Copyright is owned by the Author of the thesis. Permission is given for a copy to be downloaded by an individual for the purpose of research and private study only. The thesis may not be reproduced elsewhere without the permission of the Author.

The Stickiness Curves ol Dairy Powder

A thesis presented in partial fulfillmcnl of the requirements for the degree of Master of Technology in Bioprocess Engineering at Massey University

Jenny Y. Zuo

B. Tech (Hons)

2004

ABSTRACT

Powder s tickiness problems encoun tered during spray drying are imponam to the

dairy industry. Instantaneous stickiness is a surface phenomena that is caused by

exceeding the glass transition temperature of the amorphous sugar in the powder. usual ly

lactose in dairy powders. Instantaneous stickiness occurs at a ccnain tempernrure above

the T8 of amorphous lactose and has been denoted as the cri tical "X" value. Whether

powder panicles are sticky or not depends on whether there is enough liquid flow on the

surface between the panicles. Two pan iclcs stick 10 each other when there is enough

liquid flow to form a bridge between them after the con1,1ct. This p roject aimed to

measure the instantaneous s ticky point conditions for \'a1ious dairy powders and to relate

these lO the o perating conditions to gh·e a commcrical O\llcome for the dairy industry.

T he panicle-gun rig was dc\'clop~d to simulate the conditions in the spray drier

and the ducting pipe and cyclone. The stickiness of powder panicles occurs after a sho,1

resident time in the panicle-gun . Thu;,. stickiness is a surface phenomenon and the point

of adhesion is the instantaneous sticky point. The amount of deposit on the plate was

me:tsured at a tcmperatme. with increasing relati1·e humidity. At ,1 panicular temperamrc

and relative humid iry. the powder snick to the stainless steel p late instantaneously. This

was observed by a sudden change in % deposition on a % deposition verse RH plot. The

T-T. plot and stickiness cur\'C profi le were developed 10 determine the crit ical "X'' value

for the dairy powders.

The critical ·x· value is the temperature which exceeds the T, of amorphous

lactose when instantaneous stickiness occurs. The c ritical "X" values for various dairy

powders inc luding WMP. SMP, MPC. whey protein. buttenni lk. white cheese powder

and GUMP powder were found t0 be 33-49°C. 37-42°C. 42-51°C, 50°C. 37-39°C.

28.5°C. and 40.7°C respectively. In addition. the slope of the trend line in the T-T~ plot,

indicates how quickly the panicular powder becomes sticky once the instantaneous sticky

po int has bee n exceeded . The particle-gun rig demonstrated that powders with greater

than 30% amorphous lactose are more likely 10 cause blockage than powders with less

than 30%.

II

Both the critical ·x' value and the slope are unique to the powder. The stickiness

curve was used to relate the powder surface stickiness condition with the drier ou tlet

temperamrc and relative humidity. It was recommended to operate at conditions below

the stickiness curve for a powder to a\'oid any chamber or cyclone blockages caused by

stickiness. The slope enables a decision to be made about how close ro the critical point a

plant should be run for a panicular powder. The inlet air tempcramre or concentrate

feed ing rate can be used to 1110\'e the operating conditions towards or away from the

s tickiness curve. <1ccording to the operating s ituations.

IU

ACKNOWLEDGMENTS

The completing ofmy masterate thesis is a learning experience of skills gained from both

research work and life experience as regards the good nature of many human beings.

My gratintde goes 10 my prima,y supervisor. Tony Paterson. thank you for your

consistant suppon. guidance. patience and help towards gaining confidence (especially in

presentations). I am very grateful to have had the opponunity 10 do a Masterate under

your supervision. My secondary super,isor. John Bronlund: thank you for your help and

guidance (especially at the beginning and end of Ill)' project). Thank you 10 Mary

Paterson for proofreading my thesis.

Thanks to Technology New Zealand for their sponsorship of this project and Fonterra

Research Cemre for their assistance with this project. In pan icular. thanks to Tuan

Troung. Da\'id Pearce. Shabeshe Paramalingam. Kenneth Bidlake. and the Longbum site

drier supervisors and operators.

I would like to express my thanks 10 Ga,y Radford and Steve Glasgow of the Ins titute of

Food. i':utrition and Human Hcahh. for the use of various pieces of equipment.

Many thanks to my fiiends and colleagues: they are acknowledged for creating an

enjoyable and stimulating atmosphere that ha;, made the time spent at Massey very

pleasanl. In panicular. I would like IO thank Rachel. Anna. Kellie. Stephen. Craig. Adi.

Heath. Jeremy and John. A special thanks 10 Roland for your hclpfol discussion on my

project. and getting me involved in 1hc activities that helped me to leam more about New

Zealand cu lrure and how 10 be a real kiwi (BBQs: watching a Super 12 nigby game in the

Wellington Westpac stadium). Thanks 10 the Hodgkinson family who have provided me

love and suppon through the years that I spent in Palmerston J\onh . Also. thank you to

Sarah for he r understanding and suppon. especially during my final stages of writing up.

And last bu t 1101 leas1. I would like to than k my grandmother. mum and dad. JieJie and

aun1 Lili for both giving and encouraging me to seek for myself a demand ing and

meaningfu l education. This s1udy could not have taken p lace without that precious gift. I

love and appreciate you all.

JV

TABLE OF CONTENTS

ABSTRACT ........................................................................................................................... I I ACK NO\VLE DG~IENTS .................................................................................................. I\'

TABLE OF co:--TE'.'-TS ·····································································································" LIST OF FIGURES ......................................................................................................... V III L I ST OFT ABLES ................................................................................................................ X

Chapter 1 PROJ ECr OVERVIEW ................................................................................................... 1-1

1.1 BACKGROU,S: O ....... ..................... .... ...... ................. ............... . ........ ........... 1-1 1.2 PROPOSED SOLUTION ... ................. ......... .... .. ... ...... ... ......... .... .............. 1-1 1.3 PROJECT OBJECTIVES ....... ...................... ...................... ........ ................ .. .... 1-2 1.4 THESIS STRUCTURE .................................... ............................ ...................... 1-2

Chapter 2 LITERATURE REVIE\\I ................................................................................................. 2-1

2.1 1:--TRODUCTION ................ ............................................................................... 2-1 2.2 STICKl:S:ESS MECHA:-- ISMS ............. ......... ............ .......... .............................. 2-1

2.2.1 Viscosiry ........... ... ........................................... ........................... ................... 2-2 2 .2 .2 Liquid Bridging ........... ............................................ ... ....................... ......... 2-5

2.3 GLASS -rllA:'\SITIO~ ......................... ...................... ........................ .............. 2.12 2.3. 1 Detennination of Glass Transition Tcmpc.-antre ................... .................. 2-1 -I 2.3.2 Prediction of Glass Transition Tcmperature ............. ............................... 2-16 2.3.3 The 'T-Tg· Facmr ... ................................................................................... 2-18

2.4 DEV ICES TO MEASURE POWDER STICKl?\ESS PH E:--OME:S:A ......... 2-20 2.4. 1 Shear Cell Method .......... ................... ... ..................................................... 2-20 2.-1 .2 BlowTest .. ............. ..................................................... ................... ... ......... 2-20 2.4 .3 Sticky-point Tempcranirc Test ................................................................. 2-21 2.4.4 Fluidised-bed Rig ....................... ............ ...................... .. ........................... 2-22 2.4.5 Cyclone Stickiness Testing Device ........... ... .................................. ....... ... 2-23 2.4.6 l'aniclc-gun Rig .................... .......................... .................. .......... ............... 2-24

2.5 STICKY POI:--T CURVE ........ ..... ............................ ....................... ............... 2-25 2.6 PROJECT PLAN ......................................................... ...................................... 2-30

Chapter 3 MATERA LS AND i\l ETH ODS ....................................................................................... 3-1

3.1 , ? J.-

3.3 3.3.1

INTRODUCTION ........ .......... ...................................................... .................... ... 3-1 MATERIALS ........ ........................ ............. .... ............. .................. ...... ................. 3-2 METHODS ............ ................................. .... ............................... ......................... . 3-3

The Paniclc-Gun Rig ................. ............................... ......... ..... .................... 3-3

V

Chapter 4 M EASUREMENT A 'D PREDICTION OF T HE CRITICAL "X'', STICK! 'ESS T EI\IPERATliRE FOR DAIRY POWDERS ............................... .... ..... .... ................ 4- 1

4.1 INTRODUCTION ................ .. .... .......... ...... ........... ...... .. .. .......... .. ........ .... .. .. .4- l 4.2 AN INST ANT SKIM MILK POWDER .. ............ ........... .. ....... .. .......... .. .. .... .4-3 4.3 CRITICAL ·x· VALUE OF POWDERS....... ........ .. ... .. ........... ...... .. .. ..... .4-10

4.3.1 Skim Milk Powders (SMP) .. ............... .. ........ ..................... .. ......... ...... . 4- l 0 4.3.2 Whole Milk Powders (WMP) ....... ..... .... .. ... .... ..... ............... .. ........ .4-10 4.3.3 Milk Protein Concentrate (MPC) and Whey Protein Powder .. .. ........... 4-12 4.3.4 Specialty Powders .... ........ ................ ........ ....... ...... ....... ........ ...... ........ .4-l 2

4.4 AMORPHOUS LACTOSE .. .. .. .. . ... .............. ...................... .. ........ ........ . 4-13 4.5 PREDICTION OF THE CRITICAL "X" ...................... .......... .. ............ .4-14 4.6 COMPARISON OF STICKI NESS CURVE USING CRITICAL ·x·

APPROACH .. ....... .. .. ..... ...... ..... .. .. ...... ....... ..... .. ... .. .... .... ....... .... ... ... ... ...... ... 4- l 6 4.7 CLOSURE ...... ....... .... ...... .... ... .. ..... ......... ......... ... ... .. ....... ........ ..... .. .. .... ....... 4-19

Chapter S STICKINESS DUE TO FAT ........... .. ................ .... .................. ........................... ....... 5- I

5.1 I TRODUCTIO, ... ...... ...... ... ..... ... .. ................ ...... ........ ..... ...... .. ....... .. .5-1 5.2 \\'H !TE CHEESE POWDER ( 42.05 %TS fat) .... .. ...... .. .......... .... ......... ..... .. .. 5-1 5.3 LOW FAT CREAM POWDER (55.8-1 %TS fat) ..................... .... .... ........ ... ... 5--l 5.-1 HIGH FAT CREAM POWDER (7 1.79 %TS fat ) .... ... ............ ...... .... .. ........ .. 5-8 5.5 STICKINESS DUE TO FAT-MEL T ING MECI IANISM .. .. ... . ......... ...... .. 5-10 5.6 CLOSURE ............. .. ..... ..... .... ..... .. ..... ......... ... ...... .. ........ ....... ...... .............. .. 5-l 4

Chapter 6 l i\ lPLEI\IENTATIOJ\'. OF STICK! ·Ess Cl:RVE IK A PLANT PRELli\ t !KARY \\'ORK ............................................. .............................................. 6- t

6.1 INTRODUCTIO ...... .. ..... ... .. ........... .... ..... ....... ....... ............ ..... ...... ... .... .... .. 6- I 6.2 CASE STUDY ........ ........ .. .... .... ... .. ........ ......... .... .. .. ... ..... ... ..... .... .. ...... ...... .... 6-2

6.2.1 Determination of critical .. X .. ...................... .. ........................................ 6-2 6.2.2 Constructing st ickiness curve and mapping plant conditions .. ..... ......... .. 6-4

6.3 SPRAY DRYING PROCESS OPTIMISATION ............ ... ... ...... .. .............. 6-6 6.3.1 Effect on X via different inlet temperature and production rate ........... ... 6-8 6.3.2 Effect on X and outlet temperature via different production rate ....... ... .. 6-9

6.4 CLOSURE ............. ... ........... ...... .. ........ .. .............. ..... .... .................... ... ... .... 6- l 2

Chapter 7 CONCLUSIONS AND RECOMMENDATIONS ..................................................... 7- t

VI

REFERENCES ................................................................................................................... 8-l

NOMENCLATURE ........................................................................................................... 9-l

Appendix A l ...................................................................................................................... 10-l Appendix A2 ............................................... ....................................................................... 10-2 SMP ........................................................ .. ................ .... ........................ ............................... 10-2 WMP .... ..... ................................................. ..... ....................... .............. ........................... .... 10-8 M PC AND WHEY PROTEl:s; ................. ................................................. .................. . 10-18 SPECIALITY .......................... ........................................................................ ................. 10-28

VII

Figure 2.1 Figure 2.2

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 3 .1

Figure 3.2 Figure 3.3 Figure 3.4

Figure -I . I

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10 Figure 4.11

LIST OF FIGURES

Milk fat-melting mechani sm (taken from Foster (2002)). The relationship between to tal fat content and surface fat content expressed in terms of the specific surface area (Foster (2002)). C omparison of s ticky-point for skim milk powder measured by different techniques. Comparison ofslicky-point obtained by fluidised-bed rig and particle­gun rig for amorphous lactose (adapted from Chatterj ee (2003)). Comparison of sticky-point obtained by fluidised-bed rig {adapted from Chatterjee (2003) and particle-gun rig for whole milk powder. Stickiness curve for whey powder measured by the cyclone stickiness H•st (adapted from Boonyai (2004)) and Amorphous lactose powder measured by particle-gun rig (adapted from Chatterjee (2003)). Selected dairy powder compositions (2-0 i\l atri x) for experimental work. The r>article-gun rig. A schematic diagram of the particle-gun rig. This part of the parti cle-gun shows where powder feed from the glass funnel and firing on the sta inless s teel plate occur. The liquid-bridging mechanism caused by T • amorphous lactose being exceeded . T he % depos ition of dry instant skim milk po"der plott ed against re lative humidity at constant temperatures. The st ickiness curn of the instant Si\lP from four experimental points titted the curve abo\'e T , of amorphous lactose. % deposition plotted against T-T• for four different temperature data sets. The slope of the trend line (T-Ti plot) show how fast is the powder response to temperature and relati\·e humidity cha nge exceed the stickiness cur\'e line. % deposition plotted against T -T • for four different temperature data sets of r eplicates. The stickiness curve of the instant SI\IP in comparison with the replicat e. The stickiness curve of amorphous lactose with r eference line T, , including Chatterjee (2003)'s result. The correlation betwee n the critical '·X" value and % lactose of powder (total fat content less than 42%) tested. The internction of slope a s function of (T-T g) and % lactose in bulk. Stickiness curve of whey powder measured using cyclone stickiness test (Boonyai e t al. (2004)) and amorphous lactose and instant SMP measured by particle-gun rig.

VIII

figure 4.12

Figure 5.1

f igure 5.2

fi gure 5.3

f igure 5.4

Figure 5.5

Figure 5.6

Figure 5.7

Figure 5.8

Figure 5.9

Figure 5.10

Figure 5.11

Figure 5.12

figure 6.1 Figure 6.2

Figure 6.3

Figure 6.4 Figure 6.5 f igure 6.6

Stickiness curve of SI\I P measured by particle-gun rig (Chatterjee (2003)), nuidised-bcd rig (Chatterj ee (2003)), and stirrer test (Hennigs et al. (2001)). Deposition of the spray dried white cheese powder (42.05 %TS fat) tested on the particle-gun rig. The cheese powder (42.05 % TS fat) stickiness points at temperatures above T i of amorphous lactose. The stickiness curve for cheese powder (42.05 % TS fat) including the fitted T

0 +X line from the Gordon-and-Taylor equation and fined

T,+X line from cubic equation. Deposition of the spray dried low fat cream powder (55.84 ¾ TS fat) tested on the particle-gun rig. Close-up of the deposition of the low fat cream powder (55.84 % TS fat) tested on the particle-gun rig. The low fat cream powder (55.84 %TS fat) stickiness points at temperatures above ·1·, of amorphous lactose. The stickiness curve for low fat cheese powder (55.84 % TS fat) including the T0+X line. Deposition of the spray dried high fat cream powder (7 1. 79 % TS fat) tested on the particle-gun rig. The stickiness of high fat cream powder (7 1.79 % TS fat) caused by molten fat and amorphous lactose. Deposition due to fat in low fat cream powder tested on the particle• gun rig \\ith increas ing air tempernture a nd kee1ling R H lo" to avoid the amorphous lactose mechanism. Deposition due to fat in high fat cream powder (200 I sample and 2003 sample) tested on the particle-gun rig. 2001 hi gh fat cream powder sample w>1s checked for crystalline lactose under a polarising microscope at SOX magnifi cation. The T-T, plot of the SI\IP and the critical ·x·. determined to be 38°C. The stickiness curve of the SI\IP is fitted line through experimental points. The stickiness curve constructed from the average critical .. x•, of SI\IP with ±3 error bar. Effect of dry product throughput and inlet temperature on T -T,. Effect on dry production rate and outlet tcrnper:Hure on T -T• Demonstration of the possible changes that could effcl,t the drier outlet conditions.

lX

LIST OF TABLES

Table 2.1 The specific 'T -T g' factor for instantaneous sticking.

Table 4.1 A summary table of critical ' ·X" value ofS MP tested.

Table 4.2 A summa ry of critical "X" rnlue of WMP tested.

Table 4.3 A summary of critical .. X" value of i\'1 PC and the whey protein powders tested.

Table 4.4 A summary of critical " X-' rnlue of cheese, buttermilk and G UMP powders tested.

Table 6.1 The possible experimental matri x for variation in the so lids 1>roduction rate, inlet temperature and corresponding T -T,.

X

1.1 BACKGROUND

CHAPTER 1

PROJECT OVERVIEW

Proj<.-ct O\'CI"\ icw

The stickiness of dairy powder particles and adhesion of the particulate mass to

chamber walls and ducting surfaces are common se\"ere problems in drying operations.

Such problems include chamber. cyclone blockages and frequent downtime for cleaning:

hence it has a sign ificant economic impact in th e dairy industry. It is estimated that the

overall product loss due to the stickiness problem would be S4 mill ion per year O\"er all

Fonterra operations. Coping wi th s tic kiness in spray driers has been a matter o ftrial -and ­

error ex peri mentation to find conditions which avo id or con tro l the sticky charac teristic

of a g i\"cn composition. Therefore. it is desirable to be able to predict the stickiness

conditions and then contro l the problem d111i ng processing.

1.2 PROPOSED SOLUTION

The two mai n stickiness mechanisms ident ified by Foster (2002) are (I) Powders

that con tain mo re than -12% total fat (greater than l.95g 'm: surface fat content). when

ex posed to a tcmperamre above 40°C where the fat becomes completely mo lten and fonn

liq ui d bridges between the adjacent powder particles. and (2) The glass transition

temperattire of the amorphous sugar present in the powder is exceeded suflicient ly to

make the amorphous sugar behave as a viscous liquid and fonn liq uid bridges when

particles come into contact. Hence. it is impo11ant to be able to identify this cri tic.ii

temperature condition which exceeds the T ~ of amorphous sugar to an amount. that

allows the viscous liquid to be suflic ic ,uly liquid to enable liquid bridges to fonn whe n

two particles come together or a particle impac ts o n the wall ofa duct.

Prel iminary work earned out by Chatterjee (2003) shows promising results in using

a particle-gun rig to mimic the air cond itions in ducting and constructing the ·stickin ess

curve' to relate the measurements to the industry process conditions. T his work used the

same rig with some modification to generate stickiness cu rves in the industry process

1- 1

Project Overview

temperan,re range. Some preliminary wor1< was done in order to implement the

stickiness curve in the plant more constructively.

1.3 PROJECT OBJECTIVES

The specific objectives of this research wer~:

I) To identify conditions under which dairy powders become instantaneously s ticky

using a panicle-gun rig. The knowledge gained from the understanding of the

stickiness mechanism helps to appreciate the causes of adherence in powder

panicles.

2) To relate these sticky conditions 10 plan t operating conditions. to give a commercial

outcome for the dairy indusuy.

3) To recommend the best way to control powder stickiness during the drying process

in te1111s of its composition. To recommend changes in the operating conditions for

the spray drier.

1.4 THESIS STRUCTURE

A literature review helped this research work stay in focus and it only inc luded the

topics that relate to this project such as the stickiness mechanisms and glass transition

temperature of amorphous materials. linderstanding these fitndamental facts pro,·idcs a

good grounding for the following chapten,. Chapter three discusses the materials and

methods used and explains the dairy powder samples selected and experimental work

carried out usi ng the panicle-gun rig. Instan taneous stickiness occurs at a cenain

tcmpemtu re above the T, of amorphous lactose and has been denoted as the critical "X''

value. Chapters four and five concentrate on the identification of the critical "X" value

for powder instantaneous stickiness and use this in fonnation to consm1ct a stickiness

curve for various powders selected. A preliminary work with the aim to implement the

stick iness curve in a plant environment is di scussed in chapter s ix, with a case s tudy on

skim milk powder (SMP). Chapter seven summarises the project in a nut shell and

provides the recommendations for this research.

1-2

CHAPTER 2

LITERATURE REVIEW

2.1 INTRODUCTION

Literature Re\ icw

Powder stickiness between pan icle-pa11icle (cohesion) and panicle-wall (adhesion)

is a major problem in the dairy industry. especially in spray driers. It can cause cyclone

blockage. long downtimes for cleaning. and lowered process capacity. thereby reducing

process eflic icncy and yield. In order to carry Ott! the objectives of this work. it was

essential to understand the related areas. These areas include the dairy powder related

stickiness mechanisms. the theory on glass transition and its relation to the critical sticky

point. and the tests and mea,urements used to in,·cstigate stickiness behaviour.

2.2 STICKINESS MECHANISMS

Stickiness and caking are closely related phenomena and are commonly

encountered problems during the spray drying process and storage of dairy powders.

Stickinc,s is a time. temperature and humidity dependent phenomena [Roos el (I/.

( 1996)]. With time. stickiness can lead to crystallisation and caking which is due to the

formation of solid bridges between the paniclcs.

Stickiness due to electrostatic or molecular attraction forces is insignificant

compared to liquid bridging because of its unpredictable and irregular pattern [Peleg

( I 993 )J. In addition. Chen ( 1994) found no dlcct of charge on the panic Jes as related to

the deposition rate in the spray driers. Therefore. the initia l stage of st ickiness is almost

always due to sufficic111 flow that forms a liquid bridge between two pa11iclcs· surfuce at

their contact point. The driving force for the flow is surface tension [Downton et (If.

( 1982). Wallack and King (1988)].

Product temperature and mois ture content are two of the imponam factors that

influence the s tickiness of dairy powder and other sugar-contain ing foods [Downton et al.

(1982)). Above a panicular relative humidity and temperature condition when

undergoing spray drying. panicles like fructose from fruit juice and lactose from milk

behave in an adhesive o r cohesive fashion. Such a change from non-sticky behaviour to

2-1

Literature Rcvic,,

sticky behaviour happens with only a small change in the temperature and moisture

content [Bhandari e1 al. ( 1997). Chuy and Labuza ( 1994 ), Hennigs e1 al. (200 l )]. lt is the

state of the powder particle's surface (sticky or not) that detennines whether the particles

bounce off the walls. adhere 10 the wall or cohere to other pai1icles that are already

adhering to the walls . The temperarure and water content are related to the particle

conditions.

It has been well established that stickiness is a property of surface phenomena. It is

governed by its composition. powder exposure conditions. surface properties. viscosity.

and physical properties [Chatterjee (2003). Foster(2002)] .

The onset of powder stickiness is the focus of this project. particularly where

powder become, instantaneously sticky. The following liquid bridging mechanisms for

sticking ha\'C been identified and are the main focus of this project.

2.2.1 Viscosity

The surface properties of the drying pai1iclcs arc related to the surface \'isco,ity.

The viscosity. as a result of water remo\'al. increases rapidly as the gla,s mmsition is

approached. A typical \'isco,ity ofan amorphous material in th e glass state is equal to or

above the order of l 0 11 Pa. s. The \'iscosity decreases as the material changes from glass

to a mbbery state due to conditions that are abO\'C the glass transition tcmpcrantre (T g)

and are reported to have a ,·iscosiry ranging fi-om 10"-10' Pa.s. The viscosity of food

systems is a function of rcmpcrature. once it tran~fonns in the rubbery state associated

with stickiness and caking [Roos (1995a)). Oown1on el al. (1982) reported the critical

viscosity of stickiness of amorphous sugars at T-T • values in a range from l 0-20° C

which means that the sticky point temperature is 10-20°( higher than the glass transition

temperarurc. The magnin,de of the T-T, parameter is the fimdamcn tal indicator of

amorphous material related to s tickiness and adhesion.

Downton el al. ( 1982) proposed a model for estimating the critical viscosity (p,)

for dry particle stickiness to occur. The equation 2.1 shows that the viscosity is

influenced by the surface tension (y) . contact time (r) and the distance (KD) over which

2-2

Li1crmurc Review

flow must occur. T he critical viscosity is a constan t value of viscosity for the stickiness

to occur at short-time contacts (e .g. 1-1 Os) and is predicted from equation 2.1 to be in a

range of I 06 10 I 0~ Pa.s. The s1icky-poin1 temperature measurement test of Lazar e1 al.

( 1956) was used to measure the sticky-point temperature over a range of moisture

comcnrs for a sucroselfi-uctose mixmrc. It was found that all measured viscosity data lie

well within the predicted critical viscosity range and fall with in the range of0.32 x IO - 10

4.0 x IO 7 Pa.s. It was concluded that the proposed model provides a good understanding

of the stickiness phenomenon with a more quantitative a nd mechanistic approach.

µ , Kyr - - --KD (2.1)

Where "A·" is a dimensionless proportionality con:,tant (A-I assumed). "D " is

acmal pa11icle sizes ( 1-1 Opm),"K" is the fraction of the particle diameter required as the

bridge width (0 .01-0.00 I for a :,utlicicntly stro ng bridge).

Wallack and King (1988) used a similar model, che Frenkel model (equation 2.2).

10 predict the cricical viscosity range ofmaltodexcrin,'sucrose ' fructose and coffee extract

in a short contact time. II was found thal the measured , ·iscosity lay within the predicted

cricical viscosity range of 106-10' Pa.s. for maltodcxcrin sucrose fiuctose and 10' 107

Pa.s. for co flee exrracc. "a " is the particle radius. " x .. i, che intcrparticlc bridge size

formed by viscous flow and is cstimaced from scanning eleccron micrographs of particles

which have undergone sticking and arc then freeze dried. "alx .. is approximately equal

10 0.1. II shows chat stickiness of particles occurs by viscous flow driven by surface

energy. The Frenkel model demons1ra1es che relationship bet\\'een chc viscosity of

material with the pa11icle radius and powder surface eneq,,y. In o rde r 10 preve111 the

scicki ness. the viscosity should be raised by either increasing the particle radius of the

powder. and/o r decreasing the powder surface energy.

µ _1 rm - 2 , x-

(2 .2)

2-3

Literature Re\ icw

Both researchers D0\\~1ton e1 al. ( 1982) and Wallack and King (1988) use suiface

tension (yJ of70 m:-,1/n, for interstitial concentrate and "r" is the particle contact time I ­

I Os forihe method used by Lazar el al. (1956) for detcn11ining s tickiness. The standard

falling ball method was also used for measuremem of viscosity. The resu hs con finned

that the sticky mechanism of viscous flow is driven by smface energy. The models show

that increasing the surface-tension driving force or longer comact times would increase

the tendency toward sticking. while greater viscosity or larger particle s ize decreases the

sticking 1endency. The viscosity relates to cond itions of 1emperature and moisture

content under which stickiness occur., [Le Meste el al. (2002)].

The viscosity of ma1erial can be considered constam in the glassy state (2: 101\

but the Yiscosity changes dramatically above glass transition temperature (T ~)­

Therefore. it can be said that \'iscosity is dependen1 on 1hc 1cmpera1ure above glass

transition 1c111peranire (T>T g- e.g. T~-1 OOK) [Bhandari and Howes (1999) ]. Williams et

al. ( 1955) proposed the Williams-Landel-Ferry (\VLF) equation (equation 2.3) 10 relate

re laxa1ion 1ime of mechanical propcnies 10 temperature abo\'e its glass transition

temperan1re. The measured sucrosc 'fi11ctosc viscosity data of Downton el ol. ( 1982) was

found to fall within 1he predictions range of the \VLF equation.

(2.3)

Where JI is the viscosity and µ,, 1hc viscosity at glass 1ransi1ion temperature. T, is

the reference 1empera1urc. Tis the 1empcrarure. and C1 and c_, are 'un iversal' constants.

(17.44 and 51.6 respecti vely for many materials). Aguilera et al. (1995) used the WLF­

rype rela1ionship wi1h adjustable CI (-11. 16) and C2 ( 10. 7) coefficients to investigate

caking of 1he amorphous fish hydrolysa1e powder. It is now widely underswod that 1he

universal constams given by Williams et al. ( 1955) arc 1101 universal and shou ld be

allowed to vary for different materials.

Roos and Kare l (199 1b} and Roos (2002) have applied the WLF equa1ion 10

pred ict 1hc viscosity of the g lass as a function of temperarure for 1he crys1allisation rate of

vari ous amorphous sugars (lactose. sucrose . sucrose/ fructose).

2-4

Literature Review

2.2.2 Liquid Bridging

As introduced earlier. almost all cases of sticking and caking are caused by liquid

bridging rather than electrostatic or molecular attraction forces. It is important to pre\'Cnt

the fonnation of liquid bridges: and therefore. the caking and sticking of powders can be

a,·oided. In real situations. liquid bridging can only occur when at least some of the

pai1iclc surfuce is \'iscous (or if there is a presence of an extem,11 moisnire or a second

liquid phase) [Adhikari e1 al.(2001) ].

For the sticking process to occur. the panicle surlace has t0 become fluid at the

panicular contact point o r points. Peleg ( 1993) gave the mechanisms by which this may

be caused by:

,- Direct wetting or moisture condensation on the surface funns a film of saturntcd

solution around the paniclc.

► Re-crystallised amorphous material causes adsorbed warerto be libcrnred onto the

panicle surface.

,- The glass '1ubbcr transfonmnion of amorphous material on the surface is caused

by tempemrure exceeding glass transition temperature. The surface becomes a

vc1y \'iscous "mbbery'' state and therefore flows.

;,.. Melting on the surface (e.g. fat-containing panicles).

It is assumed that stickiness due to insufficient drying before pa11icles collide with

one another or with the walls of the drying apparatus is not the case here . The powder

panicle State is dried powder. under nonnal operation conditions.

The first case is nor appropriate for amorphous material. because it is very

hygroscopic and any surface moisture would be absorbed into the particle rather than

remaining on the surface. However. the condition of the amorphous material does

depend on the temperamre and moisrure in the surrounding air. This could depress T •.

giving the third case above. Re-crystallisation of amorphous material happens during

storage. because of its long relaxation time. Therefore. stickiness due to insufficient

2-5

Li1cn:uurc Rc\'icw

drying(~ 50% RH} and crystallisation are not applicable here. All spray-dtied powder is

assumed to be dried and contain amorphous carbohydrate present in the powder.

For this work on the sticking of daity based powder. it would appear that the two

mechanisms of liquid bridging in,·olYe either amorphous carbohydrate glass 'mbbery

transfon11ation and 'or fut-melting on the surface. panicularly in powders which have a

high fat content.

2.2.2.1 Amorphous Carbohydrate Sticking Mechanism

Amorphous lactose is fonned when water is removed so quickly that there is no

time for the lactose t0 crystallize. or by freeze-drying. Molecules either do not have time

to an-angc themselves or they arc held in the frozen water matrix and hence cannot

c1ystallize. This results in the random arrangement ofo.- and P- lactose molecules in the

solid powder matrix . Amorphous lactose has a number of imponam properties. such as

being hygroscopic: it absorbs moisture ,·ery readily from its sun-oundings. and it is

subject to a phenom~non kn own as glass transition [Roos ( 1995b) .Bhandari and Howes

( 1999). Roos et al. ( 1996}]. Lactose in amorphous rubber form is responsible for powder

stickiness in the dairy indusuy. while crystalline lactose powders can be sticky only

under extremely high tcmpcrarure and humidity conditions [Bronlund ( 1997)).

Amorphous lactose crystallisation occurs at a water content of about 6.8 g (g x I00)'t of

lactose corresponding to an equilibrium rclati\'c humidity of 37% at room temf)eraturc

I Roos (2002)]. Aguilera ,,, al. ( 1995) repo11ed that amorphous lactoi.c crystallizes at 42-

52% RH. Such change can lead to release of water and /or encapsulated milk fat in the

case of dairy powders.

Amorphous lactose becomes sticky when its glass transition tempcrarure is

exceeded suffic ientl y that the decrease in viscosity causes flow at the su,face and liquid

bridges can fon-n between panicles which causes particles to adhere together [Brooks

(2000). Foster (2002}]. The glass transition tcmperarure can be exceeded by either an

increase in temperarure to above T ¥ and/or an increase in moisrurc content and water

activity t0 suppress the T1.

2-6

Literature Review

1t is very common for speciality powders to have another carbohydrate added in

solution o r dry form. If these carbohydrates are added in solute fonn (i.e. p1ior to spray

drying) then the carbohydrate will be in the amorphous fonn in the final product.

However. the mechanism ofliquid bridging is applicable to all amorphous carbohydrates

present in dairy products. Foster (2002) investigated propenies of amorphous

carbohydrates such as sucrose. maltose. glucose. galactose. fn1c1ose and sugar mixtures

in relation to stickiness.

2.2.2.2 Fat-melting Sticking Mechanism

Milk fat contributes to sticking of dairy powders during processing and s tomge.

especial ly for high fat powders such as cream powders and cheese powders. In

processing. high fat powders cause smearing. in which the powder builds up on the inside

of the chamber. cyclones and fluidised beds [8111T (1999). Crotskcy (2000)]. Free fat

content varies with diftcrcm process conditions. For example. more fi-ee fat is observed

in roller drying than spmydrying [King ( 1965) ].

Milk fat contains a mixture of lipids. and 98.5% of its constituents are

triglyce,·icles. Differcm triglycerides (diflcring in types of fany acid. the length of chain

and the degree of saturation ) have different melting points. thus milk fat has a large

melting range. :'\orris et al. ( 1973) using a Diferential Scanning Calorimeter (DSC)

:,howed that the 1riglyce1ides in milk fat have a melting range approximately -40°C to

~40°c. A typical DSC thennogram for mi lk fat shows three melting ranges: -40°C-0°C.

0°C - 20°C. and the highest melting region 20-37°C which is the main melting peak.

Although the resolution of the peaks in the DSC thcm10gram diffi:r due to different

heating rates. the temperatures at which the peaks occur remains constant.

Fat complexes with prote in in a finely emulsified state and is said 10 be

encapsulated fat [Buma ( 1971 a). King ( I 965)]. Buma (I 97 1 a) stated that fat which is not

encapsulated is known as extrac table fat or"fi-ee·• fat. A detailed study offi-ee fat content

in relation to the powder propenics defines that fi-ee fat originates fi-om four different

sources. These are:

2-7

Litcr:i.ture Review

Surface fat. which is present as pools or patches of fat on the outer most particle surface:

Outer layer fat. consists of fat globules in the surface layer of the powder partic le and it

can be directly reached by fat solvent:

Capillary fat. consists of fat globules inside the particle. which can be reached by fat

solvents via capillary pores or cracks:

Dissolution fat. consists of fat globules. which can be reached by the solvent through

holes left by already extracted globules.

It is the free fat on the surface and some on the outer layer that contributes mostly

to the sticking mechanism. This is mainly because they have the ability 10 pool the fat

together and fonn liquid bridging when the fat is in its molten ,rnte. Some high fat

content powder. the capillmy fat near the surface also pooled by molten fat on the

smfaccs. In most cases the capillary fat and dissolution fat do not contribute to sticking

or caking. Foste,· (2002 ) used confocal la,cr scanning micro,co1)Y (CLSM) to observe

the pooling effect of fat in low fat crc,1111 powder. The observation, were caniccl out at

20°C before and after being heated to 55°C. The result supported the fat-melting

mechanism for sticking and caking. as the surfuce fat pooled at the particle contact

poinVi and fom1ed agglomerated particles.

The mechanism for sticking and cak ing due to milk fat is described by Foster

(2002) shown in Figure 2.1. Liquid bridging in high fot dairy powders occurs when there

is enough molten fat o n the surface when they are exposed to a process temperature

above 40°C.

2-8

In high temperature environment

"Sticking•

Inlow t emperature environment

·coking"

~ Surface fot

~ Powder portocl•

I M;Jk fol m<lling

and flow

J., Fot ty bridge. ~ ··~-, Milk fat crystallisation

Figure 2.1 Milk fat-melting mechanism (taken from Foster (2002)).

Litcrarurc Re\ icw

Free fat was extracted by washing the powder with an organic solvent-petro leum

ether (boiling point 40 10 60°C) for the desired time. Then the so lution was filtered

through filter pa1>cr to separate the powder and the solvent. The powder residue was

washed with the so h·cnt and dried. The solvent residue comaining the extracted fat was

allowed t0 cva1>orate until the extracted fat residue reached a constant weight. The free

fat content is calculated as a percentage of the total mass of the powder sample. It was

observed that as extraction time increases. the amount of free fat extracted increases.

Kim et al. (2002) extracted free fat from commercial SMP (0.0 I fat gig powder) and

WM P (0.266 fat gig powder) at an extmction time of 10 mins. 24 hrs and 48 hrs. The

amoun t of extracted free fat in \VM P increased from 0.019 gig powder to 0.028 gig

powder. However. the surface fat of free fat is of interest for the liquid bridge

mechanism. Foster (2002) estimated an extraction time of I Os for the surface fat ofa low

fat cream powder (55% total fat) because a constant free fut content measured with

exrraction time less than 20s was observed using the standard free fat extraction

2-9

Litcraturc Rcvic"

procedure. Buma (1971a) also suggested that an extraction time as short as 10 s or at

least I min would be needed to measure the surface fat.

The difforences in spray drying process parameters fi-om manufacturer to

manufacturer have minor imponance to the composition of the powder smfacc [Foster

(2002). Faldt and SjOholm (1996)). However. difference in particle size has an effect on

1he amount of free fut that can be extracted. the reason being that small powder particles

have much more surfuce exposed 10 the solvent than larger particles during extraction

[Buma ( 197 1 b). Fa ldt and SjOholm ( 1996)).

Su1face fat comem can be estimated from the total fat content and specific surface

area. Foster (2002) looked at the relationship between 1he surface fut content and total fin

content for different powders with varied le\'els of fot. The data was expressed in tenns

of the specific surface area in order to consider the pa11icle size cffi:cts in the free fat

extraction.

Total fat content mcasuremem used the 1--lajonnier fat extraction procedure.

Figure 2.2 taken from Foster (2002) shows a fairly linear relationship between surface fat

content and total fut content when the total fat content is abo,·e 30%.

2-10

Literature Review

6 -

~5 ~------------------------ ---.--.g • " ~4~-----------------------------1: " t 3~ -----------------------------

<J -:; ... ., 2~-----­u ~ ::,

</) 1

• --------,-----·

• • 0 .l-----♦--------------------------

0 10 20 30 40 50 60 70

Total Fat Content (g faV100g dry powder]

Figure 2.2 The relationship between total fat content and surface fat content expressed in tenns of the specific surface area (Foster (2002)).

Crot\;key (2000) examined the ef'foct of the fat-melting mechanism in lo w and

high fat cream powders. us ing an air gun to lire powder pa11iclcs at a plate and measure

the percentage of powder sn,ck to the plate . The RH was kept relatively low to avoid the

lactose changing its state while increasing temperature. Surface tension of the fat

decreases as the temperature increases. The fat is able to flow over the panicle sudacc

when its smfacc tens ion becomes equal 10 or less than the su,facc tension material of its

contact. The critical surface tension point for low fat cream powder was found 10 be

around 40°C. Funhcm1orc. as temperanire increases to a point of flow. fat viscosity is

decreasing and this pulls more fat from the capillary fat near the surface 10 the surface

and increasing the total area of fat coverage. However. for high fat cream powder. at a

temperature above 40°C. there was no increase in powder stickiness as the panic les are

covered with a layer of fat. No funher fat can be drawn from inen fat.

Although. surface fat-melting 10 fomi liquid bridges is one of the anributes to

sticking and caki ng in dairy powder. it is agreed by Faldt and Bergenstahl ( 1995) and

King ( 1965). th at lactose crystallisation is the other attri bute for gaining surfuce fat from

encapsulated fat. Th is can happen during st0rage: if amorphous lactose crystallises. as a

2 -1 1

Li1cra1urc Rc\licw

result ofan increase in temperature or moisture stresses. dismp ting the encapsulated fat.

forcing fut to spread to the powder particle surface (Faldt and Bergenst3hl ( 1995). King

( 1965)) . In the case of milk powder. King (1965) noted that when it absorbs moisture

and develops into a dry. hard and powdery texmrc. this coincides with the release of fat.

At the critical moisture content of 8 .9%-9.2%. fat was released for a powder with a fat

content of 27-28%. Lactose crystallisation was observed during this time with a

polaiizing microscope. The crystallisation mptured the continuous matrix of the SNF

(so lid not fat) and made the fat accessible to the solvent.

Researchers have used the Electron Spectroscopy for Chemical Analysis (ESCA) to

estimate the amount of surface fin on the powder in conjunc tion with free fat extraction

method. The ESCA analysis provides direct in formation about the powder surface

independent of particle size and melting point of oil phase. In addition. the powder

surface coverage ofothercomponents such as protein and carbohydrate is also detectable.

The result is the two approaches have a good correlation. For powder with a higher fat

content. the surface fat C0\'erage is high. so is the free fat [Fiildt et al. (1993). Fiildt and

SjOholm (1996). Kim et al. (2002)).

2.3 GLASS TRANSITION

In a 'glass· state. the material is solid and brittle and has an amorphous. liquid-like

structure with an extremely high viscosity. 11.:ofthc order of I 011 Pa.s [:S:oel et al. ( 1990).

Bhanda,i et al. ( 1997). Downton et al. ( I 982)]. The glass state can transfonn into a less

viscous. liquid rubbery State by escalating the tempcramre andlor water plasticization.

Such phase transfom1ation is rcfetTed to as glass transition and the tcmpcramre at which

this happens is known as glass transition cemperaturc (T.,} [Roos ( 1995b} ]. Glass

transition temperature profiles are often reponed as a function of water activity, relative

humidity or concentration (%solid). Glass transition is a second order transition and is

charac terized by a change in specific heat capacity.

Materials transfom1ing from 'glass' to a 'mbbery' state as a ftmction of temperature

is well established and this transition temperamre is dependent on moismre content

[Lillford and fryer (1998)). The glass transition temperature (T g) is being recognized as

2-12

Li,crarurc Review

a fundamental parameter to explain and control phenomenon such as stickiness. caking

and unwanted agglomeration for amorphous and semicrystnllinc foods. pa11icularly in the

case of low molecular weight carbohydrates [Bronlund (1997). Foster (2002). Brooks

(2000). Adhikari er al. (200 I). Bhandari and Howes ( 1999). Downt0n et al. ( I 982).

Bhandmi et al. (I 997). Roos and Karel ( 1991 b). Chuy and Labuza ( 1994)).

In dairy foods. the main amorphous components arc carbohydrates and proteins.

However. often the amorphous carbohydrates govern the transition temperatures

corresponding 10 changes of physical propenies . Glass transition occurs due to a water

content dependent remperarure range decreasing wirh increasing water conrem [Roos

(2002)]. Milk fat does not affect rhe glass transition temperature of milk powders and ir

is immiscible with lacrose and orher warer-soluble substances [Jouppila and Roos

( 1994a}. Jouppila and Roos ( 199-lb)J.

In general. the glass 1ransi1ion temperature of carbohydrates decreases as

molecular weigh, decreases [Soc! et al. ( 1990). Roos Y. (1993)) . Jouppila and Roos

( 1994a) found skim milk powder wirh cnzymarically hydrolysed lacrosc er. - 49°C) had

substantially lower T, rhan rhe regular milk powder (T, ~92°C). Lactose (MW 342) in

milk hydrolyzes to glucose (MW 180). and galactose (M \V 180} and this decreases the T 1

in skim milk powder and crirical water content and water activity. The T $ for anhydrous

galactose and glucose arc 30 and 31°C. respectively. bur the T g for lactose is IO 1 °C

[Roos Y. (1993}. Bhandari and Howes (2000)) . The lowT$ofthe galactosc and glucose

mixture causes more severe glass transition characterized problems rhan does anhydrous

lactose during processing and srorage.

The glass transition rempcramrc is important to this work as ir is widely

recognized that ir characterizes amorphous materials for phenomena such as stickiness.

caking and crystallisat ion when an amorphous materials" g lass transition temperature is

exceeded [Chuy and Labuza (1994). Downton et al. ( 1982). Jouppila. K. e1 al. (1997),

Jouppila and Roos (1994a). Jouppila and Roos (1994b). Lloyd e1 al. (1996). Roos and

Karel (1991a). Roos and Karel ( 1991 b)]. Bhandari el al. (1997) found a correlation

2-13

Literature Rcvic\\

between Tg and the spray drying behaviour of sugars and fiuit juices with a maltodextrin

additive .

Lloyd er al. ( 1996) showed that the caking of spray dried lactose is related to th e

onset temperature of T •· They confim1ed the viscous now of the product above T, by

compressing an amorphous lactose powder at various temperarures in a compacted

product increasing above the onset temperature ofT-T, . The temperarure change (T-Ti

factor) is the difference between the actual temperature of the powder (T) and its glass

transition temperature (T g); in rum this T-T • factor determines the rate of ,·arious changes

lBrooks (2000). Foster(2002). Roos (1995 b}. Roos e, al. (1996). Chatterjee (2003)] . It is

therefore desirable 10 know what Ti is at a panicular mo isn,rc condition. and how the T­

T, factor relates to s tickiness in dairy powders . The glass transition temperature can be

detennined either by measuring it directly or predicting it from the moi>ture content or

water activity.

2.3.1 Determination of Glass Transit ion Temperature

There are sc,·cral methods by which the glass transition tempcrawrc can be

directly measured . These analytical methods detect the sudden change in physical

propenies of the material at T •· These change of physical propenies include heat

capacity. free molecular volume. them1al expansion coetlicicm. d ielectric coefficient and

viscoe lastic prope11ies.

Ditll:rcntial them1al scanning calorimetry (DSC) is one of the most common

techniques found in the literature for determining Tg of amorphous mate,ial such as

spray-dried skim milk and freeze-dried skim milk powder [Ozmen and Langrish (2002).

Jouppila and Roos ( I 994a)] . DSC measures the temperan,rc and heat nows associated

with transitions in materia ls as a fonction of time and temperature.

Foster (2002) reviewed the literature for the possible causes for different glass

transition temperature measurements. The reasons for variation measured by DSC arc

outlined as follows:

2-14

Literature Review

► Scanning rate used (higher rate results in higher T3 value from true value). Chuy

and Labuza ( 1994) measured the glass transition temperature of carbohydrate

material and found that the glass transition temperawre was lower at slower

scanning rates and higher at higher scann ing rates. They explained that the lower

observed temperature values at the slowest heating rate may be due to the longer

residence time of the sample at any temperature during heating. which increases

the probability for viscous flow. crystallisation of the arno'l)hous carbohydrate.

and collapse . The rate of 5°C min is used by Roos and Karel ( 1991 a. 1991 b.

1991c).

► Location oft he T • on the DSC curve. Due to different interpretations ofT, values

ti-om the DSC curve. for example of amorphous lactose. some researchers use

onset. mid-point or endset for T, data [Ozmen and Langrish (2002)]. Up 10 I 0°C

diftercntial is expected when comparing the T3 value for amorphous lactose found

in the literature by ditlercnt researchers.

,- Enthalpy relaxation peaks.

► Preparation of DSC pan. It has the potential for uptake of moisture by the

powder. because amorphous carbohydrates have a hygroscopic nature. which

means they absorb moismre very quickly if the DSC pan is not fully tilled under

dry condi tions. This would raise the moisture content of the powder and result in

a lower T, value. In addition. Brooks (2000) noted that freeze-dried sugars still

hold some moisture. which can acts as a plasticiser. even after dehydration over a

strong desiccant. It is therefore the residual moisture that ha~ an elli.:ct on the

glass transition temperature and has to be accounted for in the Tgdata.

In food systems such as milk powder. the main components are carbohydrate.

prmein and fat. Jouppila and Roos (1994a) used DSC to de1em1ine the glass transition

temperature of freeze-dried milk powder products (such as 0¾fat skim milk powder.

10.7% fat milk powder. 18.6% fat milk powder. whole milk powder. and hydrolysed

lactose skim milk powder) at different water content levels. It was found that the T g

values of those milk powders (except the hydrolysed lactose skim milk powder)

2-15

Li1eta.rurc Review

containing lactose were almost equal 10 1hosc of pure lac1ose. This indicates tha1 proteins

do nor i111ernc1 wi1h lactose in 1enns of glass transition tcmperarure. and the physical stale

of S:S:F in milk powders is governed mos1ly by lactose. P,·otein does. however. have

some affect on 1he crys1allisation of amorphous lacrosc [Jouppila and Roos (1994b).

Foster (2002)]. It is also likely to affect 1hc raic of change of s1ickiness wi1h time for a

given powder. Other observations show that milk fat did not atfec1 the T • of milk

powders due to their immiscible properties with S~F or moisture. However. T11 could not

be de1em1ined from 1he DSC curves for powders containing fa1 at the RH range of23.9 to

44.4% because of the dominan1 fin-melting endothcnn in the same 1emperarure range

(.louppila and Roos ( l 994a)].

O1hcr useful 1echniques and sometimes more sensitive methods include dynamic

mechanical them1al analysis (DMA'DMTA) and mechanical spectroscopy. These

1echniqucs measure the ct1cc1 of a sinusoidally varying stress on dynamic moduli or

viscosity. ~uclear 111agne1ic resonance (NM R) spectroscopic technique and elcc1ron spin

resonance spectroscopy ( ES R) are useful techniques when smdying 1he molccul.ir

mobility and diffi1sion of 1he glas, transi1ion 1cmperature of panicles (l3handari and

Howes (2000)). Peleg (1993) extensively listed the references ofsmdies where me1hods

01her than DSC or OMA were used to de1ennine the glass transition tcmpcrarure.

2.3.2 Prediction of Glass Transition Temperature

The Gordon-Taylor equation (cq. 2.-l) is the empirical fonnula thai i, used to

prcdic1 the innucnce of moismre contcm on 1he T• (e.g. anhydrous solute and water

component [Jouppila and Roos ( 199-la)]) or to estimate 1he glass 1ransition temperamre of

a mixture of sugars [Hennigs et al. (200 I) .Roos and Karel ( 1991 a)].

_ (w,Tgt +kw2Tg 2 )

( 11·1 + kw2 )

(2.4)

Where Tg., is the glass transition temperarure of the mixmrc. Examples ofmixmre

components tha1 have been studied arc CL-lactose-water or maltoclextrin-sucrose. 111,. w;.

Tg1 and Tgi are the mass fraction and glass transition temperatures of component I

2-16

Ulcralurc Rcvie"

(lactose) and component 2 (water). respectively. k is a constant for the solid component

and it was derived experimentally.

w1 T81 + (6CP2 I 6C pl )w1 T

8 2

H'[ + (6CP2 / 6C p t )w2

(2.5)

The Couchmann-Karasz equation (eq. 2.5) is another mathematical fonnula for

detem1ining T, ofa pol>q11cr mixnire. The ratio of change in heat capac ity is similar in

form to the constant kin the Gordon-Taylor equation. The Couchmann-Karasz equation

can expand funhcr and be applied for multi-component mixnire systems such as water.

glucose and sucrose [Roos ( I 995a)]. For an ·,, · component system. the extended

Couchmann-Karasz equation is shown in equ,11io11 2.6 .

f w,6 C pi T g;

Ti;,,, - I= I II

L 11· ,6C p,

(2.6)

, - 1

For daily powders. the glass transition tcmperanire is more readily predicted from

the water activity rather than the moisture conten t of the powder. In a practical sense.

moisture contc111 of the powder includes not only the moisnire associated with sugars but

also moisnire associated with other components (such as protein) which have no cftcct on

the T, of the powder. Howeve r. water activity of dairy powders is fairly constant with

varied levels of its components (i.e. sugars. protein and milk fat). So. the T, is constant

with constant water activity. In addition. water activity can a lso be eas ily measured with

a calibrated RH probe.

Brooks (2000) fined a non-mechanistic equation to the literature d ata: a third

order model developed (eq. 2.7) to predict the T1 of amorphous lactose from the water

activity of the powder. In the same line of thought. Foster (2002) summarized the

T / moisture content and T .,twatcr activity profile for amorphous glucose. galactose.

fructose, sucrose, maltose and maltodextrin powders from data available in literamrc. and

2-17

Literature Review

fitted equations to the data. so that the pred iction ofT8 was made from ind ividual sugars

present in the powder.

T, = - 530.66(0. ) ' + 652 06(a" ) z - 366.33a, + 99.458 (O<a.<0.575) (2.7)

Foster (2002) also proposed a mathematical equatio n (eq . 2.8) for pred icting th e

Tg ofa multi-componem powder. The predic tion is based on the weighted addition o f the

individual TI' values for the amorphous sugars present. at given water activity of the

powder. Th is method was found to have an overall better fit than the Couchmann-Karasz

equation. especia lly for the predict ion of the Tg of.imorphous lactose in dairy powders.

2 .3 .3 The 'T-Tg ' Factor

II

Tg = L,X;Tgi(a,) i =I

(2.8)

Amorphous carbohydrate associated with the stickiness pheno mena is a viscosity

re lated mechanism. \\lien the tcmperamrc exceeds its glass uansition temperamre. the

viscosity decreases. allowing flow between the panicles . The rate of s ticking for

amorphous materia l is governed by how much the Tg. is exceeded (T·Tg). The individual

parameters of temperamrc and relative humid ity might not d irecting result in scnictural

changes. but the resulting T-T8 can cause dramatic s trucn,ral change lead ing to sticking.

caking. collapse and crystallisation . It is therefore, on ly T-T" that is impo11a111, not the

acrnal tcmperaf\lre and moisture cond itions required 10 ac hieve it (Brooks 2000).

Bhandari el al. ( 1997) and Roos and Karel ( 1991 b) have used the concept of T -T • to

describe the stic ki ng behavior of amorphous powder. Bhandari and Howes (1999) and

Bhandari (200 I) suggest the cri tical viscosity of sticky powder in sugar-rich food is

reached at a temperature I 0-20°C above T .. and assumed that the tempcramre of surface

of panicles during spray d rying should not reach 10-20°( above the T •·

The semilogarithmic plots of the s ticky po int temperature measu rement and the

glass transition temperature versus mo isrnre content for sucrose : fT\lctose (7: I) mixmre

2-18

Litcr.tturc RcviC\\

show the m parallel with each ocher [Adhikari er al. (200 I)]. The s ticky poim temperature

is almost 4-11 °C higher than the glass transition temperature (T 8 ,...i).

Brooks (2000) demonstrated 1ha1 the 'T-T8" factor of amorphous lac tose is a

combination of differem temperature and relative humidity changes giving this constant

value of temperature difference. resulting in the same caking strength with time. The T­

T8 of amorphous lactose for o nset instantaneous stickiness is 25°C (Table 2.1 ). However.

in order 10 avoid sticky powder. storage below glass 1ransi1ion or continuation of

processing below T -1·, of 10°C were recommended. Fos1er (2002) found the T-Ti values

of amorphous sugars where instantaneous sticking occurred (Table 2.1 ). The crical T-T~

value for ins1a111aneous stickiness would probably be less 1han 1he values given in Table

2 .1. Burr ( 1999) found 1he specific T-T ~ of cheese powder caused mainly by amo1phous

lactose is I o•c.

Keir (200 I) found the s1ickincss behm·ior of milk powders did difler from 1ha1 of

pure amorphous lactose. With regard 10 Brooks (2000)"s recommendation. 1hc T-T, of

I 0°C for safe processing is 100 conservaiive. and Keir (200 I) fonhcr s1a1cd that 1he T-T,

required for stickiness IO occur may be 30°C or higher. especially considering 1hc sho11

1ime in the driers. The s1ickiness changes in dairy powders are obviously affected by the

composi1ion of the powder.

Table 2.1 The specifi c 'T•Tv' factor for instantaneous st icking .

Amor phous Carbohydra te

Lac1ose

Sucrose Mahosc

Glucose/Lactose

Galactose/ Lactose Fnictosc/ Lactose

'T-T, ' factor (°C) 2s.o· 20.3

19.0. 25.3. 29.0 35.0, 37.4

30.8 4 1.3

*Brooks (2000) critica l instantaneous stickiness; Foster (2002) ins1an1ancous s1ickiness occurred.

2-19

2.4 DEVICES TO MEASURE POWDER STICKINESS PHENOMENA

Literature Review

Researchers have developed and reported various techniques to measure powder

cohesion and adhesion phenomena over the last 50 years but most of these measure the

s tickiness of the powder when the particles are Stationary or moving in slow motion. A

review focused on the most relevant methods follows.

2.4.1 Shear Cell Method

The powder is placed in a box and it is split in half horizontally or vertically. It is

equipped with a means of applying various nonnal stresses and shear stress to the top half

of the cell and while the lower hal fheld stationary (Jcnike. 1964 ). It was found that for

each nomial stress there is a particular shear s tress wh ich causes failure. This tensile

s trength analysis provides a direct indication of the interpart icle forces that suppon before

it yields. It was used to correlate the Oowabiltiy ofa powder especially through chutes

and hoppers.

2.4.2 Blow Test

The blow test was developed by Paterson et al. (200 I ) to measure the caking

strength of milk powder. In this test. air was passed through a small diameter tube held at

a fixed height and angle above a powder bed. The powder bed is pre-conditioned at

different water activities. The velocity of air was increased umil a channel was formed in

the bed and this flowrnte recorded as a measure of the caking strength or stickiness of the

powder. The blow test was also used to distinguish between fat caking and amorphous

lactose caking in dairy powders Foster (2002). Burr ( 1999) also used the test for the

investigation of caking strength of cheese powder in different s torage conditions.

Brooks (2000) modified the blow test by including a segmented distributor and a

glass enclosure. The segmented distributor enables more samples and multiple readings

to be taken over a period of time. The glass enclosure enables the contro l of powder

condi tions in the bed. air at desired temperature and relative humidity ready for the blow

test. Brooks (2000) found the same caking strength (a particular value ofT-T g) with time

resulted with different combinations of temperamrc and water activity. Amorphous

2-20

Literature- Re-" icw

lactose becomes instantaneously sticky a1 a T-T8 above 25°C. Fosler (2002) used 1he

same 1es1 to observe chis particular value ofT-T, for other amorphous carbohydra1es 1hat

may be prescn1 in dairy powders and 1he caking mechanism under different storage

condi1ions.

This 1es1 allows a continuous reading 10 be 1aken over time and 1he ability 10

observe the time-dependent side of sticking and caking 1>roblems under cons1ant T-T,.

Thus. i1 is possible 10 quan1ify 1he development wi1h 1ime of in1er-particlc forces caused

by 1he viscous flow of an amorphous malerial. This is usefitl in 1hc smdy of powder

condi1ions for s1orage where 1he residence 1ime is long enough for s1ickiness 10 develop

over 1ime. While i1 can be shown tha1 s1ickiness can develop very fos1 under cenain

temperature and rcla1ivc humidity condi1ions (Paterson et al. 200-1) i1 is 1101 generally a

sui1able device for measuring 1hc instantaneous stickiness needed by a particle impac1ing

1he walls ofa spray cliic,· or the ducting al\cr 1hc drier.

2.4.3 Sticky-point Temperature Test

The so-called .. s ticky-point temperamre .. 1es1 was originally developed by Lazar

et al. (1956). The principle of the 1es1er has been applied and modified by several

researchers 10 dctennine 1he s1icky point of 1he amorphous sugar powders for different

temperature and moisture contents of the material (Hennigs et <ti. (200 I ). Downton e r al.

( I 982). Chuy and Labuza ( 1994) and Wallack and King (1988)). The principle of1he 1cs1

in\'olves placing a powder sample with a known mois1ure co111en1 in a test tube. which is

immersed in a controlled temperature bath. The tube is closed to 1he aunosphere wi1h a

rotating mcrcu,y seal. and a small propeller embedded in the sample. The ba1h

1e1npcrature is raised s lowly while 1he propeller is stirred periodically (i.e. quarter rum

per second) either by hand or 1hrough mechanical means. The sticky-point 1emperature is

referred to as the temperature at which the force experienced 10 drive 1he propeller

increases sharply. The sticky point 1cmperamre is a function of moisture content. Thus

1he s1icky point decreased with increasing moisture con1ent. The 1cs1 is repealed with

01her moisture contents, and 1he sticky point tempera1ures are ploued agains1 moisture

con1en1. The combina1ion of temperature and moisture point lying above the curve

causes 1he onset of product stickiness and is to be avoided. whi le the region below 1he

2-21

Li1cra1urc Review

curve is the non-sticky safe regime. Downton el al. ( 1982) reponed that st icky-point

measurements are repca1ablc within ± I K. The bath heating rate was I °C/3 min then

1°C/5 min as the s ticky poin1 was approached [Down1on et al. (1982). Wallack and King

( 1988)]. Chuy and Labuza (1994) examined the effect of different heating rates on the

sticky point ( 1 °C/3 min. 5°Ctmin and I 0°C/min). ln general. a lower temperarurc was

observed a1 the slowest heating rates compared to faster heat ing rates for any given

powder from 1he same s torage condi1ions. A slow heating rate would provide a longer

residence time. therefore increasing the probabi lity for viscous flow and stickiness.

Downton et al. (1982) and Wallack and King (1988) have re la1ed the sticky-point

temperature to viscosity temperature data. The sticky point ofsucrose 'fructose powder is

superimposed upon 1he curves representing the \'iscosity data. The cxperimenial

viscosity value of 0.32x Io' to 4.0x IO ' mPa.s. lies within the predicted critical ,·iscosiry

range of I 06- I 0~ (Downton el al. ( 1982)]. The results of Wallack and King ( 1988) gave

similar agreement with the predic1cd \'iscosity da1a range on maltodextrin 'f\'uctose

powder.

Over.ill. the sticky-poin1 temperature test is a measure of powder s tickiness

bcha\'iour in a mildly agitated situation and the results vary with heating rate. panicle

sizes. stirrer rotational speed and tech niques.

2.4.4 Fluidised-bed Rig

The 'bench-top-scale fluidised-bed' rig was developed by Toy (2000) and was

based on a study (fluidisation test) carried out by Dixon ( 1999). This rig was used to

measure powders · cohesive stickiness propeny in a fluidised bed situa1ion. with changing

temperan1re and humidity conditions. This semp involves supplyofhumid air at d ifferent

temperarures to a small -scale fluidised-bed. The powder paniclcs surface propenies

changed as the conditions of the air were changed until the bed lost tluidisation. This

was achieved by a controlled wet-bulb temperature via a water bath (range 0-80°C) to

give 1he desired saturation humidity: this samrated air was then heated and immersed into

another hot waler bath for a desired dry bulb temperature (experimental range 40-80°C)

before de livering it into the fluid bed. The water baths' temperatures were controlled.

2-22

litcrarurc Rcvic"

monitored and recorded by setting conditions at a desired temperature and humidity. The

air veloc ity varied between 0.22 ms"1- 0.42 ms"1 and showed no effect on the stickiness

end point.

The measurement in the ·bench-top-scale fluid bed· rig was done by visual

observation of the stick iness end point or total seizure of the bed in a particular

temperan1rci\1u111idity envi ronmcn1. Chatterjee (2003) stated that the observed end points

occurred when complete seizure of the particle bed occuned. and this was categorised by

an adhered pai1icle mass which would be revitalised temporarily for a short period with

rigorous shaking or vibration . bm would collapse very soon after. Chatterjee (2003)

found lactose in an amo rphous fo1111 is the stickiest powder. and in a crystall ine fonn is

the least st ickist powder. The degree of stickiness is due to the amount of amorphous

lactose present and the different particle sizes.

This technique can be used 10 detennine the stickiness curve on a temperature

versus water activity or relative humidiry profile. but cannot be used on cohesive dairy

powders which will not fluidise such as high fat powders.

2.4.5 Cyclone Stickiness Testing Device

The stickiness cyclone testing device was developed by Boonyai et al. (2002) and

it was designed to measure partick-panicle and paniclc-wall stickiness during spray

drying. The set-up consists ofa system that provides humidified and dry air mixntre at a

specified RH. adjusted by the their tlow ratcs and a cyclone test chamber. When the

desired lc\'el of RH was researched. approximately I g of powder sample was introduced

into the cyclone chamber from the top open ing. The powder panicles travel in a rotary

fashion at the bottom of the cyclone chamber. A trave lling time of 1-2 min was suft,cient

for the powder stickiness 10 occur under appropriate temperanirc and humidity conditions

Boonyai et al. (2004). A s tickiness curve is detennincd in a similar way to the fluidised­

bed ri g.

This test has the advantage over previous methods in that the particles will be

travelling at s imi lar veloc ity to those encountered in the transport system for the powder

after a spray drier, but the conditions acntally used arc at temperatures below and relative

2-23

Li1cramrc Review

humidity conditions above those actually experienced by the particles leaving a spray

drier. There is also a question as 10 whether the history or RH ramping rate affects the

measurement of the stickiness curve by this technique.

2.4.6 Particle-gun Rig

The shear cell method. blow test and sticky-point temperamre tests measure the

development of stickiness ' caking between stationary or almost stationary pa11icles. They

are not directly applicable to measuring the instantaneous stickiness of high velocity

moving pa11icle: the situation found in spray driers and the ducts after the driers.

Research conducted by Moreya & Peleg ( 1981) has shown that physical properties such

as 'stickiness· . arc mainly characteristic prope11ies of the pai1icle surface and can occur at

much shorter times than those necessary to reach equilibrium with the interior of the

panicle. The occurrence of•instantaneous' agglomc~1te formation of incipient stickiness

in food powders has also been examined by Wallack & King ( 1988). A new technique.

the panicle-gun rig has been developed to measure the onset of ins tantaneous stickiness

with panic I es travelling and hitting walls at velocities. temperawrcs and relative humidity

similar to those encountered industrially.

O'Donnell el al. (2002) originally developed an apparatus to supply air with both

a controlled tcmperanirc and relative humidity. at a variety of now rates up to 114 I min.

Crofskey (2000) invented the rig to achieve an optimal powder delivery system where the

equilibrated particles went through the gun and were fired at a s tainless steel plate. The

percentage of powder deposit on the plate is measured and calculated for a constant

tcmperamre with varied humidity. The initial sticky point at a particular temperature and

humidity was identified from the %deposition versus relative humidity plot. Those

conditions were then used to genemte the ·stickiness curve' of that powder.

The particle-gun rig is very usefu l in tem1s of controlling air velocity. temperature

and relative humidity to mimic the actual spray drier operating condition in the plant.

Chatterjee (2003) used the rig to observe the stickiness propcnies of a wide range of

dairy powders including cream powder. cheese powder. skim and whole milk powders . It

2-24

literature Review

was con finned that the stickiness is indeed a surface phenomenon and is affected by the

surface composition ofa particle.

2.5 STICKY POINT CURVE

Chatterjee (2003) used both "bench-top-scale fluid bed" and "panicle-gun·• rigs to

generate a ·st ickiness curve' for various dairy-based powder;. A typical stickiness curve

is temperature versus relative humidity or water activity plotted that so it can directly

rel.He 10 spray drying outlet conditions. During drying. the product temperature is

generally approaching the outlet air temperature [Masters (199 1 )].

The measured s1icky-poi111 data from a st irrer-type sticky-point device Hennigs el al.

(2001 ). a fluidised-bed device Cha11erjee (2003). Bloore (2000). Boonyai e1 al. (2004)

from the cyclone tester and the paniclc-gun rig Chatterjee (2003) of skim milk powder

arc shown in Figure 2.3. There arc differences in sticky-poim obtained from the different

techniques . The stirrer-type test shows powder becomes sticky at lower tempcrarures at

the same RH and at lower RH"s at the same temperature. than the tluidiscd-bed and

panicle-gun methods. This could be because the sample was measured in an almost

static bed giving more contact points and less distance between paniclcs. Although. the

tluidiscd-bed device and the panicle-gun rig make measurements while the panicles are

moving there are different sticky-points obtained under similar RH and temperature air

conditions.

2-25

100

90

80

0 70

" e '" 60 ~ e so 3 e 8_ 40 E :!, 30

20

10

Literature Rc"icw

♦ ChatterjoCt 2003 (Particle-gun rig)

■ Chatterjee 2003(FluICllsed-bed rig)

------. ---··- •; -·· · ------- A Hennigs2001(Stlrrettest)

.... _. !'. __ .'la.. __ ~ __________ ~ A~apted fro~!~~!~'!~-~ F~ui~iution tost)

• ,. . -------- • --- . ·-----~-------.. ·······-----··•

----- - -------··· ----------------·-

• ,# • --········· ,.

• • • • •

o~--------------------~--~------~ 0.0 0.1 0.2 0.3 0,4 0.5 0.6 0.7 0,8 0.9

Water Activity

Figure 2.3 Comparison of sticky-point for skim milk powder measured by different techniques.

The s ticky-point tcmpcramre difference between the diflc rcnt techniques varied

from powder to powder. and depends on the individual powder propcnics. For example.

the stickiness-point of amorphous lactose (Figure 2.4 ) has the largest difference between

the two techniques and whole milk powder (Figure 2.5) has the least difference.

2-26

Uu..-raturc RcviC\\'

90

80

70

~ 60 e "' ~ 50 e ~

n 40

8. E 30 ~

•• .. • •

•·. - a

- - - - - - -- ___ ,.._ -- - -- - - - - -- - - --- - - - -

- - - - - -- --- - - - - - - - -~--- - - - - - - - - -- -.. ----------------- ______________ ... _

• •

• Fluidised-bod rig

■ Particle-gun rig

-----. --·--•

---------------------• 20 ---------------------------------- -------

... . --------- --- -----

10

o+----~------------------------~ 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Water Activity

Figure 2.4 Comparison of sticky-point obtained by fluidised-bed rig and particle-gun rig for amorphous lactose (adapted from Chatterjee (2003)).

The diftercnces in sticky-poim temperature detem1incd from the 1wo techniques

are probably due to 1he velociiy 1he panicles were travelling at. In the fluidised-bed rig.

the air velocity was in a range between 0.22-0.42 n1.S·1• compared 10 the paniclc-gun rig

of 20 m.s'1• Panicles travelling a1 lower speeds will have lower kinetic energy and will

require less force 10 hold 1hcm together on impac1. Panicles tral'elling a1 higher s1>eeds

mi ght knock off the panicles 1hat arc loosely adhering to each 01her. It can be concluded

that the panicle s1ickiness is aftec1ed by 1he kinetic ene rgy of the panicles. Powders

travel at high speed in the ducting pipeline and in the chamber during spray drying and

hence the paniclc-gun rig is probably more applicable as a measure of the s1ickincss of

powders during processing in 1he dairy indus1ry. Darn fi-o m all measuring techniques

needs to be compared to industrial experience to contim1 which arc the most applicable .

2-27

90

80 - - - - - - - - - - - - - - - • - - . - - - - - - - - - . - - - - - - - - - -

70 # - • --. <l 60 · • · • ~· ., ~ . :8, 50 · · - - · --- ·--- · · - -- · · - - - - - -- -- -- - - • - • . - -

•

20

10 ••. ------------- ----- ------------------

lilc.-raturc Review

♦ Fluidised-bed rig

• Partlclc--gun rig

• • • •

01---~----------~--------------~ 00 0 I 02 OJ 04

wa:e, Acttvity 05 06 07 08

Figure 2.5 Comparison of sticky-point obta ined by fluid ised-bed rig (adapted from Chatterjee (2003) and particle-gun rig for whole milk powder.

Another factor that might have an in fluence on the stickiness data gathered by

other researchers is the pre-conditioning of the powder. Once the powder has been pre­

condi tioned to a dcsir·ed tempcranrre and relati\'e humidity 0\'cr a period of time . the

internal stmcturc of the powder might have changed and this might enhance the likeliness

of stickiness to occur at a gi,·en set of conditions. This efli:ct will be looked at on the

partic le-gun in future work.

The stickiness curves of whey powder obtained from the cyclone stickiness test

Boonyai er al. (2004 ). amorphous lactose measured by the particle-gun rig arc shown in

Figure 2.6. (The composition of the whey powder· has not been given in the reference so

it has been compared to the pure lactose data of Chatterjee 2003.) There is a general trend

of constant 1cmperanrre difforence between the two stickiness curves. Amorphous

lactose would be expected to become sticky earlier than the whey powder because it has a

higher amorphous lactose content. However. this is not the case shown in Figure 2.6.

Fu rthennore. from Figure 2.6 it could predicted that if the same whey powder was tested,

the curve generated from the cyc lone stickiness test would become sticky earlier than the

2-28

Lilcra1urc Rc,1h:-,,

curve generated from the panicle-gun rig. The difference could be due 10 the fact that

powder paniclc travelling time for the cyc lone stickiness 1cs1 was about 1-2 min and it

was less than 0.05 second for panicle-gun rig. As well as this. the cyclone tester ramped

the RH values up over time. This will have given some leve l of conditioning to the

powder during the tests and probably explains the difterencc between the tests.

considering the margin of error ± s•c. Comparisons with data obtained indusnially

about where plants acmally become sticky will be needed to detcmiine which of the rwo

tests is more appropriate.

70 I ♦ Boony al (2004). (Cyclone stickiness 1est•Whey powder)

• ChatterJ" (2003), (Partlcle-gu11 rig -amorpho1.1s lactose}

60 - - - . --------·· .. --··---- --·---

G 50 -- - .. -•--- . .. . ., • • ~

f 40 • • •

~ • ::,

e 3-0 8. • E • .,

20 -·----- . ------,.. •

10

0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Water activity

Figure 2.6 Stickiness curve for whey powder measured by the cyclone stick iness test (adapted from Boonyai (2004)) and Amorphous lactose powder measured by partic le-gun rig (adapted from Chatterjee (2003)).

The stickiness curve is imponant in tenns of demonstrating the instantaneous

stickiness on panicle surfuccs which occur at a panicu lar value above Ti lactose of dairy•

based powder. It can therefore be re lated to the acnial spray drying process parameter

and be possible to predict and mon itor the process in the sense of stickiness control.

2-29

literature Review

2.6 PROJECT PLAN

Dairy powder sticking to the wall of chambers and to each o ther during spray

drying is a prevalent problem in the industry. The fo m1ation of liquid bridging caused by

decreased viscosity a t the surface was explained for how particles stick. The amorphous

sugar and fat-melting mechanisms were responsible for sticking in dairy powders. The

degree of each mechanism attributed 10 the s tickiness problem is dependent on the

composition of the powder. The introduction of the glass transi tion temperature theory

has helped researchers to im·estigate the condi tions at which the powder becomes sticky

enough to cause the problem.

Since there is a sho,1 residence time during spray drying. the conditions which are

of interest for instantaneous stickiness are those of short contact times. The pait icle-gun

rig is to be used in this work as it can provide information about powder sticky responses

to change in tcmperanire and water activity conditions. The stickiness curve is usefol 10

display a critical value when the T~ is exceeded causing instantaneous stickiness of a

powder clue to the amorphous sugar present. Chattc~ec (2003) used the rig for difterent

dairy powders a1 low temperature (28°C-52°C).

This work is to be focused on developing stickiness curves for various dai,y

powders at higher temperan,rc ranges (i.e. 60-80°C around the processing tempernturc) in

order to observe if there is a trend between powder stickiness and its composition. From

the stickiness curve. the conditions for instan taneous powder stickiness can be identified

and related to plant operating cond itions. The recommendation for an optimization of

process can then be made.

2-30

CHAPTER 3

MATERALS AND METHODS

3.1 INTRODUCTION

~fatcrials and 1\lcthods

The variery of compositions in dairy powders is 1hough1 to be an imponant factor

m relation to panicular powder stickiness curves. The materials used here are

commercial products of Fonterra. There are hundreds ofproduc1 specifications that the

company produces: the process of selecting 1he samples used in this work is explained in

the material section.

O'Donnell et al. (2002) initially developed an appamms 1hat can constantly supply

air wi1h co111rnlled 1cmpcra1ure and humidity conditions. The bubble column was

suggested as the simplest way to samrate the air supply under pressure. Crofukey (2000)

worked on the invention of the panicle-gun and used it for investigating caking problems

in cream powder,. Funher tine-tuning and devclopmct11 on the 1>a11icle-gun test rig wa,

carried out by the Fomerra Research Centre (FRC) in conjunction wi1h Massey

University. Chatterjee (2003) used 1hc rig to generate the ""stickiness curves"" for

different dairy based powders and high fat powders such as cream powder. that arc

difficult to handle by 1he bench-top-scale fluidised-bed rig at low 1cmperarures (<50°C)

and high (72.5°0) relative humidity.

The panicle-gun rig was used because i1 gave control over 1he air velocity.

1empcrature and moisn1rc that mimics 1he conditions in spray drier ducting. l1 is assumed

that the surfaces of1he pan icles leaving the drier are in equ ilibrium with the drier outlet

air temperature and relative humidity. 11 is the critical tempcranirc and humidity

condi tions al wh ich the powder becomes stic~-y in a maner of seconds (instantaneous

stickiness) that is of interest. as powder stickiness is a surface phenomenon. The

measurement of% deposi tion a t a particular temperature and relative hum idity provides

infom1ation o n the instantaneous sticky poim of powder. The experimental set-up and

protocol are explained in the methods section.

3-1

Materials and Mcchods

3.2 MATERIALS

Lactose. fut and protein are the prime components in dairy based powders.

Lactose and fat are responsible for the stickiness phenomena. The compos ition oflactose

and fat were the basis for selecting samples because they are the main components

causing powder stickiness. Fonterra Co-operative limited supplied the powders and their

bulk composi tion data (Append ix A I}. A 2-D lactose-fut composition plot is sho"TI here

to cover the desired range and also the samples that were availab le (Figure 3.1 ).

100

90

80 iii 70 ... ,t ~ 60 i= f so -i 8 "' 40 ~ u 30 X "'

i • _,

20

10 +

0 0 20

• X

40 60 80 Fat content (% TS)

♦ WMP 39(;IOMOra~

■ 1/J'IJPl'lgtlt'I ... !

W'IJP 1'119\lllar .,.,,,. X WMP~ulaf

♦ SM;> tl\S-tel"II

SMP ol"l$tant

• 51 .. P Mt<l•l/l'l'I t'INI

!J?Otow heat

♦ WIIM cneese S..ttierl"t't.

?-,CPSS

.)(CP 70

x •.•PC$6

1,1.PCTO

.,_,PCtl5

100 - Wl'lty-&987

Figure 3.1 Selected dairy powder compositions (2-0 Matrix) for experimental work.

The powders were selec ted to cover a wide range oflactose and fat compos ition .

Cream powder. cheese powder and bunennilk powder are relatively high in fat and low

in lactose content. Skim mi lk powders (SMP} arc high in lactose a nd low in fat. Whole

milk powders (WM P} have lactose and fat content in the middle range. Whey protein

po wder is high in pro tei n and low in lactose. and milk protein concentrate (M f>C} was

low in both lactose a nd fat.

Amorphous lactose was made in the pilot p lant spray drier at Massey University.

The details for preparing amorphous lactose can be found in Brooks (2000) page 3.4.

3-2

Materials Md Methods

3.3 METHODS

3.3.1 The Particle-Gun Rig

The particle-gun rig was used to simulate the drier conditions and to measure th e

point a1 which a powder becomes instantaneously s1icky. The rig is constmc1ed in two

pans: a constant humidi1y air supply system and the particle feeding system tha1 enables

the pa11icle to .. fire .. on 1he plate a1 1he desired velocity (Figure 3 .2). A constant air

temperature was used to fire dairy powder pai1iclcs at a collection plate. The relative

humidity of the air was increased and 1he % deposition recorded . Four differenl

tempemn,res be1ween 60-80°C were used as 1hey correspond 10 the spary drier outlet a ir

temperan,re tange. The ·gun · finings mean that the exposure time of powder partic le is

less than 0.05 seconds. which is a sufticicnt penetration time for the surface stickiness to

manifest.

Figure 3.2 The particle -gun rig.

O'Donnell et al. (2002) buih an apparatus that produces air with both controlled

tempemrure and relative humidity. at a desired throughput by saturating the air at one

pressure and then lowering the pressure through a pressure regulator. This principle was

3-3

Ma1crials and Methods

used by Fonterra 10 develop an equivalent apparatus. The schematic diagram of the

paniclc-gun rig is given in Figure 3.3.

Cont.tot Hu."N.dily Pow<let feui=."l& P:obt Fur.11-l

T t rtt>.:t1 '1 Box Pl P1 Hutti T3 R/ll ioi D•ttloggt: Ptrucl.t

n T4 gun

Oat.4 Logge:

A, Wl $j

F!owmttt1 Compnutd A1tln RI R.l

Au Regul.\tOI

W~dn O $$Haung

Plate

°'"" v.iv, SI $3$3

Powu St.pply

Fig ure 3.3 A schematic diagram of the particle-gun rig.

A constant air suppl y was ensured by passing compressed air through two

pressure regulator... R I eli minates any pressure fluc ruatio n fro m the compressor. R2

controls the pressure fo1· the bubble column. The air passes through an air-flow meter

before entering in the column. A steel bubble column was used to saturate the air under

higher pressure. The column was filled up with water and the temperature kept constant

through a heating element. The air stream was bubbled through the column of water at a

pressure of up 10 8 bar and then passed through an expansion valve (W3 ) 10 release the

pressure 10 a lower pressure . The relative humidity of the low pressure air stream is sci

by the ratio ofhigh (P 1) and low (P2) pressures. The air stream is then passed through an

in-line air heater which enables a desired air ou tle t tempcramre to be achieved. This

temperature change can have an effect on the relative humidiry of air, hence control of

the outlet condition. The relative humidity of air can also be adjusted by increasing the

water temperarurc in the column. allowing more water vapor in the air to be carried

3-4

>·fa1crials and Methods

through the column. The reading of water temperature in the column. the pressures and

humidity of a ir readings for the system were recorded and displayed on the computer.

Figure 3.4 This part of the particle-gun sh ows where powder feed from the glass funnel and firing on the stainless steel plate occu r.

The ,·elocity o f the air was kept constant at 20 ms'1 at 1he panicle-gun tip. Typical

velocities in industrial cyclones are in the region of20-45 ms·' Masters ( 1991 ). and it was

the velocity recommended by Crofskcy (2000). Figure 3.4 shows the panicle teed

system. The constant tcmperanire and humidity air stream entered into a vonex chamber.

A venturi effect was used to create a suction through the glass funnel which fitted in a

ho le on the centre of the vonex chamber. The degree of suction created was dependant

on the position of the glass funnel into the firing rube.

The temperature and hu midity differences between the air leaving the air-heater

and the air at the tip of the gun were found to be significant Chatterjee (2003). The heat

loss was due to the lines and fining for carrying the air and the mixing of the ambient air

through the suction of the glass fi.mnel. Modifications were made to the particle-gun for

3-5

Materials and ~lcthod,

this experimental work in order to minimize the heat losses. In addition. the air

temperature and relative humidity were measured at the tip of the gun to define the final

conditions.

3.3.1.1 Experimenta l Set-up

The operation procedure of the panicle-gun rig was 1aken mainly from Chatterjee

(2003) as 1he same procedure was used there. A detai led step-by-step descrip tion of 1he

particle-gun suin-up has been given by Chatterjee (2003 ).

A c01mol tenninal box recorded a ll the measuring devices of the process variab les

installed on the rig. All the data was then 1ransrni11ed to the computer for display. Those

variab les tha1 correspond 10 1he display on 1he computer that were included are:

P Humid: Pressure of air (P l )

P _ outlet: Pressure of air after be ing stepped-down (P2)

T_ Outlet : Temperawrc of air measured after being heated by air-heater (T3)

RH_ Ou Itel: Relative humidity of the air before 1rnvel through to the vonex chamber

(RH! )

RH_Calc: Calculated relative humidity value. The diftcre111ial pressure between Pl

and P2 is inversely proponional to the rela1ivc humidity of the outlet air.

T Hum id: Tempcr,uurc of saturated a ir in the top of the wa1cr column (T2)

Y_Cale: Calcula1ed amount of moisture in the air at P2

T Water: Temperature of the water in the co lumn (T l )

The final 1cmpera1ure (°C) and relative humid ity (% RH) of the air was measured

at 1he tip of the gun by a rotronic HYGRO PALM: a ponablc humidity and 1emperature

measuring instrument The probc·s- ..-/- I digit of the display was calibrated before use

against saturated salt solutio ns. The system was adjusted to ensure the desired condition

3-6

Marcrials and :Vk1hods

was met. The tcmpcranires and relative humidity of the air displayed on the screen was

recorded at each n,n for reference purposes.

The temperanire and humidity profile during processing was displayed on the

computer screen. This enabled better control of the system. monitoring the changes in

the system quickly in tem1s of temperature and hum idity. When the system had reached

a constant state. the experimental work was started.

The relative humidity of the outlet air can be increased by either increasing the

water temperature (wet-bulb temperature) in the column or reducing the difforential

pressure between Pl and 1'2 . (lower pressure at pressure regulator (R2)). The air-heater

can a lso be used in the control of the relative humidity: heating-up the air wou ld result in

a decrease in relative humidity.

l'or these experiments the water temperanire was set at 68°C. and the tcmpcranire

of the air was first set by the air-heater. then the relative humidity was adjusted to can)'

out the experiment. It took a long time (approx. I hour) for the system 10 reach a s tatic

condition. es1>ecially for higher temperature experiments. This experiment was carried

out by first fixing the tempernrure of the air and then increasing the relative humidity of

the air for the n111s. Four diflcrcnt temperatures between 60-80°C were used as they

correspond to the process outlet air temperature range. It worked out that the

experimental work achieved different RH"s at one temperature per day and

approx imately one powder with -I tempernrures per week. It varied from powder 10

powder. At times there was a maximum of three powders done within a week when there

was a sample available for experiments. At extreme air conditions. either high

temperature and/or high relative humidity. the powder became less free-flowing through

the fonncl as the particles tended to stick to the funnel walls. thereby taking longer to

feed the same amount of sample through. Reducing the amount of sample helped in this

regard. s ince the deposition was calculated as a percentage.

The velocity of the air was kept constant at 20ms"1 at the particle-gun tip. High

velocities in the particle-gun rig were used (20 ms·'): this is because typical velocities in

industrial cyclones are in the region of 20-45 ms•t Masters ( 1991 ). and it was the velocity

3-7

M3tcriats and tvlcthods

recommended by Crotskey (2000). The discharge valve and pressure regulator (R2) were

used to control the a ir velocity at 20ms"1• while adjusting the relative humidiry to a

desired condition. In order to have a better contro l over the fluctuation which may save

waiting ti me between each nrn. fu,iher work could change the discharge valve to a

pressure regulator.

3.3.1 .2 Experimenta l Protocol

Once the constant air velocity. temperature and RH for a test run were reached.

the conditions were recorded in a data log sheet. The protocol for ca,rying out the

experiment is described here:

A rounded stainless stee l plate of75 mm in diameter was weighed by Sanorius

BP 210S. a 5 significant figure weigh ing scale. This was placed directl y under the

panicle-gun (distance below 15cm) shown in Figure 3.-1. Chatterjee (2003)

recommended that a net of25 grams of powders be used for all test runs. The weight of

the powder used depended on the maximum deposition that could be handled on the

plate. (in some simations the powder over deposited on the plate. thus causing errors in

the experiment). and the types of powders used (some powders arc more inherent ly sticky

than o thers. so less powder was used for stickier powders). Hence. the amount of the

powder used varied from 25 grams to 5 grams and the percentage of deposition was

calculated as the measme of stickiness.

%deposition (dry weight) - (Weight deposit on the plate Total weight of feed) * I 00

The plate was weighed before and alter the experiment. so the difference is the

weight of the deposit. The beaker for containing the powder was measured before and

after each experi ment. so the exact to tal weight of the feed powder was measured. In

order to limit the time the powder sample was exposed to the ambient air where it could

pick-up the rnoismre from th e air. the sample was weighed just before the feeding started.

The ambient condit ion might have effected the powder condition at feeding: howeve r. it

was assumed that the powder panicle surface gai ned equ ilibrium with the air condition

supply once it was fed into the gun. T he effect of ambient conditions will be in vestigated

in fumre work.

3-8

Materials and Methods

The rig was sec up to supply air at 20 m.s"1 at a predctem1ined temperanire and

relative humidity. A sample of powder was fired through the gun and the % deposi1ion

recorded. The rela1ive humidity of the air was 1hen adjusted to a new higher value whi le

maintaining 1hc same dry bulb 1cmperature at the tip and a new sample of powder was

fired through the gun. This was repeated for increasing RH values.

The panicle-gun rig data of% deposition versus RH as a function of temperature

shows that the funher the critical RH is exceeded the more sticky the powders become.

Thus. the paniele-gun can identify not only the cri1ical stickiness point of a 1>owder at a

temperature. but also enables one to get a feel for how quickly the powder will become a

problem when it enters the stickiness regime. This feature is unique to this technique.

Funher analysis on the data can be found in chapter -1 and 5.

3-9

t\k a:,urcmcnl and P-rcdiclion of the Critical ··X": S1ickincss Temperature for Dairy Powdcr

CHAPTER 4

MEASUREMENT AND PREDICTION OF THE CRITICAL "X": STICKINESS TEMPERATURE FOR DAIRY

POWDERS

4.1 INTRODUCTION

The liquid-bridging mechanism 1hat occurs when dairy powders become sticky

and s tick to one another is a lso the reason that the powder particles become s ticky in the

duct lines in processing, and is caused by the glass transition temperature of amorphous

lactose being exceeded and transforming the amorphous lac tose into a rubbery state

(Figure 4. 1 ). The glass rubber s ta te change occurs at the glass transition temperamre (T .J of amorphous lactose: however. liquid-bridging fonns at some po int above the T • line

due to decreased viscosity of the amorphous lactose which therefore fl ows. The s tic ki ng

of amorphous sugars was found to be a viscosity related mechanism and th e ,-ate of

s ticking dependent on the extent 10 which the T, of powder is exceeded. Therefore. the

viscosity is a fonction of tcm1>erature d ifference (T-T0) . ··x·· is denoted as the critical

temperamre difference above T, of amorphous lactose. where the amorphous lac tose is

suftic iently liquid for the powder partic les tO become instantaneously sticky. T. and

consequcmly "X ... is a fimc tion of water activity or water con tent of the amorphous

lactose. The higher the value of··x-·. the more tolerant the powder is 10 higher RH and or

tcmpcramre cond ition with regard to stickiness. For example. it is expec ted that the "X"

of WMP will be larger than the "X-' of SMP. since there is less amorphous lactose

present in the powder.

4-1

Figure 4.1 exceeded.

Mcasurcmcnl and Prediction oftl1c Critical ·X": S1ickinC$S Temperature for Dairy Powder

Gla$Sln.ibbefy stt1•.e cnattge 81 g l8$S tran&l»l --!

Dairy powder particles

At high temperature

and/or high RH conditioris

(i .e. exceed Tg)

LJquld•bndge cau,e Slicking

The liquid-bridging mechanism caused by T • amorphous lactose being

Brooks (2000). Fos1er (2002) and Keir (200 I) have demonstrated that the ra1e of

sticking is related to the T -T• of the powder and it is no t impo11ant what RH and

temperature condition, arc used 10 obtain that T-T8 value. They also concluded thai the

higher the T-T• value. the hi gher the rate o f s tickiness. The rate of stick ing has been

graphed on a log scale due 10 the large diftcrcncc in rntes for low and high T-T • values.

Using the log scale. a fairly linear relationship can be seen between the log of the rate of

sticking and 1he T-T• of the powder. Under cc11ai11 RH and temperature combinations

the rate of stickiness development was so quick. 1ha1 it could be regarded as

instantaneous. The value ofT-T8 a1 which this occurred is an indication of where the

powder would be expected 10 caused problems in transpon lines.

The value ofT-T. enables us to correlate the rnte and degree of stickiness. as it is

dependent on temperature and RH. The prediction ofT8 ofamorphous lactose was based

on the third-order empirical equation and was calculated from the water activity or

moisture content of dairy powders [Brooks (2000)].

In this chapter. lac tose was the on ly sugar in SMP. WMP and MPC solutions:

therefore. the T 8

of amorphous lactose was used as the reference line. and the " X'' value

above the T g of lactose was observed. If there arc two or more amorphous sugars present.

the T g of multi-components can be predicted from equation 2.8 [Foster (2002)]. The

measurement and prediction of"X". the minimum T-T8 value th at causes insiantaneous

stickiness of the other specialty powders. including buttennilk, low fut/high lactose WMP

and vitamin fonified milk powder (GUM P) were also observed.

4-2

Measurement and Prediction of the Critical "X-·: Stickiness Temperature for Dairy Powder

T he value for "X'' has been shown to have a practical significance in dete1111ining

the onset of stickiness in dai,y powders rather than the glass transition temperature.

Expressing the effects o f temperamre and RH as a single parameter ofT-T, is th e most

appropriate method to represent the stickiness of dairy powders. The particle-gun rig was

developed to simulate the conditions in the spray drier and the ducting p ipe and cyclone.

The stickiness of powder particles occurs after ,1 relatively shon residence time in sp ray

driers. Thus. stickiness is a surfuce phenomenon and the point of adhesio n is the

instantaneous sticky point 'X'.

Instan t skim milk powder was first used 10 demonstrate and explain the typical

results obta ined fi-0111 the panicle-gun experiment in section 4.2. All the sample powder

tested followed the same analysis to obtain the critical '·X" value through the T-T, plot

and the stickiness curve. The samples are grouped into SMP. \\'MP. MPC. and specialty

powders and their critical .,x•· values arc represented in table fonns (section 4.3 ). The

critical ··x·· value for amorphous lactose in this work were compared with Chatterjee

(2003) in section 4.-l. Lastly. an equation was generated from multi-regression analysis

based on the samples tested. This equation predicts the critical --x· value for total fot

content less than 42% and only amoqihous lactose present in dairy po\\'dcrs.

4.2 AN INSTANT SKIM MILK POWDER

The raw data collected from the particle-gun rig was % deposi tion at a pa11icular

temperature and relat ive humidity condition for the air at the tip of the gun. Figure 4.2

shows a typical set of results for skim milk powder when the air temperature was kept

constant and the relati\'C hurnidiryofthe air was gradually increased. The data typically

show very little deposition as the RH was increased at a particular temperature. until a

critical relath·e humid ity value was reached. Al this point. namely the instantaneous

sticky point. the deposition starts and increases with increased va lues of relat ive

humidity. This infom1at ion was then used to plot the stickiness points for the powder on

a temperature versus RH graph, namely the stickiness curve (Figure 4 .3).

Any condition below the T1+ X line was identified to be a non -s ticky safe process

region and any condition above the T,+X line was identified to be a stic~-y region . This

4.3

Mcasun.."'lncnt and Prediction of the Critical .. X": S1ickincss Temperature for Dairy Powder

can be used as the chan in the plant and the operators can use it to adjust process

parameters in order to control the stickiness level. The stickiness in dairy powders is due

10 the amorphous sugars present. with the composition being used to predic t the T, cmve

based on the water activity and the temperature of the powder surface. The temperature

difference above T~ is imponam for controlling stickiness behavior of a panicular

powder.

As mentioned earlier. the raw data was first collected as % deposition of powder

on the plate at a constant temperature and increasing relative humidity. For a panicular

tem1>erature. zero% deposition occurs up to a cenain critical relati\'C humidity. At this

instantaneous sticky point. a sudden increase in % deposition occurs which increases as

RH is increased funher. These deposition data points are plotted and fi11ccl with a straight

line and the intercept i, identified 10 be the critical relative humidity or water activity for

the instantaneous s tickiness point. The % deposition vs RH (%) plot (Figure 4.2) shows

that a powders' instantaneous s tickiness condi tion occurs at reduced RH% for increasing

1emperan1rcs. Therefore. 1he powder adhesion phenomenon is a fonction of tempcr,m1rc

and humidity.

60 ♦ 80 SdegreeC

50 • • ■ 722degreec • •••• ••••••••• ••••• •••••••••• •••••• ••••••• •••••• ••

C: 40 .g ·.; & 30 ~ "i/. 20

10

.t. 68 2 degroeC

•

0 -1----.---• ..,...IO-'~ ii-l~~--l---~--­o 10 20 30 40 50 60

Relalive Humidity(%)

Figure 4.2 The % deposition of dry instant sk im milk powder plotted against relative humidity at constant temperatures .

4-4

Measurement and Prediction of1hc Critical --X": Stickiness Tcm,,craturc for Dairy Powder

The stickiness curve (Figure 4.3) shows that the Tf·X approach works. ln this

plot, the intercept RH value from raw data {Figure 4.2) at various temperatures is used 10

plot the instantaneous stickiness point for the SM P. A T 8-X line was fined to the data

using a least square of errors technique. The resulting critical ·X' value was 37.3°C. The

critical values of · x· obtained from Figure 4.3 and Figure 4.4 arc virtually the same.

considering the expe,imcntal error.

100

90 .

;; ~

80

70 g> ~

60

a> so

l 40

E 30 ,!

20

10

0 0

·••~--

- ~ ~ - ..

0 1

Tg lt1ctose ourve

0 2 03 Water aatv1ty

05 0 6

Figure 4 .3 The stickiness curve of the instant SMP from four experimental points fitted the curve above T 9 of amorphous lactose.

The T 1

of amorphous lactose was 1>rcdicted from water activity or relative

hum id ity based on the tiued empirical cubic equation [Brooks (2000)] . Figure 4.4 plots

% deposition against (T·T ,,) and this has collapsed the data from different temperatures

onto one line showing that there is one common Ts·X value at which instantaneous

stickiness for the powder occurs. The data were fitted by a regression line and the

intercept found and this was taken as the critical temperature "X" above T • of amorphous

lactose.

4.5

45

40

35

C 30 .g ·.;; 25 0 a. ~ 20 '$ 15

10

5

0 0

Measurcmcn1 and Prediction of the Critical "X"": Stickiness Temperature for Daiiy Powder

♦ 80.SdegreeC ____ _______ __ ...... _______ _ _ - ----· .. _ _ _ ___ _ _ _

■ 72 2 degreeC &682degreeC ----­• 60 Sd..nreeC

10 20 30 T-Tg

•

• ...

•

•••• • • - •·••••·-a

♦ S~e = 3 04 ~- .• ... 'R' t i O 805:f ....

40 50 60

Figure 4.4 % deposition plotted against T-T• for fou r different temperature data sets .

T-T, is the tempcrarure change of 1hc powder above its T, . At a ce11ain T-T .- 1he

powder deposit on the plate dramatically shoms up . This critical point of change in %

deposition indicates the instantaneous sticky point. and was found by extending 1he trend

line to zero % deposition. The critical T-T1 value. also known as the critical ·x· value

fonhe SMP. was found 10 be 37.9°C and slope 3.04 by this method.

The % deposition is a liinction of T-Tg. where all the dam co llapsed imo a

common intercept which is defined as the critical point ··x·. One of the advantages of

us ing T-T, plot is the distinctive stickiness point effect of amorphous lactose is

highlighted. and slope of1hc 1rend li ne indicates how fast the powder became sticky. T he

.. X" value can be used to construct 1he stickiness curve paralk l tO the T, arnoll)hOus

sugar curve.

Figure 4.4 can also be used to identi fy the temperature above T • where a given

%deposition occurs. based on the slope of the rrcnd li ne. The slope of1he trend line is a

measure of powder sensitiv ity with respect to temperature and RH changes above the

critical s ticking curve. The larger the slope, the more sensitive the powder is. meaning a

small change in temper.iturc and/or RH above the T g+X line results in large impacts on

4-6

Measurement and Prediction oflhc Cri1ical .. x··: Stickin~ss Tcmpcr.uur~ for DairyPowdcr

the stickiness of powder deposition on the plate. This is demonstrated in Figure 4.5.

Different powders will have d ifferent critical ·x· values and slopes.

100

90

G' 80 .. ~ 70 0, .,

60 :!<. ~ 50 .2 e 40 .. 0. E 30 ., ... 20

10

0 0

--T9t..1c.-.ose

.r=37 ,9 (()"(. oepos,llonJ

--xa39 5($ .. dCP0$1~10n)

--x-.:2.a (1$% dOPOS•tliOtl)

0.1 0.2 0.3 Water Activity

0.4 0.5 0.6

Figure 4.5 The slope of the trend l ine (T-T O plot) show how fast is the powder response to temperature and relative humidity change exceed the stickiness curve line.

Due to the time constraints (it t0ok about a week to generate a stickiness curve for

four temperatures} doing replicates for eve1y powder was unrealistic. However. a

replicate of the same batch of instant SM P sample was done on a different date. The

purpose of this exercise was to test the repeatability of the experiment.

The T -T, and stickiness curves of the two sets of experimental data are shown in

Figure 4.6 and -1.7.

4-7

~lcasurc,ncnt and Prediction of d1c Critical .. X": Stickiness Temperature for Dairy Powder

70

60

50

l • Instant SMP sample 1

■ Instant $MP sample replicate f- -- -• • • • • · · - · - · · · · - - • • • • • · · -

~ 40 g_ ~ 30 '#

20 -------

10

0 10 20

- ... - ... - . - ..... . Ill .

30

- T-Tg

Slope • 304 R' 2 • 0.8053

•

•

50

-skli>e ; 2 ·ss · -R'2• 0 789

60 70

Figure 4.6 repl icates.

% deposition plotted against T-T 9 for four di fferent temperature data sets of

100

90

80 6 70 ., ~ en ., ~

60

~ 50 ::,

~ ., 40 a. E

30 ., ... 20

10

0 0

Figure 4.7

T g lactose curve

0.1 0.2 0.3

Water Actici ty

0.4

Stickiness curve / Tg • X line

0.5 0.6

The stickiness curve of the instant $MP in comparison with the replicate.

4-8

Measurement and Pred iction of the Critical .. x··: Stickiness Temperature for Dairy Powder

The percentage error for the two sets of data were calcula1ed as below:

( I) The critical 'X' value from T-T • plot: 39.4±3.0 °C (7 .6%)

Sample: 37.86°C. Slope = 3

Replicate: 40.93°C. Slope - 3

(2)The cri1ical •x· value from tined s tickiness curve: 39.5±4.4 °C (I I.I)%

Sample: 37.3°(

Replicate: 41.7°(

The cri tical 'X' value obtained from the (T·T•) plo1 rnelhod has a less error than

the tined stickiness curve method. This is because the effects of RH and tempcranire are

combined by using the T~ parameter and all data points arc used to estimate the "X".

\\"hercas the fined s1ickiness curve method comains the erro r from finding the poin1 ai a

particular RH at each fixed tcmperanire. The "X" values from the rwo methods are

wi thin 1hcir e1Tor bounds and hence are not signiticamly different.

The replicate has a practical ly higher critical ·'X" va lue by both methods. The

replicate used 1hc same sample three months later. The difference maybe due 10 some of

the amorphous lactose crystallising during the 1hree months' storage. Reports show the

glass transition of amorphous lac1osc is depressed to 24°C at a critical water activity of

0.37 [Roos and Karel (1991b). Jouppi la and Roos (1994a). Bandyopadhyay et al.

(1987)]. T he water ac tivity (Aw) of both samples was measured and the replicate had a

higher Aw (0.398 @23.5°C) which is on the edge of the lactose crystall isation cond ition.

When lactose crystal iscs . it tries to am1y itself and therefore decreases in fi-ee vo lume. It

will release water as free water. hence increasing the water activity. Because it contains

crystalised lactose. it also reduces the s tickiness by having less amorphous lactose to

contribute to the instantaneous s tic~-y point. The other possible explanation of this

difference is just due to experimental error.

4-9

Mc-asurcmcnt ru1d Prcdic1ion of the Critical "X-': S1ickincss Temperature for OairyPowd<.T

4.3 CRITICAL 'X' VALUE OF POWDERS

4.3.1 Skim Milk Powders (SMP)

Three different specifications of SMP plus one replicate were tested and the

critical "X" values dctem1ined from T-T, plots and the titted stickiness curves (Table

4.1). The critical "X" value varied from 37.3°C to 42°C. The slopes varied from powder

to powder. with instant powder 2 having a distinctive ly lower slope than the other skim

milk powders tested, making it a lot less sticky than other skim milk powder.; at

equivalent conditions abO\'C their critical "X" values. The reason for this is unknown.

The composition data showed relatively small differences compared to the other skim

milk powders tested. It might ha\'e differed in pan icle size or density or other physical

propcnies that were not identified. or it could be an outlier. The data analysis for

individual SMPs in Table 4.1 can be found in Appendix A2.

Table 4.1 A summary table of c ritical " X" value of SMP tested .

S kjrn Fat Lactose Prote in Critical Slope Critical

Milk (%T S} (%TS) (%T S) "X" ~ (T•Tj) (T-T t) ' X' (On NI sticldnm cur,t)

Powders (•C) {°C)

Instant 0.62 57.84 3427 37.9 3 .04 37.3 SMP I Instant 0.62 57.84 34.27 40.9 2.95 41.7

SI\IO> I (replicate)

Instant 2 0.79 52.98 38.19 39.7 0.34 38.1

Med ium 0.83 53.01 38.05 40.2 3.33 40.3

Heat SMP

. 0.80 57.56 34.63 NIA ':-/.IA 41.2

*Chatterjee (2003) particle-gun rig expcrimc111a l result (temperature below 50°C)

4.3.2 Whole Milk Powders (WMP)

Five WMP samples were tested and the critical 'X' values found are shown in

Table 4.2. The critical ·x· value was in a range of 33.7°C to 40.2°C and slope varied

from 0.4 to 1.2. The slope characterized ind ividual powders. It plays a big ro le in

4-10

MC':1surcmcn1 nnd Predic1ion of the Critical .. x··: Stickiness Temperature for Daisy Powder

detem1ining how fast the powder will stick in response to different condit ions. The data

analysis for individual WM Ps in Table 4.2 can be found in Appendix A2.

Table 4.2 A summary of critical "X" value of WMP tested.

Whole Milk Fat Lactose Protein Critical Slope Critical

Powders (%T S) (%T S) (%T S) "X'\T-Ta} (T -Tg) "X "(fiUfd $lkkil'IC'-S$

(•C) '""'> (• C) Agglomerated 31.09 37.82 25.91 38.0 0.78 37.5

High heat 27.38 38.33 27.07 33.7 0.42 34.0

Regular I 28.44 40.68 24.9 37. 1 1.21 35.5

Instant 29.75 39.77 24.69 38.2 0.36 40.2

Regular2 26.9-l 40.93 25.91 40.0 1.07 39.4

WMP .

27 96 42.48 24.65 NIA NIA 48.6

• Chaucrjcc (2003) particle-gun rig experimental n:suh (temperature belo" 50°C).

Comparing the c1itical "X'' values from SM I' and \\'MP powders shows them to

be overlapping with each other. despite their di fti:rcnt fat contents and lactose contents.

The lactose comcnt was higher in SMP and relatively lower in WM P. Looking at the

average value from WMP (37.4±2.9°C) and SMP (39.4± 1.7°C). the ei,·or bands overlap

with each other (95% confidence interYal). These average values arc from the T-T, plot.

Despite the difference in types of powders. particle size. compositions. spray drier.

atomizing system and process scmp conditions. SMP and WMP arc essentially sticky ac

similar points. However. they are diffcrem in degrees of response to the changes in

temperacure and RH conditions. In general. the slope ofSMP is much higher than WM P:

at any particular condition that exceeds the stickiness curve, WMP deposits arc ligh ter

than SMP. This cou ld be due to the lower lactose content of WMP compared to SMP

which means that. despi te the two types of powders having simi lar X factor.. the SMP

will build deposits faster than WM P under similar cond it ions and hence can be regarded

as a "'sticker" powder.

4 -11

Mcasun.'T11cn1 ond Prcdic1ion of the Critical "X": Stickiness Temperature for D;iiry Powder

4.3.3 Mi lk Protein Concentrate (MPC) and Whey Protein Powder

The lactose conten1 in MPC was much lower than SMP and \V MP. The h igher

the concen1ration oflactose. the lower the critical "X" was found to be. Table 4.3 shows

1hat 1he lower the lac tose concentnltion . 1he less sensi1ive the powder is 10 air conditions

because of the lower slope value . T he larger ' ·X" values ind icate that a powder is more

tolerant for high 1emperanire and lor re lative hu midity cond itions. When whey protein

was fi red 1hrough the gun. 1herc was on ly a smear of powder deposited on the plate.

However. the I% lactose effect on powder stickiness was still eviden1 in whey protein

powder. with a cri1ical "X" value of 50.5°C. The data analysis for individual MPC and

whey protein powder from Table 4.3 can be found in Appendix A2.

Table 4.3 tested .

A summary of critical " X" value of MPC and the whey protein powders

MPC and Whey

protein Powdrrs M PC44

MPC56

M PC 70

MPC85

Whey protein

Fat (%TS)

0.83

1.35

1.46

1.67

0.52

Lactose (%TS)

47.56

3 1.29

18.16

4.17

1.05

4.3.4 Specialty Powders

Protein (%TS)

43.52

59.36

72.86

88.54

96.34

Criti cal Slope Critical "X" (T•Tlt) (T-T,.) "X" (flntd stk'kinl~ rurH)

c·c> c•c) 42.7 2.64 44.5

47.5 0.34 47.7

50.0 0.08 49.7

49. 1 0.04 5 1. 1

50.1 0 .04 50.5

T here arc some spec ialty powders produced by Fontem1. These include white

cheese. buttennilk. and powders containing added vitami ns. T he critical ·x· values for

the powders tested were identified as follows Table 4.4. The data analysis for individual

speciality powder from Table 4.4 can be found in Appendix A2.

4-1 2

.Mcasurcrncm and Prediction of1hc Critical ··x··: Stickinc-ss Temperature for Dairy Powder

Table 4.4 A summary of critical "X" value of cheese , buttermilk and GUMP powders tested.

Speciality f at Lactose Protein Critical Slope Critical

Dairy (%TS) (% TS) (% TS) " X ''(T·Tg) (T-Tg) "X" (f11rtd s1id,it1ts.S

Powders c•q Nno)C°C)

White cheese 42.05 27.90 21.23 28.4 2.3 28.5

White cheese .

42.05 27.90 2 1.23 NIA N/A 19.4

Snack Cheese 30.89 3388 18.92 NIA NIA 24.7

Buttcnnilk 9.33 5 1.3 3 1.92 39.3 1.3 36.9

GUMP 15.46 49.48 27.84 40.7 0.3 40.7

• Chaucrjee (2003) panick-gun rig experimental result (temperature below 50°C).

Included in table 4.4 is data from Chatterjee (2003 ). for white cheese powder and

snack cheese powder. This data was collected a t a low ternpcrantre (50°C). with RH

(60%) co11ditions. There is approximately 9°C difference in 1hese results compared to

Chatterjee (2003)'s. The reason for this difference is unknown at present. and indica1es

more work is needed.

4.4 AMORPHOUS LACTOSE

Amorphous lac1ose was made on a Jab scale spray drier using the me1hod outlined

by Brooks (2000). Brooks (2000} used the blow test method to de1errninc the critical

tempcrarure of the instantaneous stickiness poin1 above its glass transi1ion temperarure to

be 25°C. Chatterjee (2003} used the panicle-gun rig a1 low temperatures (below 48°C)

and h igh RH (60%) and fined a curve to find the critical stickiness point to be 26.1 °C.

Amorphous lac1ose was tested in this work using the particle-gun rig at high temperatures

wi1h a range of process conditions (80°C) and found the instantaneous stickiness point X

to be 24.2°C. When the two sets of data from the particle-gun were combined (Figure

4.8). the s1ickin ess curve data were best fitted with an "X" value of24.7°C.

4-13

~lcasurcmcnl and Prediction of the Critical '·X": Stickiness Temperature tOr Dairy Powder

120

100

<> 80 ,_ e il l! 60 8. E ~ 40

20

--Tg• Atnorpt,ous lactose --Tg•2.i. 2 (high tempdat.a pl)

• High temp e~ riimental points ctia:terjee (2003)

oL--------------------- +- ~--~ 0 0.1 0 .2 0.3 0.4 0.5 0.6 0.7 0.8 0 .9

Water activity

Figure 4.8 The stickiness curve of amorphous lactose with reference line T9, including Chatterjee (2003)'s result.

The result shows 1ha11he ins1an1ancous s1ickiness ofamorphous lac1ose is around

25°C. Both 1he blow 1cst and panicle-gun tes1 arc in agreement. II was confirmed that

1he amorphous lac tose instanlaneous ,ticky point is in a rnnge of2-I 10 26°C where i1 will

adhere 10 ma1erials and where instantaneous cohesion between the panicles occurs.

4.5 PREDICTION OF THE CRITICAL "X"

The origina l idea for generating the s1ickiness curves for a range of powders was

to be able 10 pred ic l 1he c ritical ·x · value of a powder from i1s kn0\\11 compositi on. The

only corre la1ion found between 1he c ri1ical "X" value of the powders and their

composition was be1ween 1hc % lactose co111en1 :md c ri1ical ··X" and between the slope of

the %deposi1ion line/ T-T1 plo1 and the % lactose content. This is shown in Figure 4.9

and 4.10. The corre la1ion was found to be stronger when the lac1ose composition was

expressed as % solid 1101 fa1 (SNF). (R = 0.85 compare 0.68). The wh ite c heese was

excluded because its fat content was more than 42% and it was obviously an outl ier.

4-14

Measurement and Prediction of the Critical .. X": Stickiness Temperature for Dairy Powder

60

0 C .g 30 j ~ ?< 20 ;; ·" ·c 0 ,o -------

y = -0.2622.x + 53.055 R1 = 0.8543

♦ TS %Lactose

• SNF % Lactose

- Linoar (SNF % Lactose) --Linear (TS %Lactose

y • -0.2362x + 50.194 R1 = 0.6801 •

oL----.---~----------------,------0 10 20 30 40 50 60 70 80 90 ,oo

% lactose

Figure 4.9 The correlation between the c ritical " X" value and % lactose of powder (total fat content less than 42%) tested .

A multiple regression analysis was used to fit the independent variable critica l 'X'

value. It was established from a set of dependent variables. %TS fat and ¾ TS lactose was

used. mainly because the two components relate to the s tickiness mech,misms. Although

%TS protein and •ns lactose can also be used. the resulting critica l ·x· \\'as calculated

to be simi lar. The regression equation worked out co be as follows:

Critical 'X'-53.91 ~(-0.20337)*%TS Fai.,.(-0.2648) • ¾ TS lactose (fatS42%TS) (4 .1)

For any dairy powder that has fat less or equal to 42%. the critical · X' value can

be predicted from equation 4.1 with ± 10°C. Although. this 10°C error margin is large .

the plot demonstrates an overall trend between the critical ·x· va lue and the %

amorphous lactose in bulk. Since. the instantaneous s tickiness of the powder has been

shown 10 be a surface phenomenon the surface composition of a powder in relations to

the c ritical ·x• is worth further work.

4-15

Measurement and Prcdic1ion of the Critical "X .. : Stickiness Temperature for Dairy Povl'dcr

Figure 4.10 shows the slopes of the %deposition T-T g plots versus %TS lactose in

the powder. The data is scanercd and for simi lar % lactose powders, there are a large

range of slopes among the powders tested. Despite the scatter. and leaving out the 100%

amorphous lactose data point. it can be seen that most powders with over 30% lactose

beco111e much more sticky quickly than powders with less than 30% amorphous lactose.

A regression of the data points in the 30-60% lactose range gives 32% with a R2

value of

0.5. It should al so be noted that some powders in the range did not become as sticky as

might have been expected from their %lac tose content. The reasons for the non-sticky

behav iours of these powders need fitrther investigation .

4 ♦WMP

■SMP

3.5 - - - - - . - - .. ,. & MPC

• " 8 u1term1lk

3 -------- ■ Whey

- e GUMP

~ 2.5 x Amorphous L.ae1 O> I-

t:. 2 Slope = 0 ·1

~ R'2 = 0.5 " 0 1.5 . - - - -

in

··---- ---------•

0.5 •• • • • 0

/ 0 20 40 60 80 100 32

Lactose (%TS)

Figure 4.10 The interaction of slope as function of (T-T9) and % lactose in bulk.

4.6 COMPARISON OF STICKINESS CURVE USING CRITICAL 'X' APPROACH

The stickiness points obtained from other researchers using different methods

were fined to a line ·X' above the T, of a111orphous lactose. The stickiness curves of

similar powders. but using the d ifferent techn iques, is compared.

4 -1 6

Mcasurcmcnl and Prcdiclion of the Critical .. X .. : Stickiness Temperature tOr Dairy Powder

Figure 4. 11 shows sticky points of amorphous lactose measured by paniclc-gun

rig are very close to the st ickiness curve of whey powder measured by the cyclone

stickiness test. The stickiness curve ofSMP (57.84 lact. %TS) is also plotted. Therefore.

the stickiness cuive of a whey powder from the panic le-gun rig can be predicted to be

somewhere between the stickiness curve of amorphous lactose and the SMP, which

would obtain a higher critical .. X" value than the cyclone stickiness test. The difference

in the critical .. X .. observe between the cyc lone stickiness test and panicle-gun rig is

probably due to the different airflows and panicle trajectories and the residence times

between the two methods resulting in diflcrcnr sticky end points. Both of the s tickiness

curve lines arc in the rubbery state of amorphous lactose. Powder pa11icles tra\'el in the

humidified air for 1-2 minutes in the cyclone sticky 1cst compared to a travelling time of

0.05 seconds in 1he pai1icle-gun rig. A longer 1ra\'elling time allows moistures to diffose

from the surfuce layer to the inner layer and therefore. more flow and more contact points

for the panic I es to adhere 10 the walls or cohese 10 each other. Particle-gun rig measures

1hc panic le stickiness at the surface o nly and on ly allows enough time for the surface of

1hc panicle to equilibrate with the sun·ounding conditions. The powder panick in the

cyclone sticky test gets stickier earlier than in the panicle-gun at the same air conditions.

4-17

90

80

~ 70 e "' .. ~ e so il E 8. E so • ...

40

30 •

Mcasurcmcnl and Prcdic1ion oflhc Critical ··x .. : Stickiness Temperature for Dairy Powder

0.1

• Cydon• sfieklMiu tut (Whey powdet stleky pts> --T g amorphous lact.

• Par1k:lt-gun rtg (Amorphous lactou sticky pts) • p&rtklt-taun rig (In.slant SMP stlcky pts)

par11Clt-taun rig (T9 + 37.3 dtgr .. C) • .. • ...... .. j.__..,c::,yc:,,1,,on:::,ec,1,:,1k:e,kl,enee,1:.,.•.:::te:.:.••,..<T-"9"•:,22"'.8e.;de,e:,,:.Ce,1c__ __

• •

0.2

X

X

··~ ······ .....•

.. , Water activity

X

•

.. , ••• •••

Figure 4.11 Stickiness curve of whey powder measured using cyclone stick iness test (Boonyai et al. (2004)) and amorphous lactose and instant SMP measured by particle -9un rig.

Fii,,>t1rc 4 .12 uses the T ~ of amorphous lactose as the reference line and shows 1he

best fit cun·e line critical X factor 1hrough the skim mi lk powder sticJ..-y end point data

obtained from vatious dilfcrcm techniques. The stirrer test results in a lo wer critical 'X'

value of I 7.5°C and the fluidised-bed rig had 1he va lue of21°C and particle-gun tig had

1he largest ctitical · X' value of -11.3°(. The increase in the cri1ical ·x· measured

indicates that the s1 ickiness of powder is a fonction of the 1>article velocity. In the stim:r

test. the powder paniclcs arc in an almost stationary state and cohesion stickiness is

measured. Although particles 1cstcd in the fluidised-bed were moving when the

stickiness end-po int was measured. the veloc ity a1 which the particles were moving were

re latively s lo w compared to the paniclc-gun rig. These differences in s ticky point critical

X measurement are due entirely to the differences in the technique used. and indicate that

the level of s tickiness is a fonction of the dynamics of the panic le as well as the air

conditions. Real blockage data from actual plants will be needed to establish which of

the techniques is most appropriate for relat ing thi s data into usable curves for industry.

4-1 8

~1casun.,nen1 and Prcdi-ction of the Critical ··x··: Stickiness Tc,npcrarurc for Dairy Powder

160

1a ----- ---- --------- --·······-

120

it e 100 "' ! ; 80

i < 60 ~

20

■ . ... • • ....... ..

______ .. _ ... ..

♦ PartleI0°gun rig (Tg♦41 .3 dogC)

■ Fluidlse-bed rig (Tg• 21 dogC)

A Slirror Iest (Tg+17.5 degC)

- Tg amorphous lactose

• •

0 L-----------------~ ~--'-.--."',---~ o.o 0.1 0.2 O.l 0.4 o.s 0.6 0 .7 0.0 •••

Water Activity

Figure 4.12 Stickiness cu rve of SMP measured by part icle-gun rig (Chatterjee (2003)), fluid ised-bed rig (Chatterjee (2003)), and s tirrer test (Hennig s et al. (2001)).

4.7 CLOSURE

This work has established the process of obtaining the instantaneous stickiness

point for various dairy powders using a particle-gun rig. The critical --x-· values for

WMP. SMI' and MPC ha,·e been found to vary between 3-1°( and 51 °C. A plot of

%deposition versus T-T,,-!l, .,0.,, succcssfolly combined the temperature and relati,·e

humidity fact0rs. This indicates that i1 is the pro perties of amorphous lac1ose causing

sticking in dairy powders. The critical ·x· value is the tcmpcranire 1hat exceeds 1he T, of

amorphous lac1osc when ins tantaneous s1ickiness occurs. In addition. the pa11icle-gun rig

demonstra1ed tha1 powders with grca1cr than 30% amorphous lactose are more likely to

cause blockages 1han powder with less 1han 30%. Both 1he critical •x· value and 1he

slope arc unique 10 the powder. The stickiness curve or Ti-X line is useful for mapping

plant operJting conditions on a chart form and the slope enables a decision to be made

about how close to the cri1ical point a plant should be run for a particular powder.

For powders that have a total fat content less than 42%, the stickiness is mainly

due to amorphous lactose when T• is exceeded by the critical "X" value of the powder. A

4-19

Measurement and Prediction of the Critical .. x··: Stickiness Temperature, for Dairy Powder

regression equation (equation 4.1 ) was generated. based on the critical · X · value obtained

from the experimental data of various dairy powders. to corrcla1e 1he effec1s of

composit ion on X. Although an overall 1rend is observed. funher smdy is required to

investigate the smface composition in re lationship wi th the cri1ical ·x·.

Comparison of the panicle-gun rig X factor wi1h the X fac1or measured by 01her

techniques shows that the foc1or is a function of1he technique used with higher kinc1ic

energy panicles giving h igh X factors for similar powders. More work is requ ired to

relate th is work to industrial experience.

4-20

CHAPTER 5

STICKINESS DUE TO FAT

5.1 INTRODUCTION

Stickiness Due to Fat

Foster (2002) showed that sticking and caking problems due t0 the fat-melting

mechanism is only significant when the total fat content is higher than or equal to -12%.

resulting in a high level of surface fat. There was a linear relationship between the total

fat content and the surface fat content expressed in tem,s of the specific smfacc area.

Whi te cheese powder and high 'low fat cream powders were tested and are

classified in the high fat content powder category. The fat ctlect of these powders is

discussed in this chapter. During the process. these powders were exposed to high

tempemturcs and all the surface fat was molten. causing smearing on the wall and

cohesion with each other.

5.2 WHITE CHEESE POWDER (42.05 %TS fat}

A similar stickiness pauem 10 other dairy powders was observed at a panicular

1emperamre with changing humidity (Figure 5.1 ). The stickiness point at an increased

tempemture occurred at a decreased relative humidity. There was a small amou111 of 1%

in itial deposition. Because the % deposition scale on Figure 5 .1 is large. the trend is not

so obvious. A constant de1>osition at tcmpcramres above 60°C shows at high

temperamres all the surfuce fat was molten. causing a constant adhesion rate. However.

there was Slill a temperature and humidity effect on the % deposition.

Chatterjee (2003) used the partic le-gun rig for white cheese powder a1 the low

tempemrurc range. below 50°C. He found that at increased 1empcramre conditions. 1hc

initial deposition increased with the same pattem. Fat-melting was aftcctcd by the

tempcran,rc changing. A linear relationship between% deposition and temperature was

found for white cheese powder at temperamrcs below 50°C.

5-1

80

70

60

* 50 C: .Q ·iii 40 0 C.

8 30

20

10

0 0

♦80 5 degree C

■ 72.2 c:egreeC 4 68 2 degreeC • 60.5 degreeC

5 10 15 20 25 RH%

30

Stickinc-ss Due to Fat

35 40 45 50

Figure 5.1 Deposit ion of the spray dried white cheese powder (42.05 %TS fat) tested on the particle-gun rig .

Since the stickiness is effected by both 1empcn11urc and humid iry of the air

condition. the T-T i parameter was used to combine those two factors as shown in Figure

5.2. The data for different air temperatures. although more scaucrcd than the low fat

dairy powders. still congTegate around a single line with a common initial stickiness

point. The initial stickiness point for the white cheese powder was obserYed at an "X" of

28.35°C. This shows the increasing stickiness is due 10 an amorphous lactose mechanism

starting at this point. A constant deposition (ranging 0.5 0.8 %deposition) before the

critical T-T, point indicates the surfuce fat w:,s molten at high temperature and was the

cause of initial adhesion.

5-2

Stickiness Due 10 f;11

♦ 80 5 degreeC _ ■ 72 2 degreeC

!_ _____________ _ • 68 2 degreeC

80

70

60 • 60 $ degreeC ___________ •••• _________ _

• replicate 80 5 degreeC

a'/. so C:

.2 -~ 40 & "' 0 30

20 •

10

• Slope = 2.255

_____ .. _R'2 = 0.7~18. ____ _

• • • ' • ••

0 ~---..... ..._- ~ - ,o_;!.a,~ -=--- ......,- --------. 0 10 20 40 50 60

T-Tg

Figure 5.2 The cheese powder (42 .05 ¾TS fat) stickiness points at temperatu res above T

9 of amorphous lactose.

The RH of the air at the individual s tickiness poi111s a1 a particul.ir temperature

were identified from Figure 5.1 and then they we,·e used to lit a curve above the T, of

lactose calculated from both the Gordon-and-Taylor equation and the cubic equation. ·x· was found 10 be 22.6°C and 28.5°C respecti\'ely (Figure 5.3 ). The data points from

Chatterjee (2003) are also planed. The curve from the Gordon-and-Taylor equation fits

both the high and low 1emperamre data sets . In most cases the cubic equation gave a

better lit. The reason it did 1101 in this case is unknown. Despite the difference between

the two T~ - X curves. for typical process cond itions the likely region lies within the

marked "box" o n Figure 5.3. The two curves a lmost supe1impose on top of each other in

this region: therefore. e ither curve can be used as the stickiness curve.

5-3

100 -

90

80

u 70 • 60 ~ 2

50 e ~ 40 E "' ... 30

20

10

0 0

Figure 5.3

..... - _ .. - - - ~ ~ ~ -

.. --- - - -~ - . --. .........

0.1

• •••

0.2

• • •••••• • .

0.3 Water Activity

Stickiness Due to Fat

--Tg Lactose ■ High temp experimenu,t Pt

• --Tg •28.5 Cu1>,c - Tg •226G-T

0.4

• Ch1;1tle1Jee {2003) T tactose,.GT

0.5 0.6

The stickiness curve for cheese powder (42 .05 ¾TS fat) including tho fitted

T 9

+X line from the Gordon-and-Taylor equation and fitted T0+X line from cubic equation.

5.3 LOW FAT CREAM POWDER (55.84 %TS fat)

Again. the % deposition of dry powder versus RH% (Figure 5.4) shows adhesion

phenomena are dependent on temperamre and relative humidity of the air. At constant

temperamre and by increasing the humidity. a poinr is reached where there is a dramatic

inc rease in % deposition. indicating instantaneous stickiness. For each temperature. a

similar trend line is observed and the intercept of the trend line was idenrified to be the

stan of instantaneous stickiness at that panicular tcmpcmrure and RH condition.

Increasing the tcmpcramre reduced the RH at which this point occurs. Since the powder

is classified as one of the high fat powders. it is expected that an initial depos ition will be

seen. This is shown in Figure 5.5 when the% deposition scale has been expanded.

5-4

50

45

40

35 C

.g 30 ·;;; g_ 25

" 0 20 ~

15

10

5

0 0

Stickiness Due 10 Fat

• 80.5 degreeC •••••• • •• •••• ___ ••••••••• • ______ ••••••••• • __ • __ • __ •

■ 72.2 degreeC • ........... ---------- -· .. ·· ....... ... . ...... ...... ... ----• 68.2 degreeC • 60.5 de reeC

5 10 15 20 25 30 RH (%)

35 40 45 50 55 60

Figure 5.4 Deposition of the spray dried low fat cream powder (55.84 ¾TS fat) tested on the particle-gun rig .

Figure 5.5 illus1n11cs the fat effect on init ial deposition . Clearly an initial 0.5%

depositio n within the temperature range between 60-80°C has occtmed. In the

tcmpcramre range. all the fat o n the surfucc will be molten. and therefore. :1 constant

deposition was observed. In contrast with low fat powders. such as \VM P or SM P. the

initial deposition was virtually zero. In those cases. amorphous lactose was the only

mechanism responsible for the po wder stickiness.

5-5

Stickiness Due- 10 fat

5

4.5

4

3.5 C: ,g 3 ·;; g_ 2.5 .. 0 2 ;!.

1.5

0.5

♦ 80.5degreeC --- -- -------• 72.2 degreeC •• ___ ______ _ .. 68.2 degreeC -""- -• 60.5 degreeC • · - - - • • • - •

• ...... .... o-1---.--------------------....--~-~

0 5 10 15 20 25 30 RH {%)

35 40 45 50 55 60

Figure 5.5 Close-up of the deposition of the low fat cream powder (55.84 %TS fat) tested on the particle-9un rig .

\\'i1h low fat c ream powder. both f.11-melting and amorphous lactose mechanisms

contribute to the powder s1ickiness. The lactose mechanisms can be dis1inc1i"ely seen in

Figure 5.6. wi1h the diftcrem temperamre data being reduced 10 one line when ploued

agains1 T-Tg. The critical T-Ti was identified to be 36.9°C. where powder pa11ic lcs stan

10 s1ick on 1he plate over and above 1hc initial deposi1ion. This critical "X" value of

36.9°C was achieved by various combinations of 1cmperantre and relative humidi1y

conditions. Comparing this critical ·x· value. the white cheese powder (28.35° C) is

much lower 1h:,n the low fat cream powder. This is due to 1he higher lactose composition

in white cheese powder than cream powder (27.9 % TS versus 24.3 7%TS). On the other

hand . more initial depos ition was expcc1ed in the cream powder because of its higher fat

con1ent. T here was no s ignificant increase within experimental cn·or in deposition caused

by the increased fot conte111. This might be due to the fact that at these higher

temperatures all the fat is molten. and for both powders there is a lready sufficient fat to

have an effect.

5-6

so

40

C .9 30 " ~ 0

"' .. o 20 .,_

10

0

♦ 30.Sdegre,e,C

■ 72.2 de9rMC - • - & 68.2 de,grHC

,i 60.Sd""9reeC

10 20 30

• ..

40 so T-Tg

Stickiness Due to Fat

Slope= 2.24 R"2 = 0.8005

60 70

Figure 5.6 The low fat cream powder (55.84 ¾ TS fat) s tickiness points at temperatures above TO of amorphous lactose.

Figure 5.7 shows the stickiness curve of T, - 35.9°( was the best fit from the

cubic equation ofT, for amorphous lactose. Th is value coJTesponds IO the point where

there is additional % deposition over the fat-melting mechanism when 1hc powder was

exposed 10 high process 1emperarnrcs. The fat mechanism is only dependent on 1he

temper.nurc. not the relative humidity. The critical ··x·• values obtained from Figure 5.6

and 5.7 are statistical ly 1hc same.

5.7

100

90

80

i'.> 70

60 i!! 2 e 50 ., Q.

E 40 ., ... 30

20

10

0

T g lactose curve

0 0.1 0.2 0.3

Water Activity

Stickiness Due to Fat

0.4 0.5 0.6

Figure 5.7 The stickiness curve for low fat cheese powder (55.84 %TS fat) including the T••X line.

5.4 HIGH FAT CREAM POWDER (71 .79 %TS fat)

The high fut cream powder contained 71. 79% TS fat and is used for

manufucruring ice cream. bakery products and powdered drinks. sou ps. and dip. A

constant initial adhesion due to fat-melting was expected in the high tempernturc range.

This ,ms because fot is completely molten at 40°C. Howe,·er. the trend is slightly

diftcrcnt from that seen for low fat cream powder (CP55). Figure 5.8 shows 10%

deposition of powder at 60°C and 5% panicle deposition above 68°C. The diftcrcnce is

clearly due to different fat content in the powder and exposure temperatures. The fat

effect on the% deposition is funher studied in the followi ng section.

5-8

80

70

60 ,e 0 50 C: 0

:-2 40 "' 0

g- 30 Q

20

10

0

0

♦ 80.5 degreeC ■ 72.2 degreeC

__ .. 68.2 degreeC • 60.5 de reeC

5 10 15

• •

20 25 RH %

30

Stickiness Due 10 Fat

..

35 40 45 50

Figure 5.8 Deposition of the spray dried high fat cream powder (71 .79 ¾TS fat) tested on the particle-gun rig .

The amorphous laccose effoct is also obvious in Figure 5 .8. A T -T~ plot (Figure

5.9) was used to combine the te 111peramre and RH eftccts on powder depositS. The darn

points are quite spread and a range of critica l T-T ~ values (25-35°C) can be detcnnined.

Although the amorphous lactose effect is observed. there was major deposition due tO the

molten fat mechanism.

5-9

Stickiness Due lO Fat

70

65 60 55 50

~ 45 0

c: 40 .Q ·ii\ 35 g_ 30 ., o 25

20 15

10 5 .

High fat cream powder 71.79 % TS fat (Feb/ 2003 sample)

• • 80.5 degreeC . • 72.2 degreeC

• 68.2 degreeC • 60.5 degreeC

• ---·- )(--

.........•• ·················-······• ·- - ·---

........ - • - •• - - - .... - . • • • - - "Ji. •......• - - - - - - - - -• •

• • .. •

• ; • - -..- -

t I,.

•

.. -.. • ♦

-·---------•

~- - - - ----. -

♦

• • .. .... •• - - ♦ •• ♦ o+--------~-----------------.

0 10 20 30

T-Tg

40 50 60

Figure 5.9 The stickiness of high fat cream powder (71 .79 %TS fat) caused by molten fat and amorphous lactose.

5.5 STICKINESS DUE TO FAT-MELTING MECHANISM

The molten fat effect of high fat powder as an initial% deposition is obse rved in

Figures 5.5 and 5.8. It is dependent on temperature only. Therefore. the fol lowing

experiment was carried out to observe 1he % deposi1ion as a fi.111 ction of 1cmperature.

while keeping the RH low to ensure there was no lac1ose cffoct. Low ' high fat cream

powders were used 10 de monstrate the molten fat effect.

For low fat c ream powder (CP 55). increasing 1he temperature from 30°C to 78°C

caused more powder panicles to stick to the plate. The relationship is linear as shown in

Figure 5.10. Possible explanations are: As temperature is increased, the fat on the

surface becomes less v iscous and more flow can occur. The increased tcmpcran1re also

decreases the surface tension of the fat. which enables more fu1 to flow out of the

capillaries in the powder panic les, increasing the amount of surface fat. These effects

combine to inc rease the % deposition tha1 occurs as temperature is increased.

5-10

Stickiness Due to Fat

0.8

0. 7 - - - - - - - • • • • • • • • • - - - • • - • · • • • • • • - • • - • • • • • • • • - • • - - - - • •

0 .6 - - - · · • · · · - - - - - - - · · • • • • • · - - - - - - - ·

C .g 0.5 .; &, 0.4

~ ~ 0.3 •

0.2

0.1

----------- -- - -------~- - ---- --

o+---"'T"----------------------0 10 20 30 40

Temperature (°C)

50 60 70 80

Figure 5.10 Deposit ion due to fat in low fat cream powder tested on the particle-gun rig w i th increasing air temperature and keeping RH low to avoid the amorphous lactose mechanism.

The •• deposirion due m fat for high fat cream powder (HFCl') is diftcrcnr from

low fat cream powder (LFCP). Overall. al the same rempernnirc. HFCP has a higher ~.

deposition than LFCP. For example. al 30°C. approximately 0.35% of LFCP snick to the

plate. compared with 5% of HFCP deposited on the plate. This is because h igh % fut in

bulk leads 10 a higher surface fat content [Buma (1971a).Foster (2002). F~ldt e1 al.

( 1993)].

The other diftcrcnce is that the shape of the HFCP st ickiness due 10 the fat curve

as a func1ion of temperature (Figure 5.11) is very difll:rcm from LFCP (Figure 5.10).

The high temperature range (above 60°() daia points. collected for the 2003 sample. arc

the in1ercep1s of1he horizontal lines from Figure 5.8. Data points al tcmpcramres below

50°C arc from Chatterjee (2003) using the same method for identifying the ini1ial %

deposition.

Due 10 the sho 11age of the 2003 sample. the s tickiness caused by molten fat for the

whole temperan,re range could not be carried out. However. a 200 I sample was fou nd in

the lab and used to do the experiment. The 200 I sample gave much higher% deposition

5-11

Stickiness Due to Fat

than the 2003 sample at the same temperature, but they have a similar shaped curve, The

curve from the 2001 sample was shifted upwards. A linear relationship between %

deposition and temperature over the range 28°C-45°C was observed. This increase in %

deposition may be explained by an increase in temperamre. causing more suiface fut tO

melt. When the tcmperamre exceeded 45°C. a decrease in % deposition was observed. A

possible explanation is that since the majority of the paniclcs were made of fat. as

temperamre increases with all the molten fat. the sm,cture of the panicle collapsed and

became liquid and less viscous. When it hit the plate. it splashed and some of the pai1iclc

splashed outside the stainless steel plate. causing less deposition to be recorded. The

higher the tempcramre. the less viscous the paniclc became and when it contacted with

the plate. it broke into smaller droplets ofliquid panicles.

55

50

45

40

.~ 35

~ 30 0

~ 25 0 ~ 0 20

15

10

5

0 0

1 ♦ 2001 s ample teste<t in 2003 , 1■ Feb/2003 samplt tested in 2003~

• •

• • 10 20 30

• •t • ••

·------• • .. - --- - . - -

40 50

Temperature ("C)

• • . ------- . •

-------• ~-------

• • • 60 70 80

Figure 5.11 Deposition due to fat in high fat c ream powder (2001 sample and 2003 sample) tested on the pa rticle-g un rig .

90

The distinctive upward shift of the curve in the 2001 sample is shown in Figure

5.11. T he best explanation of the 2001 sample having higher deposi1ion than the 2003

sample lies in its age. The wa1cr ac tivity and temperamre of the 2001 and 2003 samples

were measured before experiments too k place (0.42 at 24.0°C and 0.39 at 22.5°C

respectively). T he water activity meter is Decagon, model CX I and has been upgraded to

CX2. The repon shows the g lass transition of amorphous lactose is depressed to 24°C at

5-12

Stickiness Due to Fat

a c ritical water activity of 0.37 [Roos and Karel ( 1991 b). Jouppi la and Roos (1994a).

Bandyopadhyay et al. (1987)). T he resu lt shows that amorphous lactose in the particle.

crystal ised during s torage. which causes more fat to be present on the surface as the

lactose crystals push the fat ourwards [flildt and Bergenstahl (1996a). Faldt and

Bergenstahl (1996b)). A quick test was used to check the powder for lactose c rystals by

observing the powder under a microscope through polarizing fi lte rs. Crysta lli ne lactose

re-rotates the plane of polarized light. allowing it to be observed under two opposing

po larizing lenses. Amorphous lactose does not re-rotate the light and thus darkness is

observed. Figure 5. 12 shows there were lactose crystals in the 200 I CP 70 samples.

They were no t obsen•ed for the 2003 sample.

Figure 5.12 2001 high fat cream powder sample was checked for crystall ine lactose under a polarising microscope at SOX magnification.

5-13

Stickiness Due to Fat

5.6 CLOSURE

This work has mai nly focused on the stickiness of high fat dairy powders. Molten

fat has a significant effect. causing smear and stickiness problems d uring processing.

However. the amorphous lactose effect is still present. especially for white cheese and

low fat cream powders in T-T, p lots. A linear relationsh ip between the % deposition and

temperamrc indicates all surface fat was molten and more fat pooled from adjacent

cap illary fat as tempcran,rc increases. Extre me h igh fat in CP70 showed a s li ghtly

differe nt trend in % deposition as a function of temperature. T his was possib ly due to the

decreased fat viscosity. causing the powder to become much less viscous and flow 10 a

point that when it hit the plate. it splashed and the breakup droplets did not land o n the

plate. Over long storage time. amorphous lac tose crystaliscd and pushed encapsulated fat

to the surface. giving higher depos ition for the 200 I sample compared to the 2003

sam11le.

5- 14

lmplcmcnuition of Stickiness Curve in a Plant -Preliminary Work

CHAPTER 6

IMPLEMENTATION OF STICKINESS CURVE IN A PLANT - PRELIMINARY WORK

6.1 INTRODUCTION

The ultimate use ofa stickiness curve is to be used in the pla111 environment. The

s tickiness curve profile has temperature versus water activity / relative hum idity and is

used as a reference line to map process temperature and relative humidity ot1tlet

cond itions. his generated by fitting a best line through the experimental stickiness points

(panicular temperature and relative humidity) at a temperature ··x·· above T, for the

amorphous sugar present (usually lactose in dairy powder). This critical "X'" value can

be estimated using the composition of the powder from the regression equation obtained

in Chap ter -l.

Cp to this point. instantaneous stickiness has proven to be a surface phenomenon

with the sutTotmding air condition in equilibriu m with the powder surface. Hence. the

spray drier outle t tempcramrc and relative humid ity are vital infonnation as they

represent the powder surface temperamre and water acti\'ity. Therefore. they arc esse111ial

to control stickiness on paniclc surfaces and adhesion ofpanicles to clusters.

Using the skim milk powder (lactose 57.19%TS. fot 0.52%TS) as a case smdy.

the estimated critica l ··x·· va lue was compared with actual experimental data. It was

produced from Fontcrra. Specialty Powder Unit (SPU). Longbum. By talking to the

operators with respect to the operation procedure for controlling the process. There arc

two common operator parameters used. These are air inlet temperature and liquid

production rate that affect the outlet temperature and relati ve humidity.

Once th e plant condition has been mapped on the stickiness curve. it enables the

determination of whether the powders are sticky or not and what action can be taken to

prevent blockages in the drier and/or cyclone, due to a sticky laye r deposition on the

walls. and what action can be taken to maximize the drier capacity if it 's under processed.

An Excel spreadsheet was compi led 10 solve the mass and energy balance around the

6-1

Implementation of Stickiness Curve in a Plant -Preliminary \\'ork

spray drier and was used 10 observe rhe effect on T-T, of changes to inlet air temperature

and product flow rare. These can then be mapped on the stickiness curve .

Recommendations on what parameters to change allow optimization of the process. for

either increased throughput or lower energy costs.

6.2 CASE STUDY

This exercise was undenaken to demonstrate the process of implemen ting rhis

research imo practical use. The procedures were divided into three sections:

6.2.1 Determination of critical " X"

Instantaneous stickiness is a surface phenomena. therefore. the surface of th e

panicle leaves the drier in equilibrium with the drier outlet air temperature and relative

humidity. and these conditions detcn11ine whether a panic le becomes s ticky or not.

Temperature and relarive humidity parameters are therefore used 10 consm,ct the

stickiness cur\'e profile. Sticking is caused by the amorphous lactose being exposed to

conditions that exceed the T, of lactose . hence reducing the viscosity to a point that fonns

liquid bridging with the wall or other pa11icles. The critical "X" tcmperarure above T~

lactose is the point at which powder becomes instantaneously sticky and is detcnnined

either from experiment or by the regression equation given in Chapter 4 (for any powder

that has total fut less than 42%). This can be useful when developing a new product with

the critica l ··x" being estimated from its composition. In this case. SM P contains 0.52

%TS fat. 57.19 %TS lactose and the critical "X" and was calculated to be 38.7°C for the

p,111iclc-gu11 conditions.

The other possible way to estimate this critical "X" value is to measure it

experimentally 011 the panicle-gun rig. The same procedures were employed in which the

% deposition of powder on the plate was measured at a constant tempcrarure and

increasing RH of the air condition. The percentage deposition of dry powder was plotted

against the T-T, parameter, wh ich was used to combine the temperature and RH effects.

The T-T I

axis intercept of the trend line (38 .0°C) is identified to be the critical "X" value

for instantaneous st ickiness to occur, shown in Figure 6.1.

6-2

80

lmplcmcnta1l0n of Stickiness Cun.c in a Plant -Preliminary Work

I ♦ 80.S degreeC 1

70 - ■ 73degreeC ---- ---- ---· -· • •· - · - - - -- - - -· -- •· -- -~ • ---- - ---· • • -

• 68 2 cfegreeC 60 J • 60.5 degreeC ,. __ .

] 50 ·;;; 8_ 40 Q) 'O ,>? 0 30

20

10

0 0 10

I

20 30 T-Tg

•

50

Y: 346X· 13143 R"2 • 0 5234

60

Figure 6.1 The T-T9 plot of the SMP and the critical 'X', determined to be 38°C.

70

Four stickiness points at panicular temperatures were collected from the • •

deposition versus relative humidity plots. A best-fit stickiness curve of those four

stickiness points resulted in "X" temperan,rc difference above the T g lactose line. The

critical "X" found in this way was 40.7 °C as shown in Figure 6.2.

6-3

~ ~ => e & E ., ...

100

90

80

70

60

50 -------· 40

30 T g amorphous lactose

20

10

Implement.al ion ofSlickincss Curve in a Plant -Prcliminary \Vork

Experimental • / sdekiness points

- - - - - - . Si iCki~es·s ·curv·c ·, . __ .!9:.x lil'?_e.

Critical •x" = 40 .7 •c

OL--------.....---------------~ 0.0 0.1 0.2 0.3

Water activity

0.4 0.5 0.6

Figure 6.2 The stickiness curve of the SMP is fitted line through experimental points .

The samples were grouped in the skim milk powder category. The critical "X ..

va lue obtained from the T-Tv plot of five different specification SMP po wders used was

found 10 be 40.3±2.2 °C (95% confident interval). The critical .. X .. value for whole milk

powder was calculated w be 37.4 ±3.0 °C with 95% confidence interval. These .. X ..

values apply 10 the powder at the condition the pa11iclc-gun was run at 20 ms'1. and using

a flat stainless s1ec l plate at 90° 10 the flow.

6.2.2 Constructing stickiness curve and mapping plant conditions

Whichever way was used to de1en11ine the critical "X .. value for a powder. the

result was within the 95% confidence error band.

For the case examples being considered here. the average value of SMP and its

e rror bar arc included to construct the stickiness curve as shown in Figure 6.3. The sticky

and non-sticky regions were identified; any temperatme and/or water activity/ relative

humid ity combination which exceeds the T ,+40.3°C line possible to cause the powder to

become instantaneously sticky. In order 10 con trol the stickiness problem in the drier.

6-4

fmplcmcnr.ation of Stickiness Curve in a Plant •Preliminary Work

cyclone and ducts. the outlet air temperarure and humidity that would be in equilibrium

with the paniclc surface. have to be kept below the s tickiness line .

90

65

60

e 75 ~ ~ 70 g_ ~ 65 ...

60

55

50 0

Non-Sticky region

0 I 02 0 3 Water Act1v1ty

Sticky region

$! ck ness curve I

04 0 5 06

Figure 6.3 The stickiness curve constructed from the average critical " X"' of SMP w i th ±3 error bar.

Fomcrra 's Speciality Powder LJ nit (S PU). Longbum site. produces a wide range

of powders. including SMP and cream powders. The drier out let air temperamre is

measured and humidity is calculated fro m the nmning parameters of the spray drier.

Chatterjee (2003) outlined the step-by-step humidity calculation. The PLC displayed the

outkt temperature and relati ve humidity of the outlet air when th e SMP was made to be

8 I .8°C and 27.95%(or A": 0.2795) respectively. The values arc plotted on Figure 6.3

and show it to be in the sticky region .

The o utlet relative humidity was a lso calculated from the spreadsheet based on the

total heat and mass balance around the spray drier. The input infom1ation includes the

liqu id production rate. the solids concentration o f the feed. the inlet air flow rate.

humidity and inlet/outlet air temperature . The percentage heat loss fi-om the drier was

assu med to be 5%. however. it can be adjus ted for any panicular driers' perfonnance to

be matched when running o n water during startup. The air flow rate was no t normally

measured. and it can be adjusted by the valve. The spreadsheet enables the determination

6-5

Implementation of Stickiness Curve in a Plant -Preliminary Work

of the air flow rate . In the case ofthe SMP. the air flow rate and outlet relative humidity

were dete1111ined to be 6.7 kg.s·' and 19.6% (or A. - 0 .196) respectively. The calcu lation

does not include the flu idized bed. The calculated condition is also plotted in Figure 6.2 .

There is about 8% RH difference between the outlet relative humidity found

between the Longbum displayed value and the calculated value from the spreadsheet.

This discrepancy makes a big difference in interpreting whether the powder is sticky or

not. Therefore. the reason for 1he discrepancy needed to be detennined . A simple mass

balance calculation was carried 0111 to check the calculation base on the process inlet. 1hc

outlet air temperamre. the concentrate flow rate and tota l so lids concentrate in feed. The

cquacions programmed imo the drier PLC were traced back 10 the basic equations and

checked. One of the assumptions was 1he RH is inclepcndenl from air flow rate and

constalll values were used. Although different assu mptions we,·e made to calculate th e

outlet RH. inc luding the fluidised-bed. a s imi lar result of 15% RH was obtained both

from the spreadshee t and programmed equations. II was found 1ha1 the equat ion for

calculating 1he saturated pressure (P,.,) in the PLC had a co11stan1 value and the source of

this figure cou ld not be traced. The saturated pressure is a function of outlet temperature.

If 1he programme can be corrected to calcu late P,. conectly. the conect RH will be

d isplayed. There is a co1Tcction being made in the Longbum PLC programme while the

plant is not processing because of its being 1he oft: season period. It is recommended to

recheck RH display with calculated RH once the plant has scan to process.

6.3 SPRAY DRYING PROCESS OPTIMISATION

In order to gain some knowledge about spray drier operating control. discussions

were held with the operators involved with the powder produce ion a1 Longbum.

The percentage moisture of the final dried powder was kept below its max imum

specificacion. Produc t with a high moismre content lowers the T, of lactose and has an

effect on T-T,. It causes particles to stick to the wall or with each o ther, blocking the

cyclone or causing problems in later stages. In that case, the o perator can e ither increase

the outlet temperature or increase the SFB inlet temperature 10 lowerthe moisture content

6-6

Implementation of Stickiness Curve in a Plant -Prclimina1y \Vork

of the final product. Samples are collected from the bottom of the drier and are checked

frequently for their moisture level.

Providing every1hing else is kept constant. including other powder components.

particle size. density. drier fittings (nozzle size. angle of spray) and percent of total solids.

the outlet 1emperan1re can be increased by reducing the production throughput. (i.e.

concentrate feeding rate). Increasing the inlet temperan,re and keeping the same product

flow rate would result in the same eftcct of increasing the outlet temperature. On the

drier. the inlet 1empernn1re and outlet temperature have independent SCI points.

The exhaust fan which is taking all the air out of the drier is set at - 3mm we vacuum point and operates with the exhaust valve 90% fi1lly open. Above - 3111111 we. a

vacuum effect could 1101 be provided. Therefore. always sec the exhaust to that value.

Any dramatic reduction in vacuum means hardly any air is going through. and no change

of fan reading indicates that there is a cyclone blockage in the process line. The pressure

readings in the drier arc a measure of the level of powder product in 1he drier. However.

if the pressure reading drops fast. i1 means more powder is accumulating in 1he chamber

and therefore a blockage is occurring in the chamber.

The Excel spreadsheet enables us to visualise the effect of inlet temperature.

outlet temperature and production race on the stickiness parameter (T-T,) in a dynamic

way. Figures 6.4 and 6.5 demonstrate these effects using the SMP as an example.

6-7

lmplcmcnu1ion of Stickiness Curve in a Plant -Preliminary \\'ork

6.3.1 Effect on X via different inlet temperature and production rate

55

50

45

40 {! ,.:.

35

30

25

-+- 1.3 tonne/hr _. 1 .4 tonne/hr -+-1.5 tonne/hr ,--++- 1.6 tonne/hr ,

--- ----- ----

20 -"-~------,..----'--------~---' 150 200 250 300

Inlet Temperature ('C)

Fig ure 6.4 Effect of d ry product throughput and inlet temperature on T-T0•

Figure 6.4 shows the effects ofd1)' product throughput on T-T,. as a function of

inlet tempernturc (the dotted line at 225°( is for the SMP). whi le Figure 6.5 shows the

effect of dry product throughput on the outlet temperature. The critica l T-T • for

instantaneous stickiness of the skim milk powder fal ls between 36°(-4 1 °C with a 95%

confidence interval (Chapter 4 ). The ideal condition for the plant is m nm the drier with

a T-T• below this range. This is because with any rcmpcrnture and or humidity combined

condi tion that results in T-T• above 41 °C. the powders are definitely stick)'. and above

36°C they arc probably s ticking and will cause a problem. The drier was runni ng at 1.4

tonne per hour of dry product. and the T-T I was below the lowest bound: hence there is

an opponunity to push the drier harder. There are two ways to move the process

conditions closer to the band. It can be done either by increasing the production rate

from 1.4 tonne per hour to I .5 ton ne per hour and keeping the o ut let temperamre at 81 °C

by increasing the inlet air temperature to 235°C. or reduce th e inlet temperature from

225°C to 220°C. In this way the plant can save energy and therefore money.

6-8

Implementation ofS1ickincss Curve in a Plant •Prclimin::iry Work

6.3.2 Effect on X and outlet temperature via different production rate

60

55 ·········--··············---- ·

1 -+-1.3 tonne/hr - - - - - - - - ~ ...,_1.4 tonne/hr­

..... 1. 5 tonne/hr 50 -- ······ -------- ········· ········ ··········· ·· ....... , .6tonne/hr ·

45

"' I-; 40 ,._

35

30

25

20 50 60 70 80 90 100 110 120 130

Outlet Temperature (°C)

Figure 6.5 Effect on dry production ra te and outlet temperature on T-T 9

Increasing the production rate is the more practical thing to do tn the plan t.

because the ou tlet temperature is adjusted by the liquid production rate. as mentioned

earlier. /\ta constan t outlet tcmperantre. increasing the production rate would result in

increasing T-T 1.

6-9

90

8S

8-0

Operauon condition . - 8fCC>ngburf\ - - - ...;..___ . - ·

(calculated frcwn the --.. ■ spreodshee<) 0@@

Non-Sticky region

60 ···-------·-

ss --------

tmplcmcntation ofS1ickincss Curve in a Plant •Preliminary \Vork

Sticky region

soL---------~----~---~----..l..~ c-1----o 0 1 0.2 03

Wa1er Adrvrty

04 OS 06

Figure 6.6 Demonstration of the possible changes that could effect the drier outlet conditions.

Table 6.1 The possible experimental matrix for variat ion in the solids production rate, inlet temperature and corresponding T-T0•

DATA SOLIDS l 'iLET O UTLET T-Tc RH % A" PT. PROD. RATE T Ei\1 P. TEMP.

I 1.4 tonnelu 225 81.8 33.1 19.6 0. 196

2 1.5 tonne'hr 225 73.3 38.6 29.4 0.294 , , 1.5 tonne 'hr 235.2 81.8 34.9 20.7 0 207

4 1.6 tonne 'hr 246 818 36.6 21.7 0.217

5 1.5 tonne hr 210 60.8 48.9 51 0.5 1

6 I .4 tonne'hr 210 69.2 38.9 33.2 0.332

7 1.4 tonnc,'hr 215 73.4 36.7 27.7 0.277

The tempcrmurc and water activity data plotted on the st ickiness curve in Figure

6.6 arc from Table 6. 1. It shows the changes recommended above based on the

spreadsheet calculation and recommends the best way to optimize the process. The first

data point is the current processing condition at Longbum for the SMP. The second data

point has increased produc tion rate to 1.5 tonnes per hour and kept the same inlet

tempcrarure. which resulted in both an increase in outlet temperature and humidity. The

c ritical T-Tg is not exceeded but it does lie within the error bars for the stickiness lines.

and problems may be experienced. The plant is operating near or at its limit. The third

data point has increased inlet temperature wh ile keeping the production rate at 1.5 tonne

6-10

Implementation of Stickiness Cul'\c in a Plam -Prcliminary\Vork

per hour and outlet temperature. the outcome condition is very similar to the current

production but is closer to the sticky point lines. The founh data point has increased the

production rate 10 1.6 and uses an increased inlet temperature to produce the same outlet

temperarure as the current production. The RH % is increased but the critical T-Ts is not

exceeded. The fifth data point has 1.5 tonnc'hr and a reduced inlet temperature of2 I 0°C.

It has reduced the 0\1tle1 temperature to 60.8°C and increased the RH to 51 %. indicating

the product is ve,y moist and will be stick-y. The sixth data 1>oint has the current product

rate and a reduced inlet temperature of 210°C. which resulted in a reduced outlet

tcmperanire of 70°C and 38.9% RH and the critical T-T8 is not exceeded. but is within

the error bands and is very close to the stickiness curve. The last data point has the

current production rate and a reduced inlet temperature of 2 I 5°C (5°C higher than the

sixth data point). giving the same production as the current conditions but at lower energy

costs. The best recommended process conditions arc data points 7 and 4. It depends on

whether the company wants to save energy or increase production .

In a general trend. changing the production rate and keeping the same outlet

temperamre shifts the point sideways. Either changing the inlet temperature and keeping

the production rate the same or changing the production rate and keeping the inlet

tempemture the same shift the point venically.

If data poi111 7 was chosen. reducing the inlet temperature would mean saving

money on the energy cost. A s imple heat based calculation found approximately S 11 .34

can be saved per hour based on an assumed energy cost of $0.17 kwh. If the plant

continually processes for 20-30 hours before a nozzle wash. it would save S 340.2 of

powder. The whole production nonnally runs for 30 days.

If data point 4 was chosen. the production rate can be increased by 0.2 tonnes in

d,y production rate. a 14% increase in production throughput.

6-11

6.4 CLOSURE

Implementation of Stickiness Curve in a Plant •Preliminary \\'ork

Dairy powder stickiness duting processing is a surface phenomenon. and

therefore the outlet air temperature and relative humidity is assumed 10 be in equilibtium

with the surfuce of the powder pa11icle . For a dairy powder. there is a critical "X"" value

for instantaneous stickiness. The very first step is to be able to identify this ctitical --x·· value. There are several diftcrcm ways to do it from the data obtained with the panicle­

gun rig. All the values obtained were not significantly different for the SMP tested. It is

in a range of 36°C - 41 °C with a 95% confidence interval. The next step was to map the

plant condition on to the stickiness cur"e (T~ - ·x· 1ine). and any adjustment can be made

10 approach or move away from the stickiness curve. The main two parameters

responsible for such adjustments arc the inlet air tempcramre and the production rate. An

Excel spreadsheet enabled the solution of the problem in a dynamic way. especially 10

relate the inlet1outlet temperamre and production rate with the T-T~ parameter. For this

pa11icular skim milk powder. processing at the Longbum site. the best recommendation

for reducing energy is to reduce the inlet air by I 0 °C. hence reduce outlet temperamre 10

73.4°C and keep the same production rate. The other option is 10 increase the solid

production rate from I .-1 tonnes hr to 1.6 tonnes 11r.

6- 12

Conclusions and Recommendations

CHAPTER 7

CONCLUSIONS AND RECOMMENDATIONS

The panicle-gun rig has been used 10 measure the onset of instantaneous

stickiness of dairy powders. Its dynamic seiup has shown the stickiness phenomena is a

function of temperature. relative humidiry, and composition of the powder. It has shown

that the stickiness behaviour is a function of the surface conditions since there was

insufficient time (<0.05 s) for the pa11icles in the pai1iclc-gun to have anything but the

surface to be affected by the air conditions the particle was subjected 10. This technique

docs not need 10 pre-condition the powder. as the surf.lee was assumed to be in

equilibrium with the surrounding condition. The panicle-gun rig has enabled stickiness

data to be collected under air temperanire and RH conditions similar to those found in the

exit of spray driers. It :,!so enables the rate at which stickiness develops in a 1>owder to

be investigated as a fonction of RH and temperature. Funherstudy is required to observe

the velociry and angles of contact on eftl:cts of instamaneous stickiness in more detail.

There were two ways of identifying the c1itical --x·· value from raw data of • o

deposition versus RH at constant temperatures. The tirst was to identil)' the

instantaneous stickiness point at a particular relative humidity for each temperature

tested. Then a stickiness curve was tined to the experimental stickiness point at those

temperatures above the T~ of amorphous lactose. The best fit using a least squares error

tit of ··x· was identified to be the critical value for instantaneous stickiness. The other

way was to plot % deposition against the T-T ~ parameter for combining temperature and

relative humidiry effects. Th is was done successfully. indicating that the stickiness was

due to the Tg of amorphous lactose being exceeded and was a fonction oftcmperamre and

RH. The critical ··x-· value was identified as the x-intercept in the T -T, plot and the slope

of the trend line detects the sensitivity of the stickiness in response to condition changes.

Various dairy powders were tested for the critical ··x" value. These powders included

WMP. SMP. MPC and cheese powders. The critical ··x-• value ofSMP varied from 37°C

to 42°C: WMP varied from 33°C to 40°C: and MPC varied from 42.7°C to 51°C. The

variation between the powders depends on their composition. There was a positive linear

7-1

Conclusions and Rccommcnda1ion.s

relatio nship fonned between the critical ··x" value ofa powder and the powders' lactose

content.

Cream powders and white cheese powder are catalogued under the high fat

powders (i.e. total fat content equal to or more than 42 %). The liquid bridges fonncd by

molten fat were a function of temperature. However. amorphous lactose also contributed

to the s tickiness when the T g amorphous lactose was exceeded. Both effects are obser"ed

in T-T g plots for high fat powders.

It has been shown that the stickiness of dairy powders was caused mainly by the

amorphous lactose content. as the stickiness is a funct ion of both temperarure and RH.

The glass transition temperature of amorphous lactose correctly predicts the eftects of

temperature and RH: all the s tickiness powders as shown by the T-Tg plots collapsed

these data to a single line. This line Stans at a critical T-Tg value for a 1>owder. known as

the c ritical .. x•· \'alue. which shows that the glass transition temperature needs to be

exceeded by a constant amount in order for the powder to become instantaneously sticky.

The slope of the trend line in the T -Tg plo t shows the sensitivity of the powder to

stickiness in response to the temperature and or relative humidity changes. Both the

critica l 'X' value and the slope are unique to the powder. They also show that powders

with greater than 30% amorphous lactose become more sticky quicker once the critical X

factor above T, is exceeded.

In this work ... X .. has been measured using a panicle-gun rig at a ve locity of

20ms'1 which is typical of the ve locities of the powders in the ducting of spray driers.

For powders that have a total fat content less than 42%. the stick iness is mainly

due to amorphous lactose when Tgis exceeded by the critical ··x·· value of the powder. A

regression equation (equation 4.1) was gcnemted. based on the critical ·x· value obtained

from the experimen tal data of various dairy powders. to corre late the effects of

composition on X. Although an overall trend is observed. ftnher study is required to

investigate the su rface composition in relationship with the critical 'X'.

7-2

Conclusions and Recommendations

Once the critical "X .. for a panicular powder was identified. the s1ickincss curve

was generated on a 1emperature versus relative humidity/waler activity plot. It enables

operators to map the plant operating conditions on the stickiness curve for a particular

powder being produced. and adjustment can then be made 10 optimize the process by

moving the operating condition wward or away from the s tickiness curve. The surface of

the panicle is assumed to be in equilibrium with the drier outlet air temperature and

re lat ive humidity. and it is these conditions which are used to map the plant operating

point on the graph. The case study on the SMP has demonstr;ucd the procedure to

implement this work in an industrial environment. The drier outlet 1emperan1re and

product io n rate arc the two parameters that control the outlet conditions. An Excel

spreadsheet was consm1c1cd. based on the energy and mass balance around a spray drier.

It related the two adjustable parameters with the T-T8 parameter and was used 10 solve

the problem in a dynamic way. In the case of the SMP. the objective was tO push the drier

10 its lim it and move the point 1owards the stic kiness cun•e. The best recommendations

are either 10 reduce th e in let air by 1 o•c in order to sa\'c on energy or to increase the

concentrate feeding rate to 3.2 t0nncs 'hr (or solid production rate of 1.6 tonnes 'hr). a 14°0

increase in production throughput.

Foncerra Te Repa site imp lemented the s tickiness cun•c and uses it to monitor the

process condi1io ns so that the chamber. cyclone and ducting blockages due to s tickiness

can be a\'Oidcd. These are continuously under review and updating with the latest

knowledge bei ng used to control the process in an efficient and a dynamic way.

Several factors were kept constant throughout this experi mental work . These

included the air velocity of the particle-gun rig. sta inless steel plates and powder particle

fired 10 the plate at 90°. Funher work is recommended 10 in vestigate the fac1ors of varied

air velocity. different coating of the plate. and O!hcr angles of contact that have an effec!

on the critical .. X" value for a powder.

7-3

REFERENCES Adhikari. B .. Howes. T .. Bhandari. B.R .. & Trnong. V. (200 I}. Stickiness in foods : A

Review of Mechanisms and Test Methods. International Journal of Food Propenies.

4(1 }. 1-33.

Aguilera. J.M .. de l Valle. J. M., & Karel. M. ( 1995}. Caking Phenomena in Amorphous

food Powders. Trends in Food Science and TechnolO!!)'. 6. 1-49-155.

Bandyopadhyay. P.. Das. H.. & Shamia. G.P. ( 1987}. Moisture Adsorption

Characteristics of Casein. Lactose. Skim Milk and Chhana Powder. Journal of Food

Science and Technolo!!v. Vol. 24. Jan Feb.

Bhanda,i. B. (200 I). Glass Transition in Relation 10 Stickiness during Spray D,ying.

Food Tcchnologv International. 64-68.

Bhandari. B.R .. Dana. :S: .. & Howes. T. ( 1997). Problems Associated with Spray D1ying

ofSugar-rich Foods. Diving Tcchnologv, 15(2). 67 1-684.

Bhandari. B.R .. & Howes. T. (2000). Glass Transition in Processing and Stability of

Food. Food Tcchnologv in Australia, 52( 12). 579-585.

Bhandari. E.R .. & Howes. T. ( 1999). Implication of Glass Transition for the Drying and

Stability of D,icd Foods. Journal of Food Enginecrin!!. 40. 71-79.

Bloorc. C. (2000). Developments in Food Drying Technologv overview. International

Food Dehydration Conference - 2000 and Beyond. Carlton Crest Hotel, Melbourne.

Australia.

Boonyai. P .. Bhandari. B .. & Howes. T. (200-4). Stickiness Measurement Techniques for

Food Powders: A Review. Powder Technology. 145. 34-46.

Bronlund. J.E. ( 1997). The Modelling of Caking in Bulk Lactose. PhD thesis. Massey

University. Palmerston Nonh. New Zealand.

8-1

Brooks. G.F. (2000). The Sticking and Crvstallisation of Amomhous Lactose. Masterate

Thesis. Massey University. Pahncrston North. New Zealand.

Buma. T.J. ( 1971 a). Free Fat in Spray-dried Whole Milk. 2. An evaluation of methods

for the detem1ination of free-fat content. 3. Panic le size. its estimation . influence of

processing parameters and its re lation to free-fat content. Netherlands Milk Dairv

Journal. 25(42-80)

Buma. T.J. (197 1b). Free Fat in Spray-dried Whole Milk. 4. Significance offi-ee fat for

other propenies of practical importance. Netherland Milk Dai,y Journal. 25. 88-106.

Burr. R. ( 1999). An hwcsti~ation Into Cheese Powder Cakin~. Diploma in Dairy

Sc ience and Technology Dissertation. Massey \;niversity. Palmerston Nonh. New

Zealand.

Chatte,jee. R. (2003). Characte1ising Stickiness of Dairv Powders. Masterate Thesis.

Massey lJni\'ersity. Palmerston Nonh. New Zealand.

Chen. X.O. (199-1). Towards a Comprehensive Model based Conrrol of Milk Drying

Process. Drving Technologv. 12(5). 1105-1 130.

Chuy. L.E .. & Labuza. T.P. ( 1994 ). Caking and Stickiness of Dairy-based Food

Powders as Related to Glass-Transition. Jourtial of Food Science. 59(1 ). 43--16.

Crofskey. G.M. (2000). lnvestigatjon jnco the Caking Prob)cms Assoc iated wj1h Sprav

Dried Cream Powders 55 and 70. 4th year Food Technology Report. Massey University.

Palmerston North. New Zealand.

Dixon. A. ( 1999). Corre lat ing Food Powder Stickiness with Composition. Tcmperamre

and Relative Humid ity. Monash Universtity. Melbourne, Austra lia.

Downton. G.E .. Flores-Luna. J.L .. & King, C.J. (1982). Mechanism of Stickiness in

Hygroscop ic, Amorphous Powders. Ind. Eng. Chem. Fundam., 2 1. 447-4 5 1.

8-2

Foster. K.D. (2002). The Prediction of Sticking in Dairy Powders. PhD Thesis. Massey

University. Palmerston Nonh. New Zealand.

Faldt. P .. & Bergenstahl. B. (1996a). Changes in Surface Composition of Spray-dried

Food Powders due to Lactose Crystallisation. Lcbcnsm-Wiss. u- Techno)ogv, 29. 438-

446.

Faldt. P .. & Bergenstahl. B. (1996b). Spray-dried Whey Protein1Lactose 'Soybcan Oil

Emulsions. I. Surface Composition and Panicle Stmcture. Food Hydrocolloids. 10(4).

42 1-429.

Faldt. P .. Bergenstahl. B .. & Carlsson. G. ( 1993). The Surface Coverage of Fat on Food

Powders Analyzed by ESCA (Electron Spectroscopy for Chemical Analysis). Food

Sm1cture. I 2. 225-234.

Faldt. P .. & Bergenstihl. B. (1995). Fat Encapsulation in Spray-dried Food Powders.

JAOCS. 72(2). 171-1 76.

Faldt. P .. & SjOholm. I. (1996). Charac1erisza1ion of Spray-dried Whole Milk.

Milchwissenschaft - Milk Science International. 51(21). 88-92.

Jcnickc. A.W. (1964). Storage and flow of solids. Bulletin 123. Utah Engjnccring

Expcrimcmal Station. Uni,·crsity of Utah.

Jouppila. K .. Kansikru, . J .. & Roos. Y.H. ( 1997). Glass Transition. Water Plasticization.

and Lactose Crystallisation in Skim Milk Powder. Joumal of Dairv Science, 80(12).

3I52-3160.

Jouppila. K .. & Roos. Y.H. ( 1994a). Glass Trans itions and Crystallisation in Milk

Powders. Journal of Dairy Science, 77(10), 2907-2915.

Jouppi la. K .. & Roos, Y.H. (1994b). Water Sorption Propenies and Time-Dependent

Phenomena of Milk Powders. Joumal ofDai rv Science, 77(7), 1798-1808.

8-3

Hennigs. C.. Kockel. T.K .. & Langrish. T.A.G. (200 I). New Measurements of the

Sticky Behaviour of Skim Milk Powder. Dryine Technology 19(3&4). 471-484.

Keir. T. (2001). Milk Powder Stickiness: A l'reliminarv Studv. Masterate Thesis.

Massey IJnivcrsiry. Palmerston North. New Zealand.

Kim. E.H.J .. Chen. X.D .. & Pearce. D. (2002). On the Mechanisms of Surface

Fonnation and the Surfuce Compositions of lndustTial Milk Powders. Qiying

Tcchnologv.

King. N. ( I 965). The Physical Structure of Dried Milk. Dairv Science Abstracts, 27(3).

91-104.

Lazar. M.E .. Brown. A.H .. Smith. G.S .. Wong. F.F .. & Lindquist. F.E. ( 1956).

Experimental Production of Tomato Powder by Spray Drying. Food Technology. I 29-

133.

Le Mcstc. M .. Champion. D .. Roudaut. G .. Blond. G .. & Simatos. D. (2002). Glass

Transition and Food Technology: A Critical Appraisal. Journal of Food Science, 67(7).

2444-2-158.

Lill ford . P.J .. & Fryer. P.J. ( 1998). Food Panicles and the Problems of Hydration. illlll;i

]Chem E, 76(pan A). 797-802.

Lloyd. R.J .. Chen. X.D .. & Hargreaves. J.B. (1996). Glass Tr.msition and Caking of

Spray-dried Lactose. lmcmational Journal of Food Science and Tcchnologv, 31. 305-

3 11

Masters, K. ( 1991 ). Sprav Drving Handbook. Great Britain: Longman Scientific &

Technical.

Moreyra. R .. & Peleg. M. ( I 981) Effect of equilibrium water activiry on the bulk

properties of selected food powders. Journal of Food Science. 46.

8-4

Noel. T.R .. Ring. S.G .. & Whinam. M.A. ( 1990). Glass Transitions in Low-Moisture

Foods. Trends in Food Scjcpcc and Technology, 62-67.

Norris. G.E .. Gray. I.K .. & Dolby. R.M. ( 1973). Seasonal Variations in the Composition

and Thern,al Properties of New Zealand Milk Fat• Thennal Propcnies of Mi lk Fat and

their Re lation to Composition. Journal of Dairy Research, 40. 3 11 -321 .

O'Donnell. A.M .. Bronlund. J.E .. Brooks. G.F .. & Paterson. A.HJ. (2002). A consta111

humid ity air supply system for pilot scale applications. lntemational Journal of Food

Science and Technology. 37. 369-374.

Ozmcn. L .. & Langrish. T.A.G. (2002). Comparision of Glass Transition Temperature

and Sticky Poi111 Temperan,re for Skim Milk Powder. Dryi,rn Technology, 20(6). I 177-

11 92.

Paterson. /\.H.J.. Brooks. G.F .. & Bronlund. J.E. (200 I) The Blow Test for Measuring

the Stickiness of 1>owders. Conference of Food Engineerin~ 2001. AICHE conference

No\'crber :½-9, 200 I, Reno. Ne,·ada.

Paterson. A.HJ .. Brooks. G.F .. Fostcr. K.D .. & Bronlund. J. E. (2004 5). The development

of stickiness in amorphous lactose at constanl T-T. lcve ls. International Dairv Journal. In

Press.

Peleg. M. (1993). Glass Transitions and the Physical Stabi lity of Food Powders. In J .

Blanshard and P. Lilliford . (Editor) (Ed.). The Glassy State in Foods. 435-451.

Leicestershire : Nottingham Uni\'crsity Press.

Roos Y. ( 1993). Melting and Glass Transition of Low Molecular Weight Carbohydrates.

Carbohydrate Research, 238-248.

Roos. Y .. & Karel. M. (199 1a). Phase Transitions of Mixn, res of Amorphous

Po lysaccharides and Sugars. Biotechno l. Prog. 7. 49 -53.

Roos. Y., & Karel, M . ( 1991b). Plasticizing Effect o f Water o n T hcnnal Behavior and

Crystall isati on of Amo rphous Food Mode ls. Journal of Food Science, 56(1). 38-43.

8-5

Roos . Y .. & Karel. M. (1991c). Water and Molecular Weight Effects on Glass

Transitions in Amorphous Carbohydrates and Carbohydrate Solutions. Journal of Food

Science. 56(1). 1676-1681.

Roos. Y.H. (1995a). Phase Transitions in Foods. Academic press. INC.

Roos. Y.H. ( 1995b). Glass Tranistion-related Physicochemical Changes in Foods. Food

Technology~ 49( I 0), 97-102.

Roos. Y.H. (2002). lmpo11ance of Glass Transition and Water Activity to Spray Drying

and Stability of Dairy Powders. Lait. 82. 475-48-1.

Roos. Y.H .. Karel. M .. & Kokini. J.L. (1996 ). Glass Transitions in Low-Moisture and

Frozen Foods: Effects on Shelf Lite and Quality. Food Tcchnolouv, 50(11). 95-108.

Toy. K.I.M. (2000). Stickiness Rig : A procedure to Characterise Powder Stickiness.

~cw Zealand Dai,y Research Institute. Palmerston Nonh. New Zealand.

Wallack. D.A .. & King. C..I. ( 1988). Sticking and Agglomeration of Hygroscopic.

Amorphous Carbohydrate and Food Powders. Biotechnologv Pro~ress, 4( I ). 31-35.

Williams. M.L .. Lande!. R.F.. & Ferry. J.D. (1955). The Temperature Dependence of

Relaxation Mechaniasm in Amorphous Pol}111ers and other Glass-fom1ing Liquids.

Joumal of the American Chemical Societv, 77. 3 70 1-3 707.

8-6

a. D

k

k

T

T""'

X

x,

Greek Leners

y

t

NOMENCLATURE

Panicle radius

Water activity

Specific heat capcity

Viscous flow constant (Eq. 2. I)

Gordon and Taylor constant

Ternpcrarnre

Glass transition temperature

Glass transition temperature of mixture

Glass transition temperature of components

WL F constant

Mass fraction of components

lnterpm1ic lc bridge s ize

mass fraction o f component i

Surface tension

Time/ panic le contact time

Viscosity. Critical viscos ity.

Viscosity at glass transition temperature

m

J'k mol

~C or K

~Cor K

~C or K

~Cor K

g.1g powder

111

g. g. d,y basis

Nim

s

Pa.s

Pa .s

9-1

APPENDIXA1

COM POSITIONS OF POWDERS TESTED

I Orv Powder Product Fat Lactose Protein

(%TS) (%TS) (%TS)

Agglomerated WMP 31.09 37.82 25.91

High heat WMP 27.38 38.33 27.07

Regular WMP1 28.44 40.68 24.9

Instant WMP 29.75 39.77 24.69

Regular WMP2 26.94 40.93 25.91

Instant SMP2 0.79 52.98 38.19

Instant SMP1 0.62 57.84 34.27

Medium heat SMP 0.83 53.01 38.05

SMP 0.52 57.19 34.06

Butter milk 9.33 51 .3 31 .92

MPC 44 0.83 47.56 43.52

MPC 56 1.35 31.29 59.36

MPC 70 1.46 18.16 72.86

MPC 85 1.67 4.17 8854

Whey protein Isolates 0.52 1.05 96.34

GUMP 15.46 49.48 27.84

White cheese 42.05 27.9 21 .23

LFCP 55 55.84 24.37 16.24

HFCP 70 71 .79 13 33 12.31

Amorphous Lact. 0 100 0 ..

Product compos,11011 s were supplied by Fonterra.

I 0-1

APPENDIXA2

SMP INST A~T SMP 2

8

7

6

-~ 5 ·.;; g_ 4

" 0 3

* 2

0 0

120

100 u ., ~ 80 g> ::!,

~ 60 " e ~ 40 E ., ,...

20

0

♦ 72.2 dogreo C ■ f---------· •• ••.• -------·-• 80 5 degree C

_ • 68 2 degree Cf ••••••• • •••• _. __ • _ . • • • • • • • • ••• ••• • 60 5 degree C

10 20 30

RH %

40 50

St1duness curve / Tg• X tne

Critical?(' value = 38 t

0 0.1 0.2 0.3 Water activity

0.4 0.5

60

0.6

10-2

12

10

C 8 :~ "' 8. 6 <> .., #- 4

♦ 80,SdegreeC

■ 72.2 <legreeC A 68.2 degreeC • 605d reeC

2 -------------- --

0 10 20

100 ·-90 .. . .

···------------·--·----,1

•

---------------S1ope = 0 34 R"2 =-0 7431

30 50 60 70 80

T-Tg

80

--~ -~--- - -- -u

" ~ "' " ! a e <> C. E ., ,_

70

60

50

40

30

20

10

0

--Tg \..."tCtOSoe

•• •39 7{0¾ c.e,pos1tion)

xo:~2 6(1\.depos1t1on)

x=S4 3(5% dep0$ition)

• ...,_K=-66.9 (10% deposition) -+-x~3.S (15°4 de~Ulon

0.1 0 15 0.2 0.25

--~ .. .. .. -~ ...:~~---- - -----

- ------~ .. ..

0.3 0.35 0.4 0.45 0.5 0.55 0.6

Water Activity

10-3

MEDIUM HEAT SMP

90

80

70

♦ 80 5 degreeC ■ 72 2degreeC ~----· ·········· -· ··--·-- · --- -------- -- ----~

c: 60 .2 -~ 50 g_ ,3 40

* 30 20

,. 682degreeC1 •.••.... ..............••. . ........ ·a · . ... • 60 5 deg,eeC, • •

10 · · · · •

0

0

100

90

u 80 ., 70 &

"' ., :!:!, 60 & 3

50

~ 40 0.

30 E ., >- 20

10

0 0

10 20

0.1 0.2

30

RH%

0.3 Water Activity

40 50

0.4 0.5

60

0.6

10-4

80

70

60 C: .g 50 ·;;; 8. 40 ., 0

30 ... 20

10

♦ 80.5 de,greeC ■ 72.2 deg,eeC

.1•68 2de91..C • 605de &eC

•

---------------- - . .J

s~oe =333 __ _ R'2 • 0 7<22

0 0

y-.. -~.-.L----~10-- - ~ ~o- - ~• .... 3,.0-■■,-~-----'40._ _ __ 50 _ ___ 60 ____ ~70

100

90

80 G

70 ~ g> 60 ~ ~ 50 :,

~ &

40

E (!!.

30

20

10

0 0

•

--19 lacto$e

• x=-40 2 (0,. dtPOSll.otl)

x=o11 7(5°',. depo1i:ion)

__._x:.44 7( 15°4 ClePOS<iliOt'I)

0.1 02

T-T9

0.3 Water Activity

0.4 0.5 0.6

10-5

INSTANT SMP (REPLICATE)

80

70

60

5 50 . .,, ·;;; &.40 "' 0 * 30

20

10

0 0

100

90

80 u

70 " " ~ "' 60 .. ,, .. 50 ~ ::, .. ~ 40 Q. E 30 ., ...

20

10

0

' • 80 S deg<eeC! i ■ 73 degreeC • - - - - - - - - - - • - - - .. • • • - - - - - - - - - - - -

• 68 2 degreeC -, . 60 Sdegreec· - -- - · -- · · - - - · · · · ·" · • ·- · · · - · · -

0

10 20 30

RH(%)

········-··· .. ~ --· lg lactose curve

0.1 0.2 0.3

Water Acticity

40

0.4

50

Stickineu curvo / Tg +X lint

Critlcal -X'value= 41.7

0.5

60

0.6

10-6

C .g ·;;; 0 C.

" ,, ~ •

70

60

50

40

30

20

10

♦ 80.5 degreeC ■ 73 degreeC - - • . • • • ••• - - - - ••••• • - • - - - •• - .

• 68.2 degreeC

1 • 60.5 degreeC f- - - -- • • • • • • -• - - - - • • • • - • - • - -• •

• •

Slope= 2.95 R•2 = 0.789

0 -1---~-----◄·---......... ~ - ------

0 .0 10.0 20.0 30.0 40.0 50.0 60.0 70.0

T-Tg

10-7

_ ♦805de9reeC .. ___________________ _ 20

18 16 14

.. ■722degreeC ____ __ _ ___ .... ____ __ ___ ________ ___ .. .& .. ___ __________ _

1,, 68 2degreeC

C:

.g 12 ·.; 8. 10 " .,, 8 '#

6

4

- • GOSdegreeC - ---- ------ - -- -- ----- -- - - - ----- - - ----

• -----------$ !ope a: O 78 R"2 .0 7336 •

2 ......... .

0 -'---~- - - ...--------- '>"'-----~--....----0 10 20 30 50 60 70

r.r9

100

90

80 6 70 " e C) 60 " :!:!. e .a 50 I:: " 40 0. E

" ,_ 30 --Tglact~

20 · 1 --)(=38 (0% deposnion)

10 · l x-=44 4(5% depos11,on) . _._x=57 2(15% deposition)

0 0.1 0.15 0.2 0.25 0.3 0 .35 0.4 0.45 0.5 0 .55 0.6

Water Activity

10.9

HIGH HEAT \\IMP

14

12

_ 10

~ C 8 .2 ·~ 8. 6

c3 4

2

i3 ., ~ 80 0) ., ~ ~ 60 2 e & E .,

40

>-20

0

0

0

♦ 72.2 degree C ■ 80.S degree C .t. 68 2 degree C • 60 5 degree C • replicate 80 5 de eeC

10

0.1

20

0.2

30 RH (%)

40

-•--•

50 60

Sockiness curve I

--.... ~ /~ . Tg •X_ •~~ _______ _

0.3

Water Activity

0.4 0.5 0.6

I 0-10

G' $ g> :!e. ~ :,

"§ 8. E r!!

8

7

6

0

100

90

80

70

60

50

40

30

20

10

0

♦ 80.5 degreoC ■722degreeC - - -- -- • --·· •· ••• - - - - - - ------- • ....... " ••

:~g.~~J~::g L------- ---------- --- -- -------• . --

10 20

j--Tg Lactose

- •• 33 7(0% depos,toon)

x~45 7(5% depso1110n)

- x=69 8(15% depoS4tion)

0.1 0.2

30

T-Tg

--------------- ·

0.3 0.4

Water Activity

• Slopo: 0.<2 R11-2 • 0 6691

50

0.5

60

0.6

10-11

REGULAR WMP I

-~

35

30

25

·";i; 20 8. ,3 15

120

_ 100 u ~ g, 80 :!:!, & 60 .a t; 8. 40 g ,_ 20

I ♦ 722d&greec l ■ ■ 80.5degree C ·" · · · · · - - -- · - -- - - - - - - - - · - · · • · · · ; · - · · · · -- · - · · · ·

-'- 68.2 degree C 605de r,,eC .••••..••••••

10 20 30 40 50 60

RH(%!

---Tg lactose CUNe

o+-----------------------..::::....-0 0.1 0.2 0 .3

Water Activity

0.4 0.5 0.6

10-12

35

30

25 C

.g 20 ·.;; g_ ~ 15 ?fl.

10

5

0

♦ 80.5 degreeC ■ 72.2 degreeC • 68.2 degreeC • 60.5 degreeC

10

•

20 30

T-Tg

•

Slope= 1.2 R'2 = 0.7346

60

,= l _ 80

---------.......... . - - - --_.._ - - - - . - - - - - - ---u

t 70 a, " :;!. 60

" ~ il 50 ..

~ .. 40 Q.

E 30 " ...

20

10

0

0.1

--Tglactose --x=37 1 (0% depos1t10n)

x=• 1 2(5% deposnlon) ·x=-49.5 15% de s.ition

0.2

- ___ -:---~---------

0.3 0.4 0.5

Water Activity

70

-.

0.6

10- 13

INSTA' ff WMP

*

35

30

25

5 20 :i: <I>

& 15 ., 0

10

♦ 60 5 deg,eeC ::~.!:::~ -----------······--- ------- -----~:1:;19·-- -- -----• 682de reeC -------- ···· .. ···· -- -- -------- ········ · ·--- ·----

•

::<~ 0 -J,... _ ____ ,...,,_ _ _,...., ........ .... - --=F":::C....-::1:..- .-.e::i._ ______ ..., 5

0

100

90

u 80

" !'. 70

"' " 60 :s. ~ 50 a

40 1:: " 0. 30 E " .... 20

10

0

10 20 30 RH%

Tg lactose

40 50 60

S11okm.ess curve! ~g--X tTne··-----

0 0.1 0.2 0.3 Water Activity

0.4 0.5 0.6

10-14

9

8

7

i: 8. ~ 4

* 3

z

• ao.5 degr .. c - r---- ---- ----- -- --------- ---------- ----------------■ 72 2 <legreeC

· • 682degreoer··········· ··················· ····· ... · . ......... . ■ 605de reeC ________________ ... __ ------ ··-· · ·

Slope • 036 · · ; · ·- -- R?•t =os1s2 . ---- ---------

• oL----- - - ----..-- -:::<04 -----------.

0 10 20

100

90

80 G .. e 70

"' 60 .. :,!. e 2

50

I:! 40 " 0. E 30 " ... - TglaC1osB

20 - x=38 2(0% depos,uoo)

10 --••52 0(5% depos111on) -+-x•79 8 (15% deposi11on)

0 0.1 0.2 0.3

30

T•Tg

0.4

Waler Activity

50

0.5

60 70

0.6 0.7

10•15

REGULAR WMP 2

30

25

5

♦ 80 5 degreeC ■ 78 2 degreeC ___ • _________ ••••• ~. _ .

• 68 2 dcg,eeC • 60Sde eeC

•

0 .!.....--~- - ~ 4 _.;t.._._...L-L--~--~--0

120 l 100 ..••

G' ., ~ 80

& I!! 60 .2 e: f 40 ., >- 20

10 20 30 40 Relative Humidity{%)

•

T g lactose curve

so 60 70

Stickiness curve I Tg •X 1n.e

Cn1i,c;a1 ·x va ue = 39 ,

o.L.---~--------~----------__::.__ 0.0 0 .1 0.2 0.3

Water Activity

0.4 0 .5 0 .6

10-1 6

G' 0 !!' "' <> ~ <>

~ a; C. E " ...

30

25

5

1+ 80.Sdegreac! ■ 78 2degreeC~ - - - - - - - - • • • - .. - - - - - - - - - - - - - - - - - • • 68.2 degreeC I • ■ 605degreeC _______ _

············ .';,.a . . .... Slope= 1 07 ._ R"'2 11: 05617

• • • 0 -1--------♦---<11--------=--~------~ 0

100

90

80

70

60

50

40

30

20

10

0

10 20

- Tg Lactose - x<40(0% depos,t,on)

o x~44 7 (5% dopos11.t0n)

--x•54 (15% deposnion)

30 40 50 60 70

T-Tg

o., 0 .15 0.2 0.25 0.3 0.35 0.4 0.45 0 .5 0.55 0.6

Water Activity

10-17

MPC 44

80

70

60 C: ,g 50 ·;;; g_40 Cl> 0 ~30 0

20

10

0

0

100

90

80

u 70 e... e so :, e so &, 40 E ~ 30

20 10

MPC AND WHEY PROTEIN

• 80.5 degreeC •..•••......••• _____ •• ■ 72.2 degreeC

• • 68.2 degreeC • • - • · · · · · · · · · · -

_ ~-6_0.5 degree~J. ··--··- ....••. __ •

10 20 30 40 Relative Humidity(%)

50 60 70

Critical ·x · value== 44.5

0+------------------~--------~ 0 0.1 0.2 0.3

Water Activity

0.4 0.5 0.6

10-18

C: .2

70

60 • 80.5 degreeC ■ 72.2 degreeC

50 _ • 68. 2 degreeC • 60. 5 degreeC

'ii 40 8. ~ 30

20

10

---····· · ··-·· .... ······ y · --· · ....

•

o.._ _______ ...,..,__~--+-~+'!--~----------

90

80

70

~ 60

~ 50 :,

! 40 °' E ~ 30

20

10

0

o 10 20

• i --Tg Lactose - x=42 7 {0% depos.ruon)

~ ~ X!t44 6{S%depos1t1on)

L:_ x:48.4(15% deposition)

0.1 0.15 0.2 0.25

30 60 70 80

·.------

'

0.3 0.35 0.4 0.45 0.5 0 .55 0.6

Water Activity

10-1 9

MPC 56

8

7

6 C .g 5 ·.; 8.4 ., a *3

2

0 "' ~ a> "' :!:!.

" ;; ~ ., Q. E ., I-

♦ 80 5 degreeC - ■ 78.2oogteeC -- - - - ---- -·- - - - -- ----------- --·- ·--·. · ;,.· ·· -·-----·

_ &682degreeC __ ···········----- -------- --·-· ········-- --------• 60 5 d ...... reeC

0 10 20 30 40 50 60 70

120 l 100 - - - - -

80

60

40

20

0 0 0.1

Relative Humidity(%)

0.2 0.3

Water Activity

0,4

- -Stickiness CUN@/ Tg + X lt~

0.5 0.6

10-20

8

7

6

• 80.5 degreeC ■ 782dogreeC - -- ···· ... • ••·· · ·--- · - --- · · · · · ---·· · · ·• • ... • ·· · · -- --- ·-

-~ 5 ·i 8. 4 ,3 .... 3

- .a 68 2 degreoC I • 60 S degreeC.

2 ........ .... . .• Slope= 0 .34 ~-i ,o· 6738 · · · · · · ·

o,L_ ______ ,,_._ _ _ ___.,.._.,_..._.,L; _ __ ~------o

100

90

6 80

• e 70 "' • ~ e ? e

60

" ., E {=

50

40

30

20 0 I

10

. .

- - Tg Laciose

20

- • 1,=41 ! (00-. 0.00StllOfl) _._x,.50 4 11% oeoot11oon) ~ s.1162 , (S~ <tePOS11ion) ---s=78 !{10'1. otc)c)st.on)

0 15 02 0 25

30

0.3

40

T-Tg

0 35

Water Activity

04 0 45

60 70 80

05 0.55 06

10-21

MPC 70

120 .,--_-_-_-_-_-_-_-_-_-_---------------------------, ♦ 80 S degreeC

1 00 _ • ■ 78 2 dogreeC

-080 L _~;-~_:::~~. ---.......... _ ~ C

1::: ~:: ::: ....... :: . : : :· :: :: :: :: .. . 020- ······

0

120

100

u • • 80 .. • e. ~ i!

60

• ; .. 40 E • ...

20

0

0

10 20

Tg lactose curve

0.1 0.2

30 <O

A:Olatwc Humidity (%)

0.3

Water Activity

50 60 70

Stickiness a uvtl Tg •X llne

Critical ·x· v•tue = 49.7

0.4 o.s 0 .6

10-22

1.4

1.2

i 08

i ~ 0.6

04

02

0

140

120

6 ~ 100 g> ::!.

" =, 80 °§ 8. E

" ... 60

40

20

• 80 5 deg,eeC ■ 78.2 degreeC • • ... • ... • • • • • • - • • • • • • • • • • ·

• 68 2 degreeC • 60.Sde reeC -------------·-····

0

--------·--------Slope • 0 084 R'2 • 06345

10 20 30 40

T-T9

' ' ' ' ', ', ..............

--Tg Lac.toss - •T-Tg (O .. deposi!1on) - T-tg (2 11K, <1,tp011.1tlon)

- T-T 4%de ·-0. 1

-···-----

0.2 0.3 Water Activity

•

so

0.4

70

-0.5 0.6

10-23

MPC 85

1.4

1.2 ♦ 80.5 degreeC ■ 72.2 degf8eC - - - - - - . -• 68,2 degteeC

- • 60.Sde reeC ................. .............................. ............ -·-··· ·-C .II -,o.s 8. c3 0.6

0.4

0.2 - - - - - -• o L-----..__~ .._ _ _.___,,~ ----~ - -=-°:t---l!<'...:_ ___ _

0 10 20

120

100 u .. ~

"' 80 ..

:!!. .. 60 ~

~ ;. ; Q. 40 E " >-

20

0 0 0.1

30

0.2

40 RH %

0.3

Water Activity

50

0.4

60 70

- -Stickiness curve / Tg + X lino

Critical 'X' value= 51. 1

0.5 0.6

10-24

0.9

0.8

0.7

3 0.6

i 0.5 0 Q.

c3 0.4

* 0.3

0.2

0.1

0 0

1201

100

G " 80 & "' " ¥- 60 3 ~

;;; 0. c • ... 40

20

♦ 80.5 degreeC ■ 72.2 dcgrccC A 68.2 degreeC • 60.5 do rooC

10 20

' ' ' ' ' '

•

'

--T9 Ltd OH

- - ii1l49.t (0% Gep051bQn)

a •• 73.7 (1% oepo$tll0n)

_.,._.A•98.4 (2% deposi11on)

30

• a a a a a

-' ---

40

T-Tg

-

..

• •

.. I! ____ -~o~:_oo~i:

4 R'2=06383

r '"·- •---- '

60 70 80

• -- •

a

' ' ' a

a

' ' a

' a

---- ---······-- ------- ~----· ... ' ---• • 123.0 t3"'- oel)O&IIIOl'I)

o L.=:::::::=;::::=:=::::::=~-----------~~------~• 0 0.1 02 03 0.4

Water Activity

0.5 06 07 08

10-25

WH EY PROTEIN

0.9

0.8

07

0.6

I os 8. c3 0 4

-,_ 03

02

0 1

., ♦ 80 5 deg1eeC - - - - - - - - - - - - - - • .... - - - - - - - - - - - - - - - - - -■ 722"°9'eeC

., • 682degreeC ---··----

... • 60.Sd reeC ....................................... ~---------

10 20 30 40 50

RH %

60 70

100

90

80

Sttck1ness curve/ T9 •X hne

Cntieal ·x· value = 5-0 5

20 -- - -Tg lactose curve

10 --------------

0-1--------------~-------------~ 0 0 1 02 0.3 0 4 05 0.6

Water Activity

10-26

09

08

07

06 C

~ OS & ~ 04 $

03

02

01

0 0

140

120

~100 ~ .,, " :!,

! 80 ~

e " C. E :: 60

40

20 0 1

♦ 80.5 degre-eC - ■ 72.2 degr&eC • • • • • • • • • • • • · · · · · · · · · · · · · · • • • •

• 68.2 degreeC • 60.Sde reeC\ •• • • •••• •• • •·• · ·· ·- --- ·- •• ••• • •

• 10 20 30

02 03 04

• 40

T-Tg

0.5

Water Activity

0.6

60

Slope = 0 0433 R"2 = 0.8044

70 80

- Tg t.aciose - r. Tg (M. depo:si!>On) - T-Tg (2'!. Clepositionl ~ T-T ,, <I oon

0.7 08

10-27

SPECIALITY BUTTER M ILK POWDER

G ~

~

"' V ~ e 2 ~ 1( c ~

40

35

30

10

5

1 ♦ 80.5 d&gr&eC

■ 72 2 oogreeC • 682clegreeC 1······

......... ···· ;,··

>< 60 5 oogrooC ----~--~------------------- ------··········-

•

...

0 L_ ____ ....,,__ _____ -.,,t.::.....,,...,~ ............. L.._.,.....c._ ______ _

0 10 20

100

90

80

70

60

50

40

30 ••••.••••••••••..••.

30 RH %

4() 50

Stickiness curve I T g -+ X hne

60

20 -l-------------------,-::,,..- -----~ 0 0 1 0.2 0.3

Water Activity

04 OS 0.6

10-28

I

40

35

30

25

8. 20 ., " .. 15

♦ 80 5 degreeC ■ 72 2 degreeC - - - - - - ••••••••••••• A 68.2 degreeC • 605de reeC ...

10 · • • • • ........ -~ .... I

5 '\

•

•

..

Slope= 1.33 R0 2 = 0 7145

0-'--------- -+------ - - - """'-----------0 10 20 30 50 60 70

T-Tg

100

90

80

u 70 " e "' .. ~

60

f 50 , ~ 40 8. E " ... 30 _._ ;~T~7o~edeposuion) r--- -- -- -- -- -----

20 - r .tg (5~ <1tpo$ition) ·

10 - 1 .Tg ~10%Clt1)0$1:1on)~ ____________________ ____ ....................... ...

o.l----~---~------ ------~-----, 0 0.1 0.2 0.3

Water Activity

0.4 0.5 0.6

10-29

GUMP

7.0

6 .0

5.0

C

,g 4.0

l ~ 3.0 ...

2.0

♦ 80.5 degrecC

•. ■ 72.2 degrecC • 68.2 dcgrceC • 60.5 d-reeC

1.0 - · - · · · - · · - ·

0 10

120 ] 100

0 " e 80

'" " ~ e E

60

e • a. E 40

" ....

20

0 0 0.1

20

0.2

30

RH r/41

•

50 60

Stickiness curve /

rg•Xllne ✓----------- -

0.3

Water Activity

0.4 0 .5

70

0.6

10-30

7.0

6.0

5.0

e .2 4,0 ~ g_ ~ 3.0 ...

2.0

1.0

0

~ 90 e e, 80 ~ 0 70 ~ ~ 8_ 60

E ~

♦ 80.5 d09rceC • ■ 72.2 degreeC

& 68.2 degreeC

• 60.5 de reeC

10 20

• 30 40

T-Tg

:: ~:: :: :: . · :::::: ... : .. 30 • • .••.••.••• . •..•

20 0 .2 0.25 0.3 0.35 0 .4

Water Activity

•

so

= 0.45

..

Slope • 0.332 R"2 = 0.817

60 70

- Tg Lactose ...,_T-Tg (0%)

0.5 0 .55 0.6

10-3 I