Tackling Foam-Based Process Disruptions in Spirit Distillation by Thermal Energy Input Adaptations

Process impairing foam formation occurs regularly in batch distillation devices of the spirit industry. It negatively influences process and product quality. Up to now, such foam-related problems have not been in the focus of scientific investigations. This study aimed at preventing impairing foam formations by adapting the thermal energy input in fruit and grain mash distillations in larger scale batch distillations. The results showed that a reduction of the thermal energy input to 43 ± 1 W·L−1 during the initial heating of the mash leads to less flooding of the distillation apparatus and to a higher concentration of lower boiling compounds like methanol, acetaldehyde, and ethyl acetate as well as ethanol in the first fractions of the distillates. A standard process time and less energy consumption could be achieved by increasing the energy input again after prior reduction. However, this led to a reduction of the ethanol concentration in the distillate fractions of up to 4.3%vol, also most severe in the first fractions. A significant influence on analyzed volatile compounds in the distillate besides ethanol could not be detected. This is the first study that uses defined thermal energy input adaptations for foam management in larger scale distillation devices. The results lead the way to a more efficient distillation process with less foam formation.


Introduction
In fermentation technology, foam formations are generally considered an undesired side effect (Dorsey, 1959;Stanbury et al., 1995). Foam formation can lead to over foaming of fermentation vessels or disturbances during subsequent process steps of the fermented substrates (Murray, 2014). With respect to distillation processes, the presence of excessive foams leads to the carryover of foams onto separation trays, product contamination, and increased cleaning demands. The effects are associated with reduced separation efficiency and process disruptions (Kister, 2003;Miller, 2019;Schidrowitz & Kaye, 1906). When it comes to column malfunctions in the distillation industry, foaming plays a major role (Kister, 2014). In spirit drink production, certain raw materials are referred to be especially prone to excessive foam formation, such as cherry, wine yeast, Bartlett pear, and grain mashes (Pieper et al., 1977). Aside from this descriptive information on foam formation capacities of different mashes, Heller and Einfalt (2021) demonstrated in laboratory experiments that the polysaccharide pentosan found in rye has a major impact on foam formation in distillations of rye mashes. Additionally, it is well-known that besides polysaccharides, proteins play a crucial role in the physical stability of food foams (Dickinson, 1998(Dickinson, , 2003Nunes & Coimbra, 1998;Prins, 1988;Ye, 2008). However, in spirit drink production, comprehensive researches on substrate-based factors influencing foam formations in distillations are still missing.
To reduce foam formation during distillation, several measures have been applied. For instance, manufacturers introduced adaptations to their distillation devices such as increased pot still headspace volume and foam retention installations (Pieper et al., 1977). This, however, increases manufacturing costs. Such adaptations already have a long history. For example, the elongated swan necks in Scottish pot stills in whisky distilleries were originally invented to prevent the transfer of rising foams into the product stream (Schidrowitz & Kaye, 1906). Unfortunately, larger distillation devices have a higher surface area, which is associated with higher process costs due to additional surface area heat radiation.
In addition to these aspects, operators of distillation plants take their own actions to prevent undesired foam formations. Chemical defoamers based on silicone or mineral oil, also called anti-foam agents, are widely used for this purpose (Miller, 2019). The negative aspect of defoamers is their disposal with the stillage without special treatment. They end up in the environment, where they can cause undesired ecological effects (Routledge et al., 2014).
Detailed information on process-related solutions for proper foam management in distillation is scarce and for the most part based on heuristic recommendations. Heller and Einfalt (2021) showed that 90% of foam formation took place at mash temperatures ≥ 89.5 °C. Pieper et al. (1977) recommended "reduced thermal energy input" in order to diminish excessive foam formation. They found it beneficial to "slow down the heating process in the foam-critical temperature range." However, no additional information is given to define the thermal energy input magnitude or foamcritical temperature range. Therefore, the handling of foams is still dependent on operator experience. Extensive research is required to support new and experienced distillation plant operators in handling foam formations and in assessing the impact of anti-foam measures on their distillate. Therefore, it would be desirable to define parameters that enable a foam-resilient distillation process with low additional costs for manufacturers and operators. To close this knowledge gap, this study investigated the effects of different heating profiles on foam formation in larger scale distillation processes based on fruit and grain mashes. The study also evaluated effects on process time, energy consumption, and volatile compound composition in the resulting product.

Distillation System
A steam-heated 120-L batch copper distillation device equipped with a rectification column (Carl GmbH, Eislingen, Germany) on top was used for the distillation experiments ( Fig. 1). The attached rectification column was equipped with three sieve trays, a partial condenser at the top, and a foam retention installation at the bottom. The distillation device also contained four foam-detecting sensors (CleverLevel, Baumer GmbH, Frauenfeld, Fig. 1 Instrumentation diagram of the batch distillation system equipped with rectification column (a), copper catalyzer (b), and product condenser (c). 1: Temperature sensor "Mash." 2: Tempera-ture sensor "Headspace." 3: Temperature sensor "Tray 1." 4: Temperature sensor "Tray 2." 5: Temperature sensor "Tray 3." 6: Flow sensor "Reflux." 7: Flow sensor "Product." 8: Level sensors "Foam height" Switzerland) installed in the reboiler at different heights positioned, 5 cm, 10 cm, 15 cm, and 20 cm above the mash surface. Excessive foam formation, which reached the foam retention installation or the trays, was logged manually. The distillation device was additionally equipped with eight temperature sensors (PT100) positioned in the reboiler, at different heights of the rectification column, in the vapor line, and in the product condenser. Further technical sensors (Coriolis Micro Motion H Series, Emerson Electric Co., Ferguson, USA) measured product volume flow and reflux volume flow of the partial condenser. Cooling water flow in the product condenser and the partial condenser were measured via SM6120 flowmeters (ifm electronic GmbH, Essen, Germany).
The distillation device was regulated and monitored via an automated process control system (DPC500, Carl GmbH, Eislingen, Germany). It enabled the regulation and control of the thermal energy input via an electropneumatic steam valve (SP400, Spirax-Sarco Engineering plc, Cheltenham, UK). Besides thermal energy input, the process control system also regulated the cooling water volume flows. All technical sensors were monitored, and data were recorded every 10 s via the process control system. The measured data were averaged over 60 s.

Experimental Setup
Two test series were performed in order to investigate the effects of modified heating profiles on foam formation ( Fig. 2 and Table 1). The basis was a common heating profile (Dürr et al. 2010), which was modified at two points (adaptation points) in the test series and served as a control.
The common heating profile started off with an initial thermal energy input of 450 W·L −1 ( Q 1 ). At a defined mash temperature of 90 °C (T Mash = first adaption point), the thermal energy input was reduced to 134 W·L −1 ( Q 2 ). When the top/third tray of the rectification column reached a temperature of 75 °C (T Tray3 = second adaption point), a steady increase of thermal energy input ( Q 3 (t) and q, respectively) was applied. The thermal energy input was increased until the first product ran out of the distillation device. Subsequently, the distillation device automatically regulated the thermal energy input to maintain a constant product flow of 10 L·h −1 . Cooling water flow of the partial condenser was kept constant at 2 L·min −1 . The distillation process was terminated, as soon as the ethanol concentration of the product dropped below 25%vol.
The first test series employed reduced thermal energy input levels to inhibit foam formation. Thus, the control heating profile was modified at the first adaption point. In three experiments (HP1.1-HP1.3), the adaption point T Mash was set to a mash temperature of 75 °C and the thermal energy input ( Q 2 ) was reduced to 95, 59, and 43 W·L −1 , respectively. In a subsequent experiment (HP1.4), a reduced thermal energy input of 43 W·L −1 ( Q 2 ) at the original T Mash was applied. Because of customs restrictions concerning the duration of distillation runs, the increase in thermal energy input (q) was higher in HP1.3 and HP1.4. The second test series aimed at counteracting process prolongations induced by the thermal energy reduction in the first test series. It was based on the most promising heating profile of the first test series and introduced changes at the second adaption point. When T Tray3 was reached, the thermal energy input was raised by a certain amount ( Q 3 (0)); then, the thermal energy input (q) was increased over time as before. Distillation experiments in the second test series started with HP2.2. Subsequent distillation experiments were performed with HP2.1 or HP2.3 due to observed foam formation in HP2.2.
Distillation experiments of the first test series were performed with 100 L rye mashes in single determination. Distillation experiments of the second test series were performed with 100 L rye mashes and 100 L Bartlett pear mashes each in duplicates.

Sampling
The produced distillates were collected in 20 fractions for subsequent analysis. The first ten fractions contained 100 mL, while fractions 11-15 contained 200 mL and fractions 16-19 contained 1 L each. The final fraction contained the rest of the produced distillate. Distillations of Bartlett pears yielded less distillate. Bartlett pear distillates were therefore fractioned until fraction 18, while fraction 19 contained the rest of the produced distillate (< 1 L).

Raw Materials and Mash Preparation
Winter rye was purchased from Hahn-Mühle (Ostfildern, Germany) and Bartlett pears imported from South Tyrol were received from Kaiser Destillerie-Obstweinkellerei (Salach, Germany).
In order to quantify only the effects of the changed heating profiles on the process and the distillate composition, it was necessary to exclude influences by differing mash characteristics. Therefore, each test series was performed with the same mash batch.

Substrate Characteristics
In order to exclude temporal changes of substrate characteristics in mash batches between distillations, all drawn mashes were sampled before distillation and analyzed for specific substrate characteristics in triplicates. Samples were stored at −20 °C prior to analysis.
Substrate characteristics included quantification of ethanol concentration, performed according to Senn and Pieper (1996). Protein concentration was analyzed using the method by Bradford (1976). Total carbohydrates were quantified via phenol-sulfuric acid method (Neilson et al., 2017). pH and conductivity were measured via multimeter (HQ40D, Hach, Loveland, USA). Viscosity determination was performed after particle extraction (> 0.45 mm) with a rotary rheometer (MCR 92, Anton Paar, Ostfildern, Germany) equipped with a concentric cylinder system based on method DIN 53,019 (Deutsche Institut für Normung e.V. 2009). Viscosity determination was performed at shear rates from 0.1 to 1000 s −1 at 20 °C. In addition, dry matter (DM) and ash content of the mashes were determined according to Sluiter et al. (2008).

Statistical Analysis
Pearson correlation was applied to ascertain significant correlations of thermal energy input with volatile compound concentrations of the distillates (p ≤ 0.05). One-way analysis of variance (ANOVA) was performed to determine significant changes in substrate characteristics of mash batches between distillation experiments. All statistical analyses were performed using SPSS Statistics (V.25, IBM, Chicago, USA).

Substrate Characteristics
Analysis of mash samples taken prior to distillation showed no significant differences between mashes within a mash batch. This shows that the mashes were not significantly altered during storage between the distillation experiments. The determined substrate characteristics showed deviations of up to 6.25% (Table 3, viscosity curves not shown). We attribute these deviations to the non-homogeneous nature of the substrate, as well as to the limits of accuracy of the measurements. No alteration in ethanol content and viscosity indicated a completed liquefaction and fermentation of the mash batches prior to the distillations. Therefore, it can be assumed that differences in distillation occurred exclusively due to changes in the process conditions. The obtained results of the substrate measurements were in accordance with data from the literature (Hansen et al., 2004;Pieper et al., 1977).

Process and Foaming
Distillation with the control heating profile resulted in excessive foam formation > 20 cm and led to over foaming onto the first tray of the column (Fig. 3a). Distillations performed with heating profiles HP1.1 and HP1.2 did not result in reduced foam formation (data not shown). Both experiments showed foam formation > 20 cm. Furthermore, both heating profiles also led to excessive foam formation, which reached the first tray. In addition, both heating profiles resulted in increased process time compared to the control of 39% and 70%, respectively and in increased energy consumption of 6.8-10.9%. In HP1.3, a reduction of the thermal energy input ( Q 2 ) to 43 ± 1 W·L −1 was applied. This resulted in reduced foam formation (Fig. 3b), where the foam height fluctuated within 10 to 20 cm. However, this heating profile showed an 86% extension of the process time and resulted in an increased energy consumption of 14.6%.
To counteract the extension of the process time, the first adaption point T Mash was shifted from 75 °C mash temperature to 90 °C in heating profile HP1.4 (Fig. 4). This reduced the process time extension from 86% (HP1.3) to 63%, while maintaining a similar foaming behavior as in heating profile HP1.3. However, HP1.4 showed a higher energy consumption of 19.5% compared to the control. Excessive foam formation and over foaming onto the trays did not occur. 3-Methyl-1-butanol 6 2-Methyl-1-butanol 7 1-Butanol 8 Acetaldehyde diethyl acetal 9 2-Methyl-1-propanol 10 1-Hexanol The experiments of the first test series demonstrated that foam formation can be controlled by reduced thermal energy input during the initial boiling of the mash in larger scale distillation devices. This result confirms the previously mentioned heuristic recommendation of Pieper et al. (1977) and the results obtained in laboratory-scale experiments of Heller and Einfalt (2021). The heating profile HP1.4 showed the best results regarding foam formations. However, HP1.4 had the drawback of an increased process time (63%) and higher energy consumption (19.5%) compared to a common heating profile. Nevertheless, HP1.4 should be applied, if mashes with high foam formation potential are to be distilled. Otherwise, the usage of a common heating profile may result in over foaming of the mashes leading to a process stop, a time-demanding cleaning of the distillation system, and a restart of the process. This could cause an even longer process time and a higher energy demand compared to HP1.4.

Distillate Composition
A reduction of the thermal energy input ( Q 2 ) was negatively correlated with the ethanol, methanol, acetaldehyde, and ethyl acetate concentrations in the first fractions of the distillates, i.e., the lower the thermal energy input, the higher the concentrations of the compounds in the first fractions (Fig. 5). Significant correlations of thermal energy input and ethanol (fractions 1-13), methanol (fractions 1-14), acetaldehyde (fractions 1-5), and ethyl acetate (fractions 1-4) were found (r = −1.000-−0.951; p = 0.000-0.049), respectively. The concentrations of methanol, acetaldehyde, and ethyl   (Aylott & MacKenzie, 2010;Douady et al., 2019;Nascimento et al., 2008;Rodríguez-Bencomo et al., 2016). The reason for the increase of ethanol, methanol, acetaldehyde, and ethyl acetate concentrations in the first fractions of the distillates in the distillation experiments with a reduced thermal energy input appears to be a higher internal reflux in the column. High internal reflux before    (HP1.1, HP1.2, HP1.3, HP1.4) and around the time when the first distillate fractions are collected allows for a higher enrichment of lower boiling compounds like methanol, acetaldehyde, and ethyl acetate in the first fractions (Luna et al., 2018;Matias-Guiu et al., 2016;Scherübel, 2018;Spaho, 2017). As shown in Fig. 3, a decrease in thermal energy input increased the total volume of the reflux before the first distillate fraction was collected. The total reflux in the distillation experiment with HP1.3 was 33.8 L. The control distillation experiment showed a total reflux of 19.9 L. HP1.4 showed a total internal column reflux similar to HP1.3 (37.8 L) (Fig. 4). Because the internal reflux converges over time to a similar level of 20 L·h −1 in the different experiments, the volatile compounds no longer differed in their concentration after fraction 14.
In addition to the internal reflux, other factors, like the carryover of mash onto the first tray by excessive foaming in HP1.1 and HP1.2, could in part explain the observed differences. In essence, a separation step is lost due to the carryover of liquid onto the first tray, which results in reduced rectification performance of the column (Spaho, 2017).
All in all, the experiments show that a decrease of the thermal energy input in the heating phase of the mash affects only the composition of the first distillate fractions; the later fractions seem to be largely unaffected. Because the head fractions are normally discarded and not consumed, there is no drawback in using one of the modified heating profiles with respect to the final distillate composition. No significant changes in the concentrations of the other analyzed substances could be detected in the final distillate (not shown).

Process and Foaming
In the second test series, we considered an increase in thermal energy input, when T Tray3 was reached, to counteract process prolongations induced by prior thermal energy reduction.
Rye mash distillations with the control heating profile resulted in foam formations > 20 cm (Fig. 6a). In both distillations, the excessive foam formations reached the first tray. In contrast to the first test series, the excessive foam formations now even reached the second tray. In the two duplicate experiments, the time when over foaming occurred differed by 1.5 min. Because of this temporal shift, only one profile is shown.
Distillations with HP2.2 (data not shown) reduced foam formation and showed no overflow onto the trays. Yet, HP2.2 showed foam formations > 20 cm and reached the foam retention device. To further reduce foam formation, subsequent distillations were performed with HP2.1. In distillations with HP2.1, foam formation was limited to the still pot and did not reach the foam retention device. Initially, the foam height varied between 15 and 20 cm (Fig. 6b). Both heating profiles (HP2.1 and HP2.2) showed a significantly reduced extension of the process time and energy consumption compared to HP1.4. HP2.1 extended the process duration by 5.9% ± 0.5% and reduced the energy consumption by 1.9% ± 0.2% compared to the control. HP2.2 showed no increase in the process duration (< 1%) and reduced the energy consumption by 5.7% ± 0.1%.
Thus, the experiments show that the delay in process time introduced by a prolonged heating phase can be compensated by a higher thermal energy input in the later phase leading to a similar process time as in the control experiment.
The same experiments were performed with a different feedstock. Bartlett pear distillations with the control heating profile showed a high initial foam formation of > 20 cm, which was limited to the still pot (Fig. 7a). The initial foam collapsed and leveled off at 10 cm foam height. Distillations with the heating profile HP2.2 showed no foam formation throughout the process (data not shown). Distillation with HP2.3 showed foam formation of < 20 cm only in one of the duplicates (Fig. 7b). The two heating profiles resulted in an extension of the process time of 23% ± 1% (HP2.2) and 10% ± 1% (HP2.3). However, HP2.2 and HP2.3 reduced the energy consumption by 5.5% ± 0.1% and 12.1% ± 0.1%, respectively.
The results show that different mash types require specific heating profiles for inhibiting or reducing foam formation. Also, it became apparent that foam reduction is possible without excessive extension of process time as shown with HP2.3 in Bartlett pear and HP2.2 in rye mash distillations. Additionally, the altered heating profiles had the benefit of lower energy consumption compared to the common heating profile. We therefore concluded that the application of the altered heating profiles has no major drawbacks in terms of process efficiency, but also limits the risk of over foaming compared to a common heating profile.

Distillate Composition
Statistical analysis of the data revealed a significant negative correlation of the ethanol concentration with a higher thermal energy input after the second adaption point in rye mash as well as in Bartlett pear distillation (Fig. 8). In distillations performed with rye mashes, significant correlations were found up to fraction 6 (r = −0.940-−0.848; p = 0.005-0.033). In distillations with Bartlett pear mash, a higher thermal energy input after the second adaption point was significantly negatively correlated with the ethanol concentration up to fraction 16 (r = −0.967-−0.844; p = 0.002-0.035). The highest ethanol concentration reduction of 4.3%vol was found in the second distillate fraction of Bartlett pear distillation with HP2.3. The reduction of ethanol concentration in the drinkable fractions was 1.5-3.15%vol. No significant correlations with other analyzed volatile compounds' concentrations were found.
Previously, we attributed differences of the ethanol concentration by different heating profiles to the internal column reflux and excessive foam formation, which reached onto trays. In distillation experiments with rye mashes, the control had the highest total reflux of 55.3 ± 2.3 L before the first fractions. HP2.2 had the lowest total reflux of 36.0 ± 1.3 L. We assume that the ethanol concentration in the control distillation would have been even higher, if not for foam formations, which reached onto the trays. This probably counteracted the effects of the higher internal reflux. In distillation with Bartlett pear, the control showed a total reflux 25.0 ± 1.2 L before the first fraction. At the same time, HP2.3 showed a total internal column reflux of 20.8 ± 0.9 L. Therefore, these results were also consistent with the results of the previous experiments.
In terms of distillate composition and ethanol recovery, the altered heating profiles have the drawback of a reduced ethanol recovery compared to a common heating profile. However, the reduction in ethanol concentration was most severe in the first fraction of the distillates, which are normally discarded as heads. Considering the previously stated benefits of the altered heating profiles in terms of process and foaming, the slight reduction in ethanol recovery is economically bearable.