Extraction of sucrose from sugarbeet for sugar production on an industrial scale can be divided into two distinct types of operations: beet processing (operation of the entire factory) and syrup processing (operation of the sugar refinery only). During the beet harvesting campaign evaporator syrup is stored in large tanks for subsequent processing in the syrup campaign. It is good practice to filter the evaporator syrup prior to storage for removal of microbes and insolubles. More importantly is filtration of the syrup during the syrup campaign for food safety and to remove any solids that have precipitated during storage for sugar quality. This is normally done using a filteraid to ensure filtration to below 5 µm is achieved. The use of a filteraid is not ideal due to health concerns and the risk of causing sugar quality issues if filteraid is carried forward with the syrup. A filtration system designed to operate without the use of a filteraid was therefore trialled at one of the Tereos factories in France. Standard liquor from storage tanks was filtered through 14 – 25 µm filter material during the 2018 syrup campaign. Operational and sugar quality results are discussed.
The beet sugar harvesting campaign in Europe lasts typically from September to December. Beet is often harvested in advance and stored to avoid operational lost time due to frozen ground when harvesting becomes impossible. To extend the sugar production campaign beyond the 5-6 month harvesting period, beet sugar factories often operate a second campaign during the summer months, known as a syrup campaign. Intermediate evaporator syrup – called thick juice – can, under the right conditions, be stored in large tanks for several months without significant deterioration of the technical sucrose content (Lambrechts, 1967; McGinnis, 1982; Schrevel, 2009). The syrup campaign operates the back end of the factory independently to produce sugar from thick juice. There are numerous advantages of operating two distinct campaigns, including capital utilization, sizing of the back end in relation to the front end of the factory and overall energy usage per tonne of sugar produced (Van der Poel et al, 1998; Asadi, 2007).
During this storage time it is common for a fine precipitate to form which makes the syrup harder to process (Abdel-Rahman and Floeter, 2015; Rogé et al, 2007; Bensouissi et al, 2009). In the beet campaign this precipitation is prevented or at least delayed by the use of anti-scaling agents, typically added to thin juice to prevent calcium deposits in the evaporator and to prevent sugar quality issues (Demadis and Léonard, 2011; Doherty and Wright, 2004; Bensouissi et al, 2009). During storage, the chelated calcium is released due to the lack of efficacy of the anti-scalant to sequester calcium at lower temperatures (<22°C), often precipitating out with organic acids and/or other anions to form fine suspended solids. This can occur within the first few weeks of storage or after several months, depending on the nature of the anti-scalant used (Bensouissi et al, 2009). More importantly, the solids tend to be smaller the more anti-scalant is used, with devastating consequences to processing of the thick juice (Abdel-Rahman and Floeter, 2015; Rogé et al, 2007; Bensouissi et al, 2009). The precipitate can also be prevented by a well-operated decalcification operation.
Sugar quality is defined within the European Union in terms of its aspect (brightness), colouration and conductivity ash content. Complementary quality parameters are typically imposed by soft drinks manufacturers who require sugar without turbidity and insoluble solids as well as neutral organoleptic properties and low microbial content.
Turbidity in granulated white sugar is any fine, suspended particles that can cause a scattering of light larger than 0.45 µm. Rogé et al (2007) studied this phenomenon in detail and found a correlation between turbidity in juice and turbidity in sugar and this was also correlated well with the calcium content in both the sugar and juice. Bensouissi et al. (2009) demonstrated that sugar turbidity is mainly constituted of free calcium oxalate that is found on the surface of the crystal. They also found some calcium oxalate complexed with high molecular weight macromolecules but these were included in the sucrose crystal itself. Therefore, only about half of the turbidity can be washed off the crystal surface during centrifugation. Given that the turbidity of sugar increases with increasing thick juice storage it is likely that the majority of these complexes are formed during storage.
Calcium oxalate in sugar processing
The sugar clarification process is specifically designed to support the formation of insoluble compounds that can be removed by settling or filtration. Amongst others, oxalic acid is precipitated as calcium oxalate during the liming step in clarification. The sugar beet root naturally contains oxalate, calcium and calcium oxalate complexes (Joy, 1964; Van Der Poel et al, 1998a). Microbial infections such as fungi can also contribute to oxalic acid concentration (Dutton and Evans, 1996). Typical levels of calcium in beet are 0.24% (Drycott, 2006), depending on variety, soil and growing conditions. The clarification process adds an excess of calcium that is precipitated out as calcium carbonate leaving some residual soluble calcium in the juice.
Calcium oxalate in sugar factories have been studied at length as it was found to form one of the major constituents of evaporator scale in both cane and beet factories. Scale deposits reduce the heat transfer coefficient in the evaporators, leading to increased energy costs, more frequent requirements for maintenance and down-time for cleaning. It is important to distinguish between beet and cane scale not only because of composition but also as scale formation in a cane evaporator is orders of magnitude higher than in a beet evaporator. An evaporator train in a cane factory will typically be cleaned every 2-4 weeks while a beet evaporator train can be operated for 6 months without the need for cleaning. Nevertheless, learnings from investigations in both applications are of interest in the current discussion.
Walford and Walthew (1996) studied the solubility of calcium and oxalates in the evaporators in a cane factory and found that the solubility decreases with increase in temperature and increase in sugar concentration. Doherty and Wright (2004) confirmed this and added that the solubility increases with increase in pH. Besouissi et al (2009) indicates a solubility of 13.8 mg/L at 85°C. A large portion of the organic acids present in diffusion juice will be precipitated out during liming with calcium hydroxide milk or saccharate. Our own studies (unpublished) confirmed that double carbonatation involving progressive pre-liming removes more than 90% of the oxalates with typical levels of no more than 50-150 ppm (on dry substance) of oxalic acid in beet clarified juice.
It was suggested (Walford and Walthew, 1996; Walford, 2000) that the oxalate present in evaporator scale is likely a product of the decomposition of other organic compounds present in the juice such as aconitic acid (an intermediate compound in the degradation of citric acid). While aconitic acid is specifically a sugarcane deterioration product, it is found in only very low amounts in beet sugar: 100-200 g/kg in cane (Kanitkar et al, 2013) and <1 g/kg in beet (Norman et al, 1950). Nevertheless, Bensouissi et al (2009) confirmed that oxalic acid was indeed formed during the beet sugar evaporation process.
Van der Poel et al (1998) implicates another two potential sources of oxalic acid formation in the sugar process that pertains to cane and beet. Firstly, the decomposition of allantoic acid to glyoxylic acid and subsequently to oxalic acid, and secondly oxamic acid (half-amide of oxalic acid) the saponification of which will proceed ‘rather slowly’. They indicated that allantoic acid appears to be prevalent in beet in drought years.
Characterisation of stored thick juice or standard liquor
There are few general methods for characterization of thick juice to predict the quality of the resulting sugar in terms of filtering quality, turbidity and insoluble solids levels. The turbidity or turbidity forming compounds in thick juice can be measured by adapting the ICUMSA turbidity method for white sugars (Anon, 2013) to measure the turbidity together with colour (Anon, 2011) at an absorbance of 420 nm. Other wavelengths are often used (e.g. Anon, 2007b at 900 nm) as well as nephelometry or bespoke turbidity meters. However, none of these methods give information about the nature of the turbidity forming particles due to lack of specificity and limitation on historical data for comparisons.
The particle size analysis (PSA) gives insight into the size distribution and is critical in decisions about the size of the filter material if this is an option during operations. Figure 1 shows a typical size distribution of stored thick juice from two different tanks. The peak between 0.5 and 1.5 µm is representative of either microbes or CaCO3 while the larger portion comprises of mostly Calcium oxalate dihydrate in the bi-pyramidal form. However, particle size on its own could be misleading, firstly because it gives no indication of the level of solids loading which makes it dangerous for comparison of different juices. Secondly, the particle size is normally based on a volume distribution that skews the graphs towards the bigger particles, potentially inflating the higher size classes. PSA should, therefore, be used in conjunction with other measurements.
Figure 1: Typical thick juice particle size analysis (PSA) traces from 3 different tanks
The sugar insoluble solids test can be adapted to measure the solids loading in thick juice above a certain particle size (determined by the pore size of the membrane used during the test). In sugar, 8 µm membrane is used (Anon, 2007a) but membranes in other sizes are available to give further insight and to help determine the filtration conditions.
Other analytical tools essential for both the characterisation of thick juice and for comparison of different juices are calcium (dissolved or total after digestion) determined by EDTA titration and oxalic acid, determined by HPLC.
Standard liquor filtration
A number of authors conclude that filtration of calcium precipitate is not a technical solution to the problem of calcium precipitation during storage and that focus should remain on decalcification in the beet campaign (Abdel-Rahman and Floeter, 2015; Rogé et al, 2007; Bensouissi et al, 2009). Nevertheless, filtration of syrup after storage is widely used and, in most cases, able to control the quality parameters in terms of solids and turbidity. Traditionally, these filters rely on a filteraid (such as Perlite or Celite) to be able to get particles as small as 4 microns removed. The use of a filteraid is not ideal due to health concerns and the risk of causing sugar quality issues if fine filteraid is carried forward with the thick juice.
As a filter aid needs to be inert by definition and has a small particle size distribution (1-10 µm) it is generally classed as hazardous with long term ill health effects. It also has a potential dust explosion risk. Safety measures required may include forced ventilation, mechanical handling systems, rated zones and dampening of the dust.
A filtration system designed to operate without the use of a filteraid was therefore trialled at one of the Tereos factories in France during the 2018 syrup campaign. The system consisted of a number of automatic backwash filters type OptiFil® from Lenzing Technik GmbH. The OptiFil® is a continuous system that works according to the principle of depth, surface or cake filtration. A metal fiber fabric or fleece is used as filter material, which retains particles of different sizes either inside or on its surface.
During filtration the juice is filtered from the inside of the perforated drum (Figure 2, Unfiltrate Room or Room P1) through a single layer flat sheet stainless steel filter fabric, which can be easily exchanged to various micron ratings ranging from 100 down to 5 µm, to outside (Figure 2, Filtrate Room or Room P2).
After the maximum degree of contamination has been reached, indicated by a differential pressure over P1 and P2, the total surface of the filter material is cleaned by a backwash procedure. This takes place by one rotation (360°) of the backwash device, which forces the impurities in reversed filtration direction through the filter material (Figure 2, Concentrate Room or Room P3). The channel, shaped opening in the backwash strip, which seals to the inside surface of the perforated drum (patented design), executes the backwash with the minimum quantity of backwash liquid (filtrate), which is necessary to rinse the filter material from impurities. This process is depicted in Figure 2.
Figure 2: Cut-depiction of the Lenzing OptiFil® in backwash status
After having cleaned the whole surface, the backwash device remains in the waiting position until the differential pressure reaches the preselected value again.
Standard liquor is normally obtained from stored thick juice and several different recycling streams from the sugar end (e.g. B- and C-sugar). During the syrup campaign, standard liquor was filtered at 85°C through stainless steel woven filter material in a series of trials to systematically find the optimal balance between flow rate, pressure differentials and cycle length for filter media between 14 µm and 25 µm. Lower sizes were considered but found unnecessary as only a portion of the solids need to be removed to be able to produce a good quality sugar.
Particle size analysis (PSA)
Particle size analysis was performed using a Masterseizer 3000 with a liquid sample cell. Results are expressed graphically as volume distributions on a logarithmic scale.
The syrup sample was diluted with demineralised water and filtered under pressed (3 bar) through a 5 micron cellulose nitrate membrane. The membrane (pre-dried and pre-weighed) were dried under vacuum overnight and the increase in weight expressed on the original sample weight in mg/kg.
The ICUMSA Methods for colour (Anon, 2011) and turbidity (Anon, 2013) were modified as follows: the sample was dissolved in demineralised water. A portion of the solution was filtered under vacuum through a 0.45 µm cellulose nitrate membrane. The absorbance of both filtered and unfiltered solutions were measured on a Shimadzu uv-1800 UV-VIS spectrophotometer at 420 nm in a 1 cm cuvette against a water blank. The results were expressed as ICUMSA Units.
Scanning electron microscope with elemental detection analysis (SEM with EDA)
SEM images and elemental analyses were performed on a JEOL Electron Microscope model JSM-6010LA (In-Touc) with the aid of TouchScope 6010LA software. The surface of a piece of each of the 5 µm membranes from the solids loading test were irradiated with a fine electron beam so that secondary and backscattered electrons are emitted from the specimen surface. The surface was observed by acquisition of an image from the detected electrons. Elemental detection was used to determine the presence and quantity of elements (excluding hydrogen).
The factory setup consisted of a number of Lenzing filters in parallel, set up to allow for throughput of thick juice or standard liquor of up to 60 m3 per hour per filter. During the trials, standard liquor was filtered at 85°C, preceded by 50 µm bag filters to ensure a relatively clean juice is presented to the filters. Filter material of between 14 µm and 25 µm were tested with juice from two different storage tanks over 45 days. Production sugar samples were collected directly after drying and cooling.
Results and discussion
Tank 1 was filled during the middle of the campaign when beet quality was good and calcium levels were low (calcium results not shown). However, Tank 1 contained unusually large crystals of calcium oxalate in the tetragonal bipyramidal dihydrate form that measured as much as 20 µm although there were smaller crystals present; literature indicates the dihydrate is normally expected to be around 5-7 micron and in the elongated bipyramidal form (Rogè et al, 2007; Besouissi et al, 2009). This shape and size of crystals from Tank 1 as shown in the SEM image in Figure 3 has not previously been reported in the sugar beet literature.
Figure 3: SEM image of the solids in Tank 1 (bipyramidal calcium oxalate dihydrate)
Besouissi et al (2009) suggested that the size of the calcium oxalate crystals is dependent on the amount and type of anti-scalant used. This type of phenomenon has been studied extensively in the medical sector – calcium oxalate is commonly known as ‘kidney stones’ – and in water treatment applications (Demadis and Léonard, 2011). For example, Ishii (1991) studied the crystallisation of different calcium oxalates and found that the tetragonal bipyramidal form can be produced in the presence of malic or citric acid. After some enquiries we found that the only processing aid that was different between the filling of Tank 1 and the filling of Tank 2 was an anti-foaming agent used in the beet washing operation (cold temperature). This processing aid is not expected to carry through the process until evaporator juice; if it did, the trace amounts present during calcium oxalate precipitation could have an effect on the crystallization rate and morphology and therefore on the size of the final crystals.
Solids loading in Tank 1 was consistently around 200 mg/kg throughout the tank as measured on 5 µm membrane. This was also unusual as the solids in a tank normally tend to be higher towards the bottom part of the tank due to settling over time. Figure 4 shows the solids loading at different taps with tap 1 being one meter from the bottom of the tank.
Figure 4: Solids loading in Tank 1 in mg/kg sample. Tap 1 is at the bottom of the tank.
Tank 2 was filled during the latter part of the season when beet quality was on the decline and calcium levels were higher (calcium results not shown). Tank 2 contained calcium oxalate crystals in the elongated bipyramidal dihydrate form and sized up to 6 micron which is normal for stored beet juice calcium oxalates.
A SEM image of the solids retained on a 5 micron membrane is shown in Figure 5.
Figure 5: SEM image of the solids in Tank 2 (elongated bipyramidal calcium oxalate dihydrate)
Solids loading in Tank 2 was also more normal for stored beet juice with the lower taps showing higher solids levels compared to the others. In general the solids were somewhat higher compared to Tank 1 Solids loading as determined on 5 micron membranes are shown in Figure 6.
The first trial was started on 7/05/2018 filtering standard liquor made from thick juice from Tank 1 through 25 µm filter medium. Throughput of 60 m3 per hour per filter could be maintained. However, it was soon clear that the filter material was too large for effecting filtration as there was very little differential pressure over the inlet and outlet chambers and this did not increase significantly over the first 24 hours so the backwash sequence was done manually. The filter material was therefore changed for a smaller size.
Figure 6: Solids loading in Tank 2 in mg/kg sample. Tap 1 is at the bottom of the tank.
Trial 2 was started on 9/05/2018 filtering standard liquor made from thick juice from Tank 1 through 20 µm filter medium. Initially the operating pressure was quite low (1.9 bar) and the backwash cycles were triggered every 30 seconds so that the amount of reject was too much. In addition, the throughput was only 40 m3 per hour per filter. However, the sugar quality improved remarkably (see Section Sugar Quality). In the next few days the filters and operating parameters could be optimised to obtain the ideal operating pressure of 2.5 bars, with a throughput of 60 m3 per hour per filter and filtration cycles of 2 min 30 sec while maintaining the sugar quality.
The particle size distribution of the standard liquor, filtrate and reject during this trial is shown in Figure 7. It is clear that some of the larger particles were removed by the filters and this was enough to produce the required quality sugar.
Figure 7: PSA trace of syrup feed, filtrate and reject – juice from Tank 1 and 20µm filter medium
Trial 3 was started on 14/5/2018 filtering standard liquor made from thick juice from Tank 2 through 20 µm filter medium. A flow rate of 60 m3 per hour per filter was achieved immediately and the filtration cycles were 2 hours and 30 minutes. However, it was clear from sugar quality results that the filter media was too big to remove enough of the solids. This was anticipated as the particles in Tank 2 were much smaller than those in Tank 1. The filters were therefore stopped and the filter material changed for a smaller size.
Trial 4 was started on 17/05/2018 filtering standard liquor made from thick juice from Tank 2 through 14 µm filter medium. There was a good differential pressure over the inlet and outlet chambers of the filters with filtration cycle times around 30-40 minutes. However, the throughput achieved was on the low side at 40 m3 per hour per filter and optimisation of the filters and operating conditions could only improve this to 50 m3 per hour per filter. The operating pressures were just under 3 bars.
Trial 5 was therefore designed to improve the throughput obtained when filtering Tank 2 through 14 µm filter material by blending the thick juice from Tank 2 with that from Tank 1. The trial was started on 20/5/2018 by blending different ratios of Tanks 1 and 2 until the desired throughput of 60 m3 per hour per filter was achieved at a ratio of 1 to 5 of thick juice from Tank 1 to Tank 2. The filtration cycle lengths were about 5 minutes. Figure 8 shows the particle size distribution of the feed, filtrate and reject during Trial 5. From this trace the filtrate and feed appears similar but only because of the comparatively small amount of particles that were removed.
Figure 8: PSA trace of syrup feed, filtrate and reject – juice from Tanks 1 and 2 blended and 14 µm filter medium
The main sugar quality parameters of interest for this trial were:
- Filtering Index – an in-house measure of how well the sugar in solution performs during filtration. The index has no units but it is generally accepted that a minimum value of 25 is required and the higher the number the better the filtering quality of the sugar. Note that this method is proprietary and therefore not described in the Experimental Section.
- Solids (ppm) – the quantity of solids larger than 8 µm as determine in the laboratory using the ICUMSA Method GS2/3/9-19 (Anon, 2007a) expressed as mg/kg or ppm. The maximum allowable level (according to the bottler’s quality specification) is 7 ppm.
- Turbidity (IU) – the light absorbance due to solids at 420 nm and based on the ICUMSA Method GS2/3-18 for turbidity in white sugar (Anon, 2013).
Figure 9: Sugar quality during the syrup campaign as affected by the filtration trials
Figure 9 shows the daily average values for Filtering Index, Solids and Turbidity as well as the dates when the various trials or stops were initiated. To recap, the Trials were as follows:
- Trial 1: Tank 1 with 25 µm filtering material. No differential pressure was observed with throughput of 60 m3 per hour per filter and filter cycles were controlled manually.
- Trial 2: Tank 1 with 20 µm filtering material. Good differential pressure was achieved. After some optimisation throughput of 60 m3 per hour per filter could be achieved with filtration cycle times around 2 min 30 sec.
- Trial 3: Tank 2 with 20 µm filtering material. Throughput of 60 m3 per hour per filter was achieved with cycle times of more than 2 hours.
- Trial 4: Tank 2 with 14 µm filtering material. Throughput of 40-50 m3 per hour per filter was achieved with cycle times of 30-40 minutes.
- Trial 5: Tanks 1 and 2 blended in a 1:5 ratio with 14 µm filtering material. Good differential pressure was achieved with throughput of 60 m3/h per filter and cycle times of around 5 minutes.
In summary, Trial 1 did not appear to have an effect on the sugar quality; Trial 2 had an immediate effect: Filtering Index increased while the Solids and Turbidity reduced to acceptable levels. Trial 3 resulted in reduced Filtering Index with an increase in the Solids but Turbidity was too high. During Trial 4 the Turbidity was reduced dramatically with an increase in the Filtering Index. Trial 5 improved the Filtering Index but both the Turbidity and Solids increased slightly with solids hovering around the threshold of 7 ppm.
Precipitation of calcium salts or complexes occurs during storage of thick juice. Most prevalent is calcium oxalate. While it would be preferred to ensure very low levels of calcium to avoid the formation of oxalic acid, a factory processing an agricultural crop such as beet does not always have the luxury of having a consistent raw material quality 100% of the time. In addition an operation of the magnitude of a typical beet factory in the EU encounters mechanical, processing or other unforeseen events at least once during a campaign. The products produced during such an event, including sugar, animal feed and thick juice or other run-off syrups are either segregated or blended to improve the quality.
In the current trials filter units operating with SS filter material of between 14 and 20 µm were used to filter standard liquor to the white pans. The filters could be optimised to produce a sugar with reduced insoluble solids, turbidity and with much improved filtering characteristics.
Anon (2007a). The Determination of Insoluble Matter in White Sugar by Membrane Filtration – Official. ICUMSA Method GS2/3/9-19: 2 pp.
Anon (2007b). The Determination of Turbidity in Clarified Cane Juice, Syrups and Clarified Syrups – Accepted. Method GS7-21: 1 pp.
Anon (2011). The Determination of White Sugar Solution Colour – Official. ICUMSA Method GS2/3-10: 2 pp.
Anon (2013). The Determination of the Turbidity of White Sugar Solutions – Official. ICUMSA Method GS2/3-18: 3 pp.
Asadi M (2007). Beet Sugar Handbook. John Wiley and Sons, New York. Chapter 15: 337-344.
Abdel-Rahman E and Flöter E (2015). Reduction of Turbidity of Beet Sugar Solutions by Mechanical and Chemical Treatment. Int. J. Food Eng. 11(1): 41–49.
Bensouissi A, Rousse C, Rogé B and Mathlouthi M (2009). Effect of selected impurities on sucrose crystal growth rate and granulated sugar quality. Andrew van Hook Proceedings. 147-165.
Chairiyah N, Harijati N and Mastuti R (2016). Variation of Calcium Oxalate (CaOx) Crystals in Porang Corms (Amorphophallus muelleri Blume) at different harvest time. American Journal of Plant Sciences. 7: 306-315.
Demadis KD and Léonard I (2011). Green polymeric additives for Calcium Oxalate control in industrial waste water and process application. National Association of Corrosion Engineers International. 50(10): 40-44.
Doherty WOS and Wright PG (2004). A Solubility Model for Calcium Oxalate Formation In A Sugar Mill. Proc Aust Soc Sugar Cane Technol. 26: 8 pp.
Ishii Y (1991). Three Kinds of Calcium Oxalate Hydrates. The Chemical Society of Japan. 91: 63 – 70.
Joy KW (1964). Accumulation of oxalate in tissues of sugar-beet and the effect of Nitrogen supply. Annals of Botany. 28(4): 689-701.
Kanitkar A, Aita G and Madsen L (2013). The recovery of polymerization grade aconitic acid from sugarcane molasses. J Chem Technol & Biotechnol. 88(12): 2188-2192.
Lambrechts L, Gion, R and Simonart A (1967). Preservation of Thick Juice and its subsequent use. La Sucre Belge. 89: 139 – 149.
McGinnis RA (1982). Beet sugar technology. 3rd Ed. Beet sugar development foundation.
Norman IW, Turner JH AND Cotton RH (1950). Aconitic Acid in Sugar Beets. ASSBT vol 6 p 577.
Rogé B, Bensouissi A and Mathlouthi M (2007). Effect of calcium on white sugar turbidity. Sugar Industry / Zuckerindustrie 132(3): 170–174.
Schrevel G (2009). Beet thick juice degradation during storage. Sug Ind 134(1): 9-12.
van der Poel PW, Schiweck H and Schwartz T (1998). Sugar technology. Beet and cane sugar manufacture. Verlag Dr. Albert Martens KG, Berlin. A: 116, 129-133,150,185-186; B: Chapter 15: 919-926.
Walford SN and Walthew DC (1996). Preliminary model for oxalate formation in evaporator scale. Proc S Afr Sug Technol Ass. 70: 231-235.
Walford SN (2000). A kinetic model describing aconitic acid Isomerisation and subsequent decarboxylation to itaconic acid under factory conditions. Proc S Afr Sug Technol Ass. 74: 303-308.