foratvise
Get Started
Concentration and drying

Drying

Spray drying of whole milk concentrate to whole milk powder

Validated modelOEE bound 0.945 [0.866–0.979]20,000 Monte Carlo samples · seed 20261115 · computed November 16, 2026

In short

Drying can use at most 95 of every 100 planned hours. The rest goes to limits no plant can avoid.

  1. 95usable at most
  2. 5are a physical limit

The three limits that set it

Running time: up to 95 %

The rest are stops imposed by physics or by regulation: cleaning, changeovers, defrosting.

Speed: up to 100 %

How close to its maximum speed it can run without damaging the product or the equipment.

Good product: up to 100 %

Some product falls outside specification even with the process well under control.

See the technical detail

Technical detail

Uncertainty intervals, physical models per station and sources, for anyone who wants to check each figure.

1Scenario

Whole milk concentrated in an evaporator is atomised into a co-current spray-drying chamber. Ambient air is heated in an indirect air heater to the inlet temperature, evaporates the water of the droplets adiabatically and leaves with the powder, which is separated and packed as whole milk powder within the legal maximum moisture. Results are given per kg of powder.

Provenance of inputsDeclared assumption15Verified quote17

2Results

Point value from the nominal parameters and 95% interval from 20,000 Monte Carlo draws of the uncertain parameters.
  • Water evaporated per kg of powder0.912 [0.603–0.905] kg/kg
  • TheoreticalThermodynamic minimum: reversible work of separation (water activity), per kg of powder7.96 [7.12–9.42] kJ/kg
  • Excess heat of desorption above the latent heat (isotherm, Clausius-Clapeyron), per kg of powder26.5 [24.8–30.2] kJ/kg
  • Minimum heat without latent-heat recovery: vaporisation at the outlet state plus sensible and desorption heat, per kg of powder2,196 [1,443–2,202] kJ/kg
  • Ideal adiabatic convective dryer, outlet air in equilibrium with the powder, per kg of powder2,786 [1,809–2,850] kJ/kg
  • Ideal adiabatic convective dryer at the operating outlet temperature (no heat losses), per kg of powder3,591 [2,322–3,934] kJ/kg
  • Example-plant energy minus the generic 20 % saving of an integrated fluid bed, per kg of powder (illustrative only: the plant is already two-stage, so the saving may be counted twice; not a documented BAT)5,911 kJ/kg
  • TypicalThermal energy of drying of an example two-stage dairy plant (skimmed-milk and whey powder; its boundary may include concentration), per kg of powder: a plant-level typical value, not a like-for-like dryer value7,389 kJ/kg
  • Ideal adiabatic dryer at equilibrium outlet, per kg of water evaporated3,055 [2,866–3,313] kJ/kg
  • Lowest outlet air temperature allowed by the sorption isotherm (ideal dryer)64.0 [60.7–69.2] °C
  • Equilibrium water activity of the powder at the outlet (upper bound of the outlet relative humidity)0.434 [0.355–0.468]
  • Outlet relative humidity at the operating outlet temperature0.123 [0.070–0.165]
  • Dryer thermal efficiency bound (T_in - T_out)/(T_in - T_amb) at the equilibrium outlet0.777 [0.727–0.817]
  • Dryer thermal efficiency bound at the operating outlet temperature0.629 [0.531–0.691]
  • TheoreticalMaximum water evaporated per kg of dry air (equilibrium outlet)0.0593 [0.0507–0.0674] kg/kg
  • Water evaporated per kg of dry air at the operating outlet temperature0.0461 [0.0349–0.0546] kg/kg
  • Constant-rate period of the droplet (d-squared law at inlet air temperature)0.0480 [0.0053–0.1265] s
  • Falling-rate (diffusion-limited) period to the target moisture at the operating outlet state1.54 [0.24–10.94] s
  • TheoreticalMinimum drying time of a droplet (constant-rate plus falling-rate periods)1.59 [0.26–11.02] s
  • Highest mean moisture compatible with the specification and the declared margin0.0440 [0.0374–0.0468]
  • TheoreticalMaximum sellable-mass yield (capability-limited give-away only)0.9937 [0.9869–0.9967]
  • Best documentedSellable-mass yield at the documented moisture of commercial whole milk powder0.974
  • Solids given away by drying below the moisture specification (share of sellable mass)0.00628 [0.00331–0.01309]
  • Share of powder above the legal maximum moisture0.00135 [0.00026–0.00579]
  • TheoreticalMaximum OEE quality factor (moisture capability)0.9987 [0.9942–0.9997]
  • TheoreticalMaximum availability (dryer wet cleaning and start-up heating)0.946 [0.868–0.980]
  • TheoreticalMaximum performance factor (reference: design evaporative capacity)1.00
  • TheoreticalPhysics-bounded OEE0.945 [0.866–0.979]

3Ceilings per dimension

Each dimension shows the ceilings the model computes and, where units are comparable, documented benchmarks. Colour and marker shape identify the ceiling type throughout the Oracle.
  • Theoretical
  • Best available technique
  • Best documented
  • Typical
Energy per unit of productkJ/kg · lower is better
  • TheoreticalDOE bandwidth study, spray drying, thermodynamic minimum (set to zero for lack of chemical conversion; the work of separating water from the solids is not counted) · Benchmark0 kJ/kg
  • TheoreticalThermodynamic minimum: reversible work of separation (water activity), per kg of powder7.96 [7.12–9.42] kJ/kg
  • TypicalThermal energy of drying of an example two-stage dairy plant (skimmed-milk and whey powder; its boundary may include concentration), per kg of powder: a plant-level typical value, not a like-for-like dryer value7,389 kJ/kg
Throughputkg/kg · higher is better
  • TheoreticalMaximum water evaporated per kg of dry air (equilibrium outlet)0.0593 [0.0507–0.0674] kg/kg
Cycle or residence times · lower is better
  • TheoreticalMinimum drying time of a droplet (constant-rate plus falling-rate periods)1.59 [0.26–11.02] s
Yieldhigher is better
  • TheoreticalMaximum sellable-mass yield (capability-limited give-away only)0.9937 [0.9869–0.9967]
  • Best documentedSellable-mass yield at the documented moisture of commercial whole milk powder0.974
Qualityhigher is better
  • TheoreticalMaximum OEE quality factor (moisture capability)0.9987 [0.9942–0.9997]
Availabilityhigher is better
  • TheoreticalMaximum availability (dryer wet cleaning and start-up heating)0.946 [0.868–0.980]
Performancehigher is better
  • TheoreticalMaximum performance factor (reference: design evaporative capacity)1.00
OEEhigher is better
  • TheoreticalPhysics-bounded OEE0.945 [0.866–0.979]

4Loss cascade

What separates each ceiling from the next one, from the physical limit to typical operation.
  1. TheoreticalTypicalEnergy per unit of product
    7.96 [7.12–9.42] kJ/kg → 7,389 kJ/kg (gap 7,381 kJ/kg)

    The theoretical minimum is the reversible work needed to pull water out of the milk solids against their water activity at ambient temperature. It is small because almost all the water is free (water activity near 1); only the last few percent of moisture are bound. A real dryer instead evaporates the water and throws the latent heat away with the exhaust air, so the first and largest step of the gap is the latent heat itself (the 'minimum without heat recovery' output, about two orders of magnitude above the separation work). The second step is the sensible heat carried out by the exhaust air: the outlet air cannot be cooled below the temperature at which its relative humidity reaches the equilibrium water activity of the powder, otherwise the powder would take up water again; the 'ideal adiabatic dryer at equilibrium outlet' output adds this unavoidable exhaust loss. The rest of the gap to the typical plant level is technology: an outlet above the equilibrium bound (needed to keep a finite drying time, see the drying-time outputs), heat losses of the chamber, the efficiency of the air heater and fuel, and the residual exhaust heat that is not recovered. The typical level is the thermal energy of the two-stage drying system of an example large dairy, whose boundary may include concentration; an integrated fluid-bed dryer, which lets the spray stage leave the powder wetter at a lower outlet temperature and finishes drying in a bed with long residence time and small air flow, is reported to save about 20 %, but the example plant is already two-stage, so no BAT value is derived from it. Part of that consumption also reflects the product mix of the plant (skimmed milk and sweet whey powder, whose concentrates usually carry more water per kg of powder than the whole milk concentrate of this scenario).

  2. TheoreticalBest documentedYield
    0.9937 [0.9869–0.9967] → 0.974 (gap 0.0197)

    Solids are not destroyed by drying, so the only physical yield term is the give-away: the powder is sold by mass, and every point of moisture below the legal maximum is milk solids sold as if they were water. With a capable process (small moisture spread) the mean can sit a few standard deviations below the 5 % maximum. Commercial whole milk powder is documented at about 2.5 % water, i.e. dried far below what the specification requires, mainly for storage stability and flowability (a commercial choice, not a physical necessity). Powder losses in the exhaust (fines), deposits and start-up material are technology losses.

5Physics-bounded OEE

The highest OEE the operation can reach: availability, performance and quality limited only by unavoidable stops and physical capability.
Fraction of loading time that remains after each unavoidable loss. Axis starts at 0.80.
  • Loading time (ideal)
    1.000
  • Availability
    factor 0.946 [0.868–0.980]
    0.946 (−0.054)
  • Performance
    factor 1.00
    0.946 (−0.000)
  • Quality
    factor 0.9987 [0.9942–0.9997]
    0.945 (−0.001)
  • Physics-bounded OEE
    0.945 [0.866–0.979]

6Full derivation

Mass and energy balances, rate-limiting step, process capability and unavoidable stops, with every term explained.

Scope. Whole milk concentrate, produced in an evaporator, is atomised into a co-current spray-drying chamber. Ambient air is heated in an indirect heater to the inlet temperature, dries the droplets adiabatically and leaves the chamber together with the powder, which is separated and packed as whole milk powder. Every result is expressed per kilogram of powder unless stated otherwise. Symbols and their values, units, uncertainty distributions and sources are listed in the parameter table generated from config/physics/drying.yaml; the numbers quoted in this text are the point values of that table (Monte Carlo intervals are in the output table).

1. Mass balance — maximum yield

Let ww be the water content of the powder (mass fraction, wet basis) and TSfeedTS_{feed} the total solids of the concentrate. Drying removes water and nothing else, so the mass of solids is conserved. Per kilogram of powder:

s=1−w,F=sTSfeed,E=F−1=1−wTSfeed−1,s = 1 - w, \qquad F = \frac{s}{TS_{feed}}, \qquad E = F - 1 = \frac{1-w}{TS_{feed}} - 1,

where ss is the mass of dry solids in one kilogram of powder, FF the mass of concentrate that contains them and EE the mass of water evaporated. With TSfeed=0.50TS_{feed} = 0.50 (the lower end of the 50–60 % dry matter reported for milk thickened in an evaporator) and the mean moisture μ=4.4%\mu = 4.4\% derived in Section 4, E=0.912E = 0.912 kg of water per kg of powder.

Yield. No solids are destroyed, so the solids yield of the operation is exactly 1. Powder, however, is sold by mass, and its mass includes water. One kilogram of solids leaves as 1/(1−w)1/(1-w) kilograms of powder; the legal maximum moisture wmaxw_{max} (5 % for whole milk powder in the Codex standard) sets the largest sellable mass. Every point of moisture below wmaxw_{max} is milk solids handed over in place of water. The sellable-mass yield is therefore

Y=1/(1−μ)1/(1−wmax)=1−wmax1−μ,Y = \frac{1/(1-\mu)}{1/(1-w_{max})} = \frac{1 - w_{max}}{1-\mu},

and the give-away is 1−Y1 - Y. The physical ceiling YmaxY_{max} uses the highest mean moisture compatible with process capability (Section 4); with μ=4.4%\mu = 4.4\% the give-away is 0.63 %. The documented ceiling uses the water content of commercial whole milk powder in the USDA reference record, 2.47 %, which gives Y=0.974Y = 0.974: commercial powder is dried far below the legal limit, mainly for storage stability and flowability. Powder carried away in the exhaust air, deposits in the chamber and material produced during start-up are losses of the installed technology, not of physics.

2. Energy balance — minimum energy per unit of product

Three nested lower bounds are derived, each adding one physically unavoidable mechanism.

2.1 Absolute thermodynamic minimum: reversible work of separation

Water in milk solids is held with a chemical potential lower than that of pure liquid water. The difference is measured by the water activity awa_w, the ratio between the vapour pressure of water over the product and that of pure water at the same temperature. Removing a small mass dmwdm_w of water reversibly, at the dead-state (ambient) temperature T0T_0, and delivering it as pure liquid water requires the work

δW=RT0Mw ln⁡1aw  dmw,\delta W = \frac{R T_0}{M_w}\,\ln\frac{1}{a_w}\; dm_w ,

where RR is the molar gas constant and MwM_w the molar mass of water. Over the drying path, from the moisture of the feed X0X_0 to the final moisture XfX_f (both on a dry basis, kg water per kg of solids that sorb water),

wsep=snf∫XfX0RT0Mw ln⁡1aw(X,T0)  dX.w_{sep} = s_{nf}\int_{X_f}^{X_0} \frac{R T_0}{M_w}\,\ln\frac{1}{a_w(X, T_0)}\; dX .

Only the fat-free solids sorb water appreciably, so the moisture is expressed per kilogram of fat-free solids, snf=s (1−xfat,d)s_{nf} = s\,(1 - x_{fat,d}), with xfat,dx_{fat,d} the fat share of the dry solids of whole milk powder (USDA record). The water activity comes from the modified Oswin desorption isotherm of nonfat dry milk (desorption is the branch that applies to drying):

X=Me0(T)(aw1−aw)1/C,Me0(T)=A+B T,⟹aw=r1+r,  r=(XMe0(T))C,X = M_{e0}(T)\left(\frac{a_w}{1-a_w}\right)^{1/C}, \qquad M_{e0}(T) = A + B\,T, \qquad\Longrightarrow\qquad a_w = \frac{r}{1+r},\; r = \left(\frac{X}{M_{e0}(T)}\right)^{C},

with TT in °C, A=10.31%A = 10.31\%, B=−0.05%B = -0.05\%/K and C=2.3C = 2.3 fitted on measured isotherms at 23, 40 and 60 °C. The integral is evaluated numerically on a geometric grid. The result, 8.0 kJ per kg of powder, is small because almost all the water in the concentrate is free (awa_w close to 1, ln⁡(1/aw)\ln(1/a_w) close to 0); only the last few percent of moisture are bound strongly enough to cost noticeable work.

Difference with the DOE thermodynamic minimum. The DOE bandwidth study sets the thermodynamic minimum of spray drying of powdered dry milk to zero, "due to minimal chemical conversions": its thermodynamic minimum is the Gibbs energy of chemical reactions. The same study does use −RTln⁡aw-RT\ln a_w for the concentration and drying of sugar, which is exactly the separation work above. The value here is the dairy counterpart of that calculation; it is positive but tiny, so both values are, in practice, "close to zero" compared with any real dryer.

2.2 Minimum without latent-heat recovery

A convective dryer does not separate water reversibly: it evaporates it and lets the vapour leave with the exhaust air. If the latent heat is not recovered, the energy that must be supplied is at least

qevap=E [hv(Tout)−hl(Tfeed)]+(s cp,s+μ cp,w)(Tout−Tfeed)+qdes,q_{evap} = E\,\big[h_v(T_{out}) - h_l(T_{feed})\big] + \big(s\,c_{p,s} + \mu\,c_{p,w}\big)\big(T_{out}-T_{feed}\big) + q_{des},

where hv(Tout)h_v(T_{out}) is the specific enthalpy of water vapour at the outlet temperature, hl(Tfeed)h_l(T_{feed}) that of liquid water at the temperature of the concentrate at the atomiser (both IAPWS-95 through CoolProp, on the same reference state), cp,sc_{p,s} and cp,wc_{p,w} the specific heats of the powder solids and of water (composition model of Choi and Okos, meo.physics.foodprops), and qdesq_{des} the excess heat of desorption. The excess heat is the energy needed, above the latent heat of free water, to detach bound water; by Clausius–Clapeyron applied at constant moisture,

qst−hfg=RT2Mw(∂ln⁡aw∂T)X=RT2Mw −C BMe0(T) (1+r),q_{st} - h_{fg} = \frac{R T^2}{M_w}\left(\frac{\partial \ln a_w}{\partial T}\right)_X = \frac{R T^2}{M_w}\,\frac{-C\,B}{M_{e0}(T)\,(1+r)},

integrated over the same moisture path (B<0B < 0, so the excess is positive). The outlet temperature used is the lowest one allowed by the isotherm (Section 2.3). The result is 2 196 kJ per kg of powder, of which the excess desorption heat is 26.5 kJ; the latent heat is about 270 times the separation work. This bound already shows where the energy goes: into vaporisation, not into separation.

2.3 Ideal adiabatic convective dryer

Ambient air at T0T_0 with humidity ratio W0W_0 (from the ambient relative humidity) is heated at constant humidity ratio to TinT_{in}, enters the chamber and leaves at ToutT_{out} with humidity ratio WoutW_{out}. With no heat losses, the energy balance per kilogram of powder, with mam_a kilograms of dry air, is

ma[ha(Tin,W0)−ha(Tout,W0)]=E [hv(Tout)−hl(Tfeed)]+(s cp,s+μ cp,w)(Tout−Tfeed),m_a\big[h_a(T_{in}, W_0) - h_a(T_{out}, W_0)\big] = E\,\big[h_v(T_{out}) - h_l(T_{feed})\big] + \big(s\,c_{p,s} + \mu\,c_{p,w}\big)\big(T_{out}-T_{feed}\big),

where ha(T,W)=hda(T)+Whv(T)h_a(T,W) = h_{da}(T) + W h_v(T) is the enthalpy of moist air per kilogram of dry air (dry air and IAPWS-95 water from CoolProp, ideal mixture; the helpers are checked against CoolProp's real-gas humid-air routine HAPropsSI in the tests). This fixes mam_a for any outlet temperature, and

Wout=W0+Ema,φout=pvpsat(Tout),pv=P Woutε+Wout,W_{out} = W_0 + \frac{E}{m_a}, \qquad \varphi_{out} = \frac{p_v}{p_{sat}(T_{out})}, \qquad p_v = \frac{P\,W_{out}}{\varepsilon + W_{out}},

with φout\varphi_{out} the outlet relative humidity, pvp_v the partial pressure of water vapour, PP the total pressure and ε=Mw/Ma\varepsilon = M_w/M_a the ratio of the molar masses of water and dry air.

Outlet humidity bound. The powder leaves with the outlet air. If φout\varphi_{out} exceeded the equilibrium water activity of the powder at its target moisture, the powder would take water back instead of drying: φout≤aw(Xf,Tout)\varphi_{out} \le a_w(X_f, T_{out}). Lowering ToutT_{out} uses the heat of the air more completely but raises φout\varphi_{out}; the ideal dryer therefore leaves at the lowest temperature at which φout=aw(Xf,Tout)\varphi_{out} = a_w(X_f, T_{out}), found by bisection. At the point values Tout,min=64.0T_{out,min} = 64.0 °C and aw=0.434a_w = 0.434. The heater duty is

qdryer=ma [ha(Tin,W0)−ha(T0,W0)],q_{dryer} = m_a\,\big[h_a(T_{in}, W_0) - h_a(T_0, W_0)\big],

2 786 kJ per kg of powder (3 055 kJ per kg of water) for an inlet at 200 °C. At the declared operating outlet temperature (90 °C) the same ideal dryer needs 3 591 kJ per kg of powder.

Thermal-efficiency bound. Because the specific heat of the air changes little, the share of the heater duty that is used for evaporation is close to the classic bound

η=Tin−ToutTin−T0,\eta = \frac{T_{in} - T_{out}}{T_{in} - T_0},

the fraction of the temperature rise of the air that is given back in the chamber; the rest leaves as sensible heat of the exhaust. It is 0.777 at the equilibrium outlet and 0.629 at the operating outlet. A hotter inlet raises η\eta and lowers the energy per kilogram of water (sensitivity table), within the limit set by heat damage to the product.

2.4 Example plant, best available technique and documented operation

The BAT reference document describes an example large dairy with a two-stage drying system whose specific thermal energy consumption was 2 052.6 kWh per tonne of product, i.e. 7 389 kJ per kg of powder. The plant produced skimmed milk and sweet whey powder, and its boundary may include concentration: the same source says that about 600 kWh evaporate one tonne of water and calls the figure near the theoretical need, which implies about 3.4 t of water per t of product, far more than the 0.91 t of this scenario. The value is therefore used as the typical energy level of a plant, not as a like-for-like documented dryer value. The source also states that an integrated fluid-bed dryer reduces the drying energy by about 20 %; applied to the example plant this gives 5 911 kJ per kg, but that plant is already two-stage, so the saving may be counted twice, and the value is reported without a ceiling label. No open source documents a best-available-technique value for the dryer itself.

The DOE bandwidth study lists, for spray drying of powdered dry milk, 115 Btu/lb (current typical and state of the art) and 69 Btu/lb (practical minimum), about 267 and 160 kJ per kg. At face value these are about one tenth of the latent heat of the water that must be evaporated per kilogram of powder (Section 2.2), so their basis cannot be the full heat duty of the dryer per kilogram of powder; they are reported with compare: false. The indicative specific energy of dairies whose main product is powder (0.2–0.5 MWh per tonne of raw materials, BAT conclusions) covers the whole installation and is expressed per tonne of raw milk, so it is not comparable either.

The ordering of the energy ceilings with the real numbers is: separation work 8.0 < minimum without recovery 2 196 < ideal dryer at equilibrium outlet 2 786 < ideal dryer at operating outlet 3 591 < example plant 7 389 kJ per kg of powder (typical).

3. Rate-limiting step — minimum drying time and maximum capacity

Constant-rate period. While the droplet surface is wet it stays at the wet-bulb temperature TwbT_{wb} of the air, and evaporation is limited by heat transfer from the gas. For a droplet that moves with the gas the Nusselt number is Nu=hd/k=2Nu = h d/k = 2 (pure conduction around a sphere), so the heat flow is 2πd k (Tg−Twb)2\pi d\,k\,(T_g - T_{wb}) and the evaporated mass flow is that divided by the latent heat λ\lambda. If the volumes of water and solids are additive, only the water volume shrinks: ρw dV/dt=−2πd k (Tg−Twb)/λ\rho_w\,dV/dt = -2\pi d\,k\,(T_g - T_{wb})/\lambda, and with V=πd3/6V = \pi d^3/6 this gives the d-squared law

d(d2)dt=−8 k (Tg−Twb)ρw λ,tcr=ρw λ (d02−dc2)8 k (Tg−Twb),\frac{d(d^2)}{dt} = -\frac{8\,k\,(T_g - T_{wb})}{\rho_w\,\lambda}, \qquad t_{cr} = \frac{\rho_w\,\lambda\,(d_0^2 - d_c^2)}{8\,k\,(T_g - T_{wb})},

where d0d_0 is the initial droplet diameter, dcd_c the diameter at which the solids reach the critical content TScTS_c and a crust forms, kk the thermal conductivity of air at the film temperature (CoolProp), ρw\rho_w the density of water and λ\lambda the latent heat at TwbT_{wb}. The diameter dcd_c follows from conservation of the solids: dc3/d03=vc/v0d_c^3/d_0^3 = v_c/v_0, with v=1/ρs+(1−TS)/(TS ρw)v = 1/\rho_s + (1-TS)/(TS\,\rho_w) the volume per kilogram of solids and ρs\rho_s the density of the solids (Choi–Okos). Taking Tg=TinT_g = T_{in}, the hottest gas the droplet can meet, makes tcrt_{cr} a lower bound: 0.048 s for a 60 µm droplet.

Falling-rate period. Once the crust forms, water must diffuse to the surface. For a sphere of radius rc=dc/2r_c = d_c/2 with surface moisture in equilibrium with the air, the first term of the series solution of Fick's law gives

tfr=rc2π2Deff ln⁡ ⁣[6π2 Xc−XeXf−Xe],t_{fr} = \frac{r_c^2}{\pi^2 D_{eff}}\,\ln\!\left[\frac{6}{\pi^2}\,\frac{X_c - X_e}{X_f - X_e}\right],

where DeffD_{eff} is the effective diffusivity of water in the particle, XcX_c the moisture at the critical point, XfX_f the target moisture and XeX_e the equilibrium moisture with the surrounding air (isotherm). At the equilibrium-limited outlet of Section 2.3, Xe=XfX_e = X_f and tfrt_{fr} is infinite: the thermodynamically ideal dryer needs an infinitely long chamber. This is the physical reason why real dryers leave the air warmer and drier than the equilibrium bound. The drying times are therefore evaluated with the air at the operating outlet temperature (φout=0.123\varphi_{out} = 0.123), giving tfr=1.54t_{fr} = 1.54 s and a minimum drying time tcr+tfr=1.59t_{cr} + t_{fr} = 1.59 s, with a wide interval because DeffD_{eff} and d0d_0 are declared assumptions.

Maximum evaporative capacity. The water that one kilogram of dry air can take up is Wout−W0=E/maW_{out} - W_0 = E/m_a: 0.0593 kg per kg of dry air at the equilibrium outlet (the theoretical rate ceiling) and 0.0461 at the operating outlet. For a given air flow, this is the maximum throughput of the dryer.

4. Process capability — quality ceiling

The powder moisture under control is modelled as w∼N(μ,σw2)w \sim \mathcal N(\mu, \sigma_w^2). Powder above the legal maximum wmaxw_{max} is out of specification. Setting the mean zz standard deviations below the limit,

μ=wmax−z σw,d=Φ(−z),\mu = w_{max} - z\,\sigma_w, \qquad d = \Phi(-z),

where Φ\Phi is the standard normal distribution function and dd the out-of-specification share. With z=3z = 3 and σw=0.2\sigma_w = 0.2 percentage points, μ=4.4%\mu = 4.4\% and d=0.13%d = 0.13\%, so Qmax=1−d=0.9987Q_{max} = 1 - d = 0.9987. The margin is paid in give-away (Section 1): a wider spread forces a drier mean and more solids sold as water, which is why process capability is an economic as well as a quality ceiling.

5. Unavoidable stops — maximum availability

Milk deposits on the chamber walls, ducts and filters, and the regulation requires drying equipment to be cleaned "at intervals and by methods recommended by the manufacturer". No open numerical interval was found, so the interval between wet cleanings tcyct_{cyc}, the cleaning duration tCIPt_{CIP} and the start-up heating of the chamber on water tsut_{su} are declared scenario values:

Amax=1−tCIP+tsutcyc=0.946(tcyc=168 h, tCIP=8 h, tsu=1 h).A_{max} = 1 - \frac{t_{CIP} + t_{su}}{t_{cyc}} = 0.946 \quad (t_{cyc} = 168\ \text{h},\ t_{CIP} = 8\ \text{h},\ t_{su} = 1\ \text{h}).

The evaporator that feeds the dryer must be cleaned more often (at most every 44 h); that stop belongs to the evaporation model and stops the dryer only if there is no second evaporator or concentrate buffer.

6. Physics-bounded OEE

Scope of the bound. The theoretical label of OEEmaxOEE_{max} means physics-bounded under the declared scenario: the ceiling follows from the physics of the operation for the declared product, equipment and schedule, and it is a bound only conditional on those declarations. It follows the common definition of all models: AmaxA_{max} counts only stops required by regulation or forced by a physical mechanism, at their physical minimum duration, plus the start-up they force; PmaxP_{max} is the ratio of the physics-limited ideal cycle to the cycle of the capable process; QmaxQ_{max} is the share of conforming units, and mass losses of conforming product are reported only as yield.

OEEmax=Amax×Pmax×Qmax,Pmax=1,Qmax=1−Φ(−z).OEE_{max} = A_{max} \times P_{max} \times Q_{max}, \qquad P_{max} = 1, \qquad Q_{max} = 1 - \Phi(-z).

Pmax=1P_{max} = 1 refers to the design evaporative capacity of Section 3. The give-away of Section 1 is a material-yield loss, not an OEE quality loss (the powder is in specification), and is reported separately as yield_max. The point value is 0.945; the Monte Carlo interval over all declared uncertainties is in the output table.

7. Assumptions and validity

See the assumption and validity lists generated from the configuration file. The most important are: adiabatic chamber; only fat-free solids sorb water; the isotherm fitted at 23–60 °C is extrapolated with its linear temperature term to the outlet temperature; droplet drying with Nu=2Nu = 2 and additive volumes. The most influential inputs (sensitivity table) are the feed solids and the inlet air temperature for energy, the droplet size and diffusivity for drying time, and the cleaning interval for availability.

References

  • Codex Alimentarius Commission. Codex Standard for Milk Powders and Cream Powder, CODEX STAN 207-1999 (amendments 2010, 2013, 2014).
  • European Commission, Joint Research Centre. Best Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries, 2019.
  • U.S. Department of Energy. Bandwidth Study on Energy Use and Potential Energy Savings Opportunities in U.S. Food and Beverage Manufacturing, 2017.
  • U.S. Food and Drug Administration. Grade "A" Pasteurized Milk Ordinance, 2023 Revision.
  • U.S. Department of Agriculture, Agricultural Research Service. USDA National Nutrient Database for Standard Reference, Release 28, food group 01 report, 2015.
  • Kadas, E., Oduola, A., Rubinelli, P. M., Atungulu, G. G., Acuff, J. C. (2025). Modeling moisture sorption isotherms of milk powders at ambient and elevated temperatures using the dynamic dewpoint isotherm (DDI) method. JDS Communications, 6(5), 629–634. doi:10.3168/jdsc.2024-0683
  • Choi, Y., Okos, M. R. (1986). Effects of temperature and composition on the thermal properties of food. In Food Engineering and Process Applications, Vol. 1, Elsevier Applied Science, pp. 93–101 (coefficients cited through the open reproductions listed in config/physics/food_properties.yaml).
  • Lewis, W. K. (1921). The rate of drying of solid materials. Journal of Industrial & Engineering Chemistry, 13(5), 427–432. doi:10.1021/ie50137a021
  • Wagner, W., Pruß, A. (2002). The IAPWS formulation 1995 for the thermodynamic properties of ordinary water substance for general and scientific use. Journal of Physical and Chemical Reference Data, 31, 387–535. doi:10.1063/1.1461829
  • Bell, I. H., Wronski, J., Quoilin, S., Lemort, V. (2014). Pure and pseudo-pure fluid thermophysical property evaluation and the open-source thermophysical property library CoolProp. Industrial & Engineering Chemistry Research, 53, 2498–2508. doi:10.1021/ie4033999

7Parameters and provenance

Every input with its value, unit, role, uncertainty distribution and provenance. Open a quote to read the passage of the source that states the value.
  • TSfeedTS_{feed}0.5

    Total solids of the concentrate fed to the dryer (mass fraction)

    Technology · Uniform 0.5 to 0.6

    Verified quote
    Show verbatim quote
    After the milk has been thickened from 11 % to 50–60 % dry matter in an evaporatorBest Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries (opens in a new tab). European Commission, Joint Research Centre; Publications Office of the European Union, 2019. p. 411. EUR 29978 EN; JRC118627; doi:10.2760/243911
  • wmaxw_{max}0.05

    Legal maximum water content of whole milk powder (mass fraction)

    Regulatory · Fixed

    Verified quote
    Show verbatim quote
    Whole milk powder Milkfat Minimum 26% and less than 42% m/m Maximum water(a) 5% m/mCodex Standard for Milk Powders and Cream Powder, CODEX STAN 207-1999 (adopted 1999; amendments 2010, 2013, 2014) (opens in a new tab). Codex Alimentarius Commission, FAO/WHO Food Standards Programme, 2014. p. 2. CXS 207-1999
  • wdocw_{doc}0.0247

    Water content of commercial whole milk powder (USDA reference record)

    Benchmark · Fixed

    Verified quote
    Show verbatim quote
    2.47 496 2075 26.32 26.71 6.08 38.42USDA National Nutrient Database for Standard Reference, Release 28 (2015): Food Group 01, Dairy and Egg Products (nutrient report) (opens in a new tab). U.S. Department of Agriculture, Agricultural Research Service, Nutrient Data Laboratory, 2015. p. 556. USDA SR28, food group 01 report (sr28fg01.pdf)
  • xfatx_{fat}0.2671

    Fat content of whole milk powder (mass fraction of the powder)

    Property · Fixed

    Verified quote
    Show verbatim quote
    2.47 496 2075 26.32 26.71 6.08 38.42USDA National Nutrient Database for Standard Reference, Release 28 (2015): Food Group 01, Dairy and Egg Products (nutrient report) (opens in a new tab). U.S. Department of Agriculture, Agricultural Research Service, Nutrient Data Laboratory, 2015. p. 556. USDA SR28, food group 01 report (sr28fg01.pdf)
  • zz3

    Moisture set-point margin below the legal maximum, in standard deviations of the powder moisture

    Scenario · Uniform 2.5 to 3.5

    Declared assumption

    Operating choice, point value at the centre of the declared range 2.5-3.5, as in the pasteurization model: a larger margin removes out-of-specification powder at the cost of more solids given away.

  • σw\sigma_w0.002

    Standard deviation of the powder moisture under control (mass fraction)

    Scenario · Uniform 0.001 to 0.004

    Declared assumption

    Declared process-capability assumption (0.1 to 0.4 percentage points); no open source was found for the moisture variability of industrial dryers. Replace with the plant's measured value.

  • T0T_025 °C

    Ambient air temperature (dryer air intake and thermodynamic dead state)

    Scenario · Uniform 15 to 35 °C

    Declared assumption
  • φ0\varphi_00.6

    Relative humidity of the ambient air

    Scenario · Uniform 0.3 to 0.8

    Declared assumption
  • PP101,325 Pa

    Pressure of the drying air

    Scenario · Fixed

    Declared assumption

    Standard atmosphere; spray-drying chambers run at a slight negative pressure that does not change the result.

  • TinT_{in}200 °C

    Inlet air temperature of the spray dryer

    Scenario · Uniform 180 to 220 °C

    Declared assumption

    Declared design assumption; no open source with dairy spray-dryer inlet temperatures was verified. The theoretical results are given as functions of it (sensitivity table).

  • Tout,opT_{out,op}90 °C

    Operating outlet air temperature of the spray dryer

    Scenario · Uniform 85 to 100 °C

    Declared assumption

    Declared operating assumption; no open source verified. Must stay above the equilibrium-limited outlet temperature computed by the model.

  • TfeedT_{feed}60 °C

    Temperature of the concentrate at the atomiser

    Scenario · Uniform 45 to 75 °C

    Declared assumption
  • AA10.31 %

    Modified Oswin desorption isotherm of nonfat dry milk: constant term of Me0(T) = A + B T (percent, dry basis)

    Property · Fixed

    Verified quote
    Show verbatim quote
    Modified Oswin | 10.31 | −0.05 | 2.3 | 1.57 | 2.51Modeling moisture sorption isotherms of milk powders at ambient and elevated temperatures using the dynamic dewpoint isotherm (DDI) method (Kadas, Oduola, Rubinelli, Atungulu, Acuff), JDS Communications 6(5):629-634, open access CC BY 4.0 (opens in a new tab). Elsevier on behalf of the American Dairy Science Association; full text served by Europe PMC (open-access XML), 2025. p. 1. doi:10.3168/jdsc.2024-0683; PMCID PMC12414362
  • BB-0.05 %/K

    Modified Oswin desorption isotherm of nonfat dry milk: temperature coefficient (percent per kelvin)

    Property · Fixed

    Verified quote
    Show verbatim quote
    Modified Oswin | 10.31 | −0.05 | 2.3 | 1.57 | 2.51Modeling moisture sorption isotherms of milk powders at ambient and elevated temperatures using the dynamic dewpoint isotherm (DDI) method (Kadas, Oduola, Rubinelli, Atungulu, Acuff), JDS Communications 6(5):629-634, open access CC BY 4.0 (opens in a new tab). Elsevier on behalf of the American Dairy Science Association; full text served by Europe PMC (open-access XML), 2025. p. 1. doi:10.3168/jdsc.2024-0683; PMCID PMC12414362
  • CC2.3

    Modified Oswin desorption isotherm of nonfat dry milk: exponent

    Property · Fixed

    Verified quote
    Show verbatim quote
    Modified Oswin | 10.31 | −0.05 | 2.3 | 1.57 | 2.51Modeling moisture sorption isotherms of milk powders at ambient and elevated temperatures using the dynamic dewpoint isotherm (DDI) method (Kadas, Oduola, Rubinelli, Atungulu, Acuff), JDS Communications 6(5):629-634, open access CC BY 4.0 (opens in a new tab). Elsevier on behalf of the American Dairy Science Association; full text served by Europe PMC (open-access XML), 2025. p. 1. doi:10.3168/jdsc.2024-0683; PMCID PMC12414362
  • d0d_060 micrometer

    Initial droplet diameter at the atomiser

    Scenario · Uniform 30 to 100 micrometer

    Declared assumption

    Declared assumption for the atomiser; no open source verified. Drying time scales with its square.

  • TScTS_c0.7

    Solids content at which the constant-rate period ends (a crust forms)

    Scenario · Uniform 0.6 to 0.8

    Declared assumption

    Declared assumption; no open source verified for milk droplets.

  • DeffD_{eff}1.00 × 10⁻¹⁰ m²/s

    Effective diffusivity of water in the drying particle (falling-rate period)

    Scenario · Lognormal, σ 0.7

    Declared assumption

    Declared assumption with a wide lognormal uncertainty; no open source verified. The falling-rate time is inversely proportional to it.

  • edoce_{doc}2,052.6 kWh/t

    Specific thermal energy of the two-stage drying system of an example large dairy (per tonne of product)

    Benchmark · Fixed

    Verified quote
    Show verbatim quote
    The specific thermal energy consumption was 2 052.6 kWh/t of product or 162.5 kWh/t of raw milkBest Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries (opens in a new tab). European Commission, Joint Research Centre; Publications Office of the European Union, 2019. p. 412. EUR 29978 EN; JRC118627; doi:10.2760/243911
  • sFBDs_{FBD}0.2

    Reduction of the drying energy with an integrated fluid-bed dryer

    Benchmark · Fixed

    Verified quote
    Show verbatim quote
    if an integrated FBD is used, the energy consumption for drying can be reduced by approximately 20 %Best Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries (opens in a new tab). European Commission, Joint Research Centre; Publications Office of the European Union, 2019. p. 412. EUR 29978 EN; JRC118627; doi:10.2760/243911
  • tcyct_{cyc}168 h

    Interval between two wet cleanings of the spray dryer

    Scenario · Uniform 72 to 336 h

    Declared assumption

    Declared assumption. The PMO leaves the interval to the manufacturer and the regulator (see the unavoidable stop below); no open numerical value was found. Range 3 to 14 days.

  • tCIPt_{CIP}8 h

    Duration of the wet cleaning of the dryer

    Scenario · Uniform 4 to 12 h

    Declared assumption

    Declared assumption; replace with the validated cleaning programme of the plant.

  • tsut_{su}1 h

    Start-up heating of the chamber and stabilisation on water before product after each cleaning

    Scenario · Uniform 0.5 to 2 h

    Declared assumption

8Benchmarks

Documented values from open sources used to check the ceilings; no documented value may beat the theoretical limit.
BenchmarkCeilingValueProvenance
DOE bandwidth study, spray drying of powdered dry milk, current typical intensity (not comparable: at face value it is about one tenth of the latent heat of the water evaporated per lb of powder, so its basis cannot be the full heat duty of the dryer per lb of powder)
Show verbatim quote
Spray Drying 1,816 115 115 69 0Bandwidth Study on Energy Use and Potential Energy Savings Opportunities in U.S. Food and Beverage Manufacturing (opens in a new tab). U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, 2017. p. 57. DOE/EE-1571; OSTI ID 1512366
Typical115 Btu/lbVerified quote
DOE bandwidth study, spray drying of powdered dry milk, state-of-the-art intensity (same basis caveat as the current typical value)
Show verbatim quote
Spray Drying 1,816 115 115 69 0Bandwidth Study on Energy Use and Potential Energy Savings Opportunities in U.S. Food and Beverage Manufacturing (opens in a new tab). U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, 2017. p. 57. DOE/EE-1571; OSTI ID 1512366
Best available technique115 Btu/lbVerified quote
DOE bandwidth study, spray drying of powdered dry milk, practical minimum (R&D technologies; same basis caveat)
Show verbatim quote
Spray Drying 1,816 115 115 69 0Bandwidth Study on Energy Use and Potential Energy Savings Opportunities in U.S. Food and Beverage Manufacturing (opens in a new tab). U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, 2017. p. 57. DOE/EE-1571; OSTI ID 1512366
—69 Btu/lbVerified quote
DOE bandwidth study, spray drying, thermodynamic minimum (set to zero for lack of chemical conversion; the work of separating water from the solids is not counted)
Show verbatim quote
Spray Drying 1,816 115 115 69 0Bandwidth Study on Energy Use and Potential Energy Savings Opportunities in U.S. Food and Beverage Manufacturing (opens in a new tab). U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, 2017. p. 57. DOE/EE-1571; OSTI ID 1512366
Theoretical0 Btu/lbVerified quote
BAT conclusions, indicative specific energy of dairies whose main product is powder, lower end (whole installation, per tonne of raw materials: not comparable with a per-kg-of-powder dryer value)
Show verbatim quote
Powder 0.2–0.5Best Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries (opens in a new tab). European Commission, Joint Research Centre; Publications Office of the European Union, 2019. p. 718. EUR 29978 EN; JRC118627; doi:10.2760/243911
Best documented0.2 MWh/tVerified quote
Documented milk loss of full-cream milk powder production (share of milk processed, whole process; a loss of the whole plant, not a yield fraction of the dryer: not comparable with yield_max)
Show verbatim quote
Full-cream milk powder 0.64 0.22Best Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries (opens in a new tab). European Commission, Joint Research Centre; Publications Office of the European Union, 2019. p. 395. EUR 29978 EN; JRC118627; doi:10.2760/243911
Best documented0.64 %Verified quote
Residual moisture of the solids leaving the spray-drying stage of a multistage dryer, lower end (the fluid bed dries further)
Show verbatim quote
The solids leave the spray dryer with 3–5 % residual moistureBest Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries (opens in a new tab). European Commission, Joint Research Centre; Publications Office of the European Union, 2019. p. 411. EUR 29978 EN; JRC118627; doi:10.2760/243911
Best documented3 %Verified quote

9Unavoidable stops

Stops that physics or regulation impose; they set the maximum availability.
  • Wet cleaning of the drying equipmentVerified quote

    Milk deposits on the chamber walls, ducts and filters; drying equipment is cleaned at intervals set by the manufacturer and approved by the regulator. The interval and duration are declared scenario values.

    Show verbatim quote
    Drying equipment, cloth-collector systems, packaging equipment and multi-use dry milk products and dry whey storage containers are cleaned at intervals and by methods recommended by the manufacturerGrade "A" Pasteurized Milk Ordinance, 2023 Revision (opens in a new tab). U.S. Department of Health and Human Services, Public Health Service, Food and Drug Administration, 2023. p. 99. PMO 2023 Revision
  • Start-up heating after each cleaningDeclared assumption

    After wet cleaning the chamber must be dried and brought to the operating outlet temperature on water before product is atomised (declared duration).

10Assumptions

  1. Steady adiabatic dryer: no heat losses through the chamber walls (they only increase consumption); the powder leaves at the outlet air temperature.
  2. Air is an ideal mixture of dry air and water vapour for the outlet-state solution (CoolProp properties of air and of IAPWS-95 water); the heater duty uses the same ideal-mixture enthalpies, which the test suite checks against the real-gas humid-air routine of CoolProp (HAPropsSI).
  3. Only the fat-free solids sorb water: the desorption isotherm of nonfat dry milk is applied on a fat-free dry basis. The isotherm was fitted between 23 and 60 °C; its linear temperature term is extrapolated to the outlet temperature (about 60 to 100 °C) and used down to the final moisture.
  4. Above a water activity of 0.95 (outside the fitted range) the isotherm is extrapolated; the contribution of that region to the separation work is negligible because -ln(a_w) < 0.05 there.
  5. Droplet drying: Nu = 2 (droplet moving with the gas), film temperature between inlet air and wet bulb, additive volumes of water and solids; falling-rate period by Fickian diffusion in a sphere of the critical diameter with surface moisture in equilibrium with the operating outlet air (first term of the series).
  6. Powder moisture under control is normally distributed; powder above the legal maximum is counted as a quality loss (it is reworked or downgraded).
  7. The example-plant energy (two-stage system producing skimmed-milk and sweet-whey powder; its boundary may include the concentration step) is used as the typical energy level, not as a like-for-like documented dryer value; no documented best-available-technique value exists for the dryer, and the example value minus the generic 20 % fluid-bed saving is reported without a ceiling label because the plant is already two-stage. The wet cleaning of the spray chamber is a spray-device cleaning outside the validity of the cleaning model, so its interval and duration are declared.
  8. Performance factor 1: the dryer runs at its design evaporative capacity; the evaporator upstream is a separate unit operation (its cleaning stops are in the evaporation model).

11Range of validity

  • Co-current spray dryers for milk powders with indirect air heating; inlet air 180–220 °C, outlet 85–100 °C.
  • Whole milk powder; for skimmed milk powder set the fat share to the product value.

12Sensitivity

Elasticity: relative change of the output for a relative change of each parameter. Spearman rank correlation from the Monte Carlo draws is shown alongside.
  • tcyct_{cyc}Interval between two wet cleanings of the spray dryer
    elasticity +0.057Spearman ρ +0.83
  • tCIPt_{CIP}Duration of the wet cleaning of the dryer
    elasticity -0.050Spearman ρ -0.52
  • zzMoisture set-point margin below the legal maximum, in standard deviations of the powder moisture
    elasticity +0.014Spearman ρ +0.08
  • tsut_{su}Start-up heating of the chamber and stabilisation on water before product after each cleaning
    elasticity -0.006Spearman ρ -0.09
  • Output rises with the parameter
  • Output falls with the parameter

13Validation checks

A model is published only when every check passes.
  • Dimensional analysis28 outputs converted to their declared unitsPassed
  • Every cited value verified against its source17 verified, 0 failedPassed
  • No documented value beats the theoretical limitPassed
  • Ceilings in physical orderPassed
  • Fractions within 0 and 1Passed

14Causes of loss in this operation

Probability that each cause is documented as a relevant loss of a line, pooled over all segments.

15References

  1. Codex Standard for Milk Powders and Cream Powder, CODEX STAN 207-1999 (adopted 1999; amendments 2010, 2013, 2014) (opens in a new tab). Codex Alimentarius Commission, FAO/WHO Food Standards Programme, 2014. p. 2. CXS 207-1999
  2. Best Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries (opens in a new tab). European Commission, Joint Research Centre; Publications Office of the European Union, 2019. p. 411. EUR 29978 EN; JRC118627; doi:10.2760/243911
  3. Bandwidth Study on Energy Use and Potential Energy Savings Opportunities in U.S. Food and Beverage Manufacturing (opens in a new tab). U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, 2017. p. 57. DOE/EE-1571; OSTI ID 1512366
  4. fda_pmo_2023
  5. USDA National Nutrient Database for Standard Reference, Release 28 (2015): Food Group 01, Dairy and Egg Products (nutrient report) (opens in a new tab). U.S. Department of Agriculture, Agricultural Research Service, Nutrient Data Laboratory, 2015. p. 556. USDA SR28, food group 01 report (sr28fg01.pdf)
  6. Modeling moisture sorption isotherms of milk powders at ambient and elevated temperatures using the dynamic dewpoint isotherm (DDI) method (Kadas, Oduola, Rubinelli, Atungulu, Acuff), JDS Communications 6(5):629-634, open access CC BY 4.0 (opens in a new tab). Elsevier on behalf of the American Dairy Science Association; full text served by Europe PMC (open-access XML), 2025. p. 1. doi:10.3168/jdsc.2024-0683; PMCID PMC12414362
  7. doi:10.1021/ie4033999 — CoolProp: IAPWS-95 water, air and humid-air (HAPropsSI) properties.
  8. doi:10.1063/1.1461829 — IAPWS-95 formulation for water.
  9. doi:10.1021/ie50137a021 — Rate of drying of solids: constant-rate and falling-rate periods.