of diameter. Buffer rods are

made of Delrin (Z0 ¼ 3:45 MRayl and vL ¼ 2:43� 103

m/s). Since the lengths l1 and l2 are different, the refer-ence Aref and the measured echoes Ameas are received at

different times, Fig. 2 (bottom). The measurement is

done by determining the acoustic impedance Zb with the

help of the reflection coefficients from both interfaces,

air-Delrin, Ca, and batter-Delrin, Cb. Determination of

Ca and Cb values is done by means of two measure-

ments. First, a reference measurement is done with both

sides of the sensor in contact with air in order todetermine Ca which makes the measurement indepen-

dent of the transmitting amplitude A0, transfer function

of the piezoceramic transducer, electronics, and ampli-

tude and shape of the electric excitation pulse. Accord-

ing to Fig. 2 (top), from the reflection of the incident

acoustic wave at the left side interface

Ca ¼Aref0

A0

� aref ð1Þ

and from the right side interface

C2a ¼

Ameas0

A0

� ameas ð2Þ

where the square super index denotes the double 45�reflection at the conical tip and a accounts for the

acoustic losses in the buffer rods. Therefore,

1

a¼ aref

ameas

¼ Ameas0

Aref0 � Ca

ð3Þ

Once done this calibration measurement, the sensor is

put in contact with batter and a second measurement is

done in order to determine the Cb value. Then,

Ca ¼Aref

A0

� aref ð4Þ

and

C2b ¼

Ameas

A0

� ameas ð5Þ

From (3)–(5), and assuming that Ca � �1 (Zair ¼ 429

Rayl), then Cb can be expressed as:

Cb ¼ �

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiAmeas=Ameas0

Aref=Aref0

sð6Þ

Finally, we obtain the value of the acoustic impedance

of the batter as

Zb ¼ Z0

1þ Cb

1� Cb

ð7Þ

3. Experimental results

The sensor was tested with five different liquids and

a batter using the experimental set-up depicted in Fig. 3.

It consists of the sensor, an ultrasonic pulser receiver(Panametrics PR 5800) and a digital oscilloscope

(LeCroy LT344). Experimentally determined acoustic

impedance values of the liquids were compared with

their theoretical values published in [6]. Fig. 4 depicts

the results of this experiment. As can be seen, the sensor

is capable to accurately measure the acoustic impedance

of liquids such as water, two different measurements

were done, olive oil and methanol.However, results for glycerine and honey are not in

agreement with their theoretical values. This fact is due

to the acoustic impedance value of the buffer rod used in

the sensor, which is quite similar to those of glycerine

and honey. This makes the reflection coefficient to take

values close to zero and therefore makes the amplitude

of the signal Ameas difficult to be accurately measured. As

a consequence, there is a high percentage of errors inthese measurements. It should be noted that the sensor

Fig. 3. Block diagram of experimental apparatus.

0

0.5

1

1.5

2

2.5

3

Water_1

Water_2

OliveOil

Methan

ol

Glyceri

neHon

ey

Batter_

0

Batter_

2h

ZmeasZtheor

Z (M

Ray

l)

Acoustic Impedance

Fig. 4. Comparison of measured and theoretical acoustic impedance

values.

158 J. Salazar et al. / Ultrasonics 42 (2004) 155–159

has been designed to measure high-viscous or air filled

mixtures such as batters, materials that present much

lower values of acoustic impedance. Therefore, if the

measurement of liquids is required, a buffer rod with a

higher value of acoustic impedance should be chosen.

Furthermore, Fig. 4 also shows the measured acousticimpedance values of a batter. In this case, no theoretical

values for the acoustic impedance are available in the

literature. The batter was measured twice, just after

finishing mixing (batter_0) and after 2 h resting (bat-

ter_2 h). As yeast is present in the batter, the acoustic

impedance value is affected by the apparition of gas

bubbles in the batter during the resting. The presence of

gas bubbles makes the batter density and the acousticimpedance value to decrease, which is in agreement with

plots of Fig. 1. Values of density were measured by

weighting a batter sample of a fixed volume. However,

while density goes from 983 to 967 kg/m3, a variation of

1.6%, the acoustic impedance goes from 82.2 to 61.2

kRayl, a 25% change.

4. Conclusion

In this paper the design of an ultrasonic sensor foracoustic impedance measurements of high-viscous or air

filled mixtures such as batters has been described. The

sensor principle is based on the detection of the sound

reflection coefficient. The sensor has been successfully

proved by comparing theoretical and measured values

of the acoustic impedance of liquids. The presence of

bubbles produces changes in both density and com-

pressibility of the batter. It has been noted that varia-tions in acoustic impedance are greater than in density

and therefore easier to measure. In this way, by mea-

suring the acoustic impedance value of the batter, its

density can be controlled.

Although the results are good, more work needs to be

done. Future work includes the development of an

electronic circuitry to automatically display measure-

ments done by the sensor and a study of the sensorperformance when design aspects such as different

materials in the buffer rod and/or a piezoelectric ceramic

with different frequency are used in the construction of

the sensor.

References

[1] D.J. McClements, Ultrasonic characterization of foods and drinks:

principles, methods, and applications, Crit. Rev. Food Sci. Nutr.

37 (1997) 1.

J. Salazar et al. / Ultrasonics 42 (2004) 155–159 159

[2] M.J.W. Povey, D.J. McClements, Ultrasonics in food engineering.

Part I: Introduction and experimental methods, J. Food Eng. 8

(1988) 217.

[3] M.J.W. Povey, D.J. McClements, Ultrasonics in food engineering.

Part II: Applications, J. Food Eng. 9 (1989) 1.

[4] A. P€uttmer, P. Hauptmann, B. Henning, Ultrasonic density sensor

for liquids, IEEE Trans. Ultrason. Ferroelect. Freq. Control 47 (1)

(2001) 85.

[5] J. Salazar, J.M. �Alava, S.S. Sahi, A. Tur�o, J.A. Ch�avez, M.J.

Garc�ıa, Ultrasound measurements for determining rheological

properties of flour-water systems, Proc. IEEE Ultrason. Symp.

(2002) 855.

[6] J. David, N. Cheeke, Fundamentals and Applications of Ultrasonic

Waves, CRC series in pure and applied physics, CRC Press,

2002.