Fluidized Bed Drying of a Pharmaceutical Powder: A Parametric Investigation of

Drying of Dibasic Calcium Phosphate

Hao Chen1, Xue Liu1, Cody Bishop1, Benjamin J. Glasser1

Corresponding author Benjamin J. Glasser E-mail: [email protected]

Abstract

A parametric analysis of four fluidized bed dryer operating parameters – initial moisture

content, material loading, heating air temperature and air flux – was carried out using two

factorial experimental designs. Two different dryer scales, the Glatt GPCG-1 and

MiniGlatt, with chamber volumes of 23 L and 3 L respectively, were used to dry Dibasic

Calcium Phosphate powders. According to a Pareto analysis, initial moisture content and

air flux were the most significant variables. For the larger GPCG-1, the four variables

used in the factorial tests were found to be independent. For the smaller MiniGlatt, the

interaction of the initial moisture content and the air flow rate was also significant. From

the parametric analysis, it was observed that drying is dominated by the constant drying

stage, and zero-order kinetics, for the mild drying conditions (appropriate for many

pharmaceutical products) we considered in this work. A comparison of the two dryers

and an evaluation of energy efficiency were also carried out. It was found that drying

should be carried out with relatively low temperatures and high air fluxes for both the

large and small scale dryers in order to maximize the energy efficiency.

KEYWORDS: Fluidized bed drying; Calcium Phosphate; Energy analysis

1
 1. INTRODUCTION

Drying processes are common in the chemical, pharmaceutical and food industries [1]. It

is one of the traditional methods that removes moisture or solvents in order to provide

stable agricultural products and/or semi-finished products [2-4]. It is also known that the

drying process is energy intensive [3]. It has been reported that an average of 12 % of all

energy consumed is used on drying, and the cost of drying could reach up to 60 %-70 %

of total cost of investments [5]. Therefore, optimal operation of the drying process is

sought to meet the requirements for cost-effective manufacturing. [6]

Fluidized beds have been widely used in the chemical, pharmaceutical and food

industries as batch dryers due to their high heat and mass transfer rates as a result of the

high contact between gas and particles in the chamber [7, 8]. Successful applications of

fluidized bed dryers include drying food products such as soybeans, agricultural products

such as paddy and colza, biochemical products such as baker’s yeast and pharmaceutical

materials such as dibasic calcium phosphate powder [5, 9-12]. The reason for extensive

use of fluidized bed dryers is due to several features: 1) a fairly uniform bed temperature

in the chamber provides homogenous drying and allows for easy control of temperature

and operation [13] and 2) they can be operated at a relatively low temperature with a high

drying rate, making them favorable for heat sensitive materials [7]. In the pharmaceutical

industry, fluidized bed drying plays an important role in producing the high porosity of

pharmaceutical granules which promotes drug delivery[14-19].

2
During fluidization air flows through the bed of particles and suspends the particles in the

air stream, resulting in a dynamic fluid-like state for the particles [20]. When air passes

upwards through a bed of particles, the pressure of supplied air drops due to the

increasing resistance with increasing air flow. With a further increase in the air flow, an

expanded bed is obtained with some particles brought up into the air stream. As the

upward drag force keeps increasing with the air flow, eventually the weight of particles is

counterbalanced, the separation of the particles increases and the bed is fluidized.

Although continuous manufacturing has already been well established in many industries,

the pharmaceutical industry still predominantly relies on batch processing for several

reasons. One of the most important considerations is the issue of quality assurance. For a

continuous approach, the residence time for the apparatus needs to be measured so that

the process can be monitored appropriately. However, a batch can be accepted or rejected

based on the outcome of the quality control testing. Batch operation is suitable for heat

sensitive materials which is usually the case for biochemical and pharmaceutical products

[21] [22]. Furthermore, the flexibility of batch manufacturing makes it possible to switch

the sequence of several operations. In other words, establishing or adjusting a suitable

line with several unit operations can be easily performed in a batch process, however, this

would be considerably more difficult in a continuous process [23]. Although these

advantages have been reduced as process design and control of continuous manufacturing

improves, batch processing will most likely still play an important role in the

pharmaceutical manufacturing during the following decade.

3
Recent studies have focused on understanding the fundamentals of fluidized bed drying

by establishing models to predict the drying behavior of particles, optimizing the

operating condition by parametric studies, and designing novel fluidized bed dryers.[24]

Mortier et al. [25] reviewed the modeling of fluidized bed drying of wet porous granules.

Three different types of models, each with its own objectives, have been developed for

modeling of fluidized bed drying: porous material models, population balance models,

and computational fluid dynamics models. The porous material models, as a starting

point, investigate the dynamics of the drying process at a pore level. The continuum

approach, pore network modeling and single particle modeling are the three most often

used approaches in this area. Population balance modelling provides the particle

distribution for different properties of interest, such as particle size, moisture content and

temperature. Limited work has been done in terms of the drying kinetics of batch drying

processes. However, similar work for granulation and spray-drying processes has been

performed [26-30]. Computational fluid dynamics (CFD) models provide detailed

information about the spatial behavior of the system but requires a large amount of

computational power. The two most popular methods for performing CFD modeling are

the Eulerian-Lagrangian and the Eulerian-Eulerian approaches, where the former can

calculate the behavior of each single particle but requires significant computational effort,

while the latter considers the gas phase and solid phase as continuous and fully

interpenetrating.

Dong-Hyun Lee et al. [31] developed a mathematical model for batch drying in fluidized

bed dryers using a continuum approach. A three-phase model was used with a bubble

4
phase, an interstitial gas phase and a solid phase. A surface-stripping model was used to

describe the decrease in particle size with drying time due to attrition and entrainment.

Modified overall heat and mass transfer coefficients between the particles and the

external gas film were proposed. The model successfully predicted the mass and heat

transfer in a conventional fluidized bed dryer using an inert medium.

Several challenges of commercializing fluidized bed drying include its high energy cost,

low energy efficiency, complexity of scale-up and difficultly determining the process

end-point. Emrah Özahi et al. [32] established a model of a batch fluidized bed dryer in

terms of mass, energy and exergy balances. They found that lower initial moisture

content and higher ambient temperature led to a lower energy efficiency of drying,

whereas increasing the material loading could increase the efficiency. The flow rate of

heating air and the material loading were found to have more of an effect on drying than

other operating parameters. Sazzat et al. [5] presented an energy and exergy analysis of

paddy drying using an industrial fluidized bed dryer. They stated that higher initial

moisture content led to an increase in energy usage and energy usage ratio (EUR).

Syahrul et al. [33] conducted experiments with wheat and corn and then compared the

data with predictions from a model developed from thermodynamics. The results showed

that the effect of heating air flow rate and heating air temperature on drying varied with

different particle properties. Hematian et al. [9] studied the kinetics of drying sodium

percarbonate particles in a conical fluidized bed. They compared 15 widely used

empirical models of drying. For most operating conditions, the modified Hendersen and

Pabis model had good agreement with experimental data points.

5
Zielinska et al. [4] studied the drying behavior of carrot cubes in a fluidized bed dryer. It

was discovered that a higher air velocity of heating air was required when the moisture

content of the carrot particles was higher. As the drying process proceeded, a lower value

of air velocity was required to maintain the same height of the bed. The drying rate was

influenced by the applied heating temperature. With higher temperature the mass and

heat transfer were more intense, hence a higher drying rate. Zhao et al. [34] investigated

the effect of drying temperature, air velocity, vibration, particles size and bed height on

drying of Chinese lignite in a vibration fluidized bed. A drying curve was fitted using the

Midilli-Kucuk model [35] which is an empirical model that is suitable for a single layer

drying process. They found that the drying rate increased when higher temperature and

higher superficial air velocity were applied. Reducing the size of particles and bed height

increased the effective contact area, therefore enhance the drying. Liu et al. [36]reported

that the effect of particle size on drying is not significant for dibasic calcium phosphate

anhydrous (DCPA).

An important parameter in the design of fluidized beds is the air pattern, defined by the

geometry of the dryer and the inlet air introduction manner, which determines the

fluidization of the materials and the corresponding drying performance.[37] Michael et al.

[38] studied the hydrodynamic behavior inside different types of fluidized vessels. They

stated that the bubbling frequency decreased with time in conical beds whereas

cylindrical beds experienced an increase due to different particle circulation patterns.

These results indicated that some undesirable fluidization phenomena such as

6
defluidization or segregation could be caused by the improper choice of vessel geometry

which may lead to the failure of whole batches. Eliaers et al. [39] compared the

performance of a rotating fluidized bed in a static geometry (RFB-SG) to a conventional

fluidized bed. The RFB-SG introduced air tangentially by using a vortex chamber. Upon

contact with granules, the direction of the inlet air shifts vertically upwards. By using

such a design, the wet granules are able to form denser and more uniform beds. It was

found that RFB-SG was able to dry the biomass at a much higher feed rate but

maintained the same energy efficiency as compared to a conventional fluidized bed.

There has been a limited amount of work examining the kinetics of drying of particles in

fluidized beds and many questions remain unanswered. In batch fluidized bed drying

operations, four parameters have been considered vital for the drying performance: 1) the

initial moisture content, 2) the inlet heating air temperature, 3) the inlet heating air mass

flow rate and 4) the material loading in the dryer; these four parameters have been

considered in this work. Two batch type fluidized bed dryers with different geometry and

size were studied. The effect of operating parameters on drying was analyzed using

normal plots and Pareto charts. Our work serves to characterize the performance of these

two fluidized bed dryer systems and to provide physical insight into the fundamentals of

drying of porous materials.

2. EXPERIMENT SETUP

2.1 Materials And Methods

7
Dibasic calcium phosphate anhydrous (DCPA) particles were purchased from Fuji

Chemical Industries (USA). DCPA is a common excipient in pharmaceutical

manufacturing. The average surface area of the DCPA material is 40 m2/g (data provided

by Fuji Chemical Industries). The pore volume of the DCPA material was analyzed by

the N2 adsorption and desorption method (TriStar 3000, Micrometrics, USA). The pore

volume of the DCPA material was found to be 0.19 cm3/g. The mean size of the particles

is 120 μm. Fig. 1 shows SEM micrographs for the DCPA particles, which have a fairly

spherical shape.

2.2 Fluidized Bed Apparatus

Two commercial lab-scale batch-type fluidized bed dryers, Glatt GPCG-1 and MiniGlatt

5 (Glatt Air Techniques Inc., Ramsey, NJ, USA), were used to investigate the effect of

operating conditions on the drying kinetics. The geometries of the two fluidized bed

dryers are shown in Fig. 2. The conical drying chambers of the two dryers are made of

stainless steel. The chamber of the GPCG-1 (a) is 58 cm in height, and the diameter of

the top and the bottom are 15 cm and 30 cm, respectively. The cone entrance angle is 15°.

For the MiniGlatt (see Fig 2b), the chamber is 11 cm in height, and the diameter of the

top and the bottom are 14 cm and 6.6 cm respectively with a cone entrance angle of 37°.

The chamber volume is 23 L for the GPCG-1 and 3 L for the MiniGlatt. Compared to the

MiniGlatt dryer, the GPCG-1 dryer has a smaller relative change in radius along its

vertical axis. This leads to a relatively uniform air flow, whereas the drastic change of air

flow at the bottom of the MiniGlatt dryer results in a spout-like circulation pattern of

powders. The product temperature was measured using a thermocouple placed 2.5 cm

8
above the distributors and was immersed in the bed. A two-phase nozzle was placed in

the middle of the chamber which was used to spray atomized water into the dry powders.

Fig. 3 shows the full-cone nozzle spraying the mist. We estimate the size of the water

droplets to be about 50 microns while the DCPA particles have a mean size of

120 microns. The average DCPA particle volume was approximately 14 times larger than

the water droplet volume. A sample thief at the wall of the chambers allows sampling in

order to measure the moisture content during the drying process. A thermocouple that

allows the monitoring of inlet air temperature was attached below the distributor. Hot air

was pretreated to remove moisture through an air desiccant dryer (manufactured by Van

Air Inc., Lake City, PA, USA), and then supplied from the bottom. Filters with fine holes

were placed above the chambers to prevent the loss of particles. In order to prevent filter

blockage, the Glatt system has an automatic shaking function that was applied to the

frame of the filter. A window at the side wall of the chamber enables one to monitor the

situation inside by visual observation. Fig 2c and fig. 2d show the actual experimental

setup of the GPCG-1 and the MiniGlatt, respectively.

2.3 Operating Conditions

Prior to drying, a pre-determined amount of DCPA material was weighed and pretreated

in an oven at 80 °C for 2 days. The DCPA was then placed in the fluidized bed vessel and

the process was divided into two stages: the wetting stage and the drying stage. During

the wetting stage, cold air at room temperature flows upwards through the bed to fluidize

the dry material. Atomized water was sprayed at 20 ml/min into the bed until a pre-

determined moisture content was achieved. The drying stage was initiated immediately

9
after the wetting stage. Approximately 8 g of powder was sampled every 5 minutes to

measure the moisture content based on LOD during the drying stage. The end-point of

drying was achieved once the product moisture content dropped below 1 %.

Parametric analysis was carried out using factorial experimental design. For both dryers,

four factors were taken into consideration: 1) the initial moisture content, 2) the inlet

heating air temperature, 3) the inlet heating air mass flow rate and 4) the material loading

in the dryer. The experimental design is shown in tables 1 and 2.

The design of experiments (DOE) was conduct using Minitab statistical software. The

resulting Pareto charts and normal plots were also analyzed using Minitab. For the set of

experiment using the MiniGlatt dryer, two different loadings, 0.8 kg and 0.5 kg, were

investigated. The initial moisture content ranges from 10 % to 20 % (measured as mass of

liquid divided by mass of solid). The inlet air temperature has a high level of 80 °C and a

low level of 60 °C. The inlet air flow should exceed the minimum fluidization velocity in

order to achieve fluidization. A low level of 10 m3/h was just enough to keep the

fluidization state and a high level of 15 m3/h was chosen for comparison. Similarly, for

the GPCG-1, loadings of 3.0 kg and 4.0 kg were investigated. The same initial moisture

content as the MiniGlatt was adopted (i.e. 10 % and 20 %) for the GPCG-1. As the

GPCG-1 has much larger loadings, requiring significantly more drying time, the heating

air temperature was set at 80 °C and 100 °C. The inlet air flow was also elevated to

40 m3/h and 54 m3/h for the GPCG-1.

10
 3. RESULTS AND DISCUSSION

The drying process can be divided into three different drying stages: the preheating stage,

the constant-rate stage and the falling rate stage[40-43]. At the start of drying, the

particles are covered with excess liquid. As dry heating air flows through the material bed,

the liquid on the droplet surface is heated and evaporation starts. In this preheating stage,

the drying rate keeps increasing until the wet-bulb temperature of the drying air is

reached within a short period of time. During the constant-rate stage, the system reaches a

steady state. The drying rate is independent of the drying time which can be described

with zero order kinetics. As soon as the exterior moisture has been evaporated and the

moisture content decreases to a critical value, the drying front penetrates the droplet

surface and the drying rate drops significantly. The cause of the falling-rate stage is that

convection no longer dominates the drying process. Instead, capillary forces are

introduced as the outside surface turns into dry porous crust and water is drawn through

the pores from the wet core. It has been reported that for particle diameters greater than

2 mm, a linear curve could be used to describe the falling-rate stage[44].

The factors that affect drying in a fluidized bed are typically related to the initial moisture

content of the materials, the heating temperature of the inlet air, the flow rate of the inlet

air and the amount of material that is loaded in the chamber. Due to the different dryer

geometries, the effect of the above parameters on the drying process may vary at different

dryer scales. In this study, we compare the performance of the Glatt GPCG-1 and

MiniGlatt. The effect of each operating parameter on drying was examined. Furthermore,

the interaction of different parameters are studied using Pareto charts and normal plots.

11
At the end of this section, an energy efficiency analysis is introduced to investigate the

overall drying behavior which could lead to a deeper fundamental understanding of

drying processes.

3.1 Reproducibility Of The Drying Process

Due to time and material costs, the experiments were not replicated for every condition.

Preliminary experiments were conducted to ensure reproducibility of the drying process

in the GPCG-1 and the MiniGlatt. Fig. 4a shows 3 kg of DCPA with 20 % initial

moisture dried in the GPCG-1 fluid bed dryer at 80 °C. Three samples were taken at 5

second intervals for each data point displayed and error bars showing the standard

deviation are plotted in Fig. 4a; the error bars are less than or equal to the size of the data

points in Fig. 4a. Since the fluid bed is a well-mixed system, the samples at the thief

position should be representative of the entire fluid bed. The standard deviation for each

data point is very small, with the largest value less than 0.58 % of moisture, indicating

that the moisture content of the powder throughout the fluidized bed dryer is almost the

same for a given time point. Therefore, the drying process was not sample dependent in

the GCPG-1. For the MiniGlatt, taking three samples for each data point would result in a

loss of 10 % of the loaded materials, which may have impacted the drying results. Instead,

the drying process was replicated three times for the same operating condition. Fig. 4b

shows 0.5 kg of DCPA with 25 % initial moisture dried in the MiniGlatt fluid bed dryer

at 60 °C. All three of the drying curves in Fig. 4b overlap to a large extent. The results

show good batch to batch

12
3.2 Drying Of DCPA In The GPCG-1

First, the effect of the initial moisture content on drying was investigated as shown in Fig.

5. The bed temperature was recorded every 3 minutes (Fig. 5b) as it is an important

parameter for determining the drying status. In this work, the moisture content describes

the mass of water per mass of dry powder (Fig. 5a). The drying rate used in this study is

defined as the loss of water (kg) per 1 kg of dry powder per minute, which can be

calculated from the following equation

Mt t Mt
R
t

where R is the drying rate, and M is the moisture content of the materials at drying time t.

The three drying stages can be easily distinguished from the drying rate curves (Fig. 5c).

During the preheating stage, the bed temperature rises quickly, as shown in Fig. 5b, from

room temperature (18 °C) to around 30 °C in 6 minutes when drying at 80 °C. The

moisture content of the materials didn’t change significantly as the drying rate is slow

during the preheating stage. During the constant rate stage, the bed temperature remained

constant (Fig. 5b) and the moisture content dropped linearly with time, (Fig. 5a) with the

drying rate at its maximum (Fig. 5c). For higher initial moisture content, longer drying

times are required and thus longer constant rate stages were observed (Fig 5a and b).

When the moisture content dropped to around 1 %, the drying rate decreased significantly

to a very low level. In this falling rate stage, the bed temperature increased (Fig 5b) as the

supplied heat was no longer being used for evaporation.

13
The effect of drying temperature was studied with two levels of initial moisture content

and the results are shown in Fig. 6: panels a and b were obtained at 10 % initial moisture,

and c and d were obtained at 20 % initial moisture. In Fig. 6a the moisture content as a

function of time is shown for three drying temperatures and an initial moisture content of

10 %. Slightly difference among three curves is observed due to the experimental error.

Corresponding bed temperatures are shown in Fig. 6b. It can be seen that most of the

drying takes place in the constant rate stage with the moisture content dropping linearly

with time (Fig. 6a) and the bed temperature remaining constant (Fig. 6b). As before, once

moisture contents fall below ~1 % the bed temperature starts to increase (Fig. 6b). As the

temperature increases the drying rate increases (Fig 6a) and the bed temperature increases

(Fig. 6b). Similar results can be seen for an initial moisture content of 20 % in Fig. 6c and

6d. However, it can be seen that the drying time is greatly increased with an increase in

the initial moisture content.

The effect of air flux was studied with two levels of initial moisture content and the

results are shown in Fig. 7: panels a and b were obtained at 10 % initial moisture, and c

and d were obtained at 20 % initial moisture. In Fig. 7a and Fig. 7c the moisture content

as a function of time is shown for two air fluxes and an initial moisture content of 10 %

and 20 % respectively. Corresponding bed temperatures are shown in Figs. 7b and 7d.

As observed previously, it can be seen that most of the drying takes place in the constant

rate stage with the moisture content dropping linearly with time (Figs. 7a and 7c) and the

bed temperature remaining constant (Fig. 7b and 7d). During the constant rate stage, the

14
same temperature is observed in the bed for the two different air fluxes (see Figs. 7b and

7d).

The effect of initial loading was also studied with two levels of initial moisture content

and the results are shown in Fig. 8: panels a and b were obtained at 10 % initial moisture,

and c and d were obtained at 20 % initial moisture. In Fig. 8a and Fig. 8c the moisture

content as a function of time is shown for two initial loadings and an initial moisture

content of 10 % and 20 % respectively. Corresponding bed temperatures are shown in

Figs. 8b and 8d. As observed previously, it can be seen that most of the drying takes

place in the constant rate stage. In addition, the larger loading of material decreases the

rate of drying.

When the same drying temperature was applied, with different air fluxes (Fig. 7) and

loadings (Fig. 8), the bed temperature profiles have similar shapes, indicating that for the

ranges we have examined a change of scale or air flux does not change the general drying

behavior. Fig. 9 shows the effect of initial moisture on the end-point drying time for the

different operating conditions in Figures 6, 7 and 8. The end point drying time is taken to

be the time needed to reduce the moisture content to 1 %. Figs. 9a and 9b show the

impact of initial moisture content on the total drying time for different drying

temperatures. As the initial moisture content increased from 6 % to 10 % and 20 %, the

resulting total drying time increased accordingly. For 6 % initial moisture, the total

drying time dropped from 30 minutes at 60 oC to 24 minutes at 80 oC and to 18 minutes

at 100 oC. Similar phenomena were observed for 10 % and 20 % initial moisture. In

15
general, one can see that there is a fairly linear relationship between drying time and

initial moisture content. The drying time ratio and the moisture content ratio shown in Fig.

9b were calculated as follows. The drying time ratio is defined as the ratio of the drying

time at a given initial moisture content to the drying time at a 20 % initial moisture

content for DPCA powders under the same drying conditions. The moisture content ratio

is defined as the ratio of a given initial moisture content to the 20 % initial moisture

content. It is observed in Fig 9b that the drying time ratio is proportional to the initial

moisture content ratio, which indicates that the constant rate stage (zero order kinetics)

dominates the drying process. Furthermore, the drying time ratio didn't vary much for

drying at 60/80/100 °C – all three points lie nearly on top of one another (Fig 9b).

Therefore, the interaction between drying temperature and initial moisture content can be

neglected.

Figs. 9c and 9d illustrate the impact of initial moisture content on total drying time for

different material loadings. Once again one can see that there is a fairly linear

relationship between drying time and initial moisture content. Higher loading and initial

moisture content leads to higher drying time. The drying time ratio of different loadings

are 0.72, 0.67 and 0.80 for 6 %, 10 % and 20 % initial moisture content, respectively,

which is comparable to the loading ratio, indicating that the drying time is proportional to

the initial material loading.

The effects of initial moisture content, drying temperature, air flow rate and loading on

the drying process were further confirmed with the Pareto chart of effects and normal plot

16
shown in Fig. 10. Pareto analysis [45] can help determine the factors that are responsible

for the drying process. The drying time was regarded as the response for drying in the

Pareto analysis as shown on the horizontal axis in Fig. 10a. The listed factors on the

vertical axis represent either a single operating condition or the interaction of several

conditions. The confidence level for all intervals was set at 97.5 % (α = 0.025). Minitab

uses Lenth’s pseudo-standard error (PSE) for the error term, which assumes that random

error caused the variation in the smallest effects. The vertical red reference line indicates

the significance of factors to the drying process. It can be observed that the initial

moisture content dominates the drying time as it has the largest effect. The air flow rate

and the loading have a smaller effect on drying compared to initial moisture content. The

effect of drying temperature is the least important factor, however it is still significant.

None of the factor interactions have a significant effect on the drying time. The normal

plot is typically used to examine the distribution of data. In this study, the normal

probability plot can be interpreted as an indication of whether the operating parameters

affect the process positively or negatively. The fitted line indicates the position of the

data point if the effect of a factor is zero. The points on the right indicate a positive effect

and those on the left a negative effect. Thus, the further a point is from the fitted line, the

larger the magnitude and statistical significance of the effect on drying time. It can be

observed from the plot that increasing the initial moisture content and material loading

will increase the drying time while increasing temperature and air flow rate will decrease

the drying time. It can be concluded that all four parameters are independent of each

other and have a significant effect on the end-point of the required drying time. The

interaction effects were not found to be significant. The Pareto chart and the normal plot

17
not only allow for a quantitative understanding of the effect of different parameters on

drying but also provide a potential method of optimizing the drying process.

3.3 Drying Of DCPA In Miniglatt

In the MiniGlatt study, the material loading was reduced to 500 g and 800 g, which was

approximately 1/8 and 1/4 of the loading in Glatt GPCG-1 fluidized bed dryer. The inlet air

flow rate was 1/4 and 1/3 of that used in GPCG-1. The heating temperature and initial

moisture content were adopted at the same level as before: 60 °C/80 °C and 10 %/20 %.

The effect of initial moisture content was investigated first in the MiniGlatt. In Fig. 11,

we present results for 0.5 kg of material with different initial moisture contents (20 % and

10 %) that were dried at 80 °C with an inlet air flux of 15 m3/h. As observed in the

GPCG-1, the three drying stages can be easily distinguished from the drying rate curves

(Fig. 11c). The system reached the constant-rate stage very quickly as the bed

temperature increased from 13 °C to 24 °C (Fig. 11b). During the constant rate stage, the

bed temperature remained constant (Fig. 11b), and the moisture content dropped linearly

with time (Fig. 11a) with the drying rate at its maximum (Fig. 11c). For material with a

higher initial moisture content, longer drying times are required and thus a longer

constant rate stage was observed for the 20 % initial moisture sample (Fig. 11a and b).

When the moisture content dropped to around 1 %, the drying rate decreased significantly.

In this falling rate stage, the bed temperature increased (Fig. 11b) as the supplied heat

was no longer being used for evaporation. When the initial moisture content in the

18
material was lower, both MiniGlatt and GPCG-1 demonstrated a significant drying time

reduction.

The effect of drying temperature was studied with two levels of initial moisture content

and the results are shown in Fig. 12: panels a and b were obtained at 10 % initial moisture,

and c and d were obtained at 20 % initial moisture. In Fig. 12a the moisture content as a

function of time is shown for two drying temperatures and an initial moisture content of

10 %. Corresponding bed temperatures are shown in Fig. 12b. It can be seen that most of

the drying takes place in the constant rate stage with the moisture content dropping

linearly with time (Fig. 12a) and the bed temperature remaining constant (Fig. 12b). As

before, once moisture contents fall below ~1 % the bed temperature starts to increase (Fig.

12b). As the temperature increases the drying rate increases (Fig. 12a) and the bed

temperature increases (Fig. 12b). Similar results can be seen for an initial moisture

content of 20 % in Fig. 12c and 12d. Once again, it can be seen that the drying time is

greatly increased with an increase in the initial moisture content. Increasing the drying

temperature from 60 °C to 80 °C has only a small effect on the drying time.

The effect of air flux was studied with two levels of initial moisture content and the

results are shown in Fig. 13: panels a and b were obtained at 10 % initial moisture, and c

and d were obtained at 20 % initial moisture. In Fig. 13a and Fig. 13c the moisture

content as a function of time is shown for two air fluxes and an initial moisture content of

10 % and 20 % respectively. Corresponding bed temperatures are shown in Figs. 13b and

13d. As observed previously, it can be seen that most of the drying takes place in the

19
constant rate stage with the moisture content dropping linearly with time (Figs. 13a and

13c) and the bed temperature remaining constant (Fig. 13b and 13d). During the constant

rate stage, the same temperature is observed in the bed for the two different air fluxes

(see Figs. 13b and 13d). Increasing the air flux from 10 m3/h to 15 m3/h has a significant

effect on the drying time.

The effect of initial loading was also studied with two levels of initial moisture content,

and the results are shown in Fig. 14: panels a and b were obtained at 10 % initial moisture,

and c and d were obtained at 20 % initial moisture. In Fig. 14a and Fig. 14c the moisture

content as a function of time is shown for two initial loadings and an initial moisture

content of 10 % and 20 % respectively. Corresponding bed temperatures are shown in

Figs. 14b and 14d. As observed previously, it can be seen that most of the drying takes

place in the constant rate stage. Unlike the GPCG-1, higher material loading in the

MiniGlatt did not extend the drying time.

Pareto analysis was also carried out for the MiniGlatt. As shown in Fig. 15, the effect of

material loading on the drying time was found to be no longer significant. Unlike the

GPCG-1, moisture content and flowrate were found to have an equivalent effect on the

drying time. Interestingly, the factor interaction between moisture content and flowrate

was found to be significant for the MiniGlatt. As with the GPCG-1, increasing the

heating air temperature from 60 °C to 80 °C still only had a small effect on the drying

time.

20
3.4 Energy Efficiency Analysis

In this study, energy efficiency is defined as the ratio of energy used for water

evaporation to the input energy. The energy efficiency can thus be calculated as follows:

energy used for water evaporation

moisturecontent material loading heat of evaporation

For a given initial moisture content and material loading, both the heating temperature

and air flux can be adjusted to affect the drying process. As discussed in the previous

section, increasing the heating temperature only had a small effect on the drying time. In

order to compare different cases, we define the energy efficiency ratio as the ratio of

energy efficiencies for two different operating conditions. In Fig. 16 we show the energy

efficiency ratio for the GPCG-1 for different temperatures and moisture contents (Fig.

16a) and different air fluxes and moisture contents (Fig. 16b). Analogous results are

shown in Fig. 17 for the MiniGlatt. As shown in Fig. 16a and Fig. 17a, the energy

efficiency for drying at 60 °C is around 30 % higher than drying at 80 °C for both the

GPCG-1 and MiniGlatt indicating that a relatively low heating temperature is favorable

in terms of energy efficiency. Increasing the drying temperature from 80 °C to 100 °C

yields a decrease in the energy efficiency (Fig. 16a) for the GPCG-1. By increasing the

air flux, the energy efficiency is increased by a moderate amount as shown in Fig. 16b

and Fig.17b. Therefore, in order to achieve a high energy efficiency profile for the

parameter ranges we have examined, low temperature and high air flux should be used in

the drying process.

21
A mass and heat balance was carried out at the external surface of the wet particles

k cg ,in cg ,out H vap h T Tbed

where c g ,in and cg ,out are the inlet and outlet moisture content of the air, T is the ambient

drying temperature, Tbed is the bed temperature and h and k are the heat and mass transfer

coefficients, respectively. H vap is the latent heat of water vapor which is 2260 kJ/kg

[46]. Therefore, h/k represents the energy efficiency coefficient which can be expressed

as

h cg ,in cg ,out H vap

h/k describes the amount of moisture removed per unit of energy consumed for a given

effective surface area. Thus higher values of h/k indicate higher energy efficiencies. The

effect of the operating parameters on the energy efficiency coefficient for the MiniGlatt is

shown in Fig. 18. The h/k profiles of drying were similar to that of the drying rate curve.

Three stages can be observed: the energy efficiency coefficient increased slowly during

the preheating stage, then reached a maximum value during the constant rate stage and

finally dropped quickly to almost zero during the falling rate stage.

Fig. 18 illustrates the h/k profile for different material loadings. In the smaller MiniGlatt,

different loadings were found to have no effect on drying time, indicating that unless the

scale of the fluidized bed drying is changed and the resulting fluidization behavior of the

materials is changed during the drying, material loading does not affect the energy

efficiency coefficient. It is often difficult to characterize the fluidization status during

22
drying since the drop of moisture content will promote fluidization [25]. High loadings

can potentially increase the risk of defluidization, clumping of powders or other

undesired situations when the air flux is insufficient. Similar results were found when

different temperatures, air flux and initial moisture content were applied (figures not

shown here). Increasing the drying temperature decreases the drying times slightly,

however, the h/k ratio dropped noticeably at the higher temperature for all three drying

stages. Therefore, if energy utilization is of prime concern, drying should be kept at lower

temperatures to maximize the energy efficiency. The energy efficiency coefficient for the

constant rate stage increased up to 50 % when the air flux increased from 10 m3/h to

15 m3/h. High air flux not only helps promote mass transfer between the gas and liquid

phase, achieving a higher energy efficiency ratio, but also reduces the drying time so that

less energy is consumed overall. The initial moisture content is a measure of the amount

of water/solvent in the material, and therefore it is not surprising that it did not have an

effect on the energy efficiency coefficient. However, since the maximum energy

efficiency coefficient is reached during the constant rate stage, drying materials with a

higher initial moisture content results in a larger proportion of the drying process to occur

at the maximum energy efficiency coefficient.

3.5 Comparison Of GPCG-1 And Miniglatt

The GPCG-1 and the MiniGlatt have different bed geometries. The GPCG-1 is a

cylindrical fluidized bed dryer, providing a homogenous air flow profile along the

vertical axis. The MiniGlatt has a lower conical section in the body chamber. The inlet air

flow is much higher at the bottom than that at the top due to the cross-sectional area

23
difference. Spouted fluidization is usually observed in this type of geometry[37]. For the

two fluidized bed dryers, the initial moisture content, the drying temperature and the air

flux show a significant effect on the drying process. The initial loading is important to the

GPCG-1, however, for the MiniGlatt the effect is not significant. This could be due to the

geometry differences of the two fluidized bed dryers. Tanfara et al.[47] discovered that in

a conical fluidized bed dryer, a centralized core flow pattern could be observed at a high

air flow rate. Increasing the loading of materials will enhance the centralized flow pattern

and results in a better circulation of powders in the conical body chamber. Therefore, in

the MiniGlatt, the powders have a better chance to be in contact with the drying air.

4. CONCLUSION

The impact of the operating parameters on the drying time and energy efficiency of the

drying process were investigated in this study via parametric analysis. Initial moisture

content, material loading, heating air temperature and air flux were examined for their

effect on drying. Two fluidized bed dryers, the Glatt GPCG-1 and MiniGlatt, the former

one with a more cylindrical-like chamber (with respect to a smaller entrance angle) and

the latter one conical, were studied using DCPA powders. The constant rate stage was

found to dominate the drying process for all process conditions. Through Pareto analysis,

the effects of all parameters were reviewed quantitatively. For the small dryer, the initial

moisture content and air flux affected the drying time significantly for the conditions we

examined. For the larger dryer, the effect of loading was as important as air flux for the

present study. It was found that for both sizes of fluidized bed dryers, the heating air

temperature, though not insignificant, had the least impact on drying time for the

24
conditions we examined. For the GPCG-1, the interaction effects were found not to be

important. However, in the small scale MiniGlatt dryer, the interaction of the initial

moisture content and the air flux could not be neglected, which should be further

examined.

Energy efficiency analysis was carried out based on a simple mass and heat balance. It

was found that drying had a higher energy efficiency for low temperatures and high air

flux for the conditions we examined. The energy efficiency coefficient h/k was evaluated

by studying the mass and heat balance at the surface of the particles. The energy

efficiency coefficient is low during the preheating stage as heating air is used to raise the

bed temperature. During the constant rate stage, h/k reaches its maximum value before

the moisture content drops to a critical level. In the falling rate stage, h/k drops to almost

zero as capillary forces start to dominate the drying process. Higher initial moisture

content was found to have a higher average energy efficiency as drying remains in the

constant rate stage, and thus has the maximum h/k value, for a larger proportion of the

process.

Our work serves to characterize the performance of fluidized bed dryer systems and

provide physical insight into the fundamentals of drying of porous materials. At the same

time our results are for a limited number of drying conditions, two pieces of equipment

and only one material. Further work is needed to see if the results we have obtained here

can be generalized to other materials and drying conditions. It would also be of interest to

25
examine other fluidized bed dryers to see how equipment changes might affect the drying

process.

REFERENCES

[1] Soponronnarit, S.; Pongtornkulpanich, A.Prachayawarakorn, S. Drying

characteristics of corn in fluidized bed dryer. Drying Technology 1997, 15 (5), 1603-

1615.

[2] Meziane, S. Drying kinetics of olive pomace in a fluidized bed dryer. Energy

Conversion and Management 2011, 52 (3), 1644-1649.

[3] Sarker, M.S.H.; Ibrahim, M.N.; Aziz, N.A. et al. Drying kinetics, energy

consumption, and quality of paddy (mar-219) during drying by the industrial inclined bed

dryer with or without the fluidized bed dryer. Drying Technology 2013, 31 (3), 286-294.

[4] Zielinska, M.Markowski, M. Drying behavior of carrots dried in a spout–

fluidized bed dryer. Drying Technology 2007, 25 (1), 261-270.

[5] Sarker, M.S.H.; Ibrahim, M.N.; Abdul Aziz, N. et al. Energy and exergy analysis

of industrial fluidized bed drying of paddy. Energy 2015, 84, 131-138.

[6] Mujumdar, A.S. Research and development in drying: Recent trends and future

prospects. Drying Technology 2004, 22 (1/2), 1-26.

[7] Tatemoto, Y.; Mizukoshi, R.; Ehara, W. et al. Drying characteristics of food

materials injected with organic solvents in a fluidized bed of inert particles under reduced

pressure. Journal of Food Engineering 2015, 158, 80-85.

[8] Srinivas, G.Pydi Setty, Y. Drying behavior of uniform and binary mixture of

solids in a batch fluidized bed dryer. Powder Technology 2013, 241, 181-187.

26
[9] Hematian, S.Hormozi, F. Drying kinetics of coated sodium percarbonate particles

in a conical fluidized bed dryer. Powder Technology 2015, 269, 30-37.

[10] Emrah Özahi, H.D. Drying performance analysis of a batch type fluidized bed

drying process for corn and unshelled pistachio nut regarding to energetic and exergetic

efficiencies. Measurment 2015, 60, 85-96.

[11] Türker, M.; Kanarya, A.; Yüzgeç, U. et al. Drying of baker's yeast in batch

fluidized bed. Chemical Engineering and Processing: Process Intensification 2006, 45

(12), 1019-1028.

[12] Wan Daud, W.R. Fluidized bed dryers — recent advances. Advanced Powder

Technology 2008, 19 (5), 403-418.

[13] Assari, M.R.; Basirat Tabrizi, H.Najafpour, E. Energy and exergy analysis of

fluidized bed dryer based on two-fluid modeling. International Journal of Thermal

Sciences 2013, 64, 213-219.

[14] Ramachandran, R.; Poon, J.M.H.; Sanders, C.F.W. et al. Experimental studies on

distributions of granule size, binder content and porosity in batch drum granulation:

Inferences on process modelling requirements and process sensitivities. Powder

Technology 2008, 188 (2), 89-101.

[15] Briens, L.Bojarra, M. Monitoring fluidized bed drying of pharmaceutical

granules. AAPS PharmSciTech 2010, 11 (4), 1612-1618.

[16] Hegedűs, Á.Pintye-Hódi, K. Comparison of the effects of different drying

techniques on properties of granules and tablets made on a production scale.

International Journal of Pharmaceutics 2007, 330 (1–2), 99-104.

27
[17] Ebrahimi, A.; Saffari, M.Langrish, T. Developing a new production process for

high-porosity lactose particles with high degrees of crystallinity. Powder Technology

2015, 272, 45-53.

[18] Yazdanpanah, N.Langrish, T.A.G. Egg-shell like structure in dried milk powders.

Food Research International 2011, 44 (1), 39-45.

[19] Briens, L.Bojarra, M. Monitoring fluidized bed drying of pharmaceutical

granules. AAPS PharmSciTech 2010, 11 (4), 1612-1618.

[20] Heigl, N.; Koller, D.M.; Glasser, B.J. et al. Quantitative on-line vs. Off-line nir

analysis of fluidized bed drying with consideration of the spectral background. European

Journal of Pharmaceutics and Biopharmaceutics 2013, 85 (3, Part B), 1064-1074.

[21] Emrah Özahi, H.D. A model for the thermodynamic analysis in a batch type

fluidized bed. Energy 2013, 59, 617-624.

[22] Mustafa T¨urker, A.K., U˘gur Y¨uzgec¸, Hamdi Kapucu, Zafer S¸ enalp Drying

of baker’s yeast in batch fluidized bed. Chemical Engineering and Processing 2006, 45,

1019-1028.

[23] Séverine Thérèse F.C. Mortier, T.D.B., Krist V. Gernaey, Jean Paul Remon, Chris

Vervaet, Ingmar Nopens Mechanistic modelling of fluidized bed drying processes of wet

porous granules: A review. European Journal of Pharmaceutics and Biopharmaceutics

2011, 79, 205-225.

[24] Reyes, A.; Alvarez, P.I.Marquardt, F.H. Drying of carrots in a fluidized bed. I.

Effects of drying conditions and modelling. Drying Technology 2002, 20 (7), 1463.

28
[25] Mortier, S.T.F.C.; De Beer, T.; Gernaey, K.V. et al. Mechanistic modelling of

fluidized bed drying processes of wet porous granules: A review. European Journal of

Pharmaceutics and Biopharmaceutics 2011, 79 (2), 205-225.

[26] Immanuel, C.D.Doyle Iii, F.J. Solution technique for a multi-dimensional

population balance model describing granulation processes. Powder Technology 2005,

156 (2–3), 213-225.

[27] Hounslow, M.J.; Pearson, J.M.K.Instone, T. Tracer studies of high-shear

granulation: Ii. Population balance modeling. AIChE Journal 2001, 47 (9), 1984-1999.

[28] Pearson, J.M.K.; Hounslow, M.J.Instone, T. Tracer studies of high-shear

granulation: I. Experimental results. AIChE Journal 2001, 47 (9), 1978-1983.

[29] Peglow, M.; Kumar, J.; Heinrich, S. et al. A generic population balance model

for simultaneous agglomeration and drying in fluidized beds. Chemical Engineering

Science 2007, 62 (1–2), 513-532.

[30] Peglow, M.; Cunäus, U.Tsotsas, E. An analytical solution of population balance

equations for continuous fluidized bed drying. Chemical Engineering Science 2011, 66

(9), 1916-1922.

[31] Lee, D.-H.Kim, S.-D. Mathematical model for batch drying in an inert medium

fluidized bed. Chemical Engineering & Technology 1999, 22 (5), 443-450.

[32] Özahi, E.Demir, H. A model for the thermodynamic analysis in a batch type

fluidized bed dryer. Energy 2013, 59, 617-624.

[33] S. Syahrul, I.D., F. Hamdullahpur Thermodynamic modeling of fluidized bed

drying of moist particles. International Journal of Thermal Sciences 2003, 42, 691-701.

29
[34] Zhao, P.; Zhao, Y.; Luo, Z. et al. Effect of operating conditions on drying of

chinese lignite in a vibration fluidized bed. Fuel Processing Technology 2014, 128, 257-

264.

[35] Midilli, A.; Kucuk, H.Yapar, Z. A new model for single-layer drying. Drying

Technology 2002, 20 (7), 1503-1513.

[36] Liu, X.; Muzzio, F.J.; Glasser, B.J. et al. Fluidized bed drying of pharmaceutical

materials: Moisture measurement and effects of particle size. American Pharmaceutical

Review 2013, 16 (2)

[37] Altzibar, H.; Lopez, G.; Alvarez, S. et al. A draft-tube conical spouted bed for

drying fine particles. Drying Technology 2008, 26 (3), 308-314.

[38] Michael Wormsbecker, R.v.O., John Nijenhuis, Helen Tanfara, Todd Pugsley

The influence of vessel geometry on fluidized bed dryer hydrodynamics. Powder

Technology 2009, 194, 115-125.

[39] Eliaers, P.De Wilde, J. Drying of biomass particles: Experimental study and

comparison of the performance of a conventional fluidized bed and a rotating fluidized

bed in a static geometry. Drying Technology 2013, 31 (2), 236-245.

[40] Mezhericher, M.; Levy, A.Borde, I. Modelling of particle breakage during drying.

Chemical Engineering and Processing: Process Intensification 2008, 47 (8), 1404-1411.

[41] Mezhericher, M.; Levy, A.Borde, I. Theoretical drying model of single droplets

containing insoluble or dissolved solids. Drying Technology 2007, 25 (6), 1025-1032.

[42] Lekhal, A.; Glasser, B.J.Khinast, J.G. Impact of drying on the catalyst profile in

supported impregnation catalysts. Chemical Engineering Science 2001, 56 (15), 4473-

4487.

30
[43] Liu, X.; Khinast, J.G.Glasser, B.J. A parametric investigation of impregnation

and drying of supported catalysts. Chemical Engineering Science 2008, 63 (18), 4517-

4530.

[44] Langrish, T.A.G.Kockel, T.K. The assessment of a characteristic drying curve for

milk powder for use in computational fluid dynamics modelling. Chemical Engineering

Journal 2001, 84 (1), 69-74.

[45] Montgomery, D.C. Design and analysis of experiments; Wiley, 2012.

[46] Çengel, Y.A., and Michael A. Boles Thermodynamics: An engineering

approach. ; McGraw-Hill: Boston, 2011.

[47] Tanfara, H.; Pugsley, T.Winters, C. Effect of particle size distribution on local

voidage in a bench-scale conical fluidized bed dryer. Drying Technology 2002, 20 (6),

1273.

31
Table 1.

Tes Conditions Te Conditions

t# Loadi Initial Heating Flow st # Loadi Initial Heating Flow

ng Moistu Temperat Rate ng Moistu Temperat Rate

(kg) re (%) ure (°C) (m3/h (kg) re (%) ure (°C) (m3/h

r) r)

A1 0.8 20 % 80 °C 15 B1 0.5 20 % 80 °C 15

A2 0.8 20 % 60 °C 15 B2 0.5 20 % 60 °C 15

A3 0.8 20 % 80 °C 10 B3 0.5 20 % 80 °C 10

A4 0.8 20 % 60 °C 10 B4 0.5 20 % 60 °C 10

A5 0.8 10 % 80 °C 15 B5 0.5 10 % 80 °C 15

A6 0.8 10 % 60 °C 15 B6 0.5 10 % 60 °C 15

A7 0.8 10 % 80 °C 10 B7 0.5 10 % 80 °C 10

A8 0.8 10 % 60 °C 10 B8 0.5 10 % 60 °C 10

32
Table 2.

Test Conditions

# Loading Initial Moisture Heating Temperature Flow Rate

(kg) (%) (°C) (m3/hr)

A1 3 20 % 60 °C 40

A2 3 20 % 80 °C 40

A3 3 20 % 80 °C 54

A4 4 20 % 80 °C 40

A5 3 20 % 100 °C 40

B1 3 10 % 60 °C 40

B2 3 10 % 80 °C 40

B3 3 10 % 80 °C 54

B4 4 10 % 80 °C 40

B5 3 10 % 100 °C 40

C1 3 6% 80 °C 40

C2 3 6% 100 °C 40

C3 3 6% 60 °C 40

C4 3 6% 80 °C 54

C5 4 6% 80 °C 40

33
Figure 1

34
Figure 2

35
Figure 3

36
Figure 4

37
Figure 5

38
Figure 6

39
Figure 7

40
Figure 8

41
Figure 9

42
Figure 10

43
Figure 11

44
Figure 12

45
Figure 13

46
Figure 14

47
Figure 15

48
Figure 16

49
Figure 17

50
Figure 18

51