Density of concentrates can be accurately predicted for a specific temperature and solids content. Concentrate viscosity is also predicted using a model based on Snoeren et al. ( 1 982); however it is only accurate at low soli ds contents. Few literature values for the surface tension of concentrated milks exist, but a wide range of data is avail able for standard concentration milks over a range of temperatures. The values are scattered, which appears to relate to the measurement technique employed. The experimental
results reported here that the surface tensions of concentrated milks (fresh and
reconstituted) were within the same range as l iterature val ues below 60°C, but were significantly higher for milk above 60°C . Thi s behaviour w as exhibi ted by standard whole milks using the same technique. The Wilhelmy plate technique requires time to lower and raise the plate, which introduces problems. The sol ution must drain from the plate; this particularly affects vi scous milk concentrates. Also, the time taken allows a skin to form. The mechanism that causes this is not known preci sely but relates to disulphide cross linking for the increased tendency for skin formation at high temperatures and concentrations. Although an error analysis of the influence of surface tension on droplet size indicated less than 1 0% error, a more rapid measurement
technique than the Wilhelmy plate method may avoid problems.
This work showed that nozzle placement from the measurement device was crucial in spray characterisation. Atomisation pressure on a two fluid nozzle had only a small effect on droplet size particularly at lower air pressures. Fl uid flow rate has little effect on the mean droplet size and SMC A generally shows a l arger droplet size. Size measurements made using the high speed camera were similar to those determined by the Mastersizer S . For the droplet velocity calculated using the high peed camera no relationship between atomi sation pressure or fl uid type was seen within the error l imits.
CH APTER
5EQUIPMENT AND E XPERIMENTAL DESIGN
This chapter describes the process taken to design a mini agglomerator to study the effect of relevant variables on agglomerate properties. Agglomeration is a rapi d process and, for agglomerates to be collected, subsequent drying is also a necessity. Therefore, i t makes most sense to study agglomeration inside a commercial spray drier, which can simulate the typical moisture and temperature gradients, and the air, droplet and particle i nteractions. However, manipulating operating variables does not isol ate individual micro process or the variables that affect them ; i nstead, they provide changes that may have mul tiple influences. For this reason, industrial spray driers are ill-suited to the sort of experiments required to ful l y investigate the agglomeration process. Instead a small scale spray drier was used which had the flexibil ity to vary process parameters without having to consider the impact on production and product quality.
The variables that affect droplet-particle impact and adhesion, as identified in Chapter 2, have been l isted in Table 5 . 1 , alongside the operating parameters which influence them. While the variables in the left hand col umn represent operating parameters available using a small scale drier they are not avail able to industrial operators because of upstream recycle and downstream process considerations. Therefore in this work when the words operating parameters are used, they refer to those available in the small scale drier. The variables l isted in the right hand column relate to those identified i n Chapter 2 as important to the micro processes of agglomeration. A trade-off exists between the micro processes that occur during agglomeration and the experiments that can be conducted. Experimental ly, operating parameters can be incorporated into a statistical design while maintaining as constant the drying rate, inlet air temperature and the temperature of the droplet.
Table 5 .
I :
Key variables affecting agglomeration.Parameter Units Description Variable affected
kg S·I particle flow rate N o
x m curtain width Np
m3 S·I curtain air flow rate N p ' Up
m size of particles Dp
k g s ·1 concentrate flow rate N" , Dd
P bar Atomising air Nd , Dd
TS % total solids of N D d ' d, 11, " X
d,p subscripts = droplets and particles, IV = number flow rate [s·t U = velocity [m S·I ] , D = diameter [m], 11 = concentrate viscosity [Pa s] , X = moisture content [kg kg·I ].
Earlier discussions i dentified the interaction zone in the top of a spray drier as being the most i mportant for agglomeration ; a simplified agglomerating system was i nvestigated
Chapter 5
a curtain of particles delivered through a narrow opening in the top of a small scale Niro drier which imitated the conditions of an industrial plant (see Figure 5.2). Thi s system makes i t possible to vary and study the effect of droplet and powder properties on the degree of agglomeration. A key issue with this design is ensuring droplets are dried to produce a powder.
To construct equipment suitable for studying each variable, it was necessary to consider the proj ect in four parts. The first selected the appropriate drier and establ ished its capacity; the second focused on the design of the curtain generator, and its performance; the third involved constructing the spray device and an acceptable procedure to manufacture milk concentrate; and the final design area focused on the agglomeration experiments and the design of experimental plans. Each of these parts is discussed in the sections below.
Sheet of spray droplets
Figure 5 . 1 : Sheet of spray droplets approaching a curtain of powder particles.