The two most important objectives for a high-performance continuous vacuum pan (CVP) are good crystal quality and high exhaustions. To achieve these, the pan design needs to incorporate features that promote plug flow (a narrow crystal residence-time distribution), a high heat-transfer coefficient (HTC) and vigorous circulation. Focussing on these will also achieve an energy efficient pan that can operate on a low steam-massecuite temperature differential. Good plug flow is an essential for good crystal quality (low CV), which enables good purging with minimal washing and good exhaustions. This in turn minimises reboiling and its associated energy and sucrose losses. Low CVs also aid affination and for this reason are frequently included in raw-sugar specifications. Measures to achieve good plug flow include a good circulation profile and good inter-compartment massecuite-transfer arrangements. Achieving plug flow has presented the greatest challenge to CVP designers, but the problem is shown to have been mastered in some horizontally-configured vertical tube pans. Key requirements for high exhaustions are appropriate seed supply, good feed (supersaturation) control, vigorous circulation and a high final Brix. Various ways in which circulation can be promoted are described. Recent information is that, contrary to previous belief, longer tubes perform as well as or better than shorter tubes in CVPs. Types of pans which do or do not have the desired attributes are mentioned. It is concluded that the type of pans best suited to meeting most of these requirements are horizontally-configured vertical tube CVPs, at least one type of which is shown to come close to achieving true continuous spiral massecuite flow.
For centuries, the crystallisation stage of making sugar from cane and beet juice was performed as a batch process – whether in simple open pans or more sophisticated steam-heated vacuum pans. Despite the obvious advantages of a continuous process, it was only in the mid-1970s that the first commercially successful continuous vacuum pans (CVPs) were developed. Their introduction had been delayed by an inability to achieve good crystal quality – specifically, a uniform crystal size (low CV). Broadfoot et al. (1976) stated that “the major disadvantage of a continuous boiling system is that the crystals inherently have a spread of residence times which will result in increased CV of the product crystal”. Graham and Radford (1977) stated that “the major disadvantage associated with continuous pan boiling is the large coefficient of variation in the sugar produced”. Austmeyer (1982) commented that the main problem was “the comparatively wide crystal size distribution of the product as a consequence of the residence time pattern”. Crystal uniformity is perhaps the most important criterion for a satisfactory CVP and remains the greatest challenge for many designs. However, as will be shown, this problem has now been overcome in some CVPs.
There are also other important characteristics needed for a high performance CVP. This paper discusses the reasons for and ways to achieve three key requirements:
- Uniform crystal size
- High exhaustions
- Energy efficiency
These are not mutually independent; for example, an even crystal size permits effective centrifugation with minimum washing and hence good exhaustions, minimises reboiling and saves energy.
Crystal quality – uniform crystal size
The first CVPs in commercial operation at cane factories appear to have been produced by Langreney (Mauritius) and FCB (Réunion). Both were visited in 1975 and were producing massecuites with wide spreads of crystal size, confirming the predictions of Broadfoot and Allen (1977). For this reason, they were confined mainly to C pans from which the crystals were either remelted or ‘cleaned up’ (fines dissolved out) in batch pans if to be used for magma. Commenting on the decision to install South Africa’s first CVP, Tongaat’s FCB pan at Maidstone, Graham and Radford (1977) stated that because of the expected poor CV, it would be used for C sugar that was to be remelted.
The main reasons for the poor CVs were poor supersaturation control and flow patterns that included both short circuits and hold-up zones. These shortcomings were addressed in a new continuous pan at Maidstone in 1983 (Kruger 1983). This Tongaat-Hulett design, subsequently marketed by Fletcher Smith, featured a cross-section profile designed specifically for smooth plug flow and 12 compartments. Brix was controlled in each compartment by individual conductivity probes, subsequently replaced by RF probes (Rein 1986). Although intended to be a B pan, it was tested as an A pan and produced sugar crystals of better quality than that factory’s batch pans, with sugar of 0.66 specific grain size (SGS) (SASTA 2009), fines (through 28 mesh Tyler screen) < 25% and lower colour than that from the batch pans (Kruger 1983).
Why is a low CV important?
The importance of a good (low) CV has been emphasised by authors such as Broadfoot (1992), Journet (1994), Thelwall (2000) and Rein (2017). The reasons include better centrifuging (free-draining because the apertures between crystals are open, not blocked by smaller crystals – see Figure 1). Less washing is needed and there are no fines to be dissolved or pass through the screens. This improves exhaustion, reduces losses and reduces the energy requirement of reboiling. Affination is easier, which is why CV is often included in export specifications.
Figure 1. Drainage passages of uniform (left) and varied crystal sizes (right).
How can a low CV be achieved?
In a massecuite boiling at concentrations within the metastable range (i.e. between saturation and the maximum supersaturation before spontaneous crystal formation – Figure 7), all crystals grow at approximately the same linear rate. It follows that for uniform final crystal size out of a CVP, the input seed must have a uniform crystal size and thereafter all crystals must remain in the pan for the same length of time, i.e. there must be plug flow through the CVP. For horizontally configured CVPs, in which the massecuite progresses through a series of compartments along the pan, this means that there needs to be plug flow both laterally (within each compartment) and longitudinally (between successive compartments).
Computational fluid dynamics (CFD) simulations of various commercial CVP profiles undertaken at the Audubon Sugar Institute by Echeverri et al. (2007) confirmed that the Tongaat-Hulett / Fletcher Smith profile is close to ideal for lateral plug flow within a pan compartment (Figure 2). This predicted no significant stagnant areas, with well-balanced flow velocities across each section.
Figure 2.CFD simulation of the cross-section of the Fletcher Smith CVP.
However, this lateral or transverse flow is only one component of the overall flow through the pan. The longitudinal flow – along each compartment and between compartments – is equally important. Plug flow (equal crystal residence times) requires that there be no back-flow, hold-ups or short circuiting within or between compartments.
The widely adopted means to measure the degree of plug flow is by tracer testing, using the procedures described in sugar literature initially by Wright and Broadfoot (1977) and subsequently by Rein et al. (1985). A measured dose of tracer – usually a lithium salt – is injected into the feed at the start of the pan and its emergence tracked in a series of samples taken from the discharged massecuite. The recovery of the tracer can then be compared against a theoretical mathematical model of the flow predicted through a number of well-mixed tanks in series (t‑i‑s).
The greater the number of t‑i‑s, the closer the approach to perfect plug flow. This in turn means the better the CV that can be expected. Rein and Msimanga (1999), Thelwall (2000) and Moor (2016) have all published similar relationships between CV and t‑i‑s. That of Moor is reproduced in Figure 3. The CVs of several sugar samples from pans of known t‑i‑s values have correlated well with this curve (Moor 2016).
Figure 3. Expected CV versus number of tanks-in-series.
An example of this technique for measuring crystal residence time distribution was given by Moor (2007), who described the results from an 8-compartment Bosch C continuous pan, which tested as equivalent to 18 t-i-s (Figure 4).
Figure 4. Actual tracer versus 18 tanks-in-series model.
No tracer emerged till 4 hours after injecting the tracer, indicating no short circuiting, and the low tail values indicated minimal hold-up areas. This pattern was then compared against models for various t-i-s and found to match most closely that for 18 t-i-s. It is noteworthy that the actual performance of 18 t-i-s was considerably better than would have been achieved had each of the eight compartments behaved as a mixed tank. This confirms that the massecuite flows through the pan in a circulating spiral path with a large degree of plug flow within each compartment. This phenomenon has also been seen in results reported from other horizontally-configured vertical tube CVPs, but not from horizontally configured pans with horizontal steam tube calandrias. This can be seen from Table 1, which comprises published data.
The above results can be taken as reasonably typical for the types of pans concerned. It is interesting to consider possible reasons for good or less good performance.
The best t-i-s/ actual cells ratio in Table 1 was that from the Bosch pan at NAT&L in Vietnam. The circulation profile in this pan is very similar to the FS ‘best’ profile in Figure 2 and CFD simulation undertaken by du Plessis (2015) confirmed a good lateral (transverse) flow path with minimal stagnant areas and largely uniform velocities (see Figure 5). This pan therefore meets the requirements for a good transverse flow pattern.
Figure 5. Velocity contour plots with streamlines for Bosch CVP.
Where the pan differs most from the others tested is in its longitudinal flow. Although only an 8-compartment pan, this pan incorporates transverse partial baffles above and beneath the calandria. These baffles prevent forward or back mixing in these turbulent zones, thus inducing flow patterns equivalent to more than the actual eight compartments. Another difference from most of the others is in the massecuite transfer between compartments. This takes place through ports in the relatively quiescent downcomers, where the flow is laminar and no back-flow is likely. This was confirmed when the longitudinal flow was incorporated into the 3-dimensional CFD simulations (Figure 6). This showed clearly the overall spiral flow, with laminar flow through the transfer ports and the aligning effect of the partial baffles above and below the calandria. Virtually all the longitudinal flow along the pan occurs in the quiescent downtake area.
Note that with good spiral flow within compartments, the compartments do not perform as ‘mixed tanks’ and there is no plug flow benefit from sizing the compartments for equal residence times. Increasing compartment sizes along the pan can provide feed adjustments at regular massecuite time intervals, but the amount of feed per compartment increases significantly along the pan.
Figure 6. CFD streamlines for Bosch CVP, showing transfer and effects of partial baffles.
The pan with the second highest t-i-s/actual cell number ratio in Table 1 is the SRI-designed Mossman pan. It is noteworthy that, like the Bosch pan, this pan incorporates several internal baffles across the top of the calandria, but none below.
The third highest t-i-s/actual cell number ratio in Table 1 is the T-H Maidstone pan, whether operating as a C or an A pan. These results are significantly better than for the C and A T-H pans at Felixton. All have similar profiles, except that the Felixton pans’ tubes are shorter, the downtakes wider and massecuite transfer between compartments is in the turbulent zone above the calandria, whereas transfer in the better t-i-s Maidstone pan is through ports similar to those in the Bosch pan.
The SRI pans all exhibit t-i-s / actual cells ratios well above unity.
The exceptional number of compartments (36) in the Racecourse Mill-designed pan resulted in a high t-i-s but relatively low t-i-s /actual cells ratio. The large number of cells had been selected specifically to achieve good plug flow (McDougal and Wallace 1982), but in reviewing subsequent changes, Attard (1993) commented that the removal of cross baffles had reduced lump formation and only marginally detracted from the residence time distribution. It was felt that the extreme number of compartments was not warranted by the relatively small CV benefits (see Figure 3).
The only pans where the t-i-s is less than the number of compartments are the FCB pans with steam flow in horizontal tubes. In vertical tube pans, the massecuite is confined by the tube to a narrow up-flow path during the turbulent boiling zone, but this is not so in the horizontal tube pans and considerable to-and-fro mixing may occur here. The inter-compartment transfer arrangements on these pans are also not ideal.
Only data from horizontally configured CVPs is shown in Table 1. No t-i-s data could be found for stacked stirred pans of the BMA VKT type, but they are true ‘mixed tanks in series’ and Austmeyer (1986) confirmed that their crystal residence time behaviour “gives a good approximation” to the mathematical model for that number of ideal stirred vessels. Journet (1994) pointed out that this results in high CVs. This is a minor issue for white final sugar, whether loose or lumps, but a serious concern in cane sugar factories. To mitigate this problem, Austmeyer (1986) advocated using seed of a large crystal size and Hempelmann (1996) stated that the inferior CVs was less of a problem when a seed magma percentage of 30% or more was used. Unfortunately, these solutions require more boiling in the batch pans, limiting the benefits from continuous boiling.
A high overall exhaustion requires first that a high proportion of the sucrose in the mother liquor is deposited onto crystals and thereafter that the crystals can be recovered in centrifuging with minimal losses of fines or crystal dissolution during washing. What are the main requirements from the CVP for this?
- There must be an appropriate quality and quantity of seed. The seed crystals should be of a uniform size (lowest practically achievable CV). There must be a sufficient population of crystals to provide adequate crystallisation surfaces throughout the pan, but too many crystals sharing the growth will result in a small final crystal size. It is recommended that the seed supply rate be maintained at a fixed ratio to the total syrup or molasses feed rate, so that when the evaporation rate changes or water is boiled, the seed supply is adjusted accordingly.
- Brix control. The syrup, molasses or water feed throughout the CVP needs to be controlled to give brix levels such that the sucrose concentration in the mother liquor is within the metastable range (Figure 7), preferably near to the top of the range for rapid crystal growth. It is important that good quality instruments are used for this key function. Suitable devices are microwave transmitters (which measure water content directly but are expensive), RF probes (for high or low purity boilings) or conductivity probes (lower purities only). For accurate control, the Brix should be controlled at eight or more points along the pan.
Figure 7. Typical ranges of sucrose supersaturation in mother liquor for pans & crystallisers.
- Vigorous circulation. Broadfoot (2005) observed that “exhaustion performance is also closely related to the circulation characteristics of the pan”. To ensure crystal surfaces are presented to all the sucrose in solution, turbulence is required. However, the turbulence should not detract from the key objective of plug flow. The ideal is therefore to generate vigorous circulation within narrow channels. The first requirement for good natural circulation is a good flow profile, with an adequate downtake area (usually larger than the cross-section area of the tubes). Mechanical stirring provides good turbulence but is incompatible with plug flow. However, Attard and Doyle (1998) described a ‘hybrid solution’ for this. A stirred heavy-up module was added at the end of a conventional long flow path CVP at Racecourse Mill, to provide good turbulence in the final high Brix, high viscosity stage of the boiling.
Well-distributed jigger steam (or gas) is an effective circulation aid, especially in the final cells. SRI have developed a jigger feed system in which the incondensable gases/steam is injected through multiple laser-drilled perforations that are too small to permit back-flow of massecuite and provide an ideal distribution of fine bubbles (Rackemann and Broadfoot 2007). As an alternative, Bosch inject their jigger steam through multiple nozzles (usually 12 per compartment) evenly distributed beneath the full area of the calandria. They also use under-base heating to stimulate circulation.
Some technologists advocated short tubes for good circulation in CVPs, but Moor et al. (2018) produced evidence that longer tubes perform as well as or better than shorter tubes. The greater buoyancy from more vapour bubbles per longer tube apparently more than counters the additional friction drag.
- High brix. A high dry substance content in the mother liquor ‘forces’ sucrose out of solution and high brixes are therefore a powerful aid to high exhaustions. In this regard, the low boiling heads in CVPs provide an advantage over batch pans. It is good practice to raise the brix in the final stages of a CVP to the maximum that can be managed by the pan, crystallisers and centrifugals. The pan must therefore be designed for this, which will often require assisted circulation, e.g. by jiggers.
- Low temperature. As is clear from Figure 7, the cooler the liquid the less sucrose that can remain in solution. It is therefore desirable to boil at a low pressure, but a compromise needs to be found because viscosities increase rapidly with reducing temperature and this impedes the crystallisation rate. Vacuum systems should be provided to allow pans to be boiled at low pressures.
- Residence time. Time is needed for sucrose to deposit onto crystals, but exhaustion is more rapid in well agitated pans than in slowly stirred crystallisers. It is therefore usually cost-effective to optimise pan exhaustion by sizing CVPs generously. However, excessive retention at high temperatures in pans and crystallisers could initiate Maillard reaction, especially in C boilings. To minimise this risk, low C pan boiling pressures (12-13 kPa abs) are recommended.
Continuous pan boiling is inherently far more energy efficient than batch pan boiling. Obvious reasons are:
- Elimination of steam-outs after each batch strike, with associated wash liquid to evaporate;
- No energy to re-establish vacuum;
- Steady boiler, evaporator and vacuum system loadings;
- Low boiling heads which permit use of lower grade vapour for boiling.
These benefits apply to all continuous pans, but not all CVPs are equally energy efficient. The features and practices for good crystal quality and high exhaustions also promote energy efficiency:
- Uniform crystals → good purging → minimised reboiling → energy savings;
- High exhaustions → minimised reboiling → energy savings;
- Vigorous circulation → high HTCs → low steam-massecuite approach temperatures → allow lower grade steam → energy savings.
The lowest approach temperatures (ΔTs) can be achieved by forced circulation such as in BMA’s stirred VKTs, which are able to operate at ΔTs down to 15°C. To optimise performance in the early low brix and final high brix stages, Austmeyer (1986) has described the use of different types and speeds of stirrers for the first two and final two chambers of a VKT. However, the stirrer power requirements detract from energy efficiency and high CVs are detrimental.
Rein (2017) states that (horizontal) “continuous pans usually operate with ΔT values of 25 to 40°C”. However, Broadfoot (2018) pointed out that “horizontal CVPs have also incorporated various changes to improve operation on low pressure vapours including:
- The use of the SRI jigger tube system (Rackemann and Broadfoot 2007),
- Mechanical agitation in the heavy up cells (Attard and Doyle 1998) and
- Use of a heated pan bottom (Moor 2007).”
The first two of these have been described earlier. The third refers to the (patented) innovation in Bosch CVPs of a base heating surface below the calandria. This stimulates circulation and enables the ‘A’ CVP at F.U.E.L. in Mauritius to boil vigorously on 85 kPa abs V3 without the use of jigger steam (Moor 2007). In October 2010 tests, this pan boiled well at 14.2 kPa abs with a pressure in the calandria of 60 kPa abs, giving a ΔT of < 20oC. For this, jigger steam was used on the final two of its 12 compartments.
Where energy economy is a major consideration, Moor (2002) showed that in a boiling pan, the calandria pressure will always be sufficient for incondensable gases to be used to provide ‘free’ jigger steam.
Features and practices to provide the key CVP properties of uniform crystal size and high exhaustions have been discussed. These also contribute to low energy requirements. We conclude that the most important features for a good CVP are:
- Seed supply having low CV
- Good plug flow, both laterally and longitudinally along the pan
- Vigorous circulation
- Good supersaturation control throughout the pan
- A high final brix.
Pans differ significantly in their ability to provide these attributes. Achieving plug flow has presented the greatest challenge, but the problem is shown to have been mastered in some horizontal vertical tube pans in which continuous spiral flow is effectively achieved.
Thanks are extended to the SMRI in Durban for access to their library and to their librarian, Sam Wanda, for copies of several of the articles referenced in this paper.
*Paper presented at the XXXth Congress of the International Society of Sugar Cane Technologists, Tucuman, 2-5 September 2019 and published here with the agreement of the Society.
Arcidiacono G, Pike D, Scanlan J, Mclean RJB. 1992. The continuous B massecuite pan at Tully Mill. Proceedings of the Australian Society of Sugar Cane Technologists 14: 276–286.
Attard RG. 1993. Modifications to the Racecourse continuous pan. Proceedings of the Australian Society of Sugar Cane Technologists 15: 180–185.
Attard RG, Doyle CD. 1998. Additional heavy-up module to the Racecourse continuous low grade pan. Proceedings of the Australian Society of Sugar Cane Technologists 20: 357–361.
Austmeyer KE. 1982. Die kontinuierliche Kristallisationder Saccharose aus hentiger Sicht. Zuckerindustrie 107: 401–404.
Austmeyer KE. 1986. Analysis of sugar boiling and its technical consequences Part III – Continuous processes. International Sugar Journal 88(1047): 50–55.
Broadfoot R. 1992. Designing continuous pans for narrow crystal size distributions and improved cost performance. Proceedings of the Australian Society of Sugar Cane Technologists 14: 266–275.
Broadfoot R. 2005. Design and operating criteria for maximising the benefit of continuous vacuum pans. Proceedings of the International Society of Sugar Cane Technologists 25: 31–40.
Broadfoot R, Allen JR. 1977.Continuous low grade massecuite boiling studies. Proceedings of the International Society of Sugar Cane Technologists 16: 2667–2678.
Broadfoot R, Miller KF, Davies LW 1989. Commissioning trials on the SRI continuous high grade pan at Maryborough factory. Proceedings of the Australian Society of Sugar Cane Technologists 11:152–161.
Broadfoot R, Rackemann D, Moller D. 2018. Why the emerging strong interest in vertical continuous pans? Proceedings of the Australian Society of Sugar Cane Technologists 40: 512–525.
Broadfoot R, Wright PG. 1978. The continuous low grade pan at Mossman. Proceedings of the Queensland Society of Sugar Cane Technologists 45:171–177.
Broadfoot R, Wright PG, Miller KF, Steindl RJ. 1976. Progress in continuous boiling of low grade massecuites. Proceedings of the Queensland Society of Sugar Cane Technologists 43:171–177.
Du Plessis K. 2015. CFD Simulation of crystallisation vacuum pan. Report commissioned by Bosch Projects from ESTEQ Engineering. 26 pp.
Echeverri LF, Rein PW, Acharya S. 2007. Measurements and CFD simulation of the flow in vacuum pans. Proceedings of the International Society of Sugar Cane Technologists 26: 1341–1353.
Graham WS, Radford DJ. 1977. A preliminary report on a continuous C pan. Proceedings of the South African Sugar Technologists’ Association 51: 107–111.
Hempelmann R. 1996. The future in continuous pan boiling. International Sugar Journal 98: 231–233.
Journet G. 1994. Advantages of the FCB continuous pan. International Sugar Journal 96: 500–503.
Kruger GPN. 1983. Continuous ‘A’ pan boiling trial at Maidstone. Proceedings of the South African Sugar Technologists’ Association 57: 46–51.
McDougal EE, Wallace GA. 1982. The Racecourse continuous vacuum pan. Proceedings of the Australian Society of Sugar Cane Technologists 4: 383–388.
Moor BStC. 2002. Energy aspects of assisted pan circulation. Unpublished presentation at ISSCT Energy Management Worksop, Berlin, October 2002.
Moor BStC. 2007. Successful innovation: results from the Bosch continuous pan. Proceedings of the International Society of Sugar Cane Technologists 26: 1669–1675
Moor BStC. 2016. Modern sugar factory equipment for good recoveries, energy efficiency and low costs. Proceedings of the International Society of Sugar Cane Technologists 29: 234–246.
Moor BStC, Raghunundan A, Ramaru R. 2018. The long and short of CVP tubes. Proceedings of the South African Sugar Technologists’ Association 91: 224–238.
Rackemann DW, Broadfoot R. 2007. A new design of jigger system to improve vacuum pan performance. Proceedings of the International Society of Sugar Cane Technologists 26: 1564–1572.
Rein PW. 1986. A review of experience with continuous vacuum pans in Tongaat-Hulett Sugar. Proceedings of the South African Sugar Technologists’ Association 60: 76–83.
Rein PW, Cox MGS, Love DJ. 1985. Analysis of crystal residence time distribution and size distribution in continuous boiling vacuum pans. Proceedings of the South African Sugar Technologists’ Association 59: 58–67.
Rein PW, Msimanga MP. 1999. A review of continuous pan development in the southern African sugar industry. Proceedings of the International Society of Sugar Cane Technologists 23: 124–136.
Rein PW. 2017. Cane Sugar Engineering. 2nd Edition. Bartens, Berlin.
SASTA. 2009. SASTA laboratory Manual, 5th edition. South African Sugar Technologists’ Association, Mt Edgecombe.
Thelwall JCdeC. 2000. Features of continuous vacuum pan design. International Sugar Journal 102: 630–637.
Wright PG, Broadfoot R. 1977. The application of a lithium tracer method to residence time studies in a sugar factory. Proceedings of the International Society of Sugar Cane Technologists 16: 2569–2580.