juice circuit OA circuit
1 . Juice circuit feed vessel
2. Juice circuit pump
3. Sampling port for juice circuit, rubber septum
4. Pressure gauge, juice circuit inlet to module
5. DOC module
6. Pressure gauge, juice circuit outlet from module
7. Bleed valve for juice circuit
8. Juice circuit 9. 0 A circuit
10. 0 A reservo ir
1 1 . OA circuit pump
1 2. Pressure gauge, OA circuit inlet to module
13. 0 A balance
14. Cooling coil
1 5. OA stirrer
1 6. Water bath heater
1
68 The OA reservoir was placed o n the OA balance ( 1 3 ) in order to measure the changes in OA solution weight The inlet and outlet tubes for the were secured over the OA reservoir with clamps so they did not touch the reservoir. The OA solution in the OA reservoir was kept well mixed with the OA stirrer [ 1 5] and the circulator on the water bath heater [ 1 6] . The temperature in the OA reservoir was controlled by cool i ng with propylene glycol ( to -2°C) (chilled water i n U SA) circulated in the cooling coi l [ 1 4] at a constant flow rate and by heating from the water bath heater. The temperature in the water bath was maintained at ± O. I °C The temperature of the OA was monitored with a temperature probe attached to a display unit The temperature probe was calibrated with a mercury calibration thermometer.
A constant volume was maintained in the OA circuit, excluding the O A reservoi r which acted as a buffer tank. Water crossing the membranes and taken up in the OA circuit resulted in an i ncrease in the total weight of OA in the reservoir. This was measured on the OA balance.
3.2. 1 0. Sta n d a rd operation start up p rocedu re of DOC a p pa ratus At start up, the juice circuit was first fi l led with RO water, at the desired operating temperature, from the juice circuit feed vessel . Any air in the l i nes was released at the bleed valve. The juice circuit was flushed out with five l itres of fresh RO water, then the RO water was continuously recirculated. The feed vessel maintained a constant head of 540 mm above the base of the DOC module. The volume of water i n the juice circuit feed vessel was maintained between 5 00 and 1 000 m i l l i l itres. The volumetric water flow rate in the juice circuit was set and maintained at 4 x 1 0 -5 m3 s -1 , except when the effect of juice circuit flow rate was i nvestigated. The flow rate was set in the pilot plant DOC module to obtain the same Reynolds number as obtained in the small module in the juice circuit The flow rate was checked using a calibrated measuring cyl inder before recycling operation. The hydraulic pressure in the juice circuit was 1 4 - 1 7 kPa at the module inlet and 5 - 7 kPa at the module outlet
The juice circuit was established before any OA solution was circul ated. The OA was then slowly pumped i nto the OA circuit and air bubbles removed by tilting the DOC module from side to side. The flow rate of the OA was constant at 7 x 1 0 -6 m' s -1 for all experiments The flow rate was checked using a calibrated measuring cyl inder. The hydraulic pressure in the OA circuit was 0 - 5 kPa at the inlet to the DOC module and atmospheric at the outlet.
3.2. I 1 . Equil i b ra tion of DOC m od u le
It was necessary to equilibrate the module and membranes before collection of data. The following equil ibration operation was carried out after standard operation start up and prior to any data collection.
Firstly the juice and OA circuits were flushed out for 1 0 minutes in the small laboratory module, with RO water or intended OA, respectively. Then the OA inlet hose was transferred to a new bucket containing about 1 5 kg of fresh OA ensuring no air was introduced to the OA circuit The OA was recycled for one hour to allow the membranes to equilibrate No attempt was made to maintain the OA concentration during this time. The same equi libration procedure was used for the pilot plant DOC module. However, the OA circuit was initially flushed for 20 m inutes and the membranes were equilibrated for 90 m inutes with 2 5 - 3 0 kg of OA
3.2. 1 2. Experi m ental opera tion of DOC a p pa ratus
After the equilibration process the inlet hose of the OA circuit was transferred into a new OA solution in the OA reservoir, at the desired concentration and temperature. For the next 1 0 minutes of operation, the OA exiting the module was discarded (20 m inutes for the pilot plant module) before the outlet hose was clamped over the OA reservoir allowing OA to recycle. At the same time the juice circuit was flushed with fresh RO water. The OA was left to circulate for five m inutes before the initial weight of OA was recorded as " zero time". the small or pilot plant DOC module the O A reservoir contained approximately 25 or 56 kg of O A, respectivel y . The m ass of OA used ensured the OA concentration did not change by more than 5% over the data col lection period. Experimental trials looking at the effect of operating temperature were carried out randomly with respect to temperature. Between each trial the juice and OA circuits were equilibrated to the next desired operating temperatL're
3.2. 1 3. D etermination of wate r flux rate
The weight of the OA in the reservoir and the volume of water in the j uice circuit feed vessel were recorded at 5 m i nute intervals for 45 m inutes. The OA temperature, OA circuit inlet hydraulic pressure, j uice circuit inlet and outlet hydraulic pressures were also recorded at the same time. The temperature of the j uice circuit at the beginning and at the end of the run was recorded. There was no temperature control in the juice circuit, as the volume in the j uice circuit was small (approximately 3 . 1 8 x 1 0 -4 m3) compared
70 to the OA circuit and reservoir (approximately 0 .02 m3) Polystyrene foam sheets were placed around the module to provide insulation.
Samples of the j uice and OA circuits before and after each experimental run were col lected. The refractive index of these samples was measured and the concentrations determi ned from calibration curves. These samples were frozen and stored at -20°C for further analyses (e.g. HPLC).
3.2. 1 4. Data analysis for flu x rate
The water flux rate across the membrane was determ ined after linear regression analysis of the OA weight versus time data ( MINITAB Release 9 .2, Minitab Inc. Pennsylvania, USA). The gradi ent of the regression l ine obtained was the mass flow rate of water across the membrane (kg s -1). Except where stated, trials were carried out in triplicate. 3.2. 1 5. Cleaning of DOC modul e and m em b ranes
After each d ay the O A was drained whi le the RO water was sti l l circulating i n the juice circuit. Both the j uice and OA circuits were flushed first with warm water (3 5 - 40°C) for 30 m inutes, fol lowed by cold water ( 1 5 - 20DC) for a further 20 minutes. Each circuit was then rinsed with five litres of 0 . 1 % v/v P 3-oxonia active sanitiser. After sanitisation,
both circuits were rinsed with RO water for 1 5 minutes. The OA circuit was drained first then the juice circuit. Moisture was left in the juice circuit to keep the membranes moist In the p i lot p l ant DOC module all washing times were extended by 1 0 m inutes. When the membranes were considered to be "dirty", they were washed with a 1 0 9 1 -1 solution of Ultras i l 5 3 enzyme detergent. F ive litres were circul ated i n the juice and O A circuits for 20 minutes after the first rinse. The circuits were then rinsed again with col d water ( l 0 minutes) before sanitisation with P3-oxonia active.
3.2. 1 6. Mem brane replacement
Membranes were replaced when they had measurab le leaks . The asymmetric membranes were cut to the correct shape placed on each OA plate of the module, with the active layer of the membrane facing the required direction and kept taut with masking tape. The two OA plates were then dam ped together securing the membranes in place. S crews and bolts around the edge were tightened until the membranes were sealed between the seals and there were no water leaks from the module. The masking tape was then removed.
Water was circulated in the j uice circuit (OA circuit empty) and the amount of water lost through the membrane after one hour was determ ined. The norm al loss was 1 0 - 1 5 m ! . If more than 1 5 ml of water was lost the membranes were inspected and replaced. Water leakage from the juice circuit at the base of the module was less than 1 m l hour -I In most cases there was no water loss from the j uice circuit from the base of the module.
Each new set of membranes was tested also for salt permeabi lity to establ ish that the new set h ad sim i l ar mass transfer properties to previous sets. Following membrane repl acement the juice circuit was drained and flushed with 1 litre of 0 . 1 0 g (g solution) -1 fructose solution and then drained again. Two kilograms 0 1 0 9 (g solutionfl fructose solution were then introduced to and circulated in the juice circuit while two kilograms of 0. 1 5 g (g solutionfl NaCI solution were circul ated in the OA circuit After three hours, typical ly about 1 kg of water was lost from the juice circuit into the OA circuit The fructose solution in the juice circuit was col lected quantitatively then re-diluted back to its original concentration. The specific conductances of the initial and final fructose solutions were measured to determine the amount of NaCI which had passed through the membrane into the juice circuit
3.2. 1 7. Determining the tim e required to flush the OA circuit
The DOC apparatus was set up as outlined for standard operation [see 3 .2 . 1 0) . RO water was circulated in the juice circuit and fructose solutions (0. 1 , O J , 0 . 5 and 0.7 g (g solutionfl) were used as the OA Following the set up of flow in the juice circuit, osmotic agent was introduced to the module and the refractive index of the exiting OA was measured every m inute for the first 1 0 minutes. I t was then measured approximately every 5 minutes for the next 80 minutes. The time required to flush all the water out of the OA circuit and to obtain a steady-state concentration of OA exiting the module was determined for each concentration of OA, in at least duplicate trials. For O l and 0 . 7 g (g solutionfl fructose solutions experiments were carried out in triplicate.
3.2. 1 8. Determining the tim e required for equilibration of m e mbranes The DOC apparatus was set up as outlined for standard operation [see 3 1 0) . The OA circuit was flushed out with fresh OA for 1 0 m inutes. New OA solution was i ntroduced and the first 1 0 minutes of OA exiting the DOC module was discarded.
At a time designated as "zero time" the weight of the OA reservoir was recorded and OA exiting the module was collected in an empty tared plastic bucket (OA-out bucket). There was no recirculation of the OA solution. The weight of OA in the reservoir was
72 recorded at 5 minute intervals while the weight of the OA col lected in the OA-out bucket was recorded every 3 . 75 minutes. The water volume reduction in the juice circuit feed vessel was also recorded.
The decrease in the OA reservoir weight, over time was analysed by linear regression analysis ( MINITAB), this provided the flow rate of OA into the DOC module (OAe-J. The increase in the OA-out bucket weight over time was also analysed by linear regression analysis ( MINIT AB), this provided the flow rate of OA out of the module (OAolJr). The flux rate of water across the membrane was calculated from the difference between the O AIN and O Amn flow rates.
The water flux rate across the membrane was determined for a range of fructose solutions as OA, at concentrations 0 1 , 0.2, 0. 3 5, 0. 5, 0.6 and 0.7 g (g solutionr1 • Flux rates were determined after 1 5, 3 0, 45, 60, 75, 90 and 1 05 minutes, at 20°C . For each OA concentration the experiment was carried out i n dupl icate or triplicate.
3.2. 1 9. Visualisation of flow characteristics
To assess the juice circuit flow, amaranth dye solution \vas placed in the dye feed vessel, which was placed above the juice circuit feed vessel . The dye feed vessel was connected, with PVC tubing and nylon fittings, to the juice circuit at a tee junction just prior to the i nlet to the DOC module, as shown in Figure 3 .2(a). F low was controlled with a ball valve placed in the dye inlet PVC tubi ng. The DOC apparatus was set up as outlined for standard operation [see 3 1 0] .
A video camera was placed on a tripod and positioned approximately two metres away from the DOC module. The camera lens was pointed directly at the front flat face of the OA plate, to record the movement of the amaranth dye solution up the juice circuit in the modu le. Recording on the video camera was started first The amaranth dye solution was pulsed into the juice circuit using the bal l valve The flow of the amaranth dye solution up the juice circuit was recorded on the video camera. After each pulse of the dye the juice circuit was flushed out with clean water.
The flow conditions up the j uice circuit were recorded with the OA circuit empty and with the O A circuit ful l of 0 . 6 g (g solutionfl sucrose solution as O A. For each set of conditions the dye was pulsed in six times .
To assess the O A circuit flow, amaranth dye solution was placed in the dye feed vessel placed above the juice circuit feed vesseL The dye feeJ vessel was connected to the O A