Using dielectric specstroscopy to measure dough consistency

Researchers have developed a novel method to measure dough consistency continuously by using dielectric spectroscopy. The scientists take a wendel mixer as an example to show that the measurement method has potential.

B. Börsmann1, L. Ringer1, K. Lösche1, O. Schimmer2, T. Sokoll2, T. Koch3

The main primary constituent of dough is the natural product flour, which is subject to natural quality fluctuations depending, for example, on the cultivation method or climatic conditions. During the manufacture of doughs this gives rise, among other things, to varying rheological properties that can be detected and controlled only with difficulty [1]. If the rheological characteristic of a dough after the kneading process differs from its target specifications, this leads to considerable processing problems and quality fluctuations during downstream process steps (e.g. lamination). Thus an amount of added water that is not optimally matched to the individual flour quality can lead to “stubborn” or “shrinking” doughs [2]. In addition to recipe-dependent variables to generate optimally adjusted doughs, the kinetic energy introduced during the kneading process is of decisive importance for the production of the dough [3]. In this case the adjustable variables, namely kneading time, kneading tool geometry or kneading intensity, among others, result in complex interactions that lead to the development of the respective dough characteristics [4]. In particular the use of “intensive kneader systems”, for example double wendel mixer systems, requires accurate control of the kneading parameters, because exceeding the optimum kneading time even slightly can cause over-kneading and thus irreversible damage to the dough.

Visual and tactile impressions by experienced specialist staff are often very successful as an immediate assessment technique to ensure the consistent product quality of baked goods. However, these methods for judging the optimum dough properties are indirect and subjective. On the other hand a sufficiently specific measurement system for the in-line recording of dough properties that has proved successful in practical application does not exist. The use of process parameters such as dough temperature and/or dough warming or even the detection of the energy input as a way to control kneading time also usually yields unsatisfactory results. Kneading time control systems that make corresponding process parameters accessible based on direct dough consistency measurement, e.g. via strain gauges (SGs), have also been unsuccessful in the market. Although this technology enables a direct assessment of the dough during the kneading process, this sensor technology requires additional instruments in the kneading chamber, which significantly affects the kneader’s geometry and thus also the results of kneading. The integration of additional tools in the kneading chamber also causes increased expenditure in the area of hygiene requirements [5].

Therefore the development of a measurement and control mechanism that continuously and accurately records and controls the rheological properties of a dough during the kneading process, and in addition does not require any modifications of the geometry in the kneading chamber, may represent a novel approach to minimize quality fluctuations during the manufacture of dough and baked goods [6; 7]. In relation to this above-mentioned approach, a novel measurement method to characterize the rheological properties of wheat dough during a kneading process is being developed and tested in the context of a cooperation project funded by the German Central Innovation Program SME (Zentrales Innovationsprogramm Mittelstand, ZIM). This involves using a physical measurement method that detects the dielectric properties of dough during the kneading process. Considerable successes were recently demonstrated in the use of comparable measurement methods from the company Sequid GmbH, Bremen, Germany, based on dielectric spectroscopy to assess the quality of fresh and frozen fish, and to obtain information about the presence of water binding additives. In detail the method statistically evaluates interactions between the material to be examined and broad-band electromagnetic fields. By using multivariate analysis methods it is possible to obtain information about, among other things, the degree and nature of the water binding. This allows specific conclusions about the progress of dough structure formation as detectable by means of appropriate dielectric spectroscopy (measuring probes). By structurally integrating measurement probes of this kind into a double wendel mixer, it is possible for the first time to carry out continuous recording and determination of the dough structure formation.

Dielectric Spectroscopy – Permittivity

The application of dielectric spectroscopy as a measurement methodology to characterize the development of a dough during a kneading process is based primarily on the interactions between permanent dipoles and electric fields, whereby the totality of the substance-specific polar properties is described as permittivity. Water as a classic dipole, and/or media that contain water, e.g. dough, display particularly strongly pronounced interactions in this respect. If a medium with dielectric properties (dielectric) is placed between the plates of a plate capacitor and an electric field is applied, this leads to a change in the capacitance of the plate capacitor that depends on the dielectric properties of the medium.

At a molecular level, the applied electric field causes polarisation in the dielectric, whereby it is possible to distinguish between orientation polarisation and induced polarisation. Of particular significance in this respect is the orientation polarisation, in which the permanent electric dipoles (e.g. in H2O) that are approximately randomly aligned in the absence of an electric field become spatially reoriented along the direction of the field (see fig. 1) [8].

Chaotic arrangement of freely movable dipolar Electromagnetic fields interact with molecules without any external influence freely movable dipolar molecules


Figure 1: The principle of dielectric orientation polarisation: Interactions between electric fields and a matrix with bipolar molecules (water)

In this respect the dynamic interactions between the applied electric field and the bipolar molecule matrix that is present form the basis of the measurement principle of dielectric spectroscopy. If in turn other interactions prevent the dielectric from aligning itself to follow the electric field, this has an effect on its permittivity. This occurs for example when binding interactions with other macromolecules restrict the “freedom of movement” of the bipolar water molecules. Equally, various partial aspects of the dough formation during a kneading process (starch swelling, gluten structure formation etc.) are associated at a molecular level with the attachment or binding of free water molecules to the corresponding macromolecules (see fFigure 2) [9].


Figure 2: Dough structure formation during a kneading process as a function of water binding and permittivity during the course of the kneading time (diagrammatic illustration)

Due to the totality of the interactions described above, it is assumed that the detection of the dielectric properties of the dough may allow the establishment of a novel method that will enable the direct characterisation of the dough development during a kneading process as an inline measurement and control procedure. In addition to the actual detection of the dough permittivity and its characterisation as a function of kneading time, it is also necessary to develop the additional correlations that enable a description of the respective dough permittivity properties with the associated rheological and baking technological characteristics, so as to allow dielectric spectroscopy and its response variables to be used for the targeted generation of doughs with defined properties. In this way it is possible to use dielectric spectroscopy as a control technology to control kneading time as a function of raw material and/or flour qualities with the aim of achieving process-controlled dough consistency and/or dough quality.

Integration into mixing systems

In this respect the detection of the permittivity properties is not limited to the use of a classical plate capacitor, but can also be implemented via other measurement probe variants. One design solution that is appropriate for the project consists of using two electrical conductors, arranged parallel to one another, that can be integrated as a component of a scraper already present in the kneading chamber (see Figure 3).


Figure 3: Diagrammatic construction of the dielectric measurement probe as exemplified by a dough scraper; dough scraper with integral measuring probe (green circuit boardon the picture in the middle); Diosna double wendel mixer system with built-in measuring probe

As shown in Figure 3, the measuring probe is constructed in the form of two electrical conductors that are attached to the frame structure of the above-mentioned dough scraper before being sealed with Vulkolan. In addition the frame structure provides the precondition for integrating the signal cable without needing to modify the geometry of the bowl area or kneading chamber. The STDR-65 time domain reflectometer made by Sequid is used for the active part of the measuring system. A time domain reflectometer (TDR) consists essentially of a signal generator to produce voltage jumps with extremely short rise times, and a detector. For the present application, the voltage jump is carried via cables and the sensor’s cable connection (see figure 3) to the sensor, where it is reflected back to the STDR-65 and is detected as a function of the dielectric properties of the dough. The dimensions of the STDR-65 that is used are only 208 mm × 168 mm × 55 mm, and due to its high stability against electromagnetic interference radiation it can be integrated directly into the kneader’s housing.
The installation of the measuring probe in the lower part of the scraper ensures that it experiences a continuous incoming flow of the dough matrix that is to be characterised (see Figure 4) and can thus record the change in the permittivity properties of the dough based on the radiated input of electromagnetic waves.

Figure 4: Diagrammatic illustration of continuous dough characterisation by means of a dielectric measuring probe

Signal interpretation

In addition to the measuring probe already illustrated (in the kneading chamber or in the dough scraper), the entire measurement chain used to characterize the permittivity properties of the dough also comprises the signal cable (coaxial cable) needed to connect together the two modules.

Figure 5 shows the output of the analogue-digital converter (ADC) used in the detector as a function of time, in which each measurement point on the x axis basically corresponds to a time interval of 3.2 ps. The incoming voltage jump when it reaches the detector (ADC Output) is recognizable in the left-hand part of the figure. It is followed by the low-reflection region of the coaxial cable, shown here shortened, the transition to the sensor, and the interaction region which is essential for the evaluation. At the end of the sensor where it runs empty the voltage jump experiences total reflection, which is associated with a signal rise.

© Sequid GmbH
Figure 5: Dielectric spectroscopy signal as a function of the measuring chain (TDR => Cable => Sensor); Source: Sequid GmbH

The signal changes between the positions “transition to sensor” and “reflection at sensor end” are especially important in this respect. The description of the permittivity properties of the dough needs characteristic values that are able to describe the change in the permittivity properties of the dough during the course of the kneading time. Two empirically determined points are used for this purpose:

  • The turning point (inflection) of the signal profile at the end of the sensor (green)
  • The fixed point at the end of the sensor (red)

It can be shown experimentally that these two points mentioned above display characteristic changes depending on the changing dough matrix (as a function of kneading time). This includes firstly the signal propagation time (ΔT) and secondly the signal amplitude of the two points. Therefore these four characteristic parameters (amplitude at the turning point; ΔT at the turning point; amplitude at the fixed point; ΔT at the fixed point) are subsequently analyzed as a function of the kneading time by using proprietary analysis methods (see fFigure 6).

Figure 6: Characterisation of the permittivity properties of dough as a function of kneading time by the use of four characteristic signal values: turning point amplitude; ΔT at turning point; fixed point amplitude; ΔT at fixed point

The signal analysis illustrated in Figure 6 shows by way of example the change in the permittivity properties of the dough for a standard wheat dough during the kneading time of 6 min: at the start of the illustrated curve the scraper together with the measuring probe are still in the air, and then at the time point t ˜ -0.1 min they are immersed in the filled bowl, so the kneading process can begin at t = 0. When the process has been completed after 6 min and after a short wait time, the scraper is finally withdrawn from the dough again. Due to the way the signal is plotted on the graph, in the case of the amplitude evaluation a signal increase corresponds to a decrease in the permittivity properties of the matrix, which is equivalent to a reduction in the free water. On the other hand for the graph of the change in signal propagation time (ΔT), a decrease in the signal corresponds to a decrease in free water. This decrease in free water, i.e. the reduction in the degrees of freedom of the water contained in the matrix, is in turn explicable by the processes occurring during the formation of the dough. Thus in each case a maximum in the permittivity properties of the dough is detected in the region of a kneading time of 1–2 min, which can be interpreted through the complete homogenisation of the flour-water suspension. Among other things the swelling of the starch and above all the formation of the gluten structure begin during the further course of the kneading time – due to the input of kneading energy – as a result of which the incorporation of the free water and of the added water into the corresponding macromolecular structures commences, and a decrease in the permittivity of the material is induced. When further statistical methods are applied, it is also possible to recognize that a minimum in the dielectric properties of the material is measured after approx. 3–4 min of kneading time, which in turn can be equated to a maximum in the binding of water within the dough structure. On the other hand, continuing the kneading process for a further 5 min of kneading time leads to an increase in the permittivity properties of the dough, i.e. to an increase in the free water. In this case it is assumed that over-kneading is already occurring, and that among other things partial damage to the gluten protein structures is beginning, leading to reduced water absorbency, analogous to specific protein denaturing with, for example, changed binding conditions and spatial structures.

In the context of further validation operations, one priority task is to correlate these demonstrated changes in the permittivity properties of the dough with rheological and baking technological characteristics.

Experimental set-up

The primary aim is to correlate the permittivity properties of the dough at various kneading times, or stages towards full kneading, with rheological and baking technology parameters, so as to enable the detection of the dough permittivity to be utilized as a control instrument for the targeted generation of a wide variety of dough qualities. To achieve this objective, various standard recipes (see t Table 1) to manufacture wheat doughs are defined and manufactured using an intensive double wendel mixer system (Diosna Type: WV 24), and the rheological and baking technological characteristics of each of them after various kneading times are determined and correlated with the permittivity properties of the dough.

Table 1: Standard recipe (baker’s percentages) for the permittivity, rheological and baking technological characterisation of various intensively kneaded wheat doughs

Components Proportion [%]
Wheat flour – Roland multipurpose flour - Brema 550 100
Water 54–58
Bakery improver - Goldmalz 3
Yeast - DHW 3
Salt 1.8

After corresponding kneading times (2 min; 4 min; 6 min and 8 min), dough samples are taken and an extensograph is used to determine their elastic and plastic properties. Because the standard doughs contain yeast, among other things, shortened resting times are used: 10 min, 20 min and 30 min. Standardized baking tests based on panned bread are carried out in parallel with the rheological characterization. This involves characterizing the baking technology properties, mainly to record baking losses and specific baked product volumes.

The influence of kneading time

The validation of dielectric spectroscopy as an instrument to characterise dough consistency and the degree of complete kneading based on an inline procedure is carried out on each of various intensively kneaded doughs. The aim in doing this is to generate “under-kneaded”, “fully kneaded” and “over-kneaded” doughs by an appropriate discretisation of the kneading time (2 min, 4 min, 6 min and 8 min), to determine the typical rheological and baking technology characteristics of each of these, and to correlate them with the data from the permittivity measurement.

Figure 7 shows examples of permittivity measurements of wheat doughs with an identical recipe (dough yield 158; Roland multipurpose flour Brema 550) and the varying kneading time mentioned above.

Figure 7: Profile of dough permittivity properties during kneading processes of varying intensities (kneading duration = 4 min, 6 min and 8 min) with standard wheat doughs (dough yield 158)

The representation of the change in the dielectric properties of dough as a function of kneading time takes place based on the characteristic parameters “turning point amplitude, turning point ΔT, fixed point amplitude and fixed point ΔT” described above (see figure 7). For a better overview, only one permittivity measurement at a kneading time of 4 min, 6 min and 8 min is illustrated in each case, whereby the end of the kneading process in each case is recognizable by the abrupt rise (amplitude values) and drop (ΔT values) respectively of the curve profile, because at this point the kneader’s machine head lifts up and at this time the sensor detects only air as the interacting medium (minimal permittivity compared to an aqueous matrix). It can be seen that the curve profiles of all four characteristic parameters pass through a maximum (amplitude values) and a minimum (ΔT values) respectively at approx. 3–4 min of kneading time, whereby for both pairs of characteristic parameters these extreme values indicate maximum binding of water in the dough matrix at this point in time, since minimum permittivity is detected at each of these points. It can therefore be assumed that the formation of the gluten structure and the flour swelling during the kneading process have reached a state that displays maximum binding of the added water into the mono- and multi-layer levels of the corresponding macromolecules.

The rheology and baking technology of the doughs thus obtained are subsequently characterized with the aim of reflecting by classical offline methods the maximum water binding thus detected, so as to yield the basis for validating the dielectric spectroscopy.

Characterisation of the baking technology is carried out using standard panned bread baking tests. The baked products obtained in this way are illustrated in figure 8.

Figure 8: Wheat pan loaves made from doughs with kneading times of 2 min, 4 min, 6 min
and 8 min (l. to r.)

It is apparent in Figure 8 that the panned breads based on a kneading time of 2 min have only inadequate “windowing” (network of surface cracks), consequently this indicates non-optimum kneading or “under-kneading”, because the flour swelling and for example the subsequently occurring dextrin release, which is a precondition for the thermal hydrolysis of the starch that induces, among other things, the characteristic feature of windowing (brittle material reactions generate windowing), has taken place to only an insufficient extent. On the other hand the panned breads with kneading times of 4 min, 6 min and 8 min show pronounced windowing, and so it can be assumed that there was adequate flour swelling during the kneading process, since material reactions that are clearly more brittle are detectable. However, the panned breads with a kneading time of 6 min and 8 min show distinct fissuring. This effect is explicable as a result of damage to the gluten structure due to “over-kneading”, because the dough piece does not have the necessary structural stability during the baking process to withstand the gas pressure during the volume expansion, particularly since all the baked products have an identical proofing technique.

Against this background it is clear that only the baked products with a kneading time of 4 min show optimum backing technological characteristics, indicating an optimum kneading process. Analogous to this, the dielectric spectroscopy shows maximum water binding and a minimum in the permittivity properties of the dough at this point in the kneading. An alternative baking technology characterization of the various panned breads also yields similar results (see Figure 9).

Figure 9: Characterisation of various degrees of kneading (from left to right, 2 min, 4 min, 6 min and 8 min kneading times) based on the specific baked product volume and baking loss of panned breads (confidence interval at p = 0.05)

Determination of the baking loss and specific baked product volume was carried out using a VolScan method for several baking tests. It is apparent in Figure 9 that panned breads with a kneading time of 4 minutes are clearly distinguishable from the other kneading times. The baking loss in particular shows a minimum at a dough kneaded for 4 min. This situation can be explained as being due to “optimum” binding of water into the macromolecular dough structures, with the result that there is maximum inhibition of the release of water during the baking process. Obviously the binding forces to multilayer laminates generate a maximum under these conditions, with the result that the baking loss is minimized, with all of the consequences derivable from it (freshness retention etc.). Therefore this parameter is also in agreement with the permittivity properties of the dough.

As a supplement, the rheological properties of doughs kneaded for various periods of time are examined by means of Extensogram recordings (see Figure 10).

Figure 10: Characterisation of various degrees of kneading (l. to r. 2 min, 4 min, 6 min and 8 min kneading time) based on the elastic and plastic properties of the dough (confidence interval at p = 0.05)

Characterisation of the elastic and plastic properties of the doughs takes place for each dough (l. to r. kneading times of 2 min, 4 min, 6 min and 8 min) in each case after dough resting times of 10 min, 20 min and 30 min in the proofing compartments of the Extensograph. The data in Figure 10 reveals that doughs with a kneading time of 4 min show maximum dough elasticities. These values indicate optimum gluten structure formation, which in turn agree with the detected permittivity properties of the dough as a result of a corresponding kneading time. It is repeatedly apparent that a minimum (permittivity), i.e. maximum water binding, occurs in the range of kneading times of 3–4 min.

Summary and conclusions

Currently there is no sufficiently specific technological-sensory method existing on the market that allows the implementation of an inline measurement method to control kneading processes, taking into account the corresponding requirements profile (accuracy, hygiene etc.). Through the detection of the dough permittivity by the use of dielectric spectroscopy it is possible to show that the control of kneading time is achievable based on characterizing the binding conditions in relation to the aqueous components of a wheat dough, thus representing a novel approach to the control of kneading processes. The measurement data obtained correlate with rheological and baking technology properties (including baking loss), thus providing novel strategies for process and quality control.

It is also evident from various test series that the measurement methodology described displays more far-reaching potential. Thus pronounced characteristics were detectable through the use of dielectric spectroscopy when using different wheat flour quality classes (high gluten, low gluten), as well as with batch-dependent quality fluctuations within a flour type (change of harvest) (data not shown). Based on these relationships, it can be assumed not only that the measurement system described is usable to control kneading time to compensate for flour quality fluctuations, but also that the amount of added water can be optimized for the individual grade of flour via the inline measurement method. In addition to the applications for wheat doughs shown above, it is also expected that validation of the sensor technology will also be achievable for rye doughs.

The R&D project mentioned above is structured as a cooperation project being carried out in close cooperation between the partners ttz-BILB/EIBT, Sequid GmbH and Diosna Dierks & Söhne GmbH. Financing takes place in the context of the funding program “German Central Innovation Program SME” by the German Federal Ministry of Economics and Technology (BMWi) – Cooperation Project Funding Module (Funding Project No.: KF2010616SK0).


[1] Bolling, H., Weipert, D. (1977) Roggenqualität – erläutert an der Interdependanz zwischen Sorte, Umwelt, Mahl- und Backeigenschaften. Getreide, Mehl und Brot. 31, (12), S.307 – 312
[2] Amend, D. 1996. Grundlagen der Teigbildung bei Weizen- und Roggenteigen. Handbuch Backwaren, Kapitel 3. Behr’s Verlag Hamburg, Hamburg
[3] Belitz, H.-D., Grosch W., Schieberle, P. (2001) Lehrbuch der Lebensmittelchemie (5. Auflage) Berlin, Heidelberg: Springer Verlag
[4] Weipert, D. Beurteilung der Knetintensität der in der Teigrheologie eingesetzten registrierenden Kneter. Mühle Mischfuttertechnik 124 (12): 147-151
[5] Gerbel, A., Ahlert, D. Steuerung des Knetprozesses ueber Teigkonsistenz., Getreide-Mehl-und-Brot (Germany). (1994). v. 48(1) p. 23-26.
[6] AHLERT D.; Hohe Produktqualität über optimale Teigkonsistenz; Brot und Backwaren; 12; 10-11; 1992
[7] ZHENG H.; Rheological Properties of Dough During Mechanical Dough Development; Journal of Cereal Science; 32; 293-306; 2000
[8] Knocks, A. Dielektrische Spektroskopie an Lösungen kleiner, biochemisch relevanter Modellsysteme, Dissertation, Ruhr Universität Bochum, 2001
[9] Kieffer, R. (2004) Grundlagen der Teigbildung bei Weizen- und Roggenteigen. Handbuch Backwaren Technologie. (1. Auflage) Hamburg: Behr’s Verlag

Go back