Geof Walsh
(Techserve, South Africa)

1 Introduction

The late Maxime Rivière during his lifetime registered many patents, mainly concerned with the extraction of juice from cane. The most important, in the author's opinion, is his last patent which covered improving the efficiency of the extraction process by:- Displacement process

Maxwell (1929) wrote in a paper presented at the third congress of the International Society of Sugar Cane Technologists : "Mills in action are somewhat reminicent of the cave age; sheer brute force, huge wearing masses of metal, vast amounts of power and enormous pressures". In his "Handbook of Cane Sugar Engineering" Hugot (1972) wrote "The mill is a barbarous piece of equipment". The stage efficiency of mills is no more than 25 to 35%.

Cane diffusers now in use were developed from multi-stage counter current beet diffusers, not from scientific consideration of the desiderata. With the ever increasing processing rate of sugar factories, they are in danger of becoming mechanical monstrosities; a 10 000 tcd diffuser is 10 to 12 m wide x 60 m long. It has a mass of about 600 tons and has a headshaft weighing about 90 tons.

In both milling and conventional diffusers, the mixing/separation process is employed. The stage efficiency of cane diffusers is approximately equal to that of mills, viz. 30 to 35%.

The concept of "displacement extraction" was patented by Matthey in 1889 under English patent No. 21021 and was referred to in Noel Deerr's "Cane Sugar Handbook". (1921).

John Payne (1962 and 1968) pointed out the advantages of the displacement process. He wrote that having … "achieved the rupturing and stripping of the (juice) storage cells" ……"what is now wanted is to isolate the rich juice from the fibre. Theoretically, this can be effected efficiently by displacement with water. Such a displacement is best achieved by a counter-current flow of water and fibre-juice in which the water flows as an advancing front similar to the phenomenon of saturated flow in porous media. The advancing front of water operates like pistons through the many ruptured cell modules replacing the juice. Any macro breaks in the bed lead to piping, channeling and by-pass of water from the displacement process. Any mixing in the system reduces the effective counter-current flow. Any pressing or squeezing will expel juice concurrently as well as counter-currently and cause mixing. The ideal system therefore is an undisturbed bed of fibre and juice and a plug-type water-juice flow."

This ideal is what, in fact, is achieved in the Rivière Juice Extractor.

The advantages of displacement over mixing/separation are best demonstrated by the Ponchon - Savarit diagrams shown in Figures 1, 2 and 3.

Figure 1 is the diagram for a conventional five mill tandem with imbibition rate of 200% on fibre. It shows that the brix of the extractable juice in bagasse leaving a given stage is much higher (five to seven points) than the juice separated from the same stage.

Figure 2 is what the diagram would be like if there were 100% mixing efficiency on the five mill tandem with an imbibition rate of 200% on fibre. Number two mill bagasse would have the same brix of extractable juice as the juice (J2) coming out of the first stage. With only two stages the target brix of final bagasse composition is overshot.

Figure 3 is the Ponchon-Savarit diagram applied to a fully efficient displacement process. The brix of the juice displaced in a given stage is always higher by at least two points than the brix of the extractable juice of the bagasse leaving the same stage as was achieved in the pilot plant on which Maxime Rivière's patent was based.

If fully efficient displacement were taking place in a cane diffuser, only three stages would be enough to achieve 98% extraction using an imbibition rate of 200% on fibre. In practice, up to 18 stages are required. Buchanan (1968) calculated the ideal number of stages for a cane diffuser as 5,54 based on the stage residence time for interstage drainage determined from percolation tests and the mean particle size of prepared cane but concluded that, with a stage efficiency of 34 per cent, 17 stages would be required. In the Rivière Juice Extractor, only three stages are required.

2 Development of the Rivière juice extraction process

This process has been developed as a result of a scientific approach by Maxime Rivière to the design of an extraction plant based on the displacement principle, the structure of the raw material and its volumetric composition, as set out below:-

Bulk density of shredded cane depends on preparation and compaction and typically varies from 250 to 350 kg/m³ on a conveyor at up to 0,5 m depth. In this condition one ton of cane will occupy an apparent volume of 2,8 to 4,0 m³. The no void volume of cane being 0,89 m³/t, the void volume in the above-described bed of shredded cane is between (2,8 -0,89) =1,91 m³/t of cane and (4-0,89) = 3,11 m³/t of cane.

Assuming a fibre content of 150 kg/t of cane (F=15), the amount of juice (of say 1,05 density) necessary to fill the voids is between 1,91 x 1,05 = 2 and 3,11 x 1,05 = 3,27 t or, expressed in terms of fibre, between 2000/150 = 13,33 F and 3270/150 = 21,8 F. (In the experiments conducted by Maxime Rivière on a pilot plant in Reunion, the amount of juice required to fill the voids was 15 F to 20 F.)

In a conventional cane diffuser the 0,5 m upper layer of shredded cane has a similar density to that referred to above but, due to compaction, the density increases until, in the bottom layer of cane, the bulk density is probably two or three times higher. According to Hugot (1972), the average bulk density for cane in a diffuser is 500 to 600 kg/m³ which means that the average apparent volume occupied by one ton of cane is 1,67 to 2 m³. Deducting the no void volume of cane, this leaves a void volume of 0,78 to 1,11 m³. The filling of this void volume with 1,05 density juice will require 819 to 1165 kg of juice.

According to the operation results published by Payne (1962) for a Silver diffuser in Hawaii, the maximum amount of juice that can be present without flooding was 5,6 F. For the average fibre content of cane in Hawaii of 13,5%, the weight of juice per ton of cane was 5,6 x 135 = 756 kg. The corresponding volume of juice was 0,720 m³/t which leaves, on average, a volume of air in the mat of cane of between 0,78 - 0,72 = 0,06 and 1,11 - 0,72 = 0,39 m³ air/t of cane. But in the upper layer of the bed, the volume of air is between 2 and 3 m³/t cane.

This is surely one of the reasons for the low stage efficiency of cane diffusers, the presence of air impairing the mixing efficiency.

However, the main disadvantage of the operation of a leaching or lixiviation process through a bed of cane of 1,5 to 2 metres high, as obtains in most commercial diffusers, is the low percolation rate of about 0,1 m/min, which results in cane retention times of 50 to 60 minutes. Stage efficiency is low because of channelling and the by-passing of stage circulation juice between the stages. Percolating angles ranging from 57 degrees from the horizontal in the first stage to 14,5 degrees in the last stages have been reported. Bed-disturbing screws have to be used to facilitate percolation.

The Rivière extraction process avoids these disadvantages as it :
3 Meichage

Removal of air from the cells containing sliced beets was considered fundamental for efficient batch diffusion in the beet sugar industry before continuous diffusers were adopted. It was referred to by the ancient name "meichage" in the North of France.

Meichage was also used in Egypt in the early 1930's in the Naudet multi-cell process of bagasse batch diffusion where, after filling each cell of the diffuser with bagasse, the cell was closed, juice was gravitated into the bottom of the cell and rose through the column of bagasse until it emerged from a cock on the lid of the cell at which point the cock was closed. By this operation, the air contained in the bagasse was removed.

Of personal interest to the author is the fact that a very early commercially successful extraction of juice by cane diffusion was in a plant constructed ca 1960 by the South African firm of Patrick Murray (Pty) Ltd, with which the author was associated, at the factory of the Rhodesian Wattle Co. at Umtali. Prepared cane was processed in the multi-cell batch diffuser normally used to produce wattle bark extract.

Through combining the use of Meichage with the displacement process in a continuous cane juice extraction plant, Maxime Rivière has produced a break-through in juice extraction technology.

The validity of Maxime Rivière's juice extraction process has been demonstrated in pilot plant tests, conducted by the Sugar Research Institute, C.E.R.F., in Reunion Island and repeated independently by Copersucar in Brazil, simulating the displacement, in three stages, of the extractable juice in cane, showing that an extraction rate of 98% can be achieved in three stages followed by dewatering the megasse in a hydraulic press to a moisture content similar to that which can be achieved by a conventional last mill.

It is interesting that the measure of the thoroughness of cane preparation formerly termed the "Leachability Index" is now termed the "Displaceability Index" in many cane sugar producing countries. The design of the Rivière Juice Extractor is based upon extracting all of the displaceable juice in three stages.

4 Description of the process

Ponchon-Savarit diagram

From the results of the pilot plant experimentation, the number of stages to be used was determined by the Ponchon-Savarit diagram, Figure 3, assuming the following shredded cane analyses and operational conditions : The target of 4 brix of juice in the final megasse is reached in three stages. This corresponds to a pol extraction of 98 per cent when the purity of extractable juice in cane is 85 and the purity of extractable juice in bagasse is 60.

Flow diagram

With the above operational conditions, the flow diagram Figure 4 has been drawn. In each stage, for a 100% efficient displacement process to take place, three operations have to be implemented.

A - First Stage

Given an average typical shredded cane bulk density, the amount of Meichage juice needed to fill the voids between the cane particles completely is 15 F, giving an initial total juice content of the megasse in the first stage of 20 F at 15,55° brix, using as meichage 15 F of juice at the same brix 14,44° as in mixed juice leaving the first stage.

Displacement is performed using the juice (7,36 F, brix 9,79°) coming out of the second stage to replace the same amount (7,36 F) of juice at 15,55° brix contained in the megasse, which is thus extracted by displacement.

Before leaving the first stage the juice content of the megasse which, after displacement contains 20 F of juice at an average brix of 13,43°, is drained by gravity into a receiving tank resulting in a juice content of 6 F in the megasse leaving the first stage.

The amount of drainage juice will be 14 F at a brix of 13,85°. This juice is mixed with displaced juice (7,36 F at 15,55° brix) giving a mixed juice with brix of 14,44°. From this mixed juice, 15 F is recirculated as Meichage juice and 6,36 F leaves as extracted juice to maintain the overall balance which is:

B - Second Stage

As the megasse will have the same juice content (6 F), in and out of the stage, the Meichage juice will be 14 F in this case (instead of 15 F in the first stage).

Juice (7,36 F) coming from the third stage at a brix of 6° will displace the same amount (7,36 F) of juice at 10,58° brix.

Juice (14 F) at 9,36º brix will leave 6 F as juice content of the megasse feeding the third stage.

C - Third Stage

As in the second stage the Meichage juice (14 F) will be supplied by the drainage and displaced juice with a brix of 6º.

The dewatering mill juice (5,36 F) will be used as displacement juice, followed by imbibition water 2 F. The brix of the displaced juice (7,36 F) will be 6,54º.

Juice (14 F) at 5,71º brix will be drained from the third and last stage, giving:

5 Design of industrial unit

Two alternative configurations of the Rivière Juice Extractor have been proposed for industrial application:

(i) A horizontal drag conveyor type
This comprises a conventional type horizontal double chain and slat conveyor to transport the prepared cane delivered from the shredder to the dewatering mill(s) and having a suitably perforated deck at the Meichage, displacement and drainage sections. An industrial prototype installed by Copersucar at a sugar factory in Brazil is described in more detail under the following section.

A typical horizontal drag conveyor type of industrial unit is shown in Figure 5.

(ii) An induced flow paddle blade type
This comprises three stages, each with a paddle bladed rotor operating in a troughed casing, semi-circular in section, suitably perforated at the Meichage, displacement and drainage sections. A typical layout for an induced flow paddle blade type of industrial unit is shown in Figure 6.

In considering the operation of both types, it is important to note that, as stated previously, the displacement process can only be performed efficiently when the air has been expelled from the megasse under which condition the juice content of the megasse is at least 20 times the fibre content of the cane. This means that the fibre content of the megasse must be less than 5%.

In this condition, megasse has the hydrodynamic characteristics of a liquid. In fact, a slurry containing up to 7% fibre content can be pumped. With a juice content of 20 F, megasse will flow freely, by gravity, in an open trough. Because of this, mechanical transportation of megasse in a displacement process is not needed during Meichage and displacement. Only when drainage is taking place will the relatively drier material need to be transported mechanically.

The flow diagrams for both conveyor type and paddle blade units as shown in Figure 4.

6 Description of industrial prototype constructed in Brazil

A proportion of the cane used for the tests conducted by Copersucar was taken from near the top of a chute feeding the first mill of an existing tandem at the São Luiz A.A. factory in the state of São Paulo, Brazil, and was conveyed via a slat-type cross carrier to the prototype Rivière Juice Extractor. An opening in the deck of the cross carrier allowed the cane to fall into the extractor.

In the alternative form of feed shown in Figure 5, the prepared cane falls into a casing, semi-circular in section, in which a multi-bladed macerotor thoroughly mixes it with part of the Meichage juice, allowing the fibre-juice mixture to flow into the extractor.

The extractor incorporates a narrow steel trough 350 mm deep x ca 16 m effective length. The bottom of the trough is perforated where necessary to allow the passage of juice in the sections of the three stages where upward or downward juice flow takes place.

The width of the trough will vary with the capacity. Based on a 350 mm juice-fibre depth in the trough and a speed of 25 m/min, the width of the trough will be 650 mm per 100 tch. Speeds of up to 30 m/min are contemplated. Conventional diffusers with a width of bed in the order of 2,5 m per 100 tch, and a length of 60 m, are some four times the width and four times the length, i.e. 16 times the volume, of an equivalent Rivière Juice Extractor.

Typical trough widths of Rivière Juice Extractors will be :-

100 tch 650 mm
200 tch 1300 mm
300 tch 1950 mm
400 tch 2600 mm

These dimensions may be reduced by increasing the operating level in the trough above 350 mm and increasing the effective length proportionately. It is envisaged that, in most instances, a Rivière Juice Extractor will be capable of being installed in an existing mill house, in the space normally occupied by three or four mills in a tandem, where it would receive prepared cane directly from the existing feed system and discharge spent megasse directly into the feed chute of the dewatering / drying mill(s).

Two endless driven chains running on either side and just above the trough carry steel plate slats which effectively compartmentalize the trough into a series of boxes preventing cross-flow of juice between the sections and between the stages. The pitch of the slats is ca 400 mm.

A frequency inverter is used to set the speed of the extractor drive motor. At a speed of ca 13 m/min the calculated capacity of the unit is 25 tons cane/hour based on a cane bulk density of 250 to 300 kg/m³. The processing time is 1,25 minutes. The unit would preferably be operated at a higher speed than 13 m/min but this would have made the dimensions too small for practical operation, particularly in regard to the ease of filling the individual compartments.

Each of the three stages consists of the three individual operations described above, viz:-
  1. Meichage, which is effected by pumping juice upwards through the perforated bottom of the trough at a controlled rate such that the juice level is automatically maintained at the height of the top of the juice-fibre bed.
  2. Displacement, in which the juice in the bed is displaced downwards through the perforated bottom of the trough by a stream of lower brix juice which is pumped onto the top of the bed from the succeeding stage. In the case of the third (last) stage this displacing stream is a mixture of press water from the dewatering mill and imbibition water. In the displacement section, the juice level is automatically maintained at the top of the juice-fibre bed by controlling a valve on the displacement drain line.
  3. Drainage, where the juice in the bed is allowed to drain through the perforated bottom.
The maintenance of an unbroken column of liquid of ca 4 m is necessary to give the high liquid displacement and drainage rates on which the design is based.

The two streams from each of the three stages of the extractor, viz the displaced juice and the drainage juice, are fed from the extractor trough to a tank situated about 4 metres below the extractor trough.

The Meichage juice is pumped from this tank with: Mixed juice to processing overflows from the first tank. The displacement juice for the first stage is pumped from the second tank, that for the second stage is pumped from the third tank while that for the third stage is pumped from the press water plus imbibition water tank. The individual air displacement (Meichage), juice displacement and drainage sections of the extractor have perforated bottom plates but between the three sections of each stage and between the three stages, the bottom deck of the extractor trough is not perforated to avoid cross mixing.

Megasse from the third stage is fed through a chute to a six roller mill. The bagasse is sent to the boilers while the unscreened press water flows to the tank where it is mixed with the imbibition water. The quantity of imbibition water is measured by a magnetic flow meter and set manually to the required flow rate.

If two mills are used in tandem for dewatering/drying, imbibition may be applied between the mills, as is the case in many conventional diffuser installations, to reduce further the pol in bagasse.

Tests were conducted during the 1998 on the Copersucar extractor in Brazil but these proved to be disappointing in that the Meichage juice entering a compartment tended to displace upwards the existing liquid in the compartment, without mixing, with the result that in the immediately following displacement section, much of the Meichage juice was displaced. Reducing the inlet area for the Meichage juice to increase its velocity and induce mixing with the existing juice has largely overcome this problem. In the alternative form of induced flow paddle blade configuration, rotary mixers are provided in the Meichage section of each stage. Some form of mechanical mixing is contemplated in the drag conveyor extractor.

Further tests conducted during 1999 on the Copersucar extractor in Brazil have shown ca 10F juice in the megasse leaving each stage instead of 6F as obtained in the early batch tests indicating that the drainage time may be insufficient. This has prevented achievement of 2 points drop in brix between the juice leaving a given stage and the brix of the juice leaving the stage. However, pol extraction of 87-88% has been achieved in the three stages with overall extraction including the dewatering mill of ca 93%. Consideration is being given to introducing a light squeezing action at the drainage section. This would be more easily achievable in the paddle blade configuration of the extractor as shown in Figure 6.

7 Sand removal

An unexpected negative effect resulting from the introduction of cane diffusion in place of milling has been the retention in the cane bed of sand brought in with the cane which, in milling, was largely removed in the mill juice. This has been evident in the reduced quantity of clarifier muds in diffusion factories but, more expensively, in the increased wear and tear in boiler plants which have led to the introduction of single pass boilers with "membrane" walls. The destructive effect of sand brought in with the cane has been exacerbated by the increasing introduction of mechanical loading of cane in the fields, particularly push-piled windrowed cane.

Since the cane/juice mixture is in the liquid phase in each of the Meichage and displacement stages, it can be expected that much of the sand carried in with the cane will be decanted in the displacement stage and be carried out with the juice in the drainage stage.

It would be prudent therefore to provide for sand trapping and removal in the juice tanks beneath the extractor. The cost of such arrangements will be justified by the saving in boiler maintenance costs and in reduced capital costs of boilers on new installations.

8 Heating in the Rivière Juice Extractor

Indications are that the operating temperature has little bearing on the extraction results. In conventional diffusers the temperature is kept high enough to prevent micro-organism growth but in the Rivière Juice Extractor the retention time of ca one minute versus 50 to 60 minutes in a conventional diffuser eliminates this requirement.

In a conventional diffuser, the high temperature is also intended to assist in the dialysis process. The question is, just how much juice is extracted by lixiviation and how much by dialysis in a conventional diffuser? Not much appears to be extracted by dialysis, judging by the test results of the Rivière Juice Extractor.

It is generally accepted that starch extraction is substantially reduced by lower temperature operation. Presswater returned from the dewatering/drying mills need not be clarified or heated as in conventional diffusers. There have been many reports of difficulties experienced in feeding dewatering/drying mills with hot megasse of 80 to 85% moisture content so that the cold process employed in the Rivière Juice Extractor will be advantageous in this respect.

9 Capital Cost

For comparative purposes, the September 1999 cost of three alternative juice extracting systems, commencing after the cane preparation system which is assumed to be common to all of them, viz: a new five mill tandem, a new cane diffusion plant with dewatering/drying mills and a Rivière Juice Extractor with dewatering/drying mills, all rated for 250 tch, including civil works and installation, together with other significant differences, eg. draft juice % cane, power consumption and estimated annual maintenance costs, are set out in Table 1.

TABLE 1. Comparative costs of three juice extraction systems

Milling plant Cane diffusion plant Rivière juice extractor
Description Five 1067 x 2134 four-roller mills with 750 kW variable speed electric drives, intercarriers, Donnelly chutes, juice screens, pumps and piping.
NOTE - no allowance for additional 2250 kW power generating plant
6,5 m wide cane diffuser with dry feed conveyor, scalding juice heating, pumps and piping, draft juice screen, two 1067 x 2134 four-roller mills with 750 kW V.S. electric drives, intercarrier, presswater screens, pumps and presswater heater 1,676 m wide juice extractor with macerotor feed, pumps, tanks and piping, draft juice screen two 1067 x 2134 four roller mills with 750 kW V.S. electric drives, intercarrier and pump
Cost of extraction plant N/A $ 5 250 000 $ 1 250 000
Cost of dewatering drying mills N/A $ 4 600 000 $ 4 600 000
Total cost including civil works and erection US $ 11 400 000 US $ 9 850 000 US $ 5 850 000
Power consumption 3600 kW 1800 kW 1300 kW
Estimated annual operating and maintenance cost Processing 1 000 000 tons of cane per annum (labour and materials) US $ 942 000 US $ 437 000 US $ 295 000
Draft juice % cane 115 122.5 100
Predicted Pol extraction % 96.5 98 98

The power savings and reduced load on the evaporator station are significant, especially with the increasing importance of co-generation in the sugar industry.

10 Conclusions

Much work remains to be done to achieve the initial promise of Maxime Rivière's vision and his experimental work. The problems which arose through the lack of mixing of the Meichage juice with the residual juice from the previous stage have been largely overcome but the unexpectedly high liquid content in the megasse from each stage remains to be dealt with. The initial industrial scale trials however justify the continuation of development of the process.

11 Acknowledgements

The author expresses his thanks to Copersucar and to Techserve CC, licencees of the estate of the late Maxime Rivière, for their permission to present this paper and to Dr W S Graham for his advice and assistance in connection with it.


Buchanan, E.J., (1968) The calculation of stage efficiency and its application to diffuser design Proc. SASTA 42: 65-72.
Copersucar Co-operation de Produtores de Cana, Acucar e Alcool do Estado de São Paulo Ltda., Centro de Tecnologia, Caixa Postal 162, Piracicaba SP, Brazil
Deere, N., (1921) Cane Sugar, 2nd edition., Norman Roger, London p 250
Hugot, E., (1972) Handbook of Cane Sugar Engineering, 2nd edition, Elsevier, Amsterdam p 350
Maxwell, F., (1929) A comparative survey of milling as practised in Cuba, Hawaii and Java", Proc. ISSCT., 3: 486-496
Payne, J., (1962) The processing of sugar cane into juice and fibre, Proc. ISSCT., 11 : 971 - 991
Payne, J., (1968) Cane diffusion - The displacement process in principle and practice. Symposium on Cane Diffusion", Proc. ISSCT., 13:103-121
Rivière, M., (1921-1995) (1994) New Cane Juice Displacement Process", Patent / Pat. Appln. Nos.:Australia - 679255, Brazil - PI9404596.8, China - 96123431.8., Cuba - 22/97, E.P.O.- 94402832.3., Egypt - 123/1998, Indonesia - P97101125, Mauritius - 1157/591, Mexico -97101125, Pakistan 757/96, Phillipines I-5505, South Africa - 94/10306, U.S.A. - 08/547630, Vietnam - 0631/97, Zimbabwe - 147/96

Back To Library Page Top