foratvise
Get Started
Mixing and formulation

Mixing and blending

Batch blending of miscible liquids in a baffled stirred tank

Validated modelOEE bound 0.968 [0.855–0.983]20,000 Monte Carlo samples · seed 20261115 · computed November 16, 2026

In short

Mixing and blending can use at most 97 of every 100 planned hours. The rest goes to limits no plant can avoid.

  1. 97usable at most
  2. 3are a physical limit

The three limits that set it

Running time: up to 98 %

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 98 %

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

A beverage or dairy formulation is prepared in a 10 m3 baffled, flat-bottomed cylindrical tank with liquid height equal to the tank diameter and a single axial-flow impeller. A concentrate (syrup, flavour base) is dosed into water or milk and blended until the composition is uniform within specification; the batch is then pumped out, the tank drains by gravity and the next batch is filled. Incompatible products are separated by a cleaning-in-place.

Provenance of inputsComputed1Declared assumption17Verified quote5Reference only1

2Results

Point value from the nominal parameters and 95% interval from 20,000 Monte Carlo draws of the uncertain parameters.
  • Tank diameter (liquid height equal to diameter)2.34 m
  • Impeller speed at the installed power per volume106 [72–199] 1/min
  • Impeller Reynolds number724,267 [333,147–1,214,477]
  • Impeller tip speed4.26 [3.29–6.42] m/s
  • Blend time to 95 % homogeneity with the large impeller (D/T of the best geometry)21.7 [16.0–26.3] s
  • Blend time to 95 % homogeneity with the installed impeller24.9 [17.9–30.9] s
  • TheoreticalMinimum blend time to the specification homogeneity (large impeller)47.0 [31.9–57.3] s
  • TypicalBlend time to the specification homogeneity (installed impeller)54.0 [36.2–67.2] s
  • Batch cycle time with the large impeller (fill + blend + discharge + drainage; declared pump flows, no ceiling)45.8 [31.1–59.0] min
  • Batch cycle time with the installed impeller (declared pump flows, no ceiling)45.9 [31.2–59.1] min
  • Blended volume per hour with the large impeller (declared pump flows: a scenario value, no ceiling)13.1 [10.2–19.3] m³/h
  • TypicalBlended volume per hour with the installed impeller13.1 [10.1–19.3] m³/h
  • TheoreticalThermodynamic minimum work of blending (ideal mixing is spontaneous)0 kJ/kg
  • Blending energy to specification with the large impeller (declared geometry, not a documented BAT)0.00905 [0.00587–0.01690] kJ/kg
  • TypicalBlending energy to specification with the installed impeller0.0104 [0.0068–0.0199] kJ/kg
  • Initial coefficient of variation of the composition (segregated components)2.23 [1.75–2.94]
  • Residual coefficient of variation at the end of the blend0.00333 [0.00180–0.00655]
  • TheoreticalMaximum OEE quality factor (share of batches within the composition specification)0.984 [0.869–0.997]
  • Mean thickness of the wall film after drainage (Jeffreys)22.6 [15.1–42.1] micrometer
  • Liquid left on the wetted walls after drainage0.483 [0.324–0.901] L
  • TheoreticalMaximum yield (wall film after drainage only)0.999952 [0.999910–0.999968]
  • TypicalYield with heel and line hold-up (declared scenario)0.9970 [0.9941–0.9988]
  • TheoreticalMaximum availability (daily minimum cleaning)0.9844 [0.9791–0.9886]
  • Availability with declared changeovers (minimum cleaning between incompatible products)0.942 [0.876–0.974]
  • TypicalAvailability with typical cleanings between incompatible products0.859 [0.701–0.931]
  • TheoreticalPerformance factor against the physics-limited cycle1.00
  • TheoreticalPhysics-bounded OEE0.968 [0.855–0.983]
  • OEE with declared changeovers0.927 [0.811–0.966]

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
Cycle or residence times · lower is better
  • TheoreticalMinimum blend time to the specification homogeneity (large impeller)47.0 [31.9–57.3] s
  • TypicalBlend time to the specification homogeneity (installed impeller)54.0 [36.2–67.2] s
Throughputm³/h · higher is better
  • TypicalBlended volume per hour with the installed impeller13.1 [10.1–19.3] m³/h
Energy per unit of productkJ/kg · lower is better
  • TheoreticalThermodynamic minimum work of blending (ideal mixing is spontaneous)0 kJ/kg
  • TheoreticalDOE bandwidth study: products without significant chemical reactions have a zero thermodynamic minimum energy intensity · Benchmark0 kJ/kg
  • TypicalBlending energy to specification with the installed impeller0.0104 [0.0068–0.0199] kJ/kg
Qualityhigher is better
  • TheoreticalMaximum OEE quality factor (share of batches within the composition specification)0.984 [0.869–0.997]
Yieldhigher is better
  • TheoreticalMaximum yield (wall film after drainage only)0.999952 [0.999910–0.999968]
  • TypicalYield with heel and line hold-up (declared scenario)0.9970 [0.9941–0.9988]
Availabilityhigher is better
  • TheoreticalMaximum availability (daily minimum cleaning)0.9844 [0.9791–0.9886]
  • TypicalAvailability with typical cleanings between incompatible products0.859 [0.701–0.931]
Performancehigher is better
  • TheoreticalPerformance factor against the physics-limited cycle1.00
OEEhigher is better
  • TheoreticalPhysics-bounded OEE0.968 [0.855–0.983]

4Loss cascade

What separates each ceiling from the next one, from the physical limit to typical operation.
  1. TheoreticalTypicalCycle or residence time
    47.0 [31.9–57.3] s → 54.0 [36.2–67.2] s (gap 6.99 s)

    The time ceilings are blend times to the specification homogeneity; the batch cycle, dominated by the declared fill and discharge pump flows, carries no ceiling. At equal power per volume the blend time scales as (T/D)^(1/3) and does not depend on the impeller type, so a small impeller costs time even though it turns faster. Real blends add what the correlation excludes: density and viscosity differences between concentrate and diluent (a heavy syrup sinks and must be lifted by the flow, which can lengthen the blend several times), dosing while the impeller is uncovered, dead zones under the impeller, and the time of sampling and laboratory release.

  2. TheoreticalTypicalEnergy per unit of product
    0 kJ/kg → 0.0104 [0.0068–0.0199] kJ/kg (gap 0.0104 kJ/kg)

    Mixing ideal liquids is spontaneous: the Gibbs energy of mixing is negative, so no work is thermodynamically required. Work is needed only to make it fast: the agitator dissipates its whole power as heat by viscous friction. Blending energy is the power times the blend time; at equal power per volume a larger impeller reaches the same homogeneity sooner and therefore spends less energy. The installed impeller needs (T/D)^(1/3) more time at the same power, hence more energy. In practice the agitator often runs during the whole fill and discharge, not only for the blend time, and motor and gearbox losses are added; both multiply the energy without improving the blend.

  3. TheoreticalTypicalYield
    0.999952 [0.999910–0.999968] → 0.9970 [0.9941–0.9988] (gap 0.00300)

    Physics leaves only the liquid film that clings to the wetted walls after gravity drainage, a few tens of micrometres thick and a fraction of a litre for a 10 m3 tank. The heel left below the outlet, the liquid in the pump, valves and pipes, and the first rinse that pushes the product out are technology: their volume depends on the tank bottom, the outlet position and the line layout, and dominates the loss. Pigging or air push-out of lines and sloped or dished bottoms close this gap.

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 0.9844 [0.9791–0.9886]
    0.984 (−0.016)
  • Performance
    factor 1.00
    0.984 (−0.000)
  • Quality
    factor 0.984 [0.869–0.997]
    0.968 (−0.016)
  • Physics-bounded OEE
    0.968 [0.855–0.983]

6Full derivation

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

Scope. A beverage or dairy formulation is blended in a 10 m³ baffled, flat-bottomed cylindrical tank with liquid height equal to the tank diameter (H=TH = T) and one axial-flow impeller. A concentrate (syrup, flavour base) is dosed into water or milk and blended until the composition is uniform within specification; the batch is pumped out, the tank drains by gravity and the next batch is filled. Symbols and their values, units, uncertainty distributions and sources are listed in the parameter table generated from config/physics/mixing.yaml. Numbers quoted here are point values; the output table gives the Monte Carlo intervals.

1. Mass balance — maximum yield

Blending neither creates nor destroys mass: everything dosed into the tank leaves it as product, except what stays behind when the tank is emptied. The physics of emptying sets a lower bound on that residue: a viscous liquid draining down a vertical wall leaves a film whose thickness decreases with time but never vanishes.

Wall film after drainage (Jeffreys). For a Newtonian liquid of viscosity μ\mu and density ρ\rho draining under gravity gg from a vertical wall, the lubrication balance between weight and viscous shear gives a local film thickness, a time tt after the start of drainage and a distance xx below the top of the wetted wall,

h(x,t)=μ xρ g t.h(x,t) = \sqrt{\frac{\mu\,x}{\rho\,g\,t}} .

Averaging over the wetted height HH:

hˉ=1H∫0Hh dx=23μ Hρ g td=23 μm,\bar h = \frac{1}{H}\int_0^H h\,dx = \frac{2}{3}\sqrt{\frac{\mu\,H}{\rho\,g\,t_d}} = 23\ \mu\text{m},

for a drainage time tdt_d = 300 s and μ\mu = 1.5 mPa s. The wetted area is the wall plus the bottom, Aw=πTH+πT2/4A_w = \pi T H + \pi T^2/4 = 21 m²; the bottom is given the same mean thickness (a declared simplification). The hold-up and the theoretical yield are

Vfilm=hˉ Aw=0.48 L,Ymax=1−VfilmV=0.99995.V_{film} = \bar h\,A_w = 0.48\ \text{L}, \qquad Y_{max} = 1 - \frac{V_{film}}{V} = 0.99995 .

Technology. The heel below the outlet and the liquid in pump, valves and pipes (VhV_h, scenario 30 L) are properties of the installation, not of physics: Ytyp=1−(Vfilm+Vh)/V=0.9970Y_{typ} = 1 - (V_{film} + V_h)/V = 0.9970. The BAT reference document records that product is lost because equipment is difficult to empty before cleaning or a batch change, and reports a normal product loss of 0.5–1.5% for whole dairy plants; that boundary is wider than one tank and is not compared numerically.

2. Energy balance — minimum energy per unit of product

Thermodynamic minimum. For an ideal solution the Gibbs energy of mixing is

ΔGmix=R T∑iniln⁡xi<0,\Delta G_{mix} = R\,T \sum_i n_i \ln x_i < 0,

where nin_i and xix_i are the amount and mole fraction of each component and RR the molar gas constant. Every ln⁡xi\ln x_i is negative, so ΔGmix\Delta G_{mix} is negative for any composition: mixing is spontaneous and could, in principle, deliver work rather than consume it. The thermodynamic minimum work of blending is therefore zero (taking no credit for the recoverable work). Heats of mixing of aqueous sugar solutions are small and are neglected. This agrees with the DOE bandwidth study, which gives a zero thermodynamic minimum to products without significant chemical reactions.

Why energy is spent anyway. Molecular diffusion alone would take far longer than any production schedule: with a diffusion coefficient of order 10−910^{-9} m²/s, the diffusion time L2/DL^2/D over a metre is of the order of decades. The agitator buys speed: its shaft power PP is entirely dissipated as heat by viscous friction while it stretches and folds the fluid until diffusion can finish the job at small scales. The blending energy per kilogram is

e=P tblendρV,e = \frac{P\,t_{blend}}{\rho V},

0.0090 kJ/kg with the large impeller and 0.0104 kJ/kg with the installed impeller at 200 W/m³ (Section 3). The installed value is the typical energy ceiling and the theoretical one is zero; the large-impeller value uses a declared geometry, not a documented best available technique, so it is reported without a ceiling label.

3. Rate-limiting step — blend time and minimum batch cycle

Blend-time correlation. In the turbulent regime the time to reach 95% homogeneity, θ95\theta_{95}, of a baffled tank with H=TH = T is correlated by Grenville's equation (as presented by Grenville and Nienow, 2003, and Nienow, 1997):

N θ95=C Np−1/3(TD)2,C=5.20,N\,\theta_{95} = C\,N_p^{-1/3}\left(\frac{T}{D}\right)^2, \qquad C = 5.20,

where NN is the impeller speed, DD the impeller diameter, TT the tank diameter and NpN_p the turbulent power number of the impeller, which links power and speed through

P=Np ρ N3D5.P = N_p\,\rho\,N^3 D^5 .

Blend time at given power. Eliminating NN between the two equations:

N=(PNp ρ D5)1/3  ⇒  θ95=C T2D−1/3(ρP)1/3.N = \left(\frac{P}{N_p\,\rho\,D^5}\right)^{1/3} \;\Rightarrow\; \theta_{95} = C\,T^2 D^{-1/3}\left(\frac{\rho}{P}\right)^{1/3}.

The power number cancels: at equal power input the blend time does not depend on the impeller type, only on the geometry and the power. Because T3∝VT^3 \propto V, this is equivalent to θ95∝(P/V)−1/3 (T/D)1/3 T2/3\theta_{95} \propto (P/V)^{-1/3}\,(T/D)^{1/3}\,T^{2/3}: four times the power shortens the blend by 41/3=1.594^{1/3} = 1.59; a larger impeller at the same power blends faster. With VV = 10 m³ (TT = 2.34 m) and P/VP/V = 200 W/m³, θ95\theta_{95} = 24.9 s with D/TD/T = 0.33 and 21.7 s with D/TD/T = 0.50. The impeller speed (106 min⁻¹), Reynolds number Re=ρND2/μ=7×105Re = \rho N D^2/\mu = 7\times10^5 and tip speed (4.3 m/s) are reported to check that the flow is fully turbulent, as the correlation requires.

From 95% to the specification. Treat the decay of non-uniformity as first order: the normalised deviation from the final composition decays as e−kte^{-kt}. Since it falls from 1 to 0.05 at θ95\theta_{95}, k=ln⁡20/θ95k = \ln 20/\theta_{95}. The coefficient of variation (CoV) of composition across the tank starts, for a concentrate of volume fraction φ\varphi still segregated from the diluent, at

CoV0=1−φφ=2.23(φ=1/6),CoV_0 = \sqrt{\frac{1-\varphi}{\varphi}} = 2.23 \quad (\varphi = 1/6),

and CoV(t)=CoV0 e−ktCoV(t) = CoV_0\,e^{-kt}. The blend is stopped when the residual CoV is the composition tolerance δ\delta divided by a factor zhz_h (scenario 3), so that non-uniformity contributes little to the batch error:

tblend=θ95 ln⁡(CoV0 zh/δ)ln⁡20=2.17 θ95=47 s (large impeller), 54 s (installed).t_{blend} = \theta_{95}\,\frac{\ln(CoV_0\,z_h/\delta)}{\ln 20} = 2.17\,\theta_{95} = 47\ \text{s (large impeller)},\ 54\ \text{s (installed)}.

Minimum batch cycle. Filling and discharge are limited by the pumps (V˙f\dot V_f, V˙d\dot V_d) and the tank must drain for tdt_d before the next batch:

tcycle=VV˙f+tblend+VV˙d+td=45.8 min,V˙max=Vtcycle=13.1 m3/h.t_{cycle} = \frac{V}{\dot V_f} + t_{blend} + \frac{V}{\dot V_d} + t_d = 45.8\ \text{min}, \qquad \dot V_{max} = \frac{V}{t_{cycle}} = 13.1\ \text{m}^3/\text{h}.

With 20-minute fills and discharges the blend is a small share of the cycle; the pumps, not the agitator, are the main rate lever. Because the cycle and the rate are set by the declared pump flows, they are scenario values passed through a formula and carry no ceiling label; the physics ceiling of the time dimension is the blend time to specification. Under the common batch convention, fill, discharge and drainage belong to the ideal cycle, so they act through the performance factor (here 1, because the capable process needs no longer cycle) and not through availability. Drainage time trades cycle time against yield: the film thickness falls only as td−1/2t_d^{-1/2}.

4. Process capability — quality ceiling

The composition of a finished batch deviates from target for two independent reasons: the residual non-uniformity at the end of the blend (CoV=δ/zhCoV = \delta/z_h) and the error of the dosed quantities (relative standard deviation σd\sigma_d, load cells or flow meters), which no amount of mixing can remove. With both normally distributed, the relative composition error has standard deviation

σtot=σd2+(δ/zh)2,\sigma_{tot} = \sqrt{\sigma_d^2 + (\delta/z_h)^2},

and the share of batches within the tolerance ±δ\pm\delta is

Qmax=2 Φ ⁣(δσtot)−1=0.984,Q_{max} = 2\,\Phi\!\left(\frac{\delta}{\sigma_{tot}}\right) - 1 = 0.984,

with Φ\Phi the standard normal distribution function. A batch outside tolerance is corrected rather than destroyed, but it is not right first time and counts as a quality loss. Blending longer cannot raise QmaxQ_{max} above 2Φ(δ/σd)−12\Phi(\delta/\sigma_d) - 1: beyond a few θ95\theta_{95} the ceiling is set by dosing accuracy.

5. Unavoidable stops — maximum availability

Milk equipment must be cleaned at least once each day used. The minimum duration of the cleaning, tCIPt_{CIP}, is taken through a parameter link from the cleaning model (output cip_time_min: the physical minimum of the kinetic and displacement phases plus declared hold times). That model is valid for recirculated pipe circuits; the spray cleaning of the tank interior is outside its validity, so the linked duration is a lower bound:

Amax=1−tCIPtcyc=0.984.A_{max} = 1 - \frac{t_{CIP}}{t_{cyc}} = 0.984 .

Residues of one product (allergens, colours, flavours) may not be carried into an incompatible one, so the tank is also cleaned after each campaign of nbn_b compatible batches. That is a decision of the product mix, not physics: with the minimum duration it gives Ascen=nbtcycle/(nbtcycle+tCIP)A_{scen} = n_b t_{cycle}/(n_b t_{cycle} + t_{CIP}) = 0.942, and with the declared typical duration (1 h) the typical availability AtypA_{typ} = 0.859.

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.OEE_{max} = A_{max} \times P_{max} \times Q_{max}, \qquad P_{max} = 1 .

Pmax=1P_{max} = 1 refers to the physics-limited cycle of Section 3. The result is OEEmaxOEE_{max} = 0.968, set mainly by dosing accuracy; with the declared cleanings between incompatible products, oee_scenario = 0.927.

7. Assumptions and validity

See the assumption and validity lists generated from the configuration file. The correlation applies to liquids of similar density and viscosity; a heavy syrup that sinks to the bottom needs a buoyancy correction not included here, and its real blend time can be several times longer. The most influential inputs (sensitivity table) are the cleaning duration and the campaign length for OEE, the dosing accuracy and the tolerance for quality, and the pump flows for the cycle time.

References

  • Grenville, R. K., Nienow, A. W. (2003). Blending of miscible liquids. In Handbook of Industrial Mixing, Wiley, 507–542. doi:10.1002/0471451452.ch9
  • Nienow, A. W. (1997). On impeller circulation and mixing effectiveness in the turbulent flow regime. Chemical Engineering Science, 52, 2557–2565. doi:10.1016/S0009-2509(97)00072-9
  • Jeffreys, H. (1930). The draining of a vertical plate. Mathematical Proceedings of the Cambridge Philosophical Society, 26, 204–205. doi:10.1017/s0305004100015437
  • 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 Saving Opportunities in U.S. Food and Beverage Manufacturing, 2017.
  • 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.
  • VV10 m³

    Batch volume

    Scenario · Fixed

    Declared assumption
  • ρ\rho1,040 kg/m³

    Density of the blended product

    Scenario · Uniform 1,000 to 1,060 kg/m³

    Declared assumption
  • μ\mu0.0015 Pa·s

    Dynamic viscosity of the blended product (water at 20 °C from IAPWS through CoolProp, 1.0 mPa s, up to three times that value)

    Scenario · Uniform 0.001 to 0.003 Pa·s

    Declared assumption

    CoolProp PropsSI('V','T',293.15,'P',101325,'Water') = 1.0e-3 Pa s; the multiplier for dissolved solids is a declared assumption.

  • P/VP/V200 W/m³

    Installed agitator power per unit volume

    Scenario · Uniform 100 to 500 W/m³

    Declared assumption

    Design choice; declared assumption.

  • NpN_p1.3

    Turbulent power number of the installed impeller (four-blade 45° pitched-blade turbine)

    Scenario · Uniform 1.1 to 1.7

    Declared assumption

    No open source with a quotable power number was used; declared assumption. At fixed power per volume the blend time does not depend on it; it only sets speed, Reynolds number and tip speed.

  • CC5.2

    Constant of the turbulent blend-time correlation N theta95 = C Np^(-1/3) (T/D)^2 for H = T

    Property · Fixed

    Reference only

    Literature constant (closed-access source, cited as reference only): Grenville correlation as presented by Grenville and Nienow (2003), doi:10.1002/0471451452.ch9, and Nienow (1997), doi:10.1016/S0009-2509(97)00072-9; derivation in physics_docs/mixing.en.md, section 3.

  • (D/T)inst(D/T)_{inst}0.33

    Impeller-to-tank diameter ratio installed

    Scenario · Uniform 0.25 to 0.4

    Declared assumption
  • (D/T)L(D/T)_{L}0.5

    Largest impeller-to-tank diameter ratio considered for the best geometry

    Scenario · Fixed

    Declared assumption

    Declared design limit of the comparison; larger impellers reduce blend time and energy at equal power per volume.

  • φ\varphi0.167

    Volume fraction of the dosed concentrate in the batch

    Scenario · Uniform 0.1 to 0.25

    Declared assumption
  • δ\delta0.01

    Composition tolerance of the specification, relative to target (e.g. dissolved solids)

    Scenario · Uniform 0.005 to 0.02

    Declared assumption
  • zhz_h3

    Ratio between tolerance and residual non-uniformity at which the blend is stopped

    Scenario · Fixed

    Declared assumption

    Operating choice.

  • σd\sigma_d0.0025

    Relative standard deviation of the dosed quantities (load cells or flow meters)

    Scenario · Uniform 0.001 to 0.004

    Declared assumption
  • V˙f\dot V_f30 m³/h

    Flow rate of the filling pumps

    Scenario · Uniform 20 to 50 m³/h

    Declared assumption
  • V˙d\dot V_d30 m³/h

    Flow rate of the discharge pump

    Scenario · Uniform 20 to 50 m³/h

    Declared assumption
  • tdt_d300 s

    Gravity drainage time after the pump loses suction

    Scenario · Uniform 120 to 600 s

    Declared assumption
  • VhV_h30 L

    Heel and line hold-up left in tank bottom, outlet, pump and pipes

    Scenario · Uniform 10 to 60 L

    Declared assumption
  • nbn_b8

    Batches of compatible products between two cleanings

    Scenario · Uniform 4 to 16

    Declared assumption
  • tCIP,typt_{CIP,typ}1 h

    Duration of a typical cleaning-in-place of the tank between incompatible products

    Scenario · Uniform 0.75 to 1.5 h

    Declared assumption

    Declared typical programme, used for the typical availability.

  • tCIPt_{CIP}0.374294 h

    Minimum duration of the cleaning-in-place: physical minimum of the kinetic and displacement phases plus declared hold times (output cip_time_min of the cleaning model)

    Property · Triangular 0.242184, 0.374294, 0.536161 h

    Computed

    Output `cip_time_min` of the cleaning model: 22.46 min [14.53, 32.17].

  • tcyct_{cyc}24 h

    Maximum interval between two cleanings of milk equipment (dairy formulations)

    Regulatory · Fixed

    Verified quote
    Show verbatim quote
    all equipment is thoroughly cleaned at least once each day usedGrade "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. 98. PMO 2023 Revision

8Benchmarks

Documented values from open sources used to check the ceilings; no documented value may beat the theoretical limit.
BenchmarkCeilingValueProvenance
DOE bandwidth study: products without significant chemical reactions have a zero thermodynamic minimum energy intensity
Show verbatim quote
Most food and beverage products have a zero TM energy intensity because there are no significant chemical reactions involved.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. 41. DOE/EE-1571; OSTI ID 1512366
Theoretical0 kJ/kgVerified quote
Normal product loss in milk manufacture, percent of product (whole dairy, cleaning, start-up and changeover run-off included; wider boundary than one tank)
Show verbatim quote
Product loss during milk manufacture can be as high as 3–4 %, with normally 0.5–1.5 % of the product being wasted.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. 395. EUR 29978 EN; JRC118627; doi:10.2760/243911
Typical0.5 %Verified quote

9Unavoidable stops

Stops that physics or regulation impose; they set the maximum availability.
  • Cleaning at least once per day of use (milk equipment)Verified quote

    Hygiene regulation for milk equipment; minimum duration from the cleaning model.

    Show verbatim quote
    all equipment is thoroughly cleaned at least once each day usedGrade "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. 98. PMO 2023 Revision
  • Cleaning between incompatible products (scenario stop: availability_scenario and typical availability only)Verified quote

    Residues of one product (allergens, colours, flavours) cannot be carried into an incompatible one; the BAT reference document records the product lost because equipment is difficult to empty before cleaning or a batch change. Interval and duration are declared scenario values.

    Show verbatim quote
    Product loss to waste water caused by difficulties of emptying the equipment before cleaning/shift of production batch.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. 536. EUR 29978 EN; JRC118627; doi:10.2760/243911

10Assumptions

  1. Availability: the only physically or legally required stop is the daily cleaning of milk equipment, at the minimum duration of the cleaning model. That model is valid for recirculated pipe circuits; the spray cleaning of the tank interior is outside its validity, so the linked duration is a lower bound. Cleanings between incompatible products are a product-mix decision: they enter availability_scenario (at the minimum duration) and the typical availability (at the declared typical duration). Fill, discharge and drainage belong to the ideal batch cycle (common batch convention), so they do not reduce availability.
  2. Fully turbulent flow in a baffled vessel with H = T and a single impeller; the blend-time correlation applies to liquids of similar density and viscosity.
  3. Non-uniformity decays exponentially (first order) with a rate constant fixed by the 95 % blend time: k = ln 20 / theta95; the coefficient of variation follows the same decay from its segregated initial value.
  4. The blend is stopped when the residual coefficient of variation is the tolerance divided by z_h; dosing errors are independent and normally distributed; a batch outside tolerance is counted as a quality loss (it is corrected, not destroyed).
  5. Wall film after drainage follows Jeffreys' solution for a Newtonian liquid on a vertical plate; the bottom is assigned the same mean film thickness.
  6. Mixing is treated as ideal (no heat of mixing, no reaction); the thermodynamic minimum work is zero.

11Range of validity

  • Low-viscosity miscible liquids (up to a few mPa s) in baffled tanks with one axial or radial impeller, impeller Reynolds number above 10 000.
  • Concentrates whose density differs strongly from the diluent need a correction for buoyancy that is not included.

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.
  • zhz_hRatio between tolerance and residual non-uniformity at which the blend is stopped
    elasticity +0.071Spearman ρ —
  • δ\deltaComposition tolerance of the specification, relative to target (e.g. dissolved solids)
    elasticity +0.040Spearman ρ +0.66
  • σd\sigma_dRelative standard deviation of the dosed quantities (load cells or flow meters)
    elasticity -0.039Spearman ρ -0.63
  • tcyct_{cyc}Maximum interval between two cleanings of milk equipment (dairy formulations)
    elasticity +0.016Spearman ρ —
  • tCIPt_{CIP}Minimum duration of the cleaning-in-place: physical minimum of the kinetic and displacement phases plus declared hold times (output cip_time_min of the cleaning model)
    elasticity -0.016Spearman ρ -0.27
  • 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 source5 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. 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. 395. EUR 29978 EN; JRC118627; doi:10.2760/243911
  2. 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. 41. DOE/EE-1571; OSTI ID 1512366
  3. doi:10.1002/0471451452.ch9 — Blending of miscible liquids (turbulent blend-time correlation).
  4. doi:10.1016/S0009-2509(97)00072-9 — Impeller circulation and mixing effectiveness in turbulent flow.
  5. doi:10.1017/s0305004100015437 — Drainage of a liquid film on a vertical plate.
  6. doi:10.1021/ie4033999 — Thermophysical property library (CoolProp).