Separation method with chromatographic zones with variable length

The invention relates to a method and a device for separating at least one component of a mixture by contact between liquid and solid phases in chromatographic zones of variable length.

It applies to chiral separations and in particular to the separation of stereoisomers used in particular in the field of pharmacy.

There are different chromatographic methods that can be used for the production of chemical components on a large scale.

The publication by RM NICOUD and M. BAILLY (Choice and optimization of operating mode in industrial chromatography, Proceeding of the 9th International Symposium on preparative and industrial chromatography, PREP 92, April 1992, Nancy, p. 205-220): illustrates this back technological plan.

These processes can be classified according to several criteria: the process can be either batch or continuous, the composition of the eluent can be isocratic or a composition gradient can be carried out.

One of these possibilities is the classic 4-zone true moving bed countercurrent process where in a moving bed system for a direct countercurrent effect, solids flow continuously in a closed loop past points. fixed feed and eluent introduction alternating with raffinate and extract withdrawal points.

This process being fully known and described, only the characteristics required to understand the nature of this invention are summarized below. For the True Moving Bed mode of operation, a countercurrent contact between the liquid and solid phases is made in the column which can be divided into 4 different zones.

Zone 1: Everything between the eluent injection and extract withdrawal lines

Zone 2: Everything between the extract withdrawal and feed injection lines

Zone 3: Everything between the feed injection and raffinate withdrawal lines

Zone 4: Everything located between the raffinate withdrawal and eluent injection lines.

Due to the input / output flow rates, the liquid flow rate varies according to the zone: QpQn'Qm, Qiv being the respective flow rates in zones I, Π, ΙΠ and IV.

In 1961, the UOP Company patented a process for simulating the motion of the solid by a clever connection between interconnected columns in a closed loop, (US Patent 2,985,589 and US 3,291,726, 3,268,605). This process, called Simulated Mobile Bed (LMS), then makes it possible to practically produce the True Mobile Bed in a simple manner. It is characterized in that one periodically advances downstream (in the direction of circulation of the main fluid) the points of introduction of charge and eluent while one advances simultaneously and according to the same increment ( at least one column for example) the draw-off points of a raffinai and an extract.

All the input and output lines are therefore moved simultaneously at each period ΔΤ and the cycle time, time at the end of which they return to their initial position is equal to Ne x ΔΤ, Ne being the total number of columns.

This process has been widely described, in particular by CHARTON and NICOUD (completed design of a simulated moving bed, Journal of Chromatography 1995, 702, 97-102).

Only the minimum information necessary for a good understanding of the present invention will be recalled below.

The I / O positions are moved simultaneously at fixed intervals. It is advisable to locate the position of row by row (n), which means that at a given moment, a given input / output row is connected to the input of column n. For example in a 12 column system, Charge (9) means that the charge line is connected to the input of column 9 while Raffinai (11) means that the raffinate line is connected to the input of column 11.

Using this definition, a system can be represented by: El (3) / Ext (6) / Charge (9) / Raff (l /). For this configuration, the number of columns in zone I, II, ΙΠ and IV are respectively: 3/3/2/4. The system configuration is then completely defined by:

input / output_number of columns

At time 0: El (3) / Ext (6) / Charge (9) / Raff (l 1)

3/3/2/4

After a given time (the PERIOD) all the I / O positions are moved simultaneously by one column and the system is described as follows:

At time ΔΤ: El (4) / Ext (7) / Charge (10) / Raff (12) 3/3/274

After a new period, all the positions will again be moved simultaneously by one column and the system will then be described as follows:

At time 2 x ΔΤ: El (5) / Ext (8) / Charge (l 1) / Raff (l) 3/372/4

At time 2 x ΔΤ, the position of the raffinate has been moved from position 12 to position 1. Note that position 1 can be written as position 13 modulo 12. ([13] 12).

This presentation can be generalized to Simulated Mobile Beds comprising a number Ne of columns. For a Simulated Movable Bed composed of Ne columns, it is obvious that no position can exceed Ne. For the sake of simplicity, we will simply increase all positions by one on each move, and define all positions modulo Ne (eg [8] / v_t= 3 if Nc = 5).

If at a given moment the configuration of the Simulated Moving Bed is El (e) / Ext (x) / Charge (f) / Raff (r), a simple reasoning makes it possible to find the number of columns contained in each zone:

zone 1: AZ? / = [xe] N_VS ; zone 2: A% 2 = [fx] N_VSzone 3: M? 3 = [rf] N_VS ; zone 4: A & 4 = [er]_Nc

We can simply check that: Nbl + Nb2 + Nb3 + Nb4 = Nc.

and the system is completely defined by table 1 input / output_Nb of columns in each zone

At time 0: El (e) / Ext (x) / Feed (f) / Raff (r) Nbl / Nb2 / Nb3 / Nb4

At time ΔΤ: El ([e + l] jy_vs) / Ext ([x + l] iy_vs) / Feed ([f + Ipy_vs) / Raff ([r + l] fy_vs) Nbl / Nb2 / Nb3 / Nb4

At time η X AT: El ([e + n] / y_vs) / Ext ([x + n] / y_vs) / Feed ([f + n] ^_vs) / Raff ([r + n] N_vs) Nbl / Nb2 / Nb3 / Nb4

Table 1

The injection and withdrawal points are shifted by one column after a period ΔΤ and by Ne columns after Ne periods. The number of columns in each area remains unchanged. The injection and withdrawal points therefore return to their initial position after the cycle time Ne x ΔΤ.

The main characteristics of Simulated Mobile Bed systems (giving a practical realization of the True Mobile Bed) are defined by:

1. zones defined by the position of the input / output lines.

2. a fixed number of columns per zone.

3. zones of fixed length

4. synchronized movement of all input / output lines.

Features 2, 3 and 4 are due to the fact that the Simulated Movable Bed simulates the behavior of the true Movable Bed.

It is possible according to the FR patent. 2 721 528 to correct the disturbances in the composition of the extract and of the raffinate caused by the dead volume of the recycling pump located between the last and the first bed of the adsorption column, by increasing the duration of the adsorption column by an appropriate value. connection of an injection or withdrawal current of fluids to the system each time this current passes from a position immediately prior to a position immediately after the dead volume, then by reducing said connection time when this current passes from the position immediately after the dead volume to the next position. But once per cycle, all inputs and outputs are shifted simultaneously. This technique simply compensates for technological imperfections to make it operate as close to an ideal simulated moving bed.

In simulated moving bed separation processes using a small number of columns, it most often appears that the products recovered in the extract and in the raffinate have different purities, excellent for one of the two but insufficient for the. other. In certain types of separation, when the volume of adsorbent involved is low, the level of purity of the extract and of the raffinate may even prove to be insufficient, as will be seen in the examples.

One of the objects of the invention is to remedy these drawbacks.

Another object is therefore to increase the purity of the product withdrawn as an extract and as a raffinate.

Another object is to minimize the costs of separation.

It has thus been found that by not simultaneously moving the position of the fluid inlets and outlets during the period and during the cycle time, improved results could be obtained.

More precisely, the invention relates to a process for the separation, called VARICOL, of at least one component of a mixture containing it, in a device having a set of chromatographic columns or sections of chromatographic columns containing an adsorbent, mounted in series and in closed loop, the loop comprising at least one point of a feed injection, a point of a raffinate withdrawal, a point of an eluent injection and a point of an extract withdrawal, in which a chromatographic zone is determined between an injection point and a draw-off point or vice versa, and at the end of a given period of time, all the injection and draw-off points are offset by one. column or section of column in a given direction defined with respect to that of the flow of a main fluid flowing through the loop, the method being characterized in that, during said period, the different injection and withdrawal points of a column or column section are shifted at different times so that the length of the zones defined by said different points is variable.

The period is defined as the smallest time interval ΔΤ at the end of which each of the inputs and outputs has been shifted by one column or section of a column, the shift not taking place simultaneously for all the inputs and outputs. It should be noted that after a cycle time Ne x ΔΤ, the system has regained its initial position.

The term adsorbent is used in its most general sense. It can be an adsorbent such as a molecular sieve, for example zeolite, used in adsorption processes, or an adsorbent such as an ion exchange resin. It can also be a stationary phase based on silica, a reverse phase adsorbent and a chiral phase.

In more detail, the following succession of steps can be carried out at least once:

- Is shifted at time tl during said period ΔΤ, in a given direction, the position of the point of injection or withdrawal relative to at least one zone, of a column or section of column so as to increase the length of said zone and to reduce the length of the zone adjacent to said zone, then at an instant t2 during said period, the position of an injection or withdrawal point relative to at least another area, of a column or section of column so as to increase the length of said other zone and to decrease the length of the zone adjacent to said other zone, and the operation is repeated if necessary so that, after said period of time ΔΤ , we find the same configuration of columns as the initial configuration with an offset of all the positions of the injection points and the draw-off points of a column or of a column section.

According to a first embodiment illustrated by Table 2, the lengths of zones of a column can be made to oscillate continuously, the increase in one zone being compensated by the decrease in the next.

enter exit

Number of columns in each zone

At time 0: At time dTl: At time dT2: At time dT3: At time AT:

El (e) / Ext (x) / Feed (f) ARaff (r)

El ([e + l]_Nc) / Ext (x) / Feed (f) / Raff (r)

El ([e + 1] ^_VS) / Ext (x) / Feed (f) / Raff (l r + I] N_vs) El ([e + l]_Nc) / Ext (x) / Feed ([f + l]_Nc) / Raff ([_r+ ll_Nc)

Nbl / Nb2 / Nb3 / Nb4

Nbl-I / Nb2 / Nb3 / Nb4 + 1

Nbl-1 / Nb2-I / Nb3 / Nb4 'Nbl-1 / Nb2 + 2 / Nb3 / Nb4

El ([e + 1] Nc) / Ext ([x + 1] NcVPeed ([f +!] N_VS) / Raff ([r + 1 / yv_vs) Nbl / Nb2 / Nb3 / Nb4

At time ΔΤ + dTl: El ([e + 2] N_vs) / Ext ([x + 1] iy_vs) / Feed ([f + 1] jy_vs) / Raff ([r + l] / y_vs) Nbl-I / Nb2 / Nb3 / Nb4 + 1

At time ΔΤ + dT2: El ([e + 2] ^_vs) AExt ([x + 1 l ^_vs) / Feed ([f + 1] ^_VS) / Raff ([r + 2] y_vs) At time ΔΤ + dT3: El ([e + 2] / y_vs) / Ext ([x + l] / y_vs) / Feed ([f + 2] iy_vs) / Rqff ([r + 2J / y_vs) At time 2 X ΔΤ: El ([e + 2] ^_vs) / Ext ([x + 2] ^_vs) / Feed ([f + 2] ^_vs) / Raff ([r + 2] ^_vs)

Nbl-1 / Nb2 / Nb3 + 1 / Nb4

Nbl-1 / Nb2 + 1 / Nb3 / Nb4

Nbl / Nb2 / Nb3 / Nb4

Table 2

According to a second embodiment illustrated by Table 3, the increase in length of a zone can be compensated for by the reduction of the opposite zone.

enter exit

Number of columns in each zone

At time 0:

At dTI time:

At time AT:

At time AT + dTI At time 2 x AT:

El (e) / Ext (x) / Feed (f) / Rafflf) Nbl / Nb2 / Nb3 / Nb4

El ([e + I] N_VS) / Ext (x) / Feed (f) / Raff ([r + 1] iy_vs) Nbl-I / Nb2 / Nb3 + I / Nb4

El ([e + I 1n_vs) / Exî ([x + 1 lN_vsVFeed ([f + 1] Nc) FFaff ([r + 1 Jn_vs) Nbl / Nb2 / Nb3 / Nb4 El ([e + 2] Nc VExt ([x + I / Feed (f + I) / Raff ([r + 2]! \ '_vs) Nb I - I / Nb2 / Nb3 + 1 / Nb4

El (le + 2] NcVExt (lx + 2] NcyFeed (lf + 2] N_vsyRaff (lr + 2] N_vs Nb) / Nb2 / Nb3 / Nb4

Table 3

Several other embodiments are possible, some of them being shown in the examples.

According to one characteristic of the method, it is possible to perform during the period all the shifts of the injection or withdrawal positions with a substantially constant time phase shift and advantageously at a time phase shift at least equal to a quarter of a period.

According to one variant, it is possible, during the period, to shift the positions of the injection or draw-off points with a non-constant time phase shift.

According to another characteristic, the fluid flow rate circulating in a given zone is generally kept substantially constant.

It is advantageous to perform the offsets of the positions of the injection and draw-off points in the same direction as that of the flow in the columns or section of columns.

According to another advantageous characteristic of the method, at least one flow rate of fluid circulating in an injection or withdrawal line can be controlled by the pressure in the device. Preferably, it is that of the raffinate and / or of the extract, the other fluids then being under flow control.

A liquid can advantageously be used as eluent, but it is also possible to operate with a supercritical fluid or with a subcritical fluid.

The pressure range in which product separations are carried out can be between 0.1 and 50 MPa and preferably between 0.5 and 30 MPa. The temperature in the columns is generally between 0 ° C and 100 ° C. It was observed that the process according to the invention gave excellent results when the number of columns or sections of columns was less than 8. For values ​​greater than 8, it is very advantageous to optimize the process by studying the influence of the number and the length of the columns in each zone combined at the instant of the shift during the period of the cycle.

The invention also relates to the device in particular for implementing the method.

More precisely, it comprises a plurality of chromatographic columns or sections of chromatographic columns containing an adsorbent, mounted in series and in a closed loop, said loop comprising at least one pump for recirculating a fluid, a plurality of fluid injection lines in each column or section of column connected to at least one injection pump and a plurality of lines for withdrawing a fluid from each column or section of column connected to at least one pump of draw-off, at least one valve on each line, said loop defining at least three chromatographic zones, each of which is determined by a fluid injection point and a fluid draw-off point, the device being characterized in that it comprises means for controlling the variation over time of the length of the zones connected to said valve and adapted to shift by a column or section of column the positions of the injection and draw-off points intermittently.

The valves implemented are advantageously all or nothing valves.

The process according to the invention (VARICOL) is better explained in the examples below, but its differences compared to the Simulated Mobile Bed process appear immediately:

1. zone lengths are not constant

2. the number of columns per zone is not constant over time

3. the input / output lines are not moved simultaneously

Although the oscillation introduces a disturbance into the system, it appears surprisingly that the performance of the VARICOL process is often better than that of the Simulated Moving Bed system (see examples).

As explained, an implementation of the method according to the VARICOL invention is cyclical, so that after a given cycle time Ne x ΔΤ, the system returns to its initial configuration. During this cycle, the number of columns in each zone has been varied, and for educational purposes, it may be useful to define an average number of columns per zone:

<Nb \> = average number of columns contained in zone I during a cycle <Nb2> = average number of columns contained in zone 2 during a cycle <Nb3> = average number of columns contained in zone 3 during a cycle < Nb4> = average number of columns contained in zone 4 during a cycle.

Just as a Simulated Mobile Bed system can be presented by:

LMS

Nbl / Nb2 / Nb3 / Nb4 we can represent a periodic VARICOL process by:

VARrCOL

However, while the number of columns per area has real significance for LMS systems, average numbers which are non-integers and which have no technical significance are simply used for convenience for the VARICOL process.

EXAMPLES

Example 1:

This VARICOL process was used in order to carry out the separation of steroisomers of phytol (3,7, l, 15-tetraméthyI-2-hexadécen-l-ol, C₂₀H₄₀O). Synthetic phytol is a mixture of cis and trans isomers, the latter being used in perfumery.

The separation between the isomers of phytol is carried out on silica (Lichroprep Si 60, 25-40 micrometers from Merck KGaA, Darmstadt) with an eluent composed of heptane - ethyl acetate (75/25 v / v) at 27 ° C. . For the sake of simplicity, a solution containing 50% cis and 50% trans isomer is prepared. According to measurements carried out at the laboratory scale, the adsorption isotherms were determined and suitably adjusted on a modified Langmuir type equation model:

+

Fc.

ι + Σ * α n being the concentration of space i adsorbed on the solid λι, λ₂, Κι, K₂, Ki, K₂ are adjustable parameters

C, and Cj being the concentrations of species i and j in the mobile phase.

With:

Λ = 1.7 Ai = 1.7 K, = 0.20633 K₂ = 0.9862 K, = 0.00645 K₂ =0,0308

Knowledge of adsorption isotherms is not absolutely necessary to operate the VARICOL process but helps to find the suitable operating parameters to obtain the appropriate purities. The techniques used make use of the numerical simulation methods described for example in Fundamentals of preparative and nonlinear chromatography, G. Guiochon, S. Golsbran Shirazi and AM Katti, Academie Press, 1994.

The offsets of the point of injection or withdrawal of fluid are carried out at time t which is a fraction of period AT.

1. 5-column VARICOL systems:

Experience has shown that for a feed concentration of 6.4 g / l, an adequate set of flow rates on a system comprising 5 columns 2.6 cm in diameter and 16 cm in length corresponds to:

Q Eiuent = 24.98 ml / min Q load = 22.08 ml / min

Q Extract = 25.42 ml / min Q Recycling = 106.84 ml / min

For this set of flow rates, the optimal period for moving the positions of the inputs / outputs is: AT = 1.6 min for the Simulated Moving Bed (LMS) system.

The concentrations and purities of the extract and of the raffinate obtained by the various processes are given in Table 4. An overall purity is defined by the average value of the purities of the extract and of the raffinate.

<td></td><td colspan="2">Configuration</td><td>cis (g / L)</td><td>Extract trans (g / L)</td><td>purity%</td><td>cis (g / L)</td><td colspan="2">Raffinate trans (g / L) purity%</td><td>purity overall</td>

<td>VARICOL</td><td> 1,25 1,25</td><td> 1,25 1,25</td><td> 0,170</td><td> 2,690</td><td> 94,1</td><td> 3,060</td><td> 0,120</td><td> 96,2</td><td> 95,2</td>

<td></td><td>at t = 0</td><td> 2111</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td>

<td></td><td>at t = AT / 4</td><td> 1112</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td>

<td></td><td>at t = AT / 2</td><td> 1121</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td>

<td></td><td>at 1 = 3ΔΤ / 4</td><td> 1211</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td>

<td>LMS</td><td> 11</td><td> 12</td><td> 0,180</td><td> 2,560</td><td> 93,4</td><td> 3,050</td><td> 0,260</td><td> 92,1</td><td> 92,8</td>

<td>LMS</td><td> 11</td><td> 21</td><td> 0,280</td><td> 2,670</td><td> 90,5</td><td> 2,940</td><td> 0,120</td><td> 96,1</td><td> 93,3</td>

<td>LMS</td><td> 12</td><td> 11</td><td> 0,180</td><td> 2,500</td><td> 93,3</td><td> 3,050</td><td> 0,260</td><td> 92,1</td><td> 92,7</td>

<td>LMS</td><td> 21</td><td> 11</td><td> 0,280</td><td> 2,640</td><td> 90,4</td><td> 2,940</td><td> 0,170</td><td> 94,5</td><td> 92,5</td>

<td>VARICOL</td><td> 1,2 1,2</td><td> 1,4 1,2</td><td> 0,185</td><td> 2,694</td><td> 93,6</td><td> 3,050</td><td> 0,108</td><td> 96,6</td><td> 95,1</td>

<td></td><td>at t = 0</td><td> 1112</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td>

<td></td><td>at 1 = ΔΤ / 5</td><td> 1121</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td>

<td></td><td>at 1 = 3ΔΤ / 5</td><td> 1211</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td>

<td></td><td>at ΐ = 4ΔΤ / 5</td><td> 2111</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td>

Table 4

All the possible Simulated Moving Bed configurations (for a 5 column system) are shown in Table 4. The best average purity (93.3%) is obtained by the 1/1/2/1 configuration.

The VARICOL 4-zone 1.25-column process makes it possible to obtain an average purity of 95.2%, which is therefore approximately 2% higher than the best result obtained with the LMS process. It should be noted that this ability of the VARICOL process to obtain higher purities with columns and flow rates similar to those of LMS is extremely interesting.

The second implementation of the VARICOL method makes it possible to illustrate an operation for which the line-to-line phase shift is not identical and which also shows good results.

2. 8 column systems

In order to keep the column length and the amount of stationary phase constant in the system, the column length was reduced to 10 cm each. The same flow rates as in the 5 column system were used by adjusting the period time to ΔΤ = 1 min.

The concentrations and purities obtained in the stream of the extract and of the raffinate for the different processes are presented in Table 5. An overall purity is defined as the average value of the purities of the extract and of the raffinate.

<td rowspan="2"></td><td rowspan="2">Configuration</td><td colspan="3">Extract</td><td colspan="3">Raffinate</td><td rowspan="2">purity overall</td>

<td>cis (g / L)</td><td>| trans (g / L)</td><td>purity%</td><td>cis (g / L)</td><td>| trans (g / L) |</td><td>purity%</td>

<td>VARICOL 1</td><td>1.52.5 1.5 2.5 at t = 0 1313 at t = ΔΤ / 2 2 2 2 2</td><td> 0,080</td><td> 2,680</td><td> 97,1</td><td> 3,170</td><td> 0,130</td><td> 96,1</td><td> 96,6</td>

<td>VARICOL 2</td><td>2.52.51.51.5 at t = 0 2 3 1 2 at = AT / 2 322 1</td><td> 0,130</td><td> 2,720</td><td> 95,4</td><td> 3,110</td><td> 0,090</td><td> 97,2</td><td> 96,3</td>

<td>VARICOL 3</td><td>1.51.5 2.5 2.5 at t = 0 12 2 3 at t = ΔΤ / 2 2 13 2</td><td> 0,180</td><td> 2,770</td><td> 93,9</td><td> 3,060</td><td> 0,030</td><td> 99,0</td><td> 96,5</td>

<td>LMS</td><td> 2222</td><td> 0,110</td><td> 2,720</td><td> 96,100</td><td> 3,140</td><td> 0,070</td><td> 97,8</td><td> 97,0</td>

Table 5

In this case, the purities of the LMS are already high (approximately 97%), and it was not possible to improve these results with the VARICOL process. The way to obtain higher purities would be to increase the column lengths and / or to increase the number of columns within the framework of the process according to the invention and / or to change the flow rates used.

Example 2:

The separation between fructose and glucose was carried out on Dowex 99 monosphere (350 micrometers) in the form of calcium using water (65 ° C.) as eluent. Under these conditions, the adsorption isotherms are almost linear and the retention factors of the two sugars are given by:

E_gîu, = 0.25 K_Fruct =0.45.

1. 5 column systems:

Experience has shown that for a feed concentration of 50 g / l of each species, an adequate set of flow rates for a system consisting of columns 2.6 cm in diameter and 160 cm in length is:

Q Eiuent = 18.13 ml / min Q load = 13.76 ml / min

Q Extract = 16.57 ml / min Q Recycling = 90.36 ml / min

For this set of flow rates, the optimal period for moving the positions of the inlets / outlets is: ΔΤ = 6.4 min for the Simulated Moving Bed (LMS) system.

<td rowspan="2"></td><td rowspan="2">Configuration</td><td rowspan="2">Glucose (g / 1)</td><td rowspan="2">Extract Fructose p (g / D</td><td rowspan="2">cleanliness %</td><td colspan="3">Raffinate</td><td rowspan="2">purity overall</td>

<td>Glucose (g / D</td><td>Fructose (g / i)</td><td>purity %</td>

<td>VARICOL</td><td>1.25 1.25 1.25 1.25 att = 0 1112 at = AT / 4 1121 at = AT / 2 1211 at t = 3AT / 4 2111</td><td> 2,710</td><td> 39,380</td><td> 93,6</td><td> 41,890</td><td> 2,450</td><td> 94,5</td><td> 94,1</td>

<td>LMS</td><td> 1112</td><td> 2,860</td><td> 37,010</td><td> 92,8</td><td> 41,820</td><td> 4,870</td><td> 89,6</td><td> 92,2</td>

<td>LMS</td><td> 1121</td><td> 4,500</td><td> 39,190</td><td> 89,7</td><td> 40,040</td><td> 2,500</td><td> 94,1</td><td> 91,9</td>

<td>LMS</td><td> 1211</td><td> 2,540</td><td> 37,010</td><td> 89,7</td><td> 42,130</td><td> 4,870</td><td> 89,6</td><td> 89,7</td>

<td>LMS</td><td> 2111</td><td> 4,500</td><td> 38,660</td><td> 89,7</td><td> 40,040</td><td> 3,090</td><td> 92,8</td><td> 91,3</td>

Table 6

The improvement obtained by the VARICOL 4-zone 1.25-column process (Table 6) is also very significant here. The overall purity is almost 3% higher than that obtained in the case of a conventional LMS.

2. 6, 7 and 8 column systems

Further experiments were performed and are shown in Tables 7, 8 and 9.

The total column length as well as the flow rates were kept constant in the two types of tests of the same system. For each case, the period time and the length of the columns were adjusted.

6 column system: L = 1.33m, ΔΤ = 5.32 min

<td rowspan="2">Type</td><td rowspan="2">Configuration</td><td colspan="3">Extract</td><td colspan="3">Raffinate</td><td rowspan="2">purity overall %</td>

<td>Glucose ία / h</td><td>Fructose (o / l)</td><td>purity %</td><td>Glucose ία / Ι)</td><td>Fructose (a / l)</td><td>purity %</td>

<td>VARICOL</td><td>1.5 1.5 1.5 1.5 at t = 0 12 12 at = AT / 2 212 1</td><td> 2,11</td><td> 39,94</td><td> 95,0</td><td> 42,50</td><td> 1,88</td><td> 95,8</td><td> 95,4</td>

<td>LMS</td><td> 1 1 22</td><td> 3,29</td><td> 38,73</td><td> 92,2</td><td> 41,35</td><td> 3,00</td><td> 93,2</td><td> 92,7</td>

<td>LMS</td><td> 12 12</td><td> 1,02</td><td> 36,37</td><td> 97,3</td><td> 43,79</td><td> 5,56</td><td> 88,7</td><td> 93,0</td>

<td>LMS</td><td> 2 112</td><td> 3,29</td><td> 38,19</td><td> 92,1</td><td> 41,35</td><td> 3,59</td><td> 92,0</td><td> 92,0</td>

<td>LMS</td><td> 12 2 1</td><td> 2,97</td><td> 38,73</td><td> 92,9</td><td> 41,67</td><td> 3,00</td><td> 93,3</td><td> 93,1</td>

<td>LMS</td><td> 2 12 1</td><td> 5,05</td><td> 40,78</td><td> 89,0</td><td> 39,44</td><td> 0,78</td><td> 98,1</td><td> 93,5</td>

<td>LMS</td><td> 2 2 11</td><td> 2,97</td><td> 38,19</td><td> 92,8</td><td> 41,67</td><td> 3,59</td><td> 92,1</td><td> 92,4</td>

Table 7

7 column system: L = 1.14 m, ΔΤ = 4.56 min

<td rowspan="2">Type</td><td rowspan="2">Configuration</td><td colspan="3">Extract</td><td colspan="3">Raffinate</td><td rowspan="2">purity overall %</td>

<td>Glucose ία / L)</td><td>Fructose (or</td><td>purity %</td><td>Glucose ία / L)</td><td>Fructose fa / L)</td><td>purity %</td>

<td>VARICOL</td><td>1.75 1.75 1.75 1.75 at t = 0 12 2 2 at =^ATT / 4 2 122 att = AT / 2 22 1 2 att = 3AT / 4 2 22 1</td><td> 1,81</td><td> 40,24</td><td> 95,7</td><td> 42,83</td><td> 1,54</td><td> 96,5</td><td> 96,1</td>

<td>LMS</td><td> 12 2 2</td><td> 1,3</td><td> 38,30</td><td> 96,7</td><td> 43,49</td><td> 3,47</td><td> 92,6</td><td> 94,7</td>

<td>LMS</td><td> 2 12 2</td><td> 3,68</td><td> 40,52</td><td> 91,7</td><td> 40,93</td><td> 1,07</td><td> 97,5</td><td> 94,6</td>

<td>LMS</td><td> 2 2 12</td><td> 1,3</td><td> 37,76</td><td> 96,7</td><td> 43,49</td><td> 4,07</td><td> 91,4</td><td> 94,1</td>

<td>LMS</td><td> 2 22 1</td><td> 3,36</td><td> 40,52</td><td> 92,3</td><td> 41,25</td><td> 1,07</td><td> 97,5</td><td> 94,9</td>

Table 8

<td rowspan="2">Type</td><td rowspan="2">Configuration</td><td colspan="3">Extract</td><td colspan="3">Raffinate</td><td rowspan="2">purity overall %</td>

<td>Glucose ία / Ll</td><td>Fructose (a / Ll</td><td>purity %</td><td>Glucose ία / Ll</td><td>Fructose ία / Ll</td><td>purity %</td>

<td>LMS</td><td> 2222</td><td> 1,62</td><td> 40,20</td><td> 96,1</td><td> 43,15</td><td> 1,41</td><td> 96,8</td><td> 96,5</td>

<td>VARICOL</td><td>1.75 1.75 2.25 2.25 at t = 0 2 13 2 at =^ATT / 4 222 2 to t = 3AT / 4 1223</td><td> 1,79</td><td> 40,27</td><td> 95,7</td><td> 42,93</td><td> 1,44</td><td> 96,8</td><td> 96,2</td>

<td>VARICOL</td><td>1.75 2.25 1.75 2.25 at = O 22 2 2 at = 3AT / 4 1313</td><td> 1,26</td><td> 39,73</td><td> 96,9</td><td> 43,49</td><td> 1,99</td><td> 95,6</td><td> 96,3</td>

<td>VARICOL</td><td>1.75 2.25 2.25 1.75 at t = 0 13 22 att = M74 2222 att = 3Ay / 4 223 1</td><td> 1,67</td><td> 40,33</td><td> 96,0</td><td> 42,69</td><td> 1,42</td><td> 96,8</td><td> 96,4</td>

<td>VARICOL</td><td>2.25 1.75 2.25 1.75 àt = O 22 2 2 at = 3AT / 4 313 1</td><td> 2,17</td><td> 40,72</td><td> 94,9</td><td> 42,46</td><td> 1,00</td><td> 97,7</td><td> 96,3</td>

<td>VARICOL</td><td>2.25 1.75 1.75 2.25 at t = 0 22 1 3 àt = AT / 4 2 22 2 àt = 3AT / 4 3122</td><td> 1,74</td><td> 40,09</td><td> 95,8</td><td> 42,94</td><td> 1,55</td><td> 96,5</td><td> 96,2</td>

<td>VARICOL</td><td>2.25 2.25 1.75 1.75 at t = 0 3 2 2 1 àt = AT / 4 222 2 àt = 3âT / 4 23 1 2</td><td> 1,7</td><td> 40,24</td><td> 95,9</td><td> 42,91</td><td> 1,52</td><td> 96,6</td><td> 96,3</td>

Table 9

8 column system: L = 1 m, ΔΤ = 4 min

The analysis of the results presented in Tables 6, 7, 8 and 9 leads to the following conclusions:

• a 5-column VARICOL system is more efficient than all possible 5-column LMSs.

• a 6-column VARICOL system is more efficient than all possible 6-column LMSs.

• a 7-column VARICOL system is more efficient than all possible 7-column LMSs.

· A 5-column VARICOL system makes it possible to achieve purities equivalent to what is obtained with a 6-column LMS. The VARICOL process therefore allows a significant reduction in costs.

• the VARICOL process is more advantageous for a system with a number of columns less than 8.