DEVELOPMENT OF A LABORATORY SCALE COMPUTERIZED TEST UNIT FOR THE SIMULATION

OF DEEP BED FERMENTATION AND DRYING OF ROOIBOS TEA

ELIZABETH JOUBERT'

ARC-lnfncitec/Nietvoorbij Stellenbosch, South Africa

ZALDEUS STEENKAMP and RKUS M&LER

Department of Agriculture Western Cape Product Process Engineering

Ekenburg, South Africa

Accepted for Publication June 15, 1998

ABSTRACT

A test unit for the small-scale fermentarion and drying of rooibos tea was developed to simulate deep bed fermentation and drying of the tea in practice. m e data acquisition and control system consisted of an AT type personal computer interfaced with the test unit by a plug-in type ADDA board. Set points such as temperature, relative humidity and velocity of the inlet air could be specified for eight sequential, variable time intervals. Good control of dry bulb temperature and velocity of the inlet air was obtained, taking into account the restrictions placed on the system by the hardware. The man'mum air velocity that can be used with a bed depth of 50 mm, without formation of voids. was determined as 1.8 m/s. Precautions should be taken to ensure uniform air distribution and bed depth. Good quality tea was obtained afrer fermentm*on and drying in the test unit.

INTRODUCTION

Rooibos, an unique herbal tea, is produced from the indigenous South African plant Aspalathus linearis. The demand for high quality tea is increasing, due to growing consumption of rooibos, both in South Africa and overseas,

' Corresponding author: Dr. E. Joubert, ARC-IafruiteclNietvoorbij, Processing Technol. Div., Private Bag X5013, Stellenbosch, 7599, South Africa (E-mail address: lizettej@infruit. agric.za)

Journal of Food Process Engineering 21 (1998) 427-439. All Rights Resewed. QCopyright 1998 by Food & Nutrition Press, Inc., Tnunbull, Conneaicul 421

428 E. JOUBERT, Z. STEENKAMP and R. MULLER

especially in Japan. Processing of rooibos tea, which entails “fermentation”, i.e. oxidative changes, and drying, is carried out in the open on a drying yard, with no means of either controlling processing conditions or consistently producing tea of high quality.

Low ambient temperatures result in slow fermentation while over- fermentation occurs readily at high ambient temperatures. It is essential that the tea is dried as quickly as possible after completion of fermentation to arrest any further chemical changes, which could impair the quality of the dried product. This is not always possible during adverse weather conditions and peak processing periods, when the capacity of the drying yards is too small to dry large volumes of tea in thin layers. In extreme weather conditions fermentation is ceased for a day or two due to rain or to enable completion of drying. Frequent mechanical stirring of the tea layer is employed to increase the drying rate, but this results in formation of tea dust with no commercial value.

Taking floor fermentation and drying a step further, continuous processing is possible with a conveyor belt system with through-circulation of air for fermentation and drying. The bed depth can be varied depending on the application. Process design is only possible if the optimum processing conditions are known. It is thus important that experimental conditions should simulate the commercial scale practice. A test unit suitable for research purposes and capable of accurate control of variables such as temperature, air flow and humidity during fermentation and drying was required for such a study. A further requirement was versatility of operation by means of automatic process control, monitoring and data acquisition to ensure permanent record of events, as well as real-time data of process parameters. The availability of off-the-shelf, high accuracy analogue and digital input/output (I/O) boards with software support for personal computers gives a low cost option for automated operation of the test unit.

The objective of this study was development of a laboratory scale experi- mental unit for simulation of deep bed fermentation and drying of rooibos tea.

MATERIAL AND METHODS

Design Considerations

A test unit intended for fermentation (relatively low air flows; deep bed) and drying, with the latter as a static layer or fluidized bed (high air flows) of at least 3.5 kg tea, was designed. Cost and availability of space restricted optimal design and construction of the unit. The operating considerations determined the type of blower to be used, its location and therefore also the physical size and structural layout of the unit. A centrifugal fan was chosen for

TEST UNIT FOR ROOIBOS TEA PROCESSING 429

the test unit since it has certain advantages over an axial-flow fan e.g. better ability to cope with fluctuating operation conditions, the ability to vary its performance through speed changes and to adapt to a design for a 90' turn in air flow direction (Crocker 1980).

Structural Layout and Hardware

The test unit was designed according to operational requirements given in Table 1. A schematic outline of the unit is given in Fig. 1. The test section consisted of a rectangular container (430 mm X 336 mm x 226 mm) that was open at the top and closed at the bottom with a 30 mesh screen (U.S. Standard). The container slotted into a section of the duct which consisted of polycarbonate on the one side for viewing, and two 30 mesh screens, one directly under the container and the other approximately 776 mm above the top of the container. Two layers of fine mesh plastic netting were fastened on the lower screen. All the ducts as well as the tea container were manufactured from stainless steel (AISI 316). The duct system was fully insulated, except for the fan and the polycarbonate sheet.

TABLE 1. OPERATING REQUIREMENTS FOR A ROOIBOS TEA FERMENTATION

AND DRYING TEST UNIT Fermentation Drying

Product inlet air temperature ('C) 25-45 35-90 Tea temperature ('C) 2540 35-70 Relative humidity (product inlet air) Air velocity ( d s ) 0.2-0.7 0.2-2.0

> 95

Tea bed depth (m) 0.05-0.2 0.05-0.2

A backward-curved, centrifugal fan (Donkin Series BCC-2, Donkin Manufacturing, Port Elizabeth, South Africa), proportionally controlled by a variable frequency inverter (Varispeed VC 150, Commander Control Tech- niques, Johannesburg, South Africa), was selected to deliver a high pressure at a low speed. The selection of the fan was done after preliminary tests were carried out with moist tea (0.56 moisture ratio; wet basis) to determine the pressure drop over a bed depth of 200 mm, compensating for the resistance of the complete duct system with heaters, mixing plates etc.

Heating was supplied by electrical resistance, open-coil duct elements (12 kW) controlled by aprogrammable, proportional, multi-function thyristor control card (Phasecon P6550, Phasecon (Pty) Ltd, Northcliff, South Africa). The heaters were provided with a manual reset safety cutout to prevent overheating in case of system failure. Two sets of louvres, cranked by an actuator (Satchwell

430 E. JOUBERT, Z. STEENKAMP and R. MULLER

AX2251, Satchwell Control Systems, Slough, England) with an angular rotation of 90' and maximum rotation time of 1.7 min, facilitated recycling of the product outlet air within the unit (Fig. 1). Closing of the one louvre automatical- ly opened the other.

Cooling capacity of 1.5 kW, calculated to allow cooling of the inlet air from 35C (RH 0.6) to 26C (RH cu.0.98), was supplied by an off-the-shelf air conditioner with a direct expansion cooling coil. Humidification of the inlet air was accomplished by steam injection into the duct section between the heaters and the fan allowing the steam to be mixed in the turbulent air before it reached the test section. Two humidifiers were tested i.e. a commercially available unit (Condair LS/C 460, Condair AG, Munchenstein, Switzerland) and a custom- built steam generator. The steam capacity of the Condair (2.6 to 13.0 kgh) was controlled electronically by a 0 to 10 V signal. Steam was generated when the conductivity of the water reached the required level. The operation of the humidifier involved a filling and draining cycle. This resulted in the production of varying amounts of steam. The custom-built steam generator, giving a fixed amount of steam, heated the water by means of three 2 kW plus one 1 kW domestic immersion heaters that separately switched on and off. It operated under atmospheric pressure and the water level was regulated by a ball valve in an external reservoir. Two steam distributors were used for adequate humidifi- cation of the inlet air.

Air flow sensor

Air flow sensor

Air conditioner

Inlet air I)

T Stcam inkt (humidifier)

FIG. 1. SCHEMATIC OUTLINE OF TEST UNIT FOR FERMENTATION AND DRYING OF ROOIBOS TEA (D.B.: DRY BULB: W.B.: WET BULB)

TEST UNIT FOR ROOIBOS TEA PROCESSING 43 1

Sensors

Ungrounded thermocouples (Type T) were used to determine the tempera- ture of the tea at four points at an equal depth in the tea bed as well as for the wet (w.b.) and dry bulb (d.b.) temperatures of the product inlet and outlet air. The thermocouples were encased in stainless steel tubes and the same batch of wire was used for all the thermocouples. The mV signals given by the thermocouples were each amplified to 1 mV/ ' C by a thermocouple temperature indicator with linerarized analogue output (Digital Process Measurement model 3502, Temperature Controls (Pty) Ltd, Bellville, South Africa). The distance between the thermocouple probe and indicator was kept as short as possible. A co-axial cable was used for transmittance to the I/O board. The RH of the inlet and outlet air was calculated by the software from the respective d.b. and w.b. temperatures measured externally on the test unit. A small centrifugal fan (Fanno 21 ATXL, Giles Scientific, Johannesburg, South Africa) was used to supply air flow (>2.5 m/s) from the duct over the d.b. and w.b. thermocouples. The fan was placed downstream from the thermocouples to prevent heat transfer from the fan to the thermocouples. Measurement of air velocity was based on the calorimetric principle using a Weber Vent-captor air flow meter (type 3202.30/10; Weber Sensortechnik GmbH, Kollmar, Switzerland) with a measuring range up to 10 m/s, which was continually adjustable. The excitation voltage of the air flow meter was 24 V DC and the output signal was 4 to 20 mA. The mA signal was applied to a resistor (487 ohm) to obtain a voltage (mV) output to the analogue-to-digital (A/D) board.

Data Acquisition and Control System

The control and monitoring functions were carried out with an AT type personal computer fitted with a plug-in analogue and digital I/O board (PC30B; SA Eagle Electric, Cape Town, South Africa). The board featured sixteen 12-bit A/D input channels, two 12-bit digital-to-analogue (D/A) outputs, two 8-bit D/A outputs and 24 1/0 lines. The A/D subsystem of the board accepted analogue voltage inputs from the thermocouples and air flow sensors. The fan, heaters and humidifier were controlled by the D/A converters with the output ranges configured to 0 to 10 V. The digital I/O subsystem (0; 5 V) was used to control the air conditioner (on/off) and the louvres (on/off; direction - forwards or backwards).

Software

The data acquisition and control program for the test unit was written in Microsoft Quick Basic (V4.5). A simplified flow diagram of the control program is given in Fig. 2. Input parameters that were considered included processing

432 E. JOUBERT. Z. STEENKAMP and R. MULLER

Measure air !lor; - I- d.b. & w.b. temperature

t +

Adjust accordingly:

1. Louvres

2. Air contlitioner

3. Ileatcrs

4. Huniidifier

5. Fan

1 parometerr

h Update scrcen

1

I

satisfied?

L

No

FIG. 2. A SIMPLIFIED FLOW DIAGRAM OF THE COMPUTERIZED CONTROL SYSTEM OF THE FERMENTATION AND DRYING UNIT FOR ROOIE4OS TEA

TEST UNIT FOR ROOIBOS TEA PROCESSING 43 3

time, product inlet air temperature, RH and air flow for eight sequential variable time intervals. Additional options such as file name, directory, data saving frequency, saving of input parameters and the facility to change the set parameters during a run, were provided. Data was stored as sequential ASCII files for accessing by spreadsheet programs for further data manipulation. A graphic diqlay of the average tea temperature ('C), d.b. temperatures (product inlet and outlet air) (.C), w.b. temperatures (product inlet and outlet air) ('C), RH (product inlet and outlet air) and air velocity ( d s ) against processing time was given during a run. All the values of the temperature, RH and air velocity were displayed separately on a screen in digital form for more accurate evaluation of the process status. Each temperature data point was the average of 150 readings while the air velocity was given by an average of lo00 data samplings. The data was updated on the screen every 2.6 s. Trouble-shooting was facilitated by displaying the 12-bit number used to control the fan and heater. The action of the louvres (opening/inactive/closing) was also displayed on the screen.

Control of Set Points

Product Inlet Air Temperature. Control of the product inlet air temperature (d.b.) was achieved by proportional control of the heaters, the amount of product outlet air recirculated and the air conditioner, depending on the temperature of the inlet air. Minimum recirculation of the product outlet air was required for both fermentation and drying. Therefore the louvres were activated at the onset of a fermentatioddrying cycle to facilitate no recirculation of product outlet air. The louvres were only activated if the heaters reached the maximum output without reaching the required d.b. temperature (set point). If the d.b. temperature of the product inlet air was higher than the set point while no heating took place, the air conditioner was switched on automatically.

Relative Humidity. The humidifier reacted on the RH (product inlet air) which was calculated from the d.b. and w.b. temperatures of the product inlet air. The RH of the product inlet air was controlled only during fermentation.

Air flow. The air flow sensor controlled the fan. The fan speed was increased gradually to prevent excessive overshooting of the set point. No heating or humidification commenced before the air flow reached 0.2 m/s. This was an additional safety measure to protect the system from overheating.

Testing and Evaluating the Performance of the Test Unit. Tea fermentation and drying were carried out to assess the performance of the test unit in terms of accurate control of the d.b. temperature and air flow of the inlet

434 E. JOUBERT, 2. STEENKAMP and R. MULLER

air. The ability of both humidifiers to saturate the air to prevent drying out of the tea during fermentation was investigated at a maximum air velocity of 0.6 to 0.8 m/s and a temperature of 34 to 40C. Dry fermented tea (bulk density 240 kg/m3) at cu. 0.10 moisture ratio (wet basis) was moistened to approximate- ly 0.6 (moisture ratio; wet basis) and left overnight in a closed container for equilibration. Temperature control during drying was investigated at 40 to 70C and air velocities of 1.0 and 1.5 m/s, respectively. The maximum air velocity attainable without causing voids in the tea bed was determined for bed depths of 50 and 80 mm.

The feasibility of fermentation and drying of rooibos tea under controlled conditions in terms of quality was carried out using two batches of freshly comminuted tea. Two batches (5 kg) were harvested in the Clanwilliam area from a two-year- and six-year-old plantation, respectively. The tea was processed immediately after cutting into 3-4 mm pieces. Both batches were fermented at 36C and at a moisture ratio of 0.65 (wet basis). The tea was aerated at a rate of cu. 0.06 m3/s. Fermentation was terminated after develop- ment of a strong, sweet aroma. Drying was done immediately under the following conditions: no humidification, air inlet temperature of 40C (d.b.) and air velocity of 1 .O m/s for 3 h. The dried tea was evaluated by an expert panel of the Rooibos Tea Board according to the standardized procedure used for grading of commercially-produced rooibos tea (Anon. 1990). The quality grades, i.e. Super, Choice, Standard, each with three subdivisions (+: slightly better than norm; 0: norm; -: slightly poorer than norm) and Unacceptable were used to grade the general appearance of the leaves, the aroma, taste and color of the extract, as well as the overall grade of the tea.

RESULTS AND DISCUSSION

Software

The data acquisition and control program provided for varying fermentation and drying conditions, thus giving flexibility of operation to the test unit. Both fermentation and drying could take place during a single run which simplified the operation. The real-time graphic display of important processing parameters enabled processing conditions to be evaluated in the absence of the operator. This facility proved to be invaluable since fermentation can take up to 18 h.

Monitoring of Dry and Wet Bulb Temperature

Erratic temperature readings were obtained at first due to external electrical noise. This was eliminated effectively by amplifying the thermocouple signals and by the use of a co-axial cable. Calibration of the thermocouples was

TEST UNIT FOR ROOIBOS TEA PROCESSING 435

necessary to ensure that the same readout (f0.2C) was given by the different thermocouples. Thermocouples are less accurate than resistance thermometers, but thermocouples have the advantage of a very small size and a fast response (Anon 1982). Accurate wet bulb measurements required adequate ventilation for evaporation and thus wet bulb depression. Eastop and McConkey (1970) recommended an air velocity of at least 1.85 d s . This necessitated forced air circulation over the wick using a small centrifugal fan.

Air Flow Control Distribution and Void Formation

Set points of 1 and 1.5 m/s resulted in average air velocities of 1.OkO.13 and 1.5 +O. 19 d s , respectively. Figure 3 gives a typical example of air velocity control during fermentation. The short distance between the discharge outlet of the fan and the test section aggravated nonuniformity of the air flow. The velocity of the air discharging from the fan tends to be higher towards the outside of the scroll (Crocker 1980). More uniform air flow through the tea bed of the test unit was obtained by increasing the static pressure in section B with two layers of plastic mesh on the lower wire mesh screen (Fig. 1). Better air flow control was possible during drying than during fermentation since the humid air affected operation of the sensor. In isolated cases high humidity during fermentation caused condensation on the air flow sensor which resulted in erratic readings.

There was considerable resistance to air movement through the moist tea bed due to the surface water on the small particles (< 10 mesh) blocking the interstices of the bed during fermentation and the initial stages of drying. A rapid increase in fan speed to reach the air velocity set point resulted in the formation of voids in the tea bed as the bed pushed upward. This blow-through resulted in air velocities substantially higher than the set point, which was then automatically corrected. However, since the moist tea particles did not formed a free flowing mass, the voids channelled the air flow through the bed, resulting in uneven aeration and drying during the fermentation and drying cycles, respectively. The object was thus to avoid void formation during both fermentation and drying. Since no bubbling of the bed took place after void formation, fluidization of the bed during drying would only have been possible if a stirring device was installed for mechanical breakup of the bed. The moistening of the tea before fermentation resulted in extraction of soluble solids to the leaf surface, which imparted a stickiness to the leaves. The control algorithm for the fan was subsequently altered to give a very slow increase in air velocity that ensure air flow passing through the interstices of the bed and removal of surface moisture during drying. Voids also formed in the tea bed when presumably the static pressure and thus pressure drop over the bed exceeded a critical value due to a combination of tea mass, density of the tea

436 E. JOUBERT, Z. STEENKAMP and R. MULLER

1

0.8

n

3 0.6 .C( 0 0 0 4

$ 0.4 3 4

0.2

0

bed and air velocity. The pressure drop determined the maximum air flow attainable before bubbling of a particle bed occurs (Hovmand 1987). The maximum air flow through a 50 mm layer of moist tea was 1.8 m / s while the maximum for an 80 mm layer was 1.4 m / s .

5

.

0 100 200 300 400 500 600 700 Fermentation time (min)

FIG. 3. A TYPICAL AIR VELOCITY-TIME PROFILE DURING FERMENTATION AT 38C AND AIR VELOCITY SET POINT OF 0.6 M/S

Temperature and Humidity Control of the Product Inlet Air During Fermentation

The temperature and humidity of the product inlet air were controlled during fermentation. Humidification by the Condair humidifier complicated temperature control since humidification resulted in fluctuating product inlet air temperatures (d.b. and w.b.) during fermentation. The normal operation of the humidifier involves periodic filling up with tap water or draining of some of the water. Steam production varied or ceased altogether during this cycle causing the temperature to drop (> 5C) due to the decrease in energy input. The heaters balanced the energy loss but as soon as steam production commenced, the temperature overshot the set point by more than 5C, decreasing again to the set point as the heater output decreased.

TEST UNIT FOR ROOIBOS TEA PROCESSING 437

The variation in steam production also resulted in drying of the tea during the fermentation stage. This is undesirable since water is necessary for development of the characteristic rooibos tea flavor. Drying of tea took place at 0.98 RH (set point), 40C d.b. temperature of product inlet air and an air velocity of 0.7 d s . Under these conditions the moisture ratio of the tea (cu. 4 kg) decreased from 0.60 to 0.53 (wet basis) within 80 min due to a fluctuation in RH (0.75 to 1.0) caused by the refill and drain cycles of the humidifier.

These problems were overcome by ensuring adequate and constant steam production which eliminated the varying energy input and fluctuating air humidity. This was achieved by using a steam generator which delivered a fixed amount of steam depending on the number of elements switched on. Adequate steam production to prevent drying of the tea was obtained with fermentation carried out at 40C (product inlet air) and an air velocity of 0.6 d s . The d.b. temperature of the product inlet air was controlled within 1.4C of the set point. The average temperatures obtained with fermentation at a set point of 40C were 39.7+0.7C, 39.6f1.3C and 39.6f1.4C, respectively, for the tea, d.b. and w.b. temperatures of the product inlet air. The temperature of the tea approxi- mated the product inlet air (d.b. and w.b.), if the air was saturated, otherwise the tea temperature corresponded to the w.b. temperature of the product inlet air.

Temperature Control of the Inlet Air During Drying

A typical temperature-time profile obtained during drying of tea at 40C is depicted in Fig. 4. It was found that better control of the d.b. temperature was obtained if no humidification took place. The d.b. temperature of the product inlet air was controlled at 40C during drying. It is evident from Fig. 4 that the tea temperature decreased at first due to the evaporation of moisture on the leaf surface. It increased as drying took place eventually reaching the d.b. temperature of the product inlet air after cu. 120 min. This period will depend on the drying conditions. The d.b. temperature of the product outlet air was lower than the product inlet temperature due to evaporative cooling and heat loss. The conditions used for drying i.e. temperature and air velocity determined the heat loss which took place at the polycarbonate section and at the front section of the tea container which was not insulated. The difference increased with increasing inlet air temperature.

Approximately 34 kW of heating was needed to increase the inlet air temperature from 27C to 90C for an air flow of 0.448 m3/s. Partial recycling of the outlet air was used as a trade-off against the cost in order to reduce the heating capacity to 12 kW. It was assumed that high temperatures would only be necessary during falling rate drying. The increase in the moisture content of the outlet air would therefore be negligible in the latter stages of drying with the

438 E. JOUBERT, Z. STEENKAMP and R. MULLER

b 1 5 -

10

5 -

high volume of air passing through the tea. The louvres were activated to recycle the minimum product outlet air necessary to maintain the required product inlet air temperature. The position of the louvres was changed slowly to prevent excessive temperature fluctuation around the set point. Fixed bed drying at a product inlet air temperature (d.b.) of 55C at 1.5 m/s and ambient temperature of 18C could be accomplished without recirculation of the outlet air.

-

FIG. 4. A TYPICAL TEMPERATURE-TIME PROFILE OF TEA TJMF'ERATURE AND DRY BULB TEMPERATURE OF THE INLET AND OUTLET AIR OBTAINED

DURING DRYING AT 4OC WITH AN AIR VELOCITY OF 1 .O M I S WITH THE EXPERIMENTAL TEST UNIT

Tea Quality

The evaluation of the two batches of tea fermented and dried in the test unit is given in Table 2. The tea was graded Super and Choice grade, with Super grade indicating the better quality tea. The individual quality parameters were acceptable, taking into account that processing was not necessarily terminated at optimum flavor development.

CONCLUSIONS

A small fermentation and drying test unit interfaced with a personal computer afforded flexibility of operation with automated control and data

TEST UNIT FOR ROOIBOS TEA PROCESSING 439

acquisition for the fermentation and drying of a small batch (3.5 kg) of rooibos tea. The obtained results showed that, not only can processing conditions be adequately controlled in the test unit, but that good quality tea can be obtained using the unit. This test unit will be applied in a further study to determine the effect of deep bed fermentation and drying conditions on the quality of rooibos tea.

TABLE 2. QUALITY GRADING' OF TWO BATCHES OF ROOIBOS TEA FERMENTED AND

DRIED UNDER CONTROLLED DEEP BED CONDITIONS IN TEST UNIT Batch' Quality Parameters

Appearance Aroma TasteExhact Color Overall Grading

A S s' Sf K+ S- B K- K K K K

1 S-,S,S+

K-,K

= Differentiation of Super grade (+: strong aroma and taste; -: weak aroma and

= Differentiation of Choice grade (+ : strung aroma and taste; -: weak aroma and taste)

taste) 2 Batch A: Tea from a six-year-old plantation

Batch B: Tea from a two-year-old plantation

ACKNOWLEDGMENTS

This research was supported by the Rooibos Tea Board. The authors are indebted to Jim Swiegers for supplying the initial program modules, which were subsequently extended.

REFERENCES

ANON. 1982. Temperature Sensing with Thermocouples and Resistance

ANON. 1990. Gradering. Die Rooibos Nr. 90, 7-9. CROCKER, B.B. 1980. Fans and blowers. In Kirk-Othmer. Encyclopedia of

Chemical Technology, 3rd Ed., Vol. 9, (G.I. Bushy, C.I. Eastmond, A. Klingsberg and I. Spiro, eds.) pp. 768-794, John Wiley & Sons, New York.

EASTOP, T.D. and MCCONKEY, A. 1970. Applied Thennodynamics for Engineering Technologists, 2nd Ed., p. 474, Longman, London.

HOVMAND, S. 1987. Fluidized bed drying. In Handbook of Zndusm'al Drying. (A.S. Mujumdar, ed.) pp. 165-225, Marcel Dekker, New York.

Thermometers. A Practical Handbook, 2nd Ed. Labfacility , Middlesex.