foratvise
Get Started
Filling and packaging line

Filling

Rotary isobaric (counter-pressure) gravity filling of beverages into bottles

Validated modelOEE bound 0.995 [0.986–0.998]20,000 Monte Carlo samples · seed 20261115 · computed November 16, 2026

In short

Filling can use at most 99 of every 100 planned hours. The rest goes to limits no plant can avoid.

  1. 99usable at most
  2. 1are a physical limit

The three limits that set it

Running time: up to 99 %

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

A carbonated soft drink is filled into 0.5 L PET bottles on a rotary filler with N valves. At each valve the bottle is pressed against the valve, pressurised with CO2 from the bowl head-space, filled by gravity while the displaced gas returns to the bowl, and decompressed to atmosphere through a snift orifice before it leaves the carousel. The fill volume is set by the level of the return-gas tube (level filling). With zero carbonation the same model describes a still liquid filled by gravity at atmospheric pressure.

Provenance of inputsComputed4Declared assumption27Verified quote10Reference only3

2Results

Point value from the nominal parameters and 95% interval from 20,000 Monte Carlo draws of the uncertain parameters.
  • Tolerable negative error (tolerable deficiency) T for the nominal quantity15.0 mL
  • Minimum target mean fill that satisfies the prepackage rules502.6 [501.4–505.1] mL
  • Give-away per bottle (target mean minus nominal quantity)2.58 [1.38–5.05] mL
  • Give-away as a share of the nominal quantity0.00515 [0.00277–0.01010]
  • Capability index of the fill volume against the lower limit Qn - T2.93 [1.49–5.65]
  • Largest fill standard deviation for which the average requirement alone binds (T / z for the T1 share)7.65 mL
  • Share of bottles with a T1 error at the target mean7.61 × 10⁻¹⁹ [1.17 × 10⁻⁶⁴–4.03 × 10⁻⁶]
  • Share of bottles rejected for a T2 error (below Qn - 2T)6.00 × 10⁻⁶⁰ [2.77 × 10⁻²⁰⁸–2.93 × 10⁻¹⁷]
  • TheoreticalMaximum yield: nominal volume sold per volume filled (give-away and T2 rejects)0.9949 [0.9900–0.9972]
  • Best documentedYield including the documented product loss per changeover of a traditional filling operation0.9948 [0.9899–0.9971]
  • TheoreticalMaximum OEE quality factor (bottles without a T2 error)1.00
  • Equilibrium CO2 pressure of the product at the filling temperature (Henry's law)2.46 [1.70–3.26] bar
  • Counter-pressure (absolute) used for filling3.46 [2.54–4.83] bar
  • Minimum work to lift the liquid to the filler bowl0.0196 [0.0105–0.0385] kJ/kg
  • Minimum (isothermal) work to compress the counter-pressure gas vented at the snift0.0223 [0.0104–0.0447] kJ/kg
  • TheoreticalThermodynamic minimum work of filling (lift + vented counter-pressure gas)0.0419 [0.0269–0.0746] kJ/kg
  • Free liquid area of the neck (bore minus gas-return tube)342 [292–375] mm²
  • Minimum pressurisation time of the bottle (ideal nozzle, adiabatic charging)0.0817 [0.0465–0.1781] s
  • Minimum liquid fill time (Torricelli discharge, discharge coefficient 1)0.525 [0.417–0.699] s
  • Minimum snift (head-space blowdown) time through the snift orifice0.267 [0.093–0.784] s
  • TheoreticalMinimum valve process time (pressurise + fill + snift)0.874 [0.651–1.438] s
  • TypicalValve process time with throttled liquid flow, real nozzles, controlled snift and valve actuation (declared scenario)3.10 [2.28–5.03] s
  • Maximum bottles per hour per valve (physics minimum times)3,089 [1,837–4,287] 1/h
  • TheoreticalMaximum bottles per hour of the rotary filler (physics minimum times)222,435 [132,241–308,691] 1/h
  • TypicalBottles per hour of the rotary filler with the declared typical valve times62,649 [37,444–86,980] 1/h
  • Availability with declared changeovers (minimum CIP plus product changeovers)0.974 [0.922–0.991]
  • TypicalAvailability with a typical CIP programme (no changeovers)0.972 [0.916–0.989]
  • TheoreticalMaximum availability (minimum filler CIP)0.995 [0.986–0.998]
  • TheoreticalPerformance factor against the physics-limited rate1.00
  • TheoreticalPhysics-bounded OEE0.995 [0.986–0.998]
  • OEE with declared changeovers0.974 [0.922–0.991]

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
Yieldhigher is better
  • TheoreticalMaximum yield: nominal volume sold per volume filled (give-away and T2 rejects)0.9949 [0.9900–0.9972]
  • Best documentedYield including the documented product loss per changeover of a traditional filling operation0.9948 [0.9899–0.9971]
Qualityhigher is better
  • TheoreticalMaximum OEE quality factor (bottles without a T2 error)1.00
Energy per unit of productkJ/kg · lower is better
  • TheoreticalThermodynamic minimum work of filling (lift + vented counter-pressure gas)0.0419 [0.0269–0.0746] kJ/kg
  • Best available techniqueDOE bandwidth study, state-of-the-art on-site energy intensity of fluid milk packaging (whole packaging step: filling, container handling and closing; wider boundary than the filler) · Benchmark27.9 kJ/kg
  • TypicalDOE bandwidth study, current typical on-site energy intensity of fluid milk packaging (whole packaging step) · Benchmark34.9 kJ/kg
Cycle or residence times · lower is better
  • TheoreticalMinimum valve process time (pressurise + fill + snift)0.874 [0.651–1.438] s
  • TypicalValve process time with throttled liquid flow, real nozzles, controlled snift and valve actuation (declared scenario)3.10 [2.28–5.03] s
Throughput1/h · higher is better
  • TheoreticalMaximum bottles per hour of the rotary filler (physics minimum times)222,435 [132,241–308,691] 1/h
  • TypicalBottles per hour of the rotary filler with the declared typical valve times62,649 [37,444–86,980] 1/h
Availabilityhigher is better
  • TheoreticalMaximum availability (minimum filler CIP)0.995 [0.986–0.998]
  • TypicalAvailability with a typical CIP programme (no changeovers)0.972 [0.916–0.989]
Performancehigher is better
  • TheoreticalPerformance factor against the physics-limited rate1.00
OEEhigher is better
  • TheoreticalPhysics-bounded OEE0.995 [0.986–0.998]

4Loss cascade

What separates each ceiling from the next one, from the physical limit to typical operation.
  1. TheoreticalTypicalCycle or residence time
    0.874 [0.651–1.438] s → 3.10 [2.28–5.03] s (gap 2.23 s)

    The theoretical valve time adds three ideal steps: adiabatic charging of the bottle through a loss-free nozzle, frictionless Torricelli discharge of the liquid through the full free annulus of the neck, and an unrestricted blowdown of the head-space. In operation the liquid flow is throttled on purpose: a carbonated product released at high velocity entrains gas, nucleates bubbles on the wall and foams, so the effective discharge coefficient is a fraction of one. The snift is not a free blowdown either: when the head-space pressure drops below the equilibrium CO2 pressure the product is supersaturated and degasses violently, so decompression is stepped and lasts about a second instead of a few hundredths. Valve actuation, the calming time before the snift and real gas-path losses add the rest.

  2. TheoreticalTypicalThroughput
    222,435 [132,241–308,691] 1/h → 62,649 [37,444–86,980] 1/h (gap 159,786 1/h)

    The machine rate is the number of valves times the share of the revolution available for the process divided by the valve time. The gap between the theoretical and the typical rate is therefore the inverse of the time gap above (foaming-limited liquid flow, controlled snift, actuation), at equal number of valves and equal angular share.

  3. TheoreticalBest documentedYield
    0.9949 [0.9900–0.9972] → 0.9948 [0.9899–0.9971] (gap 1.02 × 10⁻⁴)

    The theoretical yield already contains the give-away imposed by the prepackage rules for the declared fill dispersion and the drift of the lot mean, plus the bottles with a T2 error. The documented yield adds the product pushed out and rinsed away at each changeover in a traditional filling operation (75–100 L per changeover in the BAT reference document); techniques such as product recovery pigs, shorter product paths and late product diversification close this gap.

  4. TheoreticalBest available techniqueEnergy per unit of product

    The thermodynamic minimum work of filling is tiny: lifting the liquid a couple of metres and compressing the few tens of millilitres of CO2 that are vented at the snift cost a few hundredths of a kilojoule per kilogram. The state-of-the-art and typical energy intensities of a packaging step are three orders of magnitude larger because they include drives of the carousel, conveyors, closing and labelling, compressed air, container rinsing and losses of motors and pneumatics; the energy of filling is set by machinery, not by the physics of the liquid.

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 0.95.
  • Loading time (ideal)
    1.000
  • Availability
    factor 0.995 [0.986–0.998]
    0.995 (−0.005)
  • Performance
    factor 1.00
    0.995 (−0.000)
  • Quality
    factor 1.00
    0.995 (−0.000)
  • Physics-bounded OEE
    0.995 [0.986–0.998]

6Full derivation

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

Scope. A carbonated soft drink filled into 0.5 L PET bottles on a rotary level filler with NN valves, by isobaric (counter-pressure) gravity filling. With zero carbonation the same equations describe a still liquid filled by gravity at atmospheric pressure. Symbols and their values, units, uncertainty distributions and sources are listed in the parameter table generated from config/physics/filling.yaml. Numbers quoted in this text are point values; the output table gives the Monte Carlo intervals.

1. Mass balance — maximum yield

Filling neither creates nor destroys product: every millilitre that leaves the bowl ends up in a bottle. The losses of a filler are therefore not physical but legal and statistical. A prepackage must carry at least its nominal quantity on average, and only a small share of packages may fall short by more than a tolerance. Because the individual fill volume scatters, the filler must aim above the nominal quantity, and the product filled above the nominal is given away: it is sold without being paid for. Bottles that fall short beyond twice the tolerance cannot be sold at all and are rejected.

The legal rules. For a nominal quantity QnQ_n the EU Directive 76/211/EEC (Annex I, point 2.4) and OIML R 87 (Table 1) fix the tolerable negative error (OIML: tolerable deficiency) TT; for QnQ_n from 500 to 1 000 mL it is T=15T = 15 mL. Three requirements follow:

  1. Average requirement — the mean actual content of the lot is at least QnQ_n (OIML R 87, 3.2; Directive, Annex I, 1.1).
  2. T1 errors — at most pT1=2.5%p_{T1} = 2.5\% of the packages may hold less than Qn−TQ_n - T (OIML R 87, 3.3.2). The Directive states the same rule as a proportion "sufficiently small" to pass its reference sampling test.
  3. T2 errors — no package may hold less than Qn−kT2TQ_n - k_{T2}T, with kT2=2k_{T2} = 2 (Directive, Annex I, 1.3; OIML R 87, 3.3.3).

NIST Handbook 133 (US) expresses the individual limit differently, as a "Maximum Allowable Variation" (MAV) beyond which a deficiency is an unreasonable minus error; it is cited for context and not modelled.

Fill-volume distribution. In a level filler the volume in each bottle is the brim-full capacity of the bottle minus the head-space left above the end of the return-gas tube. Both vary from bottle to bottle, so the individual fill volume is modelled as V∼N(μ,σ2)V \sim \mathcal N(\mu, \sigma^2), where μ\mu is the target mean and σ\sigma the standard deviation of individual fills (a declared scenario value, 2 mL, range 1–4 mL). Between control checks the mean itself drifts (temperature of the product, change of bottle lot, wear of valves); this drift is described by the standard deviation of the lot mean σμ\sigma_\mu.

Minimum target mean. Each rule sets a lower bound on μ\mu:

μavg=Qn+Δavg+zμ σμ,μT1=Qn−T+zT1 σ,zT1=Φ−1(1−pT1)=1.960,\mu_{avg} = Q_n + \Delta_{avg} + z_\mu\,\sigma_\mu, \qquad \mu_{T1} = Q_n - T + z_{T1}\,\sigma, \qquad z_{T1} = \Phi^{-1}(1 - p_{T1}) = 1.960, μ=max⁡(μavg, μT1).\mu = \max(\mu_{avg},\ \mu_{T1}).

Here Δavg=0\Delta_{avg} = 0 is the offset of the average requirement (the mean may not be below the nominal), zμ=2.576z_\mu = 2.576 is the one-sided 99.5% safety factor the operator applies so that each lot, not just the long-run average, meets the average requirement, and Φ\Phi is the standard normal distribution function. μT1\mu_{T1} is the mean at which exactly 2.5% of the bottles would carry a T1 error. The T1 rule binds only when σ>(T−zμσμ)/zT1\sigma > (T - z_\mu\sigma_\mu)/z_{T1}; the output sigma_legal_max =T/zT1=7.65= T/z_{T1} = 7.65 mL shows the dispersion beyond which the individual rule, not the average rule, sets the mean. With the scenario values the average rule binds: μ=502.6\mu = 502.6 mL.

Give-away, rejects and yield.

g=μ−Qn,r=Φ ⁣(Qn−kT2T−μσ),Ymax=(1−r) Qnμ.g = \mu - Q_n, \qquad r = \Phi\!\left(\frac{Q_n - k_{T2}T - \mu}{\sigma}\right), \qquad Y_{max} = (1 - r)\,\frac{Q_n}{\mu}.

gg is the give-away per bottle (2.6 mL, 0.52% of the nominal), rr the share of bottles with a T2 error, which are rejected at the filler and whose content is lost, and YmaxY_{max} the ratio of the nominal volume sold to the volume filled. YmaxY_{max} is the theoretical yield ceiling for the declared capability: it falls only when the dispersion grows (more give-away, more rejects).

Capability index. The fill is compared with the lower legal limit LSL=Qn−TLSL = Q_n - T through

Cpk=μ−(Qn−T)3σ=2.93.C_{pk} = \frac{\mu - (Q_n - T)}{3\sigma} = 2.93 .

A capable filler (CpkC_{pk} well above 1) has negligible rejects; its yield loss is entirely the give-away imposed by the average rule and the drift of the mean.

Documented ceiling. At each product changeover the previous product is pushed out of bowl, valves and product line and the system is rinsed. The BAT reference document reports that "Normally, 75–100 litres" are lost per changeover in a traditional filling operation (liquid dairy). With VcoV_{co} litres lost per changeover and VrunV_{run} litres filled between two changeovers,

Ydoc=Ymax−VcoVrun.Y_{doc} = Y_{max} - \frac{V_{co}}{V_{run}} .

2. Energy balance — minimum work per unit of product

A filler changes neither the temperature nor the composition of the product; the only energy it must supply is mechanical.

Lifting. The liquid is raised from the buffer tank to the bowl, a height Δz\Delta z. The reversible work per kilogram is wlift=g Δzw_{lift} = g\,\Delta z (gg = 9.81 m/s²): 0.020 kJ/kg for 2 m. The fall from the bowl into the bottle is driven by this potential energy and needs no further work.

Counter-pressure gas. In isobaric filling the bottle is first pressurised with CO2 from the bowl head-space to the filling pressure pfp_f; while the liquid flows in, the displaced gas returns to the bowl through the vent tube, so it is not consumed. What is consumed is the gas left in the head-space Vhs=Vb−μV_{hs} = V_b - \mu (brim-full capacity minus fill), which is vented to atmosphere at the snift. Replacing it requires compressing that amount of gas from atmospheric pressure p0p_0 to pfp_f; the least work is the reversible isothermal compression

Wgas=pf Vhs ln⁡pfp0,wgas=Wgasρ μ,W_{gas} = p_f\,V_{hs}\,\ln\frac{p_f}{p_0}, \qquad w_{gas} = \frac{W_{gas}}{\rho\,\mu},

where ρμ\rho\mu is the mass of product in the bottle: 0.022 kJ/kg at pfp_f = 3.46 bar.

Thermodynamic minimum.

wmin=g Δz+wgas=0.042 kJ/kg.w_{min} = g\,\Delta z + w_{gas} = 0.042\ \text{kJ/kg}.

This is tiny — and it is meant to be shown as tiny. The dissolution of CO2 (carbonation) happens upstream in the carbonator and is outside the boundary; so is the refrigeration that keeps the product cold. The DOE bandwidth study sets the thermodynamic minimum of bottling to zero because it counts only chemical conversions; its boundary ignores mechanical work and is not comparable with wminw_{min}. The documented energy intensities of a whole packaging step (DOE, fluid milk: 12 Btu/lb state of the art, 15 Btu/lb current typical, that is 28 and 35 kJ/kg) are three orders of magnitude above wminw_{min}: they include drives, conveyors, closing, compressed air and motor losses, which belong to machinery, not to the physics of filling.

3. Rate-limiting step — minimum valve time and maximum throughput

Each valve performs three physical steps whose durations can be bounded: pressurisation of the bottle, liquid flow, and decompression (snift).

Minimum counter-pressure (Henry's law). The product must not release CO2 while it flows, so the filling pressure must exceed the equilibrium pressure of the dissolved gas. Carbonation is expressed in "volumes": vCO2v_{CO_2} volumes of gas at 0 °C and 1 atm per volume of liquid, i.e. a concentration c=vCO2 pSTP/(R TSTP)c = v_{CO_2}\,p_{STP}/(R\,T_{STP}) mol/m³ (RR = 8.314 J/(mol K), pSTPp_{STP} = 101 325 Pa, TSTPT_{STP} = 273.15 K), or c/ρc/\rho mol/kg. The solubility of CO2 in water is given by the Weiss correlation

ln⁡K0=A1+A2 100T+A3ln⁡T100,\ln K_0 = A_1 + A_2\,\frac{100}{T} + A_3 \ln\frac{T}{100},

with K0K_0 in mol/(kg atm), TT in kelvin and A1=−58.0931A_1 = -58.0931, A2=90.5069A_2 = 90.5069, A3=22.2940A_3 = 22.2940 (Weiss, 1974; K0K_0 = 0.034 mol/(kg atm) at 25 °C). Henry's law gives the equilibrium pressure peq=(c/ρ)/K0p_{eq} = (c/\rho)/K_0: 2.46 bar for 3.5 volumes at 6 °C. The filler runs at pf=peq+Δpmp_f = p_{eq} + \Delta p_m, with a margin Δpm\Delta p_m (scenario, 1 bar). For a still liquid pf=p0p_f = p_0 and the gas steps vanish.

Pressurisation (compressible flow). Gas flows from the bowl at pfp_f through the gas path of area AgA_g (the vent tube, diameter dvd_v) into the bottle of volume VbV_b. For an isentropic orifice the mass flow is

m˙=Cd A pupRsT0 ψ(r),ψ(r)={κ(2κ+1)κ+12(κ−1)r≤r∗2κκ−1(r2/κ−r(κ+1)/κ)r>r∗\dot m = C_d\,A\,\frac{p_{up}}{\sqrt{R_s T_0}}\,\psi(r), \qquad \psi(r) = \begin{cases}\sqrt{\kappa}\left(\frac{2}{\kappa+1}\right)^{\frac{\kappa+1}{2(\kappa-1)}} & r \le r^* \\ \sqrt{\frac{2\kappa}{\kappa-1}\left(r^{2/\kappa} - r^{(\kappa+1)/\kappa}\right)} & r > r^*\end{cases}

where r=pdown/pupr = p_{down}/p_{up}, r∗=(2/(κ+1))κ/(κ−1)=0.546r^* = (2/(\kappa+1))^{\kappa/(\kappa-1)} = 0.546 is the critical (choked) pressure ratio, κ=1.30\kappa = 1.30 the ratio of specific heats of CO2 (CoolProp), Rs=R/MR_s = R/M the specific gas constant, T0T_0 the gas temperature and CdC_d the discharge coefficient (1 for an ideal nozzle). The energy balance of a rigid volume charged adiabatically from a reservoir at T0T_0 gives V dp/dt=κRsT0 m˙V\,dp/dt = \kappa R_s T_0\,\dot m. Separating variables with r=p/pfr = p/p_f:

tpress=Vbκ Cd Ag RsT0∫p0/pfreqdrψ(r),t_{press} = \frac{V_b}{\kappa\,C_d\,A_g\,\sqrt{R_s T_0}} \int_{p_0/p_f}^{r_{eq}} \frac{dr}{\psi(r)},

integrated numerically up to req=0.99r_{eq} = 0.99, the pressure equalisation at which the liquid valve opens. Adiabatic charging requires less gas than isothermal charging (the gas heats on compression), so this is the shortest possible time: 0.082 s.

Liquid flow (Torricelli). With bowl and bottle at the same pressure, the driving force is the liquid head HH between the bowl level and the valve outlet. The ideal (frictionless) exit velocity is v=2gHv = \sqrt{2gH} (Bernoulli). The liquid flows through the neck bore dnd_n minus the return-gas tube dvd_v:

Afree=π4(dn2−dv2),tfill=μCd,l Afree 2gH.A_{free} = \frac{\pi}{4}(d_n^2 - d_v^2), \qquad t_{fill} = \frac{\mu}{C_{d,l}\,A_{free}\,\sqrt{2gH}} .

With Cd,l=1C_{d,l} = 1, AfreeA_{free} = 342 mm² and HH = 0.40 m, the minimum fill time of 502.6 mL is 0.53 s. No publicly available drawing of the 28 mm PET neck finish could be retrieved, so dnd_n is a declared scenario value (21.7 mm, range 20.5–22.5 mm).

Snift (head-space blowdown). After filling, the head-space at pfp_f is vented through the snift orifice dsd_s down to (1+ϵs) p0(1+\epsilon_s)\,p_0. For an adiabatic blowdown the gas temperature falls as T=T0(p/pf)(κ−1)/κT = T_0 (p/p_f)^{(\kappa-1)/\kappa} and V dp/dt=−κRsT m˙V\,dp/dt = -\kappa R_s T\,\dot m, which with s=p/pfs = p/p_f gives

tsnift=Vhsκ Cd As RsT0∫send1dss1+(κ−1)/(2κ) ψ(p0/(pfs))=0.27 s.t_{snift} = \frac{V_{hs}}{\kappa\,C_d\,A_s\,\sqrt{R_s T_0}} \int_{s_{end}}^{1} \frac{ds}{s^{1+(\kappa-1)/(2\kappa)}\,\psi(p_0/(p_f s))} = 0.27\ \text{s}.

This gas-dynamic time is a strict lower bound. In practice the snift is longer and stepped: once the head-space pressure falls below peqp_{eq} the product is supersaturated by the ratio peq/p0p_{eq}/p_0 (about 2.4), CO2 nucleates on the bottle wall and the bottle foams over if decompression is fast. No physical law fixes that duration without knowledge of the nucleation sites, so the controlled snift time is a declared scenario value (1 s, range 0.5–2 s).

Valve time and machine rate.

tvalve,min=tpress+tfill+tsnift=0.87 s,t_{valve,min} = t_{press} + t_{fill} + t_{snift} = 0.87\ \text{s}, tvalve,typ=tpress(Cd,g)+tfill(Cd,l)+max⁡(ts, tsnift(Cd,g))+ta=3.1 s,t_{valve,typ} = t_{press}(C_{d,g}) + t_{fill}(C_{d,l}) + \max(t_s,\ t_{snift}(C_{d,g})) + t_a = 3.1\ \text{s},

where Cd,gC_{d,g} and Cd,lC_{d,l} are the discharge coefficients of gas and liquid paths in operation (the liquid one includes deliberate throttling against foaming), tst_s the controlled snift and tat_a the valve actuation and settling time. On a rotary filler each valve has the share ϕ\phi of a revolution for its process (the rest is infeed and discharge of the bottle), so

n˙valve=ϕtvalve,n˙machine=N ϕtvalve.\dot n_{valve} = \frac{\phi}{t_{valve}}, \qquad \dot n_{machine} = N\,\frac{\phi}{t_{valve}} .

With ϕ\phi = 0.75 and NN = 72 valves: 3 090 bottles per hour per valve and 222 000 bottles per hour at the physical limit; 62 600 bottles per hour with the typical valve times.

4. Process capability — quality ceiling

The capability model is the one of Section 1. Bottles with a T2 error are the only quality loss counted by OEE (they cannot be marketed), so

Qmax=1−r=1−Φ ⁣(Qn−kT2T−μσ).Q_{max} = 1 - r = 1 - \Phi\!\left(\frac{Q_n - k_{T2}T - \mu}{\sigma}\right).

At CpkC_{pk} = 2.9 this is 1 to many decimals; QmaxQ_{max} falls visibly only when σ\sigma approaches the tolerance (for σ\sigma = 16 mL it is 0.998). Give-away is not an OEE quality loss — the bottle is good — but a yield loss, reported in YmaxY_{max}.

5. Unavoidable stops — maximum availability

One stop cannot be avoided: cleaning-in-place of the filler (bowl, valves, gas paths) at the interval ΔCIP\Delta_{CIP} set by the hygiene plan (declared; no open source gives it for soft-drink fillers). Its minimum duration 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 the recirculated product pipes and valves; the spray cleaning of the bowl interior is outside its validity, so the linked duration is a lower bound:

Amax=1−tCIPΔCIP=0.995.A_{max} = 1 - \frac{t_{CIP}}{\Delta_{CIP}} = 0.995 .

Product changeovers (push-out and rinse, duration tcot_{co} every Δco\Delta_{co}) are a decision of the product mix, not physics. They are reported separately as Ascen=Amax−tco/ΔcoA_{scen} = A_{max} - t_{co}/\Delta_{co} = 0.974 and enter only oee_scenario. The typical availability uses a declared typical CIP programme of 1.5–3 h: Atyp=1−tCIP,typ/ΔCIPA_{typ} = 1 - t_{CIP,typ}/\Delta_{CIP} = 0.972.

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 rate of Section 3: a machine running at the rate set by its own minimum valve times has no performance loss by definition. The gap between the theoretical and the typical rate (a factor of about 3.5) is a loss of the installed technology against physics, reported as the rate cascade. The result is OEEmaxOEE_{max} = 0.995, set by the minimum cleaning stop; with the declared changeovers, oee_scenario = AscenPmaxQmaxA_{scen} P_{max} Q_{max} = 0.974.

7. Assumptions and validity

See the assumption and validity lists generated from the configuration file. The most influential inputs (sensitivity table) are the cleaning and changeover intervals and durations for OEE; the drift of the lot mean and its safety factor for yield; and the liquid discharge coefficient and controlled snift time for the typical rate.

References

  • Council Directive 76/211/EEC on the making-up by weight or by volume of certain prepackaged products, consolidated text of 11.04.2009. Publications Office of the European Union.
  • International Organization of Legal Metrology. OIML R 87 Quantity of product in prepackages, Edition 2016.
  • National Institute of Standards and Technology. NIST Handbook 133, Checking the Net Contents of Packaged Goods, 2026 Edition. doi:10.6028/NIST.HB.133-2026
  • 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.
  • Weiss, R. F. (1974). Carbon dioxide in water and seawater: the solubility of a non-ideal gas. Marine Chemistry, 2, 203–215. doi:10.1016/0304-4203(74)90015-2
  • 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.
  • QnQ_n500 mL

    Nominal quantity stated on the bottle

    Scenario · Fixed

    Declared assumption

    0.5 L bottle of the scenario.

  • TT15 mL

    Tolerable negative error (tolerable deficiency) for nominal quantities from 500 to 1 000 mL

    Regulatory · Fixed

    Verified quote
    Show verbatim quote
    from 500 to 1 000 — 15 from 1 000 to 10 000 1.5Council Directive 76/211/EEC of 20 January 1976 on the approximation of the laws of the Member States relating to the making-up by weight or by volume of certain prepackaged products (consolidated text of 11.04.2009) (opens in a new tab). Publications Office of the European Union, 2009. p. 5. CELEX 01976L0211-20090411
  • kT2k_{T2}2

    Multiple of the tolerable negative error beyond which no prepackage may be marketed (T2 error)

    Regulatory · Fixed

    Verified quote
    Show verbatim quote
    no prepackage having a negative error greater than twice the tolerable negative error given in the table in 2.4 may bear the EEC signCouncil Directive 76/211/EEC of 20 January 1976 on the approximation of the laws of the Member States relating to the making-up by weight or by volume of certain prepackaged products (consolidated text of 11.04.2009) (opens in a new tab). Publications Office of the European Union, 2009. p. 5. CELEX 01976L0211-20090411
  • pT1p_{T1}2.5 %

    Maximum share of prepackages with a T1 error in a homogeneous group

    Regulatory · Fixed

    Verified quote
    Show verbatim quote
    A homogenous group of prepackages shall contain no more than 2.5 % of packages having T1 errors.OIML R 87 Quantity of product in prepackages, Edition 2016 (E) (opens in a new tab). International Organization of Legal Metrology (OIML), 2016. p. 13. OIML R 87:2016 (E)
  • Δavg\Delta_{avg}0 mL

    Average requirement: the mean content may not be below the nominal quantity (offset zero)

    Regulatory · Fixed

    Verified quote
    Show verbatim quote
    The average actual quantity of product in prepackages shall be at least equal to the nominal quantity.OIML R 87 Quantity of product in prepackages, Edition 2016 (E) (opens in a new tab). International Organization of Legal Metrology (OIML), 2016. p. 13. OIML R 87:2016 (E)
  • σ\sigma2 mL

    Standard deviation of the individual fill volume (bottle capacity spread plus level-setting spread of the valves)

    Scenario · Uniform 1 to 4 mL

    Declared assumption

    Process-capability assumption; no open source with measured fill-volume dispersion of PET level fillers was found. Replace with the measured value of the line.

  • σμ\sigma_{\mu}1 mL

    Standard deviation of the lot mean between control checks (temperature, bottle lot, level drift)

    Scenario · Uniform 0.5 to 2 mL

    Declared assumption

    Declared assumption for the drift of the controlled mean.

  • zμz_{\mu}2.576

    Safety factor on the lot mean so that each lot meets the average requirement (one-sided 99.5 %)

    Scenario · Fixed

    Declared assumption

    Operating choice: Phi^-1(0.995). A larger factor raises give-away linearly.

  • ρ\rho1,040 kg/m³

    Density of the sweetened carbonated soft drink

    Scenario · Uniform 1,000 to 1,050 kg/m³

    Declared assumption

    Range from water (diet product) to a regular sweetened drink; declared assumption.

  • Δz\Delta z2 m

    Height from the product buffer tank level to the liquid level of the filler bowl

    Scenario · Uniform 1 to 4 m

    Declared assumption
  • vCO2v_{CO_2}3.5

    Dissolved CO2 expressed as volumes of gas at 0 °C and 1 atm per volume of liquid (0 for still liquids)

    Scenario · Uniform 2.5 to 4

    Declared assumption

    Carbonation level of the product; no open source with a regulated or reference value was used.

  • TfT_f6 °C

    Product temperature at the filler

    Scenario · Uniform 2 to 12 °C

    Declared assumption
  • A1A_1-58.0931

    Constant A1 of the Weiss correlation for the solubility of CO2 in pure water, K0 in mol/(kg atm)

    Property · Fixed

    Reference only

    Literature constant (closed-access source, cited as reference only): Weiss (1974), ln K0 = A1 + A2 (100/T) + A3 ln(T/100); gives K0 = 0.0340 mol/(kg atm) at 25 °C. Derivation in physics_docs/filling.en.md, section 3.

  • A2A_290.5069

    Constant A2 of the Weiss correlation

    Property · Fixed

    Reference only

    Literature constant (closed-access source, cited as reference only): Weiss (1974).

  • A3A_322.294

    Constant A3 of the Weiss correlation

    Property · Fixed

    Reference only

    Literature constant (closed-access source, cited as reference only): Weiss (1974).

  • Δpm\Delta p_m1 bar

    Counter-pressure margin above the equilibrium CO2 pressure (keeps the gas in solution while filling)

    Scenario · Uniform 0.5 to 2 bar

    Declared assumption

    Operating choice of the filler; declared assumption.

  • p0p_0101,325 Pa

    Atmospheric pressure

    Property · Fixed

    Computed

    Standard atmosphere.

  • VbV_b0.53 L

    Brim-full (overflow) capacity of the 0.5 L PET bottle

    Scenario · Uniform 0.52 to 0.54 L

    Declared assumption

    No open bottle drawing was used; declared assumption.

  • κ\kappa1.3

    Ratio of specific heats of CO2 (ideal gas at the filling temperature)

    Property · Fixed

    Computed

    CoolProp: cp0 = 823.8 J/(kg K) at 279 K; kappa = cp0/(cp0 - R/M) = 1.298.

  • MM0.0440098 kg/mol

    Molar mass of CO2

    Property · Fixed

    Computed

    CoolProp PropsSI('M','CO2').

  • dnd_n21.7 mm

    Inner diameter (bore) of the 28 mm PET neck finish

    Scenario · Uniform 20.5 to 22.5 mm

    Declared assumption

    No publicly available neck-finish drawing could be retrieved (the standard drawings are distributed to members and suppliers); declared assumption with a range.

  • dvd_v6 mm

    Outer diameter of the return-gas (vent) tube inside the neck; also the gas path for pressurisation

    Scenario · Uniform 4 to 8 mm

    Declared assumption
  • dsd_s1 mm

    Diameter of the snift (decompression) orifice

    Scenario · Uniform 0.6 to 1.5 mm

    Declared assumption
  • HH0.4 m

    Liquid head from the bowl level to the valve outlet (driving head with equalised pressures)

    Scenario · Uniform 0.25 to 0.6 m

    Declared assumption
  • reqr_{eq}0.99

    Pressure ratio bottle/bowl at which the liquid valve opens (pressure equalised)

    Scenario · Fixed

    Declared assumption

    Model choice: equalisation within 1 %.

  • ϵs\epsilon_s0.01

    Residual over-pressure (fraction of atmospheric pressure) at the end of the snift

    Scenario · Fixed

    Declared assumption

    Model choice: decompression to within 1 % of atmospheric pressure.

  • Cd,lC_{d,l}0.35

    Effective discharge coefficient of the liquid path in operation (valve losses plus deliberate throttling against foaming of carbonated products)

    Scenario · Uniform 0.2 to 0.6

    Declared assumption

    No open source with measured fill rates per valve was found; declared assumption.

  • Cd,gC_{d,g}0.8

    Discharge coefficient of the gas paths (pressurisation and snift)

    Scenario · Uniform 0.6 to 0.9

    Declared assumption
  • tst_{s}1 s

    Duration of the controlled (stepped) snift used to avoid foam-over of the supersaturated product

    Scenario · Uniform 0.5 to 2 s

    Declared assumption

    Not bounded by an open source; the gas-dynamic minimum is computed, the controlled duration is declared.

  • tat_{a}0.5 s

    Valve actuation and settling times per cycle (lift, opening, closing, calming before snift)

    Scenario · Uniform 0.3 to 1 s

    Declared assumption
  • ϕ\phi0.75

    Share of a carousel revolution available for the valve process (the rest is infeed and discharge)

    Scenario · Uniform 0.65 to 0.85

    Declared assumption
  • NN72

    Number of filling valves of the rotary filler

    Scenario · Fixed

    Declared assumption
  • ΔCIP\Delta_{CIP}72 h

    Calendar interval between two cleanings-in-place of the filler

    Scenario · Uniform 24 to 168 h

    Declared assumption

    No open source states a cleaning frequency for soft-drink fillers; the frequency is set by the plant's hygiene plan. Declared assumption.

  • tCIPt_{CIP}0.374294 h

    Minimum duration of the filler CIP: 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].

  • tCIP,typt_{CIP,typ}2 h

    Duration of a typical filler CIP programme (rinse

    Scenario · Uniform 1.5 to 3 h

    Declared assumption

    Declared typical programme, used for the typical availability.

  • Δco\Delta_{co}24 h

    Calendar interval between two product (flavour) changeovers

    Scenario · Uniform 8 to 72 h

    Declared assumption
  • tcot_{co}0.5 h

    Duration of a product changeover (product push-out and water rinse of bowl and valves)

    Scenario · Uniform 0.25 to 1 h

    Declared assumption
  • VcoV_{co}75 L

    Product lost per changeover in a traditional filling operation (lower documented value

    Benchmark · Uniform 75 to 100 L

    Verified quote
    Show verbatim quote
    Normally, 75–100 litres of milk can easily be lost because of the changeover of production in a traditional filling operation.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. 422. EUR 29978 EN; JRC118627; doi:10.2760/243911

8Benchmarks

Documented values from open sources used to check the ceilings; no documented value may beat the theoretical limit.
BenchmarkCeilingValueProvenance
DOE bandwidth study, state-of-the-art on-site energy intensity of fluid milk packaging (whole packaging step: filling, container handling and closing; wider boundary than the filler)
Show verbatim quote
Packaging 15 12 12 0 2.9 2.3 2.3 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. 56. DOE/EE-1571; OSTI ID 1512366
Best available technique12 Btu/lbVerified quote
DOE bandwidth study, current typical on-site energy intensity of fluid milk packaging (whole packaging step)
Show verbatim quote
Packaging 15 12 12 0 2.9 2.3 2.3 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. 56. DOE/EE-1571; OSTI ID 1512366
Typical15 Btu/lbVerified quote
DOE bandwidth study sets the thermodynamic minimum of bottling to zero because no chemical conversion takes place; its boundary ignores mechanical work, so it is not comparable with the lift and gas-compression work computed here
Show verbatim quote
Bottling, storage and delivery Set to zero due to minimal chemical conversionsBandwidth 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. 65. DOE/EE-1571; OSTI ID 1512366
Theoretical0 Btu/lbVerified quote
Product lost per changeover in a traditional filling operation, litres (BAT reference document, dairy)
Show verbatim quote
Normally, 75–100 litres of milk can easily be lost because of the changeover of production in a traditional filling operation.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. 422. EUR 29978 EN; JRC118627; doi:10.2760/243911
Best documented75 LVerified quote

9Unavoidable stops

Stops that physics or regulation impose; they set the maximum availability.
  • Cleaning-in-place of the fillerDeclared assumption

    Product residues in bowl, valves and gas paths support microbial growth; the hygiene plan of the plant fixes the interval. Duration and interval are declared scenario values.

  • Product (flavour) changeover (scenario stop: availability_scenario only)Verified quote

    A new product requires pushing out the previous one and rinsing the bowl and valves; the BAT reference document documents the product lost in a traditional filling changeover.

    Show verbatim quote
    Normally, 75–100 litres of milk can easily be lost because of the changeover of production in a traditional filling operation.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. 422. EUR 29978 EN; JRC118627; doi:10.2760/243911

10Assumptions

  1. Individual fill volumes are normally distributed around the target mean; the lot mean drifts with the declared standard deviation between control checks.
  2. Bottles with a T2 error are rejected at the filler and their content is lost; bottles with a T1 error are legal up to the 2.5 % share and are sold.
  3. Isobaric filling: bowl and bottle are at the same pressure while the liquid flows, so the driving force is the liquid head only; the displaced gas returns to the bowl and is not consumed.
  4. Gas flows are isentropic orifice flows (choked below the critical pressure ratio); the bottle is charged and the head-space is discharged adiabatically, which gives the shortest times (lower bounds).
  5. CO2 dissolution (carbonation) happens upstream in the carbonator and is outside the boundary; the solubility is that of CO2 in pure water (Weiss correlation), the effect of dissolved sugar on solubility is neglected.
  6. Availability: the filler CIP is the only physically required stop; its minimum duration comes from the cleaning model, valid for the recirculated product pipes and valves; the spray cleaning of the bowl interior is outside that validity, so the linked duration is a lower bound. Product changeovers are a product-mix decision, not physics: they enter only availability_scenario and oee_scenario. The typical availability uses the declared typical CIP programme.
  7. Bottle rigidity: the PET bottle volume does not change during pressurisation (the expansion of the bottle under pressure is neglected).

11Range of validity

  • Level (vent-tube) gravity fillers for liquids of water-like viscosity, carbonated or still, in rigid or semi-rigid bottles of 0.2 to 2 L.
  • Prepackage rules of the EU Directive 76/211/EEC and OIML R 87; NIST Handbook 133 uses a different (maximum allowable variation) scheme that is not modelled here.

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.
  • ΔCIP\Delta_{CIP}Calendar interval between two cleanings-in-place of the filler
    elasticity +0.005Spearman ρ +0.94
  • tCIPt_{CIP}Minimum duration of the filler CIP: physical minimum of the kinetic and displacement phases plus declared hold times (output cip_time_min of the cleaning model)
    elasticity -0.005Spearman ρ -0.31
  • Output rises with the parameter
  • Output falls with the parameter

13Validation checks

A model is published only when every check passes.
  • Dimensional analysis31 outputs converted to their declared unitsPassed
  • Every cited value verified against its source10 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. Council Directive 76/211/EEC of 20 January 1976 on the approximation of the laws of the Member States relating to the making-up by weight or by volume of certain prepackaged products (consolidated text of 11.04.2009) (opens in a new tab). Publications Office of the European Union, 2009. p. 5. CELEX 01976L0211-20090411
  2. OIML R 87 Quantity of product in prepackages, Edition 2016 (E) (opens in a new tab). International Organization of Legal Metrology (OIML), 2016. p. 13. OIML R 87:2016 (E)
  3. nist_hb133_2026
  4. 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. 422. EUR 29978 EN; JRC118627; doi:10.2760/243911
  5. 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. 56. DOE/EE-1571; OSTI ID 1512366
  6. doi:10.1016/0304-4203(74)90015-2 — Solubility of CO2 in water (Weiss correlation).
  7. doi:10.1021/ie4033999 — Thermophysical property library (CoolProp).