You are on page 1of 11

Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

Tongaat Hulett, Private Bag 3, Glenashley, 4022, South Africa

Abstract A new design continuous vacuum pan has been developed by Tongaat Hulett, and has been installed at their Xinavane factory in Moambique. The pan is unlike any previous continuous pan design in that it is of a circular (round) construction with two internal floating calandrias stacked one above the other in a common vapour space. Although the calandrias share this common vapour space, each can be boiled independently of the other. This allows for on the run boiling-out and cleaning of the pan, thus alleviating an inherent problem of high grade (A-massecuite) continuous pans which require regular shut down and boiling-out to remove encrustation above the boiling surface and/or in the tubes. The design of the pan is such that it is possible to convert an existing batch pan into a continuous vessel or to install a continuous pan on the footprint of an existing batch pan. This paper describes the design of the pan and lists its advantages and reasons for the development of the design.
Keywords: factory process, continuous pan, circular pan, crystallisation, pan boiling, vacuum pan

Introduction The first continuous pan installed in South Africa was at Tongaat in 1977 (Graham and Radford, 1977). This was an FCB design with horizontal tubes, and was installed on Cmassecuite duty. The first Tongaat Hulett designed continuous pan was installed at Maidstone in 1982 (Kruger, 1983). This was originally a B-pan, later converted to A duty. Based on the success of the design, two A-, two B- and two C-pans were installed at Felixton. This factory was the first sugar mill with continuous pan boiling on all three massecuite grades. The original Tongaat Hulett continuous pan has been very successful, and over 170 units have been installed throughout the world1. The inverse cardioidal design of the pan body (i.e. an upside-down heart shape) has been reproduced on many competing pan designs. This design can be seen in Figure 1. Although the traditional horizontal pan has been very successful, the design does have a few drawbacks. These are discussed below. Large footprint required A typical 130 m3 continuous pan will require approximately 68 m2 of floor space. The length of the pan is typically 16.2 m and the width 4.2 m. This excludes space around the pan for

This continuous pan design is referred to as the horizontal continuous pan, so as to differentiate it from the new circular continuous pan design.


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

walkways. In green-field installations this is not usually a problem, but space constraints can emerge when existing pan-floors are expanded.

Vapour Dome

Calandria Massecuite Discharge


Figure 1. The inverse cardioidal shape of the TH pan.

Cannot be boiled-out whilst still in production A-massecuite continuous pans typically require draining and boiling-out every 10 to 14 days. This is due to encrustation that takes place in the pan, both above the massecuite surface as well as in the tubes. The boil-out procedure generally requires that the pan be isolated for a period of three to five hours (Montocchio, 1988). This can result in production interruptions, depending on the factorys crush strategy and stop-day schedules. Cannot be installed in an existing batch pan position A new continuous pan cannot generally be installed in the area where a batch pan has been removed. This need might arise if the capacity of a pan-floor is being increased or the energy utilisation of the factory optimised. Energy cost savings can be realised if batch pan boiling operations are replaced by continuous pan operations making use of a lower grade of vapour. In the event of a batch pan being replaced with a continuous pan, whether for energy reasons, pan volume capacity constraints or simply replacing a pan at the end of its life, substantial cost savings can be achieved if the new continuous pan can be located in the original batch pan position; no new steel work may be required, as well as fewer pipe work modifications. Background to the design The possible debottlenecking of the Amatikulu factory was investigated in 1999, and one of the constraints identified was the pan-floor. It was realised that the replacement of one of the 85 m3 batch A-pans with a 120 m3 continuous pan would relieve the constraint. Note that the replacement of an 85 m3 batch pan with a 120 m3 continuous pan results in an effective batch pan volume increase of 77 m3, because continuous A-pans are 35% more efficient than the


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

equivalent batch pan. This increase in efficiency has been empirically determined by Tongaat Hulett, and is due to a combination of the following factors: Batch pans are unproductive in terms of crystallisation for substantial periods (primarily between each strike) due to striking, cutting and steaming out. The nominal massecuite volume of a batch pan is only utilised at the end of each strike. For most of the time the pan is only partially full and thus the full, nominal pan volume is not continuously being used for crystallisation. Higher crystallisation rates can normally be achieved in continuous pans because they achieve better circulation and can be more easily controlled at optimum conditions. Direct replacement of a batch pan with a continuous pan would not have been easily achievable with a conventional continuous pan design. The possibilities of a vertical continuous pan were then debated, and the beginnings of the current design took place. In discussions regarding the possibility of a vertical pan, the following issues were considered: The conversion of an 85 m3 batch pan to a continuous version was carried out at Amatikulu in 1983 as part of A-continuous pan boiling trials. This conversion took the form of installing vertical baffles to create six pie segments. The baffles extended from the pan wall, across the top of the calandria, through the downtake and along the bottom of the calandria back to the pan wall, forming seals between the bottom of the calandria and the saucer of the pan. Although high crystal deposition rates were obtained with this pan, heavy encrustation took place in the tubes and pan circulation was poor. The pan circulation was improved through the installation of steam jiggers, but it was subsequently realised that the poor circulation was due to a restriction in flow down the downtake resulting from the installation of the six radial baffles. Although the baffles had no effect on the flow area of the downtake, they reduced the effective hydraulic diameter2 available to the massecuite. The low hydraulic diameter is indicative of the excessive friction in the downtake, inhibiting the movement of massecuite and reducing circulation. It was apparent that the central downtake concept with a vertically baffled pan was not ideal. The specific example of the batch pan, converted to a continuous pan at Amatikulu described above, highlights the general difficulty of designing efficient pans. Although vacuum pans of various designs have been constructed for well over 100 years, the attempts to apply quantitative scientific and engineering techniques to guide pan design have had limited success. The incorporation of unsubstantiated assumptions and theories to compensate for the lack of knowledge or data appears to have been a major failing. Early work in pan design was

The hydraulic diameter is defined as four times the cross-sectional area for flow, divided by the wetted perimeter. This concept is useful when considering flow in non circular tubes or channels, because it allows many quantities (for example the Reynolds number) to be calculated in the same way as for a round tube. For circular geometry, the hydraulic diameter is equal to the tube diameter. The hydraulic diameter is an appropriate parameter to consider when evaluating friction during flow through a pan downtake of unusual shape.


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

misled by Webre (1932) who asserted that no bubble formation took place within the pan tubes. This was questioned by Allan (1962) and subsequently disproved by the test work of Skyring and Beale (1967) and was confirmed locally by Rouillard (1986). More recently advances in the techniques of computational fluid dynamics have been applied to modelling circulation in vacuum pans (Echeverri and Rein, 2007; Rackemann et al., 2006). This has provided some extra insight into how pan design affects pan circulation, but the predictions have not yet been fully validated. Pan designs incorporating floating calandrias3 have been particularly contentious, with a local paper on the appropriate positioning of a pan downtake (Hulett, 1965) being motivated by the great controversy existing last year with the re-introduction of the floating calandria pan to our industry. The paper used simple friction loss calculations to support the circular downtake of a conventional calandria against the annular downtake of a floating calandria, but the recorded discussions of this paper highlight the influence of unsubstantiated opinions on support for particular pan designs (e.g. A floating calandria pan does not have dead zones; It was far better to allow the circulation to take its natural course out to the circumference and thus have the downtake on the circumference of pan). In Australia, floating calandria pan designs have traditionally not been very successful (Wright, 1966; 1974). Wright (1974) states, The early enthusiasm for the floating calandria design waned, following some adverse boiling results in certain conditions. One of the reasons for these concerns is the difficulty of installing mechanical circulators. Miller (2001) also reported on the limited support for floating calandria designs in Australia and the fact that a number of these designs had been either relegated to less onerous boiling duties or even converted to conventional central downtake pans. Given the lack of proven scientific design calculations, pan design remains largely empirical, and practical experience with past designs provides the most important guidance for the design of new pans. The poor experiences with floating calandria pans means that any new pan design incorporating a floating calandria would require very careful consideration. In the original Tongaat Hulett continuous pan design the calandria runs centrally down the length of the pan with the downtake at the outside of the pan (refer to Figure 1). It was realised that, in this respect, the traditional horizontal continuous pan be considered to be, in effect, a floating calandria pan. Due to the poor experiences previously reported with floating calandria pans, it was accepted that care would have to be taken in designing the downtake in a circular pan of this type, despite the success of the traditional horizontal continuous pan. As discussed above, the initial concept was to replace an existing batch pan with a continuous pan, so the existing footprint of an 85 m3 batch pan was used as the basis for sizing the pan. A segmented batch pan with a floating calandria and peripheral downtake was considered. In order to obtain the increased massecuite volume on an existing batch pan footprint, it was realised that a second calandria would have to be added above the first calandria. The addition

The floating calandria term refers to a vessel in which the calandria is centrally situated in the pan, with the downtake at the periphery of the vessel. This is opposed to the fixed calandria design which incorporates a central downtake with the vertical tubes at the periphery of the vessel.


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

of a second calandria also has the advantage of increasing the size of the individual compartments, given that a fixed number of compartments are necessary to achieve the required pan performance. In order to produce crystals of a uniform size during pan boiling, it is necessary for all crystals to stay in the pan for the same length of time. In continuous pan boiling, this is achieved by ensuring that the flow of massecuite through the pan approaches true plug flow. Rein et al. (1985) modelled continuous pan operations using a tanks in series residence time distribution model and showed that increasing the number of pan compartments leads to a closer approximation to plug flow. Around 10 to 12 compartments are generally necessary to produce an acceptably uniform crystal size in the resulting sugar. However, after carrying out conceptual work, the debottlenecking at Amatikulu never took place, and the new design, although patented, was shelved. The Xinavane expansion The Xinavane factory in Moambique has undergone a number of debottlenecking exercises and expansions since being purchased by Tongaat Hulett in the late 1990s. In 2004 the factory was expanded to 150 tch (Phase-1 expansion), and the design was carried out to facilitate an envisaged Phase-2 expansion to 250 tch. During Phase-1, a 75 m3 horizontal continuous A-pan was installed, as well as a 42 m3 horizontal continuous C-pan. The plan for Phase-2 was to install a new 100 m3 horizontal continuous A-pan, convert the existing A-pan to B- duty, and extend the 42 m3 C-pan to 63 m3. Only eight compartments of the C-pan were originally installed, and the intention was to add a further four compartments to form a traditional 12 compartment pan. The new A and C-continuous pans for Phase-1 were installed on a new pan-floor, and the layout was designed to allow space for the future 100 m3 horizontal continuous pan, located between these two continuous pans. However, the 250 tch expansion did not take place, and instead the factory has been expanded to 380 tch. Unfortunately this resulted in the planned pan-floor layouts being incompatible with actual requirements; a further two continuous pans were required, and not one as originally planned. Extending the pan-floor to accommodate two continuous pans was possible, but would have incurred substantial extra costs. It was realised that if vertical continuous pans rather than horizontal continuous pans were installed, these two new pans could be installed on the same footprint that had been allocated for the single continuous pan of the existing design. A decision was made to follow this route, and the designs for the new pans were completed. Two of the new design circular continuous pans have now been installed at the Xinavane factory; a 120 m3 A-pan and a 100 m3 B-pan. Both of these have been located in the area of the pan floor that was originally allocated for the installation of a single 100 m3 horizontal continuous pan. It has been decided that the term circular pan rather than vertical pan - should be applied to the new design to avoid confusion with the BMA VKT design (Hempelmann, 2006).


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

General description of the pan The pan consists of a cylindrical vessel with two stacked floating calandrias in a common vapour space. Each of the calandrias has six segments or compartments, with seed entering at the top calandria in the first compartment, passing through the remaining five compartments, and then overflowing a weir to the first compartment of the bottom calandria; in actual terms the seventh compartment. The massecuite then passes through compartments seven to twelve, before overflowing the outlet weir in the twelfth compartment into a discharge leg. The calandrias are of the floating type and the down-takes are at the periphery of the vessel rather than in the centre. Each of the calandrias can be bypassed, either by allowing the massecuite from the sixth compartment to flow directly to the strike receiver, or by pumping seed directly into the seventh compartment. This latter arrangement would result in the top calandria being bypassed. Both calandrias share a common vapour space, requiring a single condenser and vacuum control system. Vapour generated by the bottom calandria is vented to the vapour space of the top section of the pan via six semi-circular ducts at the periphery of the pan. This results in the top and bottom sections of the pan having a common vapour space. Steam enters each calandria through a central steam pipe running vertically through the centre of the pan. Each calandria has its own separate steam supply and control valve. This facility allows different vapours to be used in each of the calandrias if so desired. This also allows one half of the pan to be completely stopped if necessary. It is also conceivable that two different grades of massecuite could be boiled in the same vessel; for example, B-massecuite in the top calandria and C-massecuite in the bottom calandria, each with its own calandria steam pressure and condensate flow control loops. The absolute pressure would be common to both top and bottom units. Massecuite passes from one compartment to the other in a similar manner to the Tongaat Hulett horizontal pan; a small exit opening is provided in the inter-compartment baffles to allow the massecuite to pass through. A slight difference from the horizontal pan is that the crossover point is at the edge of the calandria near the downtake rather than at the furthest point from the downtake. The basic principles of the pan design can be seen in Figure 2. Further pan enhancements A number of additional enhancements have been incorporated into the pan design that are not currently features of the horizontal continuous pan. The cold wall The vertical calandria wall in the downtake of a conventional pan is at the same temperature as any of the tubes. Heating of the massecuite at this surface will occur, with concomitant evaporation. The natural convection flow direction at this point will be upwards (as in the


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

tubes). This upward flow directly opposes the desired direction of massecuite flow down the down-take. Ideally, the wall of the calandria should be at a reduced temperature to avoid this effect. A relatively cold calandria wall that cannot initiate evaporation will not retard downwards convection flow of massecuite in the downtake, and thus not hinder the normal massecuite circulation pattern.

Massecuite Transfer Ports Between Compartments

Vapour Dome

Sight Glasses

Upper Calandria

Massecuite Transfer From Top To Bottom Calandria

Compartment Baffles

Vapour Transfer Ducts Lower Calandria

Massecuite Discharge

Steam Inlets

Figure 2. Basic principles of the pan design.

In the circular pan, a double wall in the calandria at the downtake has been provided, creating a narrow annular channel running around the outside face of the calandria. This channel has small holes drilled in it at the top and bottom connecting it to the calandria. The channel is also connected to the vapour space of the pan. The holes drilled into the calandria have been sized to remove the correct volume of incondensable gas. Because of the pressure drop across these holes, the pressure in the channel will approach that of the relatively low pressure vapour space of the pan. This compartment therefore serves the purpose of extracting heavy and light incondensable gases from the calandria, and at the same time due to the channel being under reduced pressure provides a cold wall that is presented to the massecuite, thereby promoting the correct massecuite circulation pattern. Circulation ratio optimisation An important aspect of pan design is the pan circulation ratio. This is defined as the ratio of the area of upward flow of massecuite to that for the downward flow through the centre well (Smith, 1936). Smith specified a ratio of less than three to ensure adequate massecuite circulation, and modern pans generally have a ratio of less than two (Jenkins, 1958).


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

The circulation ratio of the Amatikulu 85 m3 batch pans is 2.6 whilst that of the 130 m3 Tongaat Hulett horizontal continuous A-pan installed at Komati is 1.1. According to Smiths (1936) definition (which was originally developed for batch pan technology), the circulation ratio refers to the relative cross sectional areas at the top of the tubeplate. In designing the shape of the downtake and the saucer area under the tubeplate in the continuous pan, this concept was taken a step further, as one of the concerns is the massecuite flow into the wedge of the compartment created by the division plates underneath the bottom tubeplate. In a conventional continuous pan, the rectangular shape of each compartment ensures that the relative open area of the tubes to the downtake remains constant at all points progressing along the calandria (i.e. between division plates). In the circular pan, each compartment has a wedge shape, and there was some concern regarding the circulation of massecuite up the tubes at the apex of the wedge where the massecuite flow is at its narrowest. The concern about the wedge shape can be seen in Figure 3.

Figure 3. Section through a calandria, showing the wedge shape compartments.

The circulation ratio concept was therefore carried through to the area under the calandria. In order to optimise circulation and ensure that all the tubes receive an adequate massecuite flow, the circulation ratio was calculated for the feed path into each individual row of tubes beneath the bottom tubeplate. In other words, a cross section of the massecuite flow area


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

under the calandria was ratioed to the cross sectional area of tubes available to the massecuite at each row of tubes. The shape of the saucer was altered to ensure that this ratio remained approximately constant all the way to the final row of tubes in the apex of the wedge. In this manner, circulation of massecuite under the calandria should not be impeded as the massecuite progresses deeper into the wedge. This concept can be seen in Figure 4, where the circulation ratio is calculated across the width of the calandria, both for the new circular continuous pan design and for two existing horizontal continuous pans built by Tongaat Hulett. The circulation ratio is well below the target value of 2 at all points in the massecuite flow path underneath the calandria.





Circulation Ratio [-]





New Design Typical TH Continuous Pan 1 Typical TH Continuous Pan 2

0 0 10 20 30 40 50 60 70 80 Relative Distance Across Calandria [%]



Figure 4. Circulation ratio under the calandria.

Vacuum system Traditionally, the reduced pressure in a vacuum pan is achieved by the combination of a counter-current direct contact condenser (to condense the water vapour) followed by a vacuum device (to remove incondensable gases). The most common vacuum device is a liquid ring vacuum pump, but may be a steam ejector or even a water eductor. The absolute pressure in the pan is normally controlled by varying the flow of cooling water entering the pan condenser, with an increased water flow leading to a reduced absolute pressure. To ensure that excessive quantities of water are not used in this way, a tailpipe temperature over-ride is sometimes employed, which limits the water flow when the temperature of the hot water leaving the condenser falls too low.


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

This strategy prevents the vacuum device from lowering the absolute pressure below the desired value by loading it with uncondensed vapour. Under certain conditions (specifically a new pan with very low air leakage, cold injection water and a low pan evaporation rate) the quantity of uncondensed vapour leaving the top of the condenser with the incondensable gases may be substantial. This can have two negative consequences viz. cavitation damage to the rotors of liquid ring vacuum pumps due to condensation within the pump, and the carryover of entrained injection water into the incondensable line due to the high vapour velocities in the head-box of the condenser resulting from the uncondensed vapour. The carryover of injection water will cause instabilities in the absolute pressure control and can also cause severe damage to liquid ring vacuum pumps. An improved control system, which overcomes this problem, has thus been developed for use with the circular continuous pan. In terms of the improved control system, the flow of cooling water into the condenser is varied to maximise condensation. Maximum condensation is achieved when the temperature of the incondensable gases leaving the condenser are cooled to the injection water temperature; but, since this true maximum condensation is a limiting condition only achieved with infinite water usage, a practical compromise is required. This compromise is achieved by controlling the incondensable approach temperature (incondensable gas temperature minus injection water temperature) to a practically achievable low value (approximately 2 to 5C). This target value is selected so as to approach full condensation without excessive injection water usage. As a further precaution against excessive injection water usage a conventional tailpipe temperature override is also part of the control strategy. The absolute pressure within the pan is separately controlled by a second control loop that bleeds atmospheric air into the suction of the vacuum device. The excess capacity of the vacuum device is thus loaded with air rather than uncondensed vapour, eliminating the possibility of water carryover from the condenser resulting from high vapour velocities in the top of the condenser. Conclusion The new Tongaat Hulett circular continuous pan represents a distinct change in continuous pan technology, offering some advantages over the conventional horizontal design. The new pan configuration allows for the direct replacement of an existing batch pan with a continuous pan. The two-calandria arrangement of the circular continuous pan can accommodate the boiling of two separate massecuite grades in the same vessel, and also facilitates boiling-out on the run, thereby improving the time efficiency of continuous pan operations. Improvements in the new design include a cold downtake wall to enhance massecuite circulation, as well as an improved vacuum control system. Two of these pans have been installed at the Xinavane factory in Moambique, and will be commissioned during the 2009 season. REFERENCES
Allan CJ (1962). Some suggested revisions of pan circulation theory. Proc Qld Soc Sug Cane Technol 21: 89-94.


Schorn PM et al

Proc S Afr Sug Technol Ass (2009) 82: 258 - 268

Echeverri, L.A. and Rein, P.W. (2007). Measurements and CFD simulation of the flow in vacuum pans. Proc Int Soc Sug Cane Technol 26: 1341-1353. Graham WS and Radford DJ (1977). A preliminary report on a continuous C Pan. Proc S Afr Sug Technol Ass 51: 107-111. Hempelmann R (2006). The VKT continuous vacuum pan more than 20 years of experience. Int Sug J 108(1296): 704-710. Hulett D (1965). Where should the downtake be? Proc S Afr Sug Technol Ass 39: 30-33. Jenkins GH (1958). Heating surface arrangement and circulation in vacuum pans. Proc Qld Soc Sug Cane Technol 17: 199-207. Kruger GPN (1983). Continuous A pan boiling trial at Maidstone Sugar Factory. Proc S Afr Sug Technol Ass 57: 46-51. Miller KF (2001). Productivity improvement in batch pans. Proc Aust Soc Sug Cane Technol 23: 346350. Montocchio G (1988). Les cuites continues a Felixton Proc ARTAS 3: 198. Rackemann DW, Broadfoot R and Stephens DW (2006). Improved CFD modelling of natural circulation in vacuum pans. Proc Aust Soc Sug Cane Technol 28: Rein PW, Cox MGS and Love DJ (1985). Analysis of crystal residence time distribution and size distribution in continuous boiling vacuum pans. Proc S Afr Sug Technol Ass 59: 58-67. Rouillard EEA (1986). A study of boiling parameters under conditions of laminar non-Newtonian flow with particular reference to massecuite boiling. PhD Thesis, University of Natal. Skyring A and Beale RF (1967). Boiling in massecuites. Proc Qld Soc Sug Cane Technol 34: 49-53. Smith N (1936). Notes on the design of vacuum pan. Proc Qld Soc Sug Cane Technol 1936: p 91. Webre AL (1932). Experiences of the working of the calandria pan. Proc Assoc Cuban Sug Technol pp 124-140. Wright PG (1966). The modelling of crystallisation schemes in a raw sugar factory. Proc Qld Soc Sug Cane Technol 25: 324-333. Wright PG (1974). Vacuum Pans Batch and Continuous. Sugar Technology Reviews 2. Elsevier Scientific Publishing Company, Amsterdam. pp 137-163.