Empirical modeling for spray drying process of sticky and non-sticky products Academic research paper on "Agriculture, forestry, and fisheries"

Procedía

Food Science

ELSEVIER

Procedía Food Science 1 (2011) 690 - 697

The 11th International Congress on Engineering and Food (ICEF11)

Empirical modeling for spray drying process of sticky and

non-sticky products

Lee Woun Tana, Mohd. Nordin Ibrahima, Raja Kamilb, Farah Saleena Taipa*

aDepartment of Process and Food Engineering, Faculty of Engineering, Universiti Putra Malaysia, 43400 Serdang, Selangor,

Malaysia.

bDepartment of Electrical and Electronic Engineering, Faculty of Engineering, Universiti Putra Malaysia, 43400 Serdang,

Selangor, Malaysia.

Abstract

Spray drying is a common drying technique in food industries to convert liquid to powder form. A good understanding on the dynamic behavior of the process is important to ensure proper control. The aim of this study is to develop empirical models for spray drying of whole milk powder and orange juice powder using a nozzle atomizer spray dryer. Maltodextrin was used as the drying agent material in spray drying of orange juice powder as it is considered as sticky powder. A preliminary study on the effect of several inputs such as inlet air temperature, feed flow rate and maltodextrin concentration on the product quality was studied. The selection of suitable inputs is important to ensure the desired quality of final products (moisture content). It was found that inlet air temperature gave more significant effect on outlet air temperature and powder moisture compared to other two inputs. Inlet air temperature and outlet temperature were selected as the manipulated variable and controlled variables respectively. Empirical models were developed by applying step change in the inlet air temperature. For spray drying of orange juice powder, its response was faster to achieve steady state because maltodextrin inhibited sticky behaviour. Both empirical models can be represented by first order plus time delay (FOPTD) and valid because R2>0.6.

© 2011 Publishedby ElsevierB.V.Selectionand/orpeer-review under responsibility of11thInternational Congress onEngineeringandFood (ICEF 11) Executive Committee.

Keywords: spray drying; empirical modeling; first order plus time delay.

* Corresponding author. Tel.: +603-89466357; fax: +603-89454440. E-mail address: saleena@eng.upm.edu.my.

2211-601X © 2011 Published by Elsevier B.V. Selection and/or peer-review under responsibility of 11th International Congress on Engineering

and Food (ICEF 11) Executive Committee.

doi:10.1016/j.profoo.2011.09.104

1. Introduction

Spray drying is a common process in food industries. It involves the transformation of liquid feed through a hot medium (air) in order to produce product in powder form. The spray dried powders have longer shelf life and resemble the quality of the original liquid feed.

The spray dried products can be classified into two groups, nonsticky and sticky. Examples of nonsticky products are dairy powders, micro-encapsulated powders and egg powders. These powders can be dried using a simple dryer and remain free flowing [1-2]. Fruit and vegetable juice powders, honey powders and lactose powders belong to sticky products. Sticky products are difficult to spray dried under normal conditions and exhibit sticky behaviour due to rich in sugars (sucrose, glucose, lactose and fructose) and organic acids (citric, malic and tartaric acid) [3-4]. In order to achieve a successful drying, high molecular weight drying agent materials such as maltodextrin has been used in spray drying of sticky products [1, 5-6].

The quality of the spray dried product depends greatly on the process parameters. Moisture content is one the effective factors in determining the product quality [7]. For nonsticky products such as dairy powders and egg powders, the final moisture content are less than 5% [8] and 4-9% [4, 8] respectively. In contrast, the final moisture content for sticky products such as fruit juice powders are around 2-4% moisture [2]. Therefore, controlling the parameters that have direct significant effect on the product quality is very important in spray drying process.

The design of a good control system rests on appropriate understanding of the dynamic behavior of spray dryer. The development of effective dynamic model is necessary and can be developed by theoretical or empirical analysis. This paper focuses on the development of dynamic models empirically for control purposes in spray drying process. The aim of this study is to develop the empirical models for non sticky product (whole milk powder) and sticky product (orange juice powder) using a nozzle atomizer spray dryer.

2. Materials and Methods

Two types of samples were used in this study. They were whole milk (nonsticky, NS) manufactured by Dutch Lady Malaysia and orange juice concentrate (sticky, S) manufactured by Barkath Co-Ro Manufacturing Sdn. Bhd., Malaysia. Whole milk was heated up to 42±2°C before feeding to the spray dryer. For spray drying of orange juice powder, maltodextrin DE 12-15 with 6% moisture content (Epic Chemicals Sdn. Bhd, Malaysia) was used as the drying agent material. An aqueous solution with 30 °Brix was prepared by dispensing maltodextrin in warm water (about 50 °C) with constant stirring. Then, this solution was added to orange juice concentrate.

A nozzle atomizer spray dryer, model Lab-Plant SD 05 Laboratory Scale Spray Dryer (L.P. Technology LTD, Huddersfield, U.K) was employed. The air flow rate was kept at 62 m3/h and the atomizer pressure was 2.1 bar. A preliminary study on the effect of several inputs on the outputs was studied. For spray drying of whole milk powder, inlet air temperature (165-185 °C) and feed flow rate (400-800 ml/h) were investigated. For spray drying of orange juice powder, inlet air temperature (130150 °C) and maltodextrin concentration (7-21%) were investigated. These experiments were carried out in triplicate. Outlet air temperature and ambient air temperature with relative humidity were recorded each 10 seconds. Relative humidity and humidity for inlet and outlet air were determined using CYTSoft Psychrometric Chart Version 2.2. The powder moisture content was determined using Ohaus MB-45 Moisture Analyzer (Ohaus Corporation, Switzerland) at 105°C. Upon investigation, the most influential input on the outlet air temperature and moisture content will be selected as the manipulated variable. Drying air parameters and the characteristics of liquid feed were observed in both processes.

For empirical modeling, the sample was fed into spray dryer at constant manipulated variable for 10 minutes, with 2.1 bar of atomizer pressure and 62 m3/h of air flow rate. Then, a step change of the

manipulated variable was applied. Outlet air temperature was recorded continuously until the process reached steady state. The experiments were carried out in three replications. The data obtained was plotted and the model obtained resembles first order process with time delay (FOPTD). The obtained model needs to undergo diagnostic evaluation and was verified with additional data before being used for process control.

3. Results and discussion

The effects of input variables on the output variables were investigated. The input variables for spray drying of whole milk powder (NS) were inlet air temperature and feed flow rate. The inputs variables for spray drying of orange juice powder (S) were inlet air temperature and maltodextrin concentration. Thus, drying air parameters and the characteristics of liquid feed were observed in both processes.

Table 1 presents the spray drying parameters and moisture content at different inlet air temperatures. This operating parameter showed similar effects for both spray dried products. Hot dry air has constant humidity and very low relative humidity. When hot air is contacted with liquid droplet, the heat required to vaporize the moisture comes from the sensible heat. It caused the reduction of inlet air temperature to outlet air temperature. As increase in the inlet air temperature resulted in significant increase in the outlet air temperature. As the outlet air temperature increased, the outlet air humidity also increased but relative humidity reduced [4]. Moreover, high inlet air temperature led to greater efficiency of heat and mass transfers. It provided greater driving force for moisture evaporation, and produced powder with low moisture content. Similar findings were observed by [9]; [10]; [11]; [12]; and [13]. Therefore, adjusting inlet air temperature could regulate outlet air temperature and it could be an indirect measurement of moisture content.

Table 1. Spray drying parameters and moisture content of powder at different inlet air temperature

(a) Whole milk powder (NS)

Inlet air Outlet air Moisture content

Temperature (°C) Temperature (°C) Humidity (kgH2O/kg dry air) Relative humidity (%) (%)

165 83.7± 0.1 0.0474± 0.0001 13.1± 0.1 4.33± 0.04

175 91.8± 1.9 0.0482± 0.0007 9.7± 0.8 3.11± 0.20

185 93.8± 0.8 0.0514± 0.0002 9.6± 0.4 2.99± 0.02

Operating conditions: air flow rate=62m3/hr and inlet air humidity=0.0131 kgH2O/kg dry air

(b) Orange juice powder (S)

Inlet air Outlet air Moisture content

Temperature (°C) Temperature (°C) Humidity (kgH2O/kg dry air) Relative humidity (%) (%)

130 83.3± 3.0 0.0298± 0.0004 7.3± 0.4 3.07± 0.23

140 91.3± 1.4 0.0335± 0.0001 7.0± 0.4 2.46± 0.10

150 95.3± 0.1 0.0359± 0.0001 6.5± 0.1 2.07± 0.08

Operating conditions: air flow rate=62m3/hr and inlet air humidity=0.0131 kgH2O/kg dry air

The effect of feed flow rate was only investigated for whole milk powder. Table 2 shows the effect of feed flow rate on the spray drying parameters and powder moisture content. As feed flow rate increased, more milk was atomized into the drying chamber and produced larger liquid droplets. Larger droplets had small surface area may reduced the outlet air temperature and hence, outlet air humidity and relative humidity became higher. Larger droplets increased the distance to travel the heat to the centre of the droplets and also increased the distance to travel the moisture from the centre of the droplets the surface. Hence, the moisture is difficult to evaporate and caused the powder cannot dry enough. Studies conducted by other researchers also confirmed these findings [9-11, 14-15].

Table 2. Spray drying parameters and moisture content of whole milk powder (NS) at different feed flow rate

Feed Flow rate (ml/h) Outlet air Moisture content

Temperature (°C) Humidity (kgH2O/kg dry air) Relative humidity (%) (%)

400 99.1± 0.7 0.0448± 0.0004 7.0± 0.1 3.06± 0.10

600 91.8± 1.9 0.0482± 0.0007 9.7± 0.8 3.11± 0.20

800 76.5± 1.2 0.0555± 0.0008 20.2± 1.3 3.49± 0.16

Operating conditions: air flow rate=62m3/hr, inlet air temperature=175°C and inlet air humidity=0.0131 kgH2O/kg dry air

For sticky products, high molecular weight drying agents such as maltodextrin are used to achieve a successful drying. Thus, the effect of maltodextrin concentration is more prominent than that of feed flow rate [9]. Orange juice with three different concentration of maltodextrin (7%, 14% and 21%) was spray dried at 140°C inlet air temperature. The feed concentration will reduce when the maltodextrin concentration added was increased. The effect of maltodextrin concentration on the process outputs is presented in Table 3. Increment in the maltodextrin concentration led to lower outlet air temperature, higher outlet air humidity and relative humidity. Goula and Adamopoulos [1] reported that the addition of maltodextrin may lower the drying rate because the moisture is difficult to diffuse in larger maltodextrin molecules. Higher maltodextrin concentration also resulted in higher feed moisture content and consequently higher product moisture content. This is probably due to the reduction of total solid in the feed.

Table 3. Spray drying parameters and moisture content of orange juice powder at different maltodextrin concentration

Maltodextrin concentration (%) Outlet air Moisture content

Temperature (°C) Humidity (kgH2O/kg dry air) Relative humidity (%) (%)

7 101.8± 2.9 0.0290± 0.0001 4.2± 0.6 1.84± 0.04

14 97.9± 1.3 0.0305± 0.0001 5.1± 0.2 2.15± 0.02

21 91.3± 1.4 0.0335± 0.0001 7.0± 0.4 2.46± 0.10

Operating conditions: air flow rate=62m3/hr, inlet air temperature=140°C and inlet air humidity=0.0131 kgH2O/kg dry air

Among the output variables, outlet air temperature was defined as controlled variable because it can be continuously measured. The final moisture content is related to the to the outlet air temperature [13, 16].

According to Tables 1, 2 and 3, inlet air temperature was the input variable that showed the greatest influence on both nonsticky and sticky powders moisture content. Besides, inlet air temperature had effects on the outlet air temperature. Inlet air temperature was the manipulated variable in order to control the outlet air temperature and also indirectly control powder moisture content.

The dynamic models for spray drying processes of whole milk powder (NS) and orange juice powder (S) were developed by carrying out step test procedure. Process reaction curve was generated in response to a step change of inlet air temperature and illustrated in Figure 1. All processes showed exponential responses and pure time delay, hence the predicted curves can be represented by FOPTD models. The response of spray drying for orange juice powder was faster to reach steady state compared with the response of spray drying for whole milk powder. It is because of the addition of maltodextrin in orange juice powder production.

The transfer function that simplified in Laplace domain for spray drying of whole milk powder (NS) and orange juice powder (S) are given in Eq.1 and 2 respectively. Both equations present the effect of inlet air temperature on outlet air temperature and describing the dynamic behavior of spray drying of nonsticky and sticky products.

A OC — 0.16s

Gp (s) = 086e--(()

p 2.44s +1

G ( ) 0.74e—010s m

Gp(s) = —r" (2)

F 0.53s +1

The evaluation of the model was required to determine how well the model fits the experimental data used for parameter estimation. The coefficient of determination (R2) is a measure of the degree of fit. Table 4 shows the coefficient of determination, R2 were between 0.6121 and 0.8256. Both models can be considered valid and used in process control because Gong et al. [17] mentioned that the model with R2 > 0.6 can be considered as a valid model. Correlation coefficient, R, is a measure of the degree of linear relationship between the experimental and predicted data. R for both processes was above +0.8562 which indicates a stronger degree of positive linear relationship. Model verification is the final check on the model and comparing it with new or additional data that not had been used in the parameter estimation [18]. The comparisons of one new experiment and predicted data for both spray drying processes were performed. The coefficient of determination, R2 were 0.6801 (NS) and 0.6667 (S). Figure 2 shows R values were positive relationship with 0.8981 (NS) and 0.8515 (S).

Fig. 1. Process reaction curves for spray drying of (a) whole milk powder and (b) orange juice powder

Lee Uouo Tan et al. U ProcediaFoodSciencel (2011)690 - 697 Table 4: Coefficient of determination, R2 and correlation coefficient, R

Whole milk powder (NS) Orange juice powder (S)

Experiment coefficient of correlation coefficient of correlation

determination, R2 coefficient, R determination, R2 coefficient, R

1 0.6622 0.9086 0.7837 0.9297

2 0.8256 0.9657 0.6209 0.9231

3 0.7197 0.8562 0.6121 0.8805

86 87 88 89 90 91 92 93 94 95 Predicted outlet air temperature (°C)

88 89 90 91 92 93 94 95 Predicted outlet air temperature (°C)

Fig. 2. A correlation of experimental data versus predicted data for (a) whole milk powder and (b) orange juice powder

4. Conclusion

Among the inputs, inlet air temperature gave more significant effect on outlet air temperature and powder moisture compared to feed flow rate and feed concentration. Thus, this parameter was selected as manipulated variable in order to control the outlet air temperature and also indirectly to control moisture content. Two empirical models of spray drying of non sticky and sticky products were determined based on process reaction curve. These models had similar dynamic behavior and can be represented as first order plus time delay (FOPTD) models. However, the process response of non sticky product was slower compared to sticky product. Both models were valid (R2>0.6) not only for experiment data that used in calculating its parameter, but also for the additional data that are not used in the parameter estimation.

Acknowledgments

The authors gratefully acknowledge the Department of Process and Food Engineering, Faculty of Engineering, Universiti Putra Malaysia and the Ministry of Science, Technology and Innovation (MOSTI) for the financial support under FRGS grant (5523399).

References

[1] Goula A.M. and Adamopoulos K.G. 2008. Effect of maltodextrin addition during spray drying of tomato pulp in dehumidified air:1. Drying kinetics and product recovery. Drying Technology, 26(6), 714-725.

[2] Sudhagar M., Jaya S., and Das H. 2002. Sticky Issues on Spray Drying of Fruit Juices. in ASAE. 2950, Niles Road, St. Joseph, MI 49085-9659 USA.

[3] Adhikari B., et al., 2004. Effect of addtion of maltodextrin on drying kinetics and stickiness of sugar and acid-rich foods during convective drying: experiments and modelling. Journal of Food Engineering, 62: 53-68.

[4] Bhandari B.R., Patel K.C., and Chen X.D. 2008. Spray drying of food materials- process and product characterics. In Chen X.D. and Mujumdar A.S. (Eds.). Drying Technologies in Food Processing, Blackwell, Oxford.

[5] Langrish T.A.G., Chan W.C., and Kota, K. 2007. Comparison of maltodextrin and skim milk wall deposition rates in a pilot-scale spra dryer. Powder Technology, 179, 83-89.

[6] Martinelli L., Gabas A.L., and Telis-Romero J. 2007. Thermodynamic and quality properties of lemon juice powder as affected by maltodextrin and arabic gum. Drying Technology, 25, 2035-2045.

[7] Knegt R.J. and van den Brink H. 1998. Improvement of the drying oven method for determination of the moisture content of milk powder. International Dairy Journal, 8, 733-738.

[8] Codex Alimentarius, Codex Standard for Milk Powders and Cream Powder, FAO/WHO Food Standards.

[9] Tonon R.V., Brabet C., and Hubinger M.D. 2008. Influence of process conditions on the physochemical properties of acai (Euterpe oleraceae Mart.) powder produced by spray drying. Journal of Food Engineering, 88, 411-418.

[10] Chiou D. and Langrish T.A.G. 2008. A comparison of cystallisation approaches in spray drying. Journal of Food Engineering, 88, 177-185.

[11] Chegini G.R. and Ghobadian B. 2005. Effect of spray-dried conditions on physical properties of orange juice powder. Drying Technology, 23, 657-668.

[12] Cai Y.Z. and Corke H. 2000. Production and properties of spray-dried Amaranthus Betacyanin Pigments. Journal of Food Science, 65(6), 1248-1252.

[13] Masters K. 1985. Spray Drying Handbook. 4th edition ed. England: Longman Scientific and Technical.

[14] Krishnaiah D., et al. 2009. Optimal operating conditions of spray dried Noni Fruit extract using k-carrageenaan as adjuvant. Journal of Applied Sciences, 9(17), 3062-3067.

[15] Souza A.S., et al. 2009. Influence of spray drying conditions on the physical properties of dried pulp tomato. Cienc. Tecnol. Aliment., Campinas, 29(2), 291-294.

[16] Patel R.P., Patel M.P., and Suthar A.M. 2009. Spray drying technology: an overview. Indian Journal of Science and Technology, 2(10), 44-47.

[17] Gong W.J., et al. 2007. Optimization strategies for separation of sulfadiazines using Box-Behnken design by liquid chromatography and capillary electrophoresis. Journal of Central South University Techonology, 2, 196-201.

[18] Marlin T.E. 2000. Process control: Designing Processes and Control Systems for Dynamic Performance 2nd Edition ed., New York: McGraw-Hill. 268-275.

Presented at ICEF11 (May 22-26, 2011 - Athens, Greece) as paper MCF1028.