Spray Dryer Powder Agglomeration

What Agglomeration Does to a Dairy Powder

A spray dryer running without deliberate agglomeration produces near-spherical primary particles. For skim milk they are small — commonly a volume-median well under 100 μm — dense, and difficult to reconstitute: dropped into water they float, wet slowly from the outside and form lumps with a dry core. An agglomerated powder is made of the same primary particles bonded into open, irregular clusters. The difference in behaviour comes from the pore structure between the particles rather than from the particles themselves.^[1][1]

How Agglomeration Happens Inside the Dryer

Two mechanisms operate in a modern multi-stage dairy dryer, and most plants use both.

1. Spray-to-spray agglomeration (primary, or “straight-through”)

Overlapping sprays can create collisions between droplets at different stages of drying. Liquid droplets may coalesce; suitably sticky surfaces may adhere; dry surfaces may rebound. Nozzle spacing, angle, feed conditions and the drying environment determine the outcome. Assess the actual spray pattern and powder result.^[7][8][9]

2. Fines-return agglomeration (secondary, or “nozzle-zone”)

Fine powder recovered from the cyclones or bag filter and from the fluid bed is blown back into the chamber close to the atomiser. These dry fines collide with drying droplets and act as collectors: each successful collision leaves a fine particle bonded to the surface of a droplet, and repeated collisions build the cluster. This is the dominant lever on an industrial instant-powder dryer, and it is the reason the position of the fines-return duct relative to the nozzles is a design decision rather than a piping convenience.^[1][6][7]

The Massey University and Fonterra work by Williams and co-workers is the most-cited experimental study of the mechanism. Running skim milk at small and pilot scale with and without a fines curtain, they confirmed that fines addition promotes agglomeration and that the extent of agglomeration depends principally on the mass flux ratio of fines to spray, followed by particle size.^[6] Fröhlich and co-workers at Hohenheim found that the total solids of the feed strongly affects agglomerate properties in nozzle-zone agglomeration,^[10] and van Boven and co-workers showed by response-surface trials that a higher droplet-collision frequency increases the degree of agglomeration and improves rehydration.^[8]

Merge, Stick or Bounce: The Physics of a Collision

Every collision between a fine particle and a drying droplet has one of three outcomes, and only one of them builds an agglomerate.^[7][9]

The single-droplet collision experiments at Wageningen quantified the window. Colliding glass beads with drying maltodextrin droplets and timing the collision against the droplet’s locking point (the moment a solid skin forms), Eijkelboom and co-workers found that sticking occurred only between about 0.75 and 1.5 times the locking-point time, regardless of the maltodextrin grade; before that window collisions merged, after it they bounced. Collision speed, in the 0.3–3 m/s range typical of a spray dryer, had no clear effect within that window.^[9] These experiments demonstrate mechanisms rather than a universal operating window for dairy powders: agglomeration is controlled by when in its drying history a droplet meets a fine, which is set by droplet size, chamber temperature and humidity, and fines-return position.

Why surface stickiness follows the glass transition

The sticky window exists because the surface of a drying milk droplet passes through a rubbery state on its way from solution to glass. Amorphous lactose is the component that governs it. Dry amorphous lactose has a glass transition temperature (T_g) of around 100 °C, but water is a powerful plasticiser and a few percent of moisture brings T_g down by tens of degrees.^[3] Roos showed that dairy powders become sticky once their surface temperature exceeds T_g by roughly 10–20 °C, and Hennigs, Kockel and Langrish measured the sticky-point curve for skim milk powder directly as a function of moisture content; reported T − T_g values for skim milk powder range from the mid-teens to the low thirties of degrees depending on the test method.^[2][4] Palzer’s analysis of amorphous food powders puts the same rule to work for both the agglomeration you want in the dryer and the caking you do not want in the silo.^[5]

Two consequences follow. First, the surface composition of the droplet matters as much as its bulk composition: Kim, Chen and Pearce showed that spray-dried milk powders are fat- and protein-enriched at the surface relative to the bulk, and that drying conditions change the surface composition.^[14] Fat-rich surfaces on whole milk powder are the reason it needs lecithination to become instant, while protein at the surface promotes sticking — the Wageningen group found that adding protein to a maltodextrin feed increased the fraction of agglomerated particles.^[9] Second, the same T_g physics that builds agglomerates in the chamber will build wall deposits and cake the powder if the outlet temperature and humidity are too high or the fluid bed does not cool the powder below its sticky point before packing.^[2][5]

The Atomiser Is the Primary Control

On a nozzle dryer the droplet size distribution leaving the atomiser sets everything downstream: how quickly each droplet reaches its locking point, how many collision partners exist per kilogram of feed, and how much surface area is available for evaporation. A litre of concentrate atomised to 50 μm droplets contains around 15,000 million droplets; at 100 μm it contains about 2,000 million. Droplet count scales with the inverse cube of diameter, and surface area with the inverse of diameter, so a 10% reduction in droplet size gives roughly 37% more droplets and 11% more evaporating surface.

Pressure nozzles: orifice, swirl chamber and pressure

A pressure-swirl nozzle uses its internal passages and outlet geometry to form a liquid sheet and spray. Pressure, flow, fluid properties and geometry interact. Compare the required duty and spray pattern using suitable measured data or a validated nozzle model; there is no universal best cone angle or fixed capacity increase for a change of internal component.^[12][13]

The estimator below uses a published academic scaling relation for simplex pressure-swirl atomisation. For unchanged liquid and air density and an unchanged discharge coefficient, combining its flow term with flow proportional to orifice area times the square root of pressure gives the relative form shown. These assumptions must be checked; the calculation is not a validated prediction for every milk-concentrate nozzle.^[12][13]

What moves feed viscosity

Concentrate solids, temperature, formulation, heat treatment and holding history can change viscosity. Measure viscosity at conditions relevant to the lance and compare it with stable production records. Also investigate nozzle condition, feed pressure, air handling and fines return; a change in powder properties need not have a single cause.^[1][1]

Droplet size versus powder size: fines return changes the whole distribution

Droplet size at the nozzle and particle size after drying, agglomeration and conveying describe different stages. Measure them using clearly stated methods where needed. A change in the final distribution can arise from atomisation, agglomeration, breakage or segregation; a bimodal distribution alone does not identify the cause.

Compare nozzle or internal-component changes through controlled trials at the same formulation and duty. Record flow, pressure, temperature, particle-size distribution, density, reconstitution properties, deposits and saleable-solids recovery.

Cluster nozzles

Cluster heads distribute a lance duty among several outlets and create interacting spray patterns. Any benefit in droplet size, pressure, agglomeration or capacity is design- and duty-specific. Verify spray overlap, wall clearance, operating range, cleanability and the required powder properties through representative trials.

Two-fluid nozzles for viscous feeds

Two-fluid atomisers use a gas stream to assist breakup. Compare liquid and atomising-gas duties, spray distribution, gas consumption, clogging and cleaning for the intended feed. Establish starting, stopping and protective sequences from the approved equipment design. Rotary atomisers provide another arrangement with different operating and maintenance requirements.^[12][13]

Rotary atomisers

Rotary (disc) atomisers set droplet size by wheel peripheral speed rather than pressure, handle viscous and abrasive feeds without orifice wear, and give a wide, flat spray. They are used on many dairy dryers, particularly older single-stage plants and for feeds that are hard on nozzles. Their spray geometry makes nozzle-zone fines return less effective than on a multi-nozzle dryer, so agglomeration on rotary dryers leans more heavily on the fluid bed.^[1][1]

Beyond the Atomiser: The Other Levers

Measuring Agglomeration

Agglomeration is not measured directly; it is inferred from particle size, bulk density and reconstitution behaviour. The standard methods are:^[15][16]

Interactive Atomisation Change Estimator

The estimator below applies the droplet-size and flow scaling rules from the atomisation section to a change you are considering — a different orifice disc, a pressure change, a different number of lances, or a concentrate that has come in at a different viscosity — and reports the expected change in droplet size, flow per nozzle, droplet count and evaporating surface area. It scales from your current duty, so it does not need to know your swirl chamber or dryer geometry, but it does need an honest starting point.

How to use it

1. Fill in the left-hand panel with how the dryer runs today. Take the nozzle pressure from the gauge at the high-pressure pump or lance, the orifice size from the number stamped on the orifice disc, the number of lances that are actually spraying, and the total concentrate feed from the dryer feed flowmeter. Viscosity and surface tension are the concentrate’s properties at the lance; if you do not have measurements, use the typical values in the table below.
2. Enter a verified baseline droplet size. Use a suitable measurement or validated model for the current feed and nozzle. The initial form values only illustrate the calculation; they do not establish actual milk-concentrate performance.
3. Change only what you are proposing to change in the right-hand panel. Leave everything else equal to the baseline. Total feed is held constant, so if you change pressure, orifice or nozzle count the tool checks whether the nozzles can actually pass the feed at that pressure.
4. Read the note. It tells you whether the flow balance works, what pressure the pump will really settle at if it does not, and which way the sticky window and agglomeration will move.

Related calculator: Spray Dryer Fines-Risk Indicator — screen inlet/outlet conditions and feed solids for fines generation, and Milk Powder Mass Balance Calculator for evaporator and dryer water removal.

Troubleshooting Guide

Nozzle maintenance is agglomeration control

Inspect the orifice, internal passages, sealing faces and components under the approved maintenance procedure. Wear or damage can alter flow and spray pattern. Trend pressure, flow, powder results and service hours, and investigate changes before assigning the cause to either wear or control settings.

Frequently Asked Questions

What is the difference between agglomeration and granulation?

In spray drying, agglomeration means joining primary particles into porous clusters during or immediately after drying, using the particles’ own sticky surfaces as the binder. Granulation usually refers to a separate wet process in which a binder liquid is sprayed onto dry powder in a fluid bed or mixer to build granules. Spray-fluid-bed granulation is used for some food and pharmaceutical powders, but dairy instant powders are made by in-dryer agglomeration with fines return, finished in a fluid bed.

Why does whole milk powder need lecithin to be instant?

Spray-dried whole milk powder has a fat-enriched surface, and fat is hydrophobic, so even a well-agglomerated WMP wets poorly. A small amount of lecithin, usually dissolved in butter oil, is sprayed onto the warm powder in the fluid bed; the amphiphilic lecithin coats the fat surface and lets water in. Skim milk powder has little surface fat and can be instant on agglomeration alone.^[1][1][14]

Does a higher nozzle pressure always give more agglomeration?

No. Higher pressure gives smaller droplets and more of them, which increases collision frequency and moves the sticky window closer to the nozzle. Whether that increases agglomeration depends on whether the fines-return point and the chamber conditions still put fines into that window. Raising pressure without re-tuning fines return can reduce agglomeration and raise the fines fraction.

How much fines should be returned?

There is no universal figure. The Massey work found that the fines-to-spray mass flux ratio is the dominant variable and that more fines gives more agglomeration up to the point where the extra recycle load costs capacity and energy.^[6] Most plants set it by trial against bulk density and wettability targets. What matters more than the number is that the ratio is stable, because a dryer whose fines return fluctuates cannot make a consistent powder.

Can agglomeration be modelled and controlled automatically?

Partly. CFD models coupled with population balances now predict agglomeration and coalescence in spray dryers,^[11] and single-droplet drying models can predict the locking point that defines the sticky window.^[9] Industrially, the controllable inputs are still nozzle pressure, fines-return rate, outlet temperature and humidity, and the measured outputs are bulk density, PSD and wettability with a lag of minutes to hours. Closing that loop with inline particle sizing and humidity measurement is where the practical gains are; claims of large energy or uniformity improvements from automated control should be checked against the specific plant and product before being relied on.

Can agglomeration be measured online?

Inline laser-diffraction and image-analysis particle sizers are available for powder lines and are used on some dairy dryers to track PSD in near-real time. Bulk density, wettability and dispersibility remain laboratory measurements. Because agglomerates break during sampling and conveying, any online result has to be correlated against the laboratory method at the packing point before it is used for control.

See our related spray dryer training, spray dryer fines, spray dryer crack testing, milk powder production, infant formula and milk powder, evaporator training, milk powder science information and dairy factory design pages, or browse all consultancy services.