foratvise
Get Started
Whole plant or unspecified

Cleaning in place (CIP) and sanitation

Cleaning in place (CIP) of closed process equipment — minimum programme duration, water, chemicals and energy

Validated modelOEE bound 0.951 [0.933–0.969]20,000 Monte Carlo samples · seed 20261115 · computed November 16, 2026

In short

Cleaning in place (CIP) and sanitation 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

A pasteurizer circuit (plate heat exchanger and product piping) fouled with milk-protein deposit is cleaned by a recirculating CIP programme: pre-rinse with water, alkaline wash with hot sodium hydroxide, intermediate rinse, acid wash for the mineral scale, final rinse and hot-water sanitisation. All phases run at the minimum turbulent velocity required for pipe cleaning. The cleaning programme is the main unavoidable stop of the host process; its minimum duration (output cip_time_min) is the figure other models cite.

Provenance of inputsComputed5Declared assumption16Verified quote14

2Results

Point value from the nominal parameters and 95% interval from 20,000 Monte Carlo draws of the uncertain parameters.
  • Reynolds number at the minimum velocity (rinse water)76,240
  • Wall shear stress at the minimum velocity5.34 Pa
  • Minimum duration of one rinse displacement1.73 [0.99–2.82] min
  • Minimum alkaline circulation time (deposit removal kinetics)2.26 [1.22–4.79] min
  • TheoreticalMinimum duration of a full CIP programme: physical minimum of the kinetic and displacement phases plus declared hold times22.5 [14.5–32.2] min
  • Capable programme: minimum duration plus the declared alkaline time margin (a capability margin, not a documented BAT)23.6 [15.6–33.8] min
  • TypicalTypical full CIP programme (declared assumption)90.0 [61.6–118.6] min
  • Water volume of one rinse displacement0.319 [0.181–0.518] m³
  • Minimum fresh water per cycle with final-rinse recovery and solution reuse0.686 [0.398–1.096] m³
  • TheoreticalMinimum fresh water per tonne of product0.00416 [0.00185–0.00997] m³/tonne
  • Water per tonne without recovery (three rinses and single-use solutions)0.0119 [0.0057–0.0273] m³/tonne
  • NaOH lost per tonne of product with solution reuse (mixing zones only)7.42 × 10⁻⁴ [5.94 × 10⁻⁴–4.67 × 10⁻³] kg/tonne
  • NaOH per tonne of product with single-use solution0.0153 [0.0125–0.1022] kg/tonne
  • Friction work of pumping at the minimum velocity over the minimum programme0.259 [0.113–0.561] MJ
  • TheoreticalThermodynamic minimum energy per kg of product (pumping friction work)0.00157 [0.00057–0.00454] kJ/kg
  • Best available techniqueEnergy per kg of product with reused hot solutions (pumping, steel heating, mixing zones)0.477 [0.197–0.981] kJ/kg
  • Energy per kg of product with single-use solutions2.07 [1.00–4.28] kJ/kg
  • Share of cycles that fail verification and must be repeated0.0524 [0.0010–0.2123]
  • TheoreticalMaximum availability of the host process with the capable programme0.953 [0.937–0.971]
  • TypicalAvailability of the host process with a typical programme0.842 [0.809–0.893]
  • TheoreticalQuality factor of cleaning: cycle share kept after repeat cleans0.998 [0.990–1.000]
  • TheoreticalShare of the production cycle not consumed by cleaning (physics-bounded)0.951 [0.933–0.969]

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 timemin · lower is better
  • TheoreticalMinimum duration of a full CIP programme: physical minimum of the kinetic and displacement phases plus declared hold times22.5 [14.5–32.2] min
  • TypicalTypical full CIP programme (declared assumption)90.0 [61.6–118.6] min
Otherm³/tonne · lower is better
  • TheoreticalMinimum fresh water per tonne of product0.00416 [0.00185–0.00997] m³/tonne
  • TypicalCIP water of market-milk dairies, lower end: 50 % of total water (example installation) times the lower BAT level of waste water (0.3 m3/t) · Benchmark0.150 m³/tonne
Energy per unit of productkJ/kg · lower is better
  • TheoreticalThermodynamic minimum energy per kg of product (pumping friction work)0.00157 [0.00057–0.00454] kJ/kg
  • Best available techniqueEnergy per kg of product with reused hot solutions (pumping, steel heating, mixing zones)0.477 [0.197–0.981] kJ/kg
Availabilityhigher is better
  • TheoreticalMaximum availability of the host process with the capable programme0.953 [0.937–0.971]
  • TypicalAvailability of the host process with a typical programme0.842 [0.809–0.893]
Qualityhigher is better
  • TheoreticalQuality factor of cleaning: cycle share kept after repeat cleans0.998 [0.990–1.000]
OEEhigher is better
  • TheoreticalShare of the production cycle not consumed by cleaning (physics-bounded)0.951 [0.933–0.969]

4Loss cascade

What separates each ceiling from the next one, from the physical limit to typical operation.
  1. TheoreticalTypicalCycle or residence time
    22.5 [14.5–32.2] min → 90.0 [61.6–118.6] min (gap 67.5 min)

    The physical minimum adds three rinse displacements (plug flow plus Taylor dispersion to the conductivity end point), the alkaline removal time of the deposit (two-stage kinetics: NaOH diffusion and reaction through the layer, then peeling once the alkali reaches the steel), the acid and the sanitising holds. A capable programme must extend the alkaline phase above the mean required time because deposit thickness and removal rate scatter from run to run; without that margin half of the cycles would fail verification. Typical programmes of 1–2 h are dominated by what the physical minimum excludes: heating the solution and the steel in the circuit, draining between phases, filling of tanks and vessels (spray devices), branched circuits with dead legs cleaned in sequence, timers set instead of conductivity end points, and fixed safety times inherited from validation.

  2. TheoreticalBest available techniqueEnergy per unit of product
    0.00157 [0.00057–0.00454] kJ/kg → 0.477 [0.197–0.981] kJ/kg (gap 0.475 kJ/kg)

    Heat given to the solution and the circuit is recoverable in principle, so the thermodynamic minimum is only the friction work of pumping at the required velocity. Best practice reuses the hot detergent, so it loses only the mixing zones at each phase change, but the steel of the circuit is heated every cycle and its heat leaves with the next rinse.

  3. Best available techniqueTypicalEnergy per unit of product

    Single-use solutions heat the whole inventory twice per cycle; whole dairies clean tanks, silos, separators and fillers too, with tank losses, long pipe runs and heat losses of centralised CIP stations, which is why the dairy-level share (about 20 % of the energy) is far above the circuit-level physics.

  4. TheoreticalTypicalOther

    The minimum water is two rinse displacements (the final rinse is recovered as the next pre-rinse) plus the make-up of the mixing zones of the reused solutions. Real dairies use far more: rinses on timers, vessels and tanks, dead legs, single-use solutions and manual cleaning; CIP is about half of the water of an example dairy.

  5. TheoreticalTypicalAvailability
    0.953 [0.937–0.971] → 0.842 [0.809–0.893] (gap 0.111)

    With a fouling-limited run of 8–9 h, every hour of cleaning costs about 11 % of the production cycle; the gap between the minimum programme and a typical one is the largest single availability loss of heat-treatment lines.

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 0.90.
  • Loading time (ideal)
    1.000
  • Availability
    factor 0.953 [0.937–0.971]
    0.953 (−0.047)
  • Performance
    factor 1.00 · implied by OEE = A × P × Q
    0.953 (−0.000)
  • Quality
    factor 0.998 [0.990–1.000]
    0.951 (−0.002)
  • Physics-bounded OEE
    0.951 [0.933–0.969]

6Full derivation

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

Scope. A pasteurizer circuit — plate heat exchanger and product piping, equivalent length LL and inner diameter DD — is fouled with a milk-protein deposit and cleaned by a recirculating programme: pre-rinse with water, alkaline wash with hot sodium hydroxide, intermediate rinse, acid wash for the mineral scale, final rinse and hot-water sanitisation. Cleaning is the main unavoidable stop of the heat-treatment models, so this model's central output is cip_time_min, the minimum duration of a full programme — the physical minimum of the kinetic and displacement phases plus declared hold times — with its Monte Carlo interval; the models of closed circuits cleaned in place (pasteurization, filling, mixing, evaporation) take it through a parameter link. Symbols and their values, units, uncertainty distributions and sources are listed in the parameter table generated from config/physics/cleaning.yaml.

1. Hydrodynamic requirement

Cleaning of pipes relies on turbulence: the wall shear stress carries reacted soil away and renews the alkali at the deposit surface. The minimum mean velocity commonly applied, together with a 0.5 % NaOH solution at about 80 °C, is vminv_{min} = 1.5 m/s. At that velocity the circuit carries V˙=vmin πD2/4\dot V = v_{min}\,\pi D^2/4 and

Re=ρ vmin Dμ,f=0.316 Re−1/4,τw=f8 ρ vmin2,u∗=vminf/8.Re = \frac{\rho\,v_{min}\,D}{\mu}, \qquad f = 0.316\,Re^{-1/4}, \qquad \tau_w = \frac{f}{8}\,\rho\,v_{min}^2, \qquad u^* = v_{min}\sqrt{f/8}.
  • ρ\rho, μ\mu — density and viscosity of the rinse water at 20 °C (IAPWS-95 through CoolProp); the cold rinse has the lowest Reynolds number, the conservative case for dispersion.
  • ff — Darcy friction factor of a smooth pipe (Blasius); τw\tau_w — wall shear stress; u∗u^* — friction velocity.

For a 51 mm line, Re≈76 000Re \approx 76\,000 (fully turbulent) and τw≈5.3\tau_w \approx 5.3 Pa. Water alone does not clean: a ten-minute water flush at 80 °C removed only 0.3 mm of a protein deposit, because the soil must first be swollen and broken down chemically.

2. Rate-limiting steps — minimum duration of each phase

Rinse displacement. A rinse pushes one liquid out of the circuit with another. With plug flow alone it would last one loop time L/vminL/v_{min}; axial dispersion spreads the interface. Taylor's dispersion coefficient for turbulent pipe flow is K=10.1 (D/2) u∗K = 10.1\,(D/2)\,u^*, and after one pass the interface has a Gaussian width σx=2KL/vmin\sigma_x = \sqrt{2 K L / v_{min}}. The rinse ends when the displaced liquid at the outlet falls to a residual fraction set by the conductivity end point, zrz_r standard deviations behind the mean interface:

tr=L+zr σxvmin,Vr=πD24 (L+zr σx),Vmix=πD24 2 zr σx.t_r = \frac{L + z_r\,\sigma_x}{v_{min}}, \qquad V_r = \frac{\pi D^2}{4}\,(L + z_r\,\sigma_x), \qquad V_{mix} = \frac{\pi D^2}{4}\,2\,z_r\,\sigma_x .
  • zrz_r — standard-normal quantile of the end point (3.09 for a residual fraction of 10⁻³, declared range 2.33–3.72).
  • VrV_r — water volume of one rinse; VmixV_{mix} — volume of the mixing zone created at each change of liquid, which cannot be returned to either tank.

With the point values a rinse lasts about 1.7 min.

Alkaline phase: deposit removal kinetics. Open measurements on whey-protein deposits cleaned with 0.5 % NaOH show two stages. In the first, NaOH diffuses into the swollen layer and reacts with it; the thickness falls at a nearly constant rate, independent of position, from h0h_0 to a transition thickness hτh_\tau = 0.94 mm. Once the alkali reaches the steel, adhesion is lost and a drag force peels the layer from its leading edge; this second stage is β\beta = 2.86 times faster. The first stage takes 79 % of the total time, which was 5.8 min for a 2.44 mm deposit in turbulent flow (Re 5500) at 70 °C. The first-stage rate is therefore

r1=αk h0,lab−hτs1 tend,lab,r_1 = \alpha_k\,\frac{h_{0,lab} - h_\tau}{s_1\,t_{end,lab}},

and the alkaline time for a deposit of thickness h0h_0 is

talk=h0−hτr1+hτβ r1,talk≥Lvmin.t_{alk} = \frac{h_0 - h_\tau}{r_1} + \frac{h_\tau}{\beta\,r_1}, \qquad t_{alk} \ge \frac{L}{v_{min}} .
  • h0h_0 — initial deposit thickness on the hottest plates (declared uniform range 1.0–2.44 mm, the upper end being the laboratory deposit; the point value is the centre, 1.72 mm).
  • s1s_1 — first-stage share of the time; tend,labt_{end,lab} — measured total cleaning time.
  • αk\alpha_k — acceleration at plant conditions: rate increases with velocity and temperature, but no scaling law is given, so αk\alpha_k = 1–2 is declared, with the centre of the range, 1.5, as point value.
  • The alkali must also reach every point of the circuit at least once (one loop time).

Applied to the laboratory deposit with αk=1\alpha_k = 1, the formula returns 5.6 min against the 5.8 min measured, a check of the internal consistency of the three cited figures.

Acid and sanitising holds. Mineral scale left by the alkali is dissolved by nitric acid; hot water sanitises at the end. No open source with dissolution kinetics of milkstone was found, so tacidt_{acid} = 5–15 min and tsant_{san} = 0–10 min are declared scenarios.

Minimum duration of a full programme. Solutions are pre-heated in their tanks (that can run in parallel with the pre-rinse), so the minimum duration is the physical minimum of the kinetic and displacement phases (three rinses and the alkaline removal) plus the declared acid and sanitising hold times, whose point values are the centres of their ranges:

tCIP,min=3 tr+talk+tacid+tsan.t_{CIP,min} = 3\,t_r + t_{alk} + t_{acid} + t_{san}.

With the point values tCIP,min≈22t_{CIP,min} \approx 22 min, of which 15 min are the declared holds.

3. Process capability — repeat cleans

Deposit thickness and removal rate scatter from run to run; the required alkaline time is modelled as lognormal with logarithmic standard deviation σln\sigma_{ln} (declared 0.15–0.35, of the order of the relative scatter of the measured second-stage speed-up). A programme that extends the alkaline phase by a fraction mm above the mean required time fails verification (conductivity, ATP swabs) in a share

prep=Φ ⁣(−ln⁡(1+m)σln),p_{rep} = \Phi\!\left(-\frac{\ln(1+m)}{\sigma_{ln}}\right),

and those cycles are repeated. The capable programme lasts tprog=tCIP,min+m talkt_{prog} = t_{CIP,min} + m\,t_{alk} (about 24 min, with prep≈5p_{rep} \approx 5 % at mm = 0.5). Without margin half of the cycles would fail.

4. Mass balance — minimum water and chemicals

The circuit inventory is Vinv=πD2L/4+VheelV_{inv} = \pi D^2 L/4 + V_{heel}, where VheelV_{heel} is the minimum tank and pump-suction volume needed to recirculate. Three rinses and two chemical inventories are needed per cycle. The final rinse is clean enough to be recovered as the next pre-rinse, and hot solutions can be reused, losing only their mixing zones. The minimum fresh water per cycle is then

Vw,min=2 Vr+2 Vmix,V_{w,min} = 2\,V_r + 2\,V_{mix},

against 3 Vr+2 Vinv3\,V_r + 2\,V_{inv} without recovery. Per tonne of product, divide by the production of one run, mp=V˙p ρp trunm_p = \dot V_p\,\rho_p\,t_{run}. Sodium hydroxide lost per cycle is cNaOH ρ Vmixc_{NaOH}\,\rho\,V_{mix} with reuse and cNaOH ρ Vinvc_{NaOH}\,\rho\,V_{inv} with single use. With the point values the minimum is about 0.004 m³ of water and less than 0.001 kg of NaOH per tonne of product, against 0.15 m³/t estimated for CIP in market-milk dairies (half of the lowest documented dairy waste-water level) and a lowest documented NaOH use of 0.2 kg/t. The gap is the cleaning of tanks, silos and fillers, rinses on timers, dead legs and single-use solutions; an example dairy reduced its water to 55 % and its detergent to 34 % in six years of CIP optimisation.

5. Energy balance — minimum energy per unit of product

Heat given to the solution and to the steel can be recovered in principle (the solution returns to its tank, the steel cools into a rinse that could preheat the next), so the thermodynamic minimum is only the mechanical work dissipated by friction while circulating at the required velocity:

Wpump=Δp V˙ tCIP,min,Δp=f LD ρ vmin22,emin=Wpumpmp.W_{pump} = \Delta p\,\dot V\,t_{CIP,min}, \qquad \Delta p = f\,\frac{L}{D}\,\frac{\rho\,v_{min}^2}{2}, \qquad e_{min} = \frac{W_{pump}}{m_p}.

Best practice reuses the hot detergent, so it loses only the mixing zones, but it reheats the steel of the circuit every cycle:

eBAT=Wpump+mst cp,st (Tc−Tw)+2 ρ Vmix cp,s (Tc−Tw)mp,e_{BAT} = \frac{W_{pump} + m_{st}\,c_{p,st}\,(T_c - T_w) + 2\,\rho\,V_{mix}\,c_{p,s}\,(T_c - T_w)}{m_p},

and single-use solutions heat both inventories, 2 ρ Vinv cp,s (Tc−Tw)2\,\rho\,V_{inv}\,c_{p,s}\,(T_c - T_w), instead of the mixing zones.

  • mstm_{st}, cp,stc_{p,st} — mass and specific heat of the stainless steel (declared); cp,sc_{p,s} — specific heat of the dilute solution (water at 80 °C); TcT_c — cleaning temperature; TwT_w — fresh-water temperature.

With the point values emin≈0.002e_{min} \approx 0.002 kJ/kg, eBAT≈0.5e_{BAT} \approx 0.5 kJ/kg and single use about 2 kJ/kg of product. At dairy level CIP is reported at about 20 % of total energy, which on the lowest BAT level for market milk (0.1 MWh/t) is about 70 kJ/kg: the difference is the cleaning of everything else in the plant, tank and pipe heat losses, and programmes far longer than the physical minimum.

6. Unavoidable stops and the "OEE" of cleaning

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.

Hygiene regulation requires milk equipment to be cleaned at least once each day used, but fouling sets a shorter physical interval: indirect heat exchangers are cleaned no more than 8–9 h after start-up. With that run length trunt_{run}, cleaning defines the availability ceiling of the host process:

Amax=truntrun+tprog,OEEmax=truntrun+tprog (1+prep),Qmax=OEEmaxAmax.A_{max} = \frac{t_{run}}{t_{run} + t_{prog}}, \qquad OEE_{max} = \frac{t_{run}}{t_{run} + t_{prog}\,(1 + p_{rep})}, \qquad Q_{max} = \frac{OEE_{max}}{A_{max}} .

The "OEE" of a cleaning operation is defined here as the share of the production cycle (run plus cleaning, including repeats) not consumed by cleaning; QmaxQ_{max} is the share kept after repeat cleans. With the point values Amax≈0.95A_{max} \approx 0.95 against about 0.84 with a typical 1.5 h programme: each hour of cleaning costs about 11 % of an 8–9 h production cycle.

7. Assumptions and validity

See the assumption and validity lists generated from the configuration file. The declared acid and sanitising holds and the deposit thickness dominate cip_time_min; the circuit length dominates water and the steel mass dominates energy.

References

  • Zhang, Y., van den Berg, F. W. J., Andersen, M. L., Portela, L. M., Parjikolaei, B. R., Bakalis, S. (2025). Cleaning in place of whey protein fouling — mechanisms of removal. Journal of Food Engineering, 400, 112636. doi:10.1016/j.jfoodeng.2025.112636
  • European Commission, Joint Research Centre. Best Available Techniques (BAT) Reference Document for the Food, Drink and Milk Industries, 2019.
  • U.S. Food and Drug Administration. Grade "A" Pasteurized Milk Ordinance, 2023 Revision.
  • Taylor, G. I. (1954). The dispersion of matter in turbulent flow through a pipe. Proceedings of the Royal Society A, 223, 446–468. doi:10.1098/rspa.1954.0130
  • 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.
  • vminv_{min}1.5 m/s

    Minimum mean velocity of the cleaning liquid in the pipes

    Technology · Fixed

    Verified quote
    Show verbatim quote
    It is typically a 0.5 wt% NaOH solution applied at temperatures around 80 ◦C, with a liquid velocity around 1.5 m s−1Cleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 2. doi:10.1016/j.jfoodeng.2025.112636
  • TcT_c80 °C

    Temperature of the alkaline (and acid) cleaning solution

    Technology · Uniform 70 to 90 °C

    Verified quote
    Show verbatim quote
    It is typically a 0.5 wt% NaOH solution applied at temperatures around 80 ◦C, with a liquid velocity around 1.5 m s−1Cleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 2. doi:10.1016/j.jfoodeng.2025.112636
  • cNaOHc_{NaOH}0.5 %

    Mass concentration of sodium hydroxide in the alkaline solution

    Technology · Uniform 0.5 to 2 %

    Verified quote
    Show verbatim quote
    It is typically a 0.5 wt% NaOH solution applied at temperatures around 80 ◦C, with a liquid velocity around 1.5 m s−1Cleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 2. doi:10.1016/j.jfoodeng.2025.112636
  • h0h_01.72 mm

    Initial thickness of the protein deposit on the hottest surfaces

    Scenario · Uniform 1 to 2.44 mm

    Declared assumption

    Declared range from 1.0 mm (assumed plant deposit after one run) to the laboratory type-A whey-protein deposit of 2.44 mm (zhang_2025_cip_whey p. 8: 'an initial fouling layer thickness h0 of 2.44 ± 0.045 mm'); the point value is the centre of the range, not the pessimistic upper end.

  • hτh_\tau0.94 mm

    Deposit thickness at which removal switches from diffusion–reaction to peeling

    Property · Fixed

    Verified quote
    Show verbatim quote
    an initial fouling layer thickness h0 of 2.44 ± 0.045 mm is reduced to a transition thickness hτ of 0.94 ± 0.027 mmCleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 8. doi:10.1016/j.jfoodeng.2025.112636
  • h0,labh_{0,lab}2.44 mm

    Initial deposit thickness of the laboratory cleaning runs

    Property · Fixed

    Verified quote
    Show verbatim quote
    an initial fouling layer thickness h0 of 2.44 ± 0.045 mm is reduced to a transition thickness hτ of 0.94 ± 0.027 mmCleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 8. doi:10.1016/j.jfoodeng.2025.112636
  • s1s_179 %

    Share of the total cleaning time taken by the first (diffusion–reaction) stage

    Property · Fixed

    Verified quote
    Show verbatim quote
    The first stage accounts for 79 ± 0.15 % of total cleaning timeCleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 8. doi:10.1016/j.jfoodeng.2025.112636
  • β\beta2.86

    Ratio of the removal rate of the second (peeling) stage to that of the first

    Property · Normal, sd 0.81

    Verified quote
    Show verbatim quote
    the removal rate during the second stage is 2.86 ± 0.81 times faster than that of the first stageCleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 8. doi:10.1016/j.jfoodeng.2025.112636
  • tend,labt_{end,lab}5.8 min

    Total cleaning time of the deposit in turbulent flow at 70 °C (Reynolds number 5500)

    Property · Fixed

    Verified quote
    Show verbatim quote
    with a total cleaning time of 5.8 min (at a Reynolds number of 5500 and 70 ◦C)Cleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 2. doi:10.1016/j.jfoodeng.2025.112636
  • αk\alpha_k1.5

    Acceleration of the removal rate at plant conditions (higher velocity and temperature) relative to the laboratory run

    Scenario · Uniform 1 to 2

    Declared assumption

    The cited study reports that the cleaning rate increases with velocity and temperature but gives no scaling law; the factor 1-2 is a declared assumption and the point value is the centre of that range (not the conservative lower end).

  • σln\sigma_{ln}0.25

    Cycle-to-cycle scatter of the required alkaline time (standard deviation of its logarithm)

    Scenario · Uniform 0.15 to 0.35

    Declared assumption

    Declared assumption; of the same order as the relative scatter of the second-stage speed-up reported in the cited study (0.81/2.86).

  • mm0.5

    Extension of the alkaline phase above the mean required time, as a fraction

    Scenario · Uniform 0.25 to 0.75

    Declared assumption

    Operating choice: a larger margin lowers the repeat-clean share at the cost of a longer stop.

  • tacidt_{acid}10 min

    Minimum circulation time of the acid wash (mineral scale removal)

    Scenario · Uniform 5 to 15 min

    Declared assumption

    Declared hold time, point at the centre of its range: no open source with dissolution kinetics of milkstone was registered. It is part of cip_time_min as a declared hold, not as a physical minimum.

  • tsant_{san}5 min

    Hot-water sanitisation hold at the end of the programme

    Scenario · Uniform 0 to 10 min

    Declared assumption

    Declared hold time, point at the centre of its range 0-10 min; it is part of cip_time_min as a declared hold, not as a physical minimum.

  • DD0.051 m

    Inner diameter of the cleaned circuit

    Scenario · Fixed

    Declared assumption

    Same product line as the pasteurization model.

  • LL150 m

    Equivalent length of the cleaned circuit (supply and return)

    Scenario · Uniform 80 to 250 m

    Declared assumption
  • ρw\rho_w998.2 kg/m³

    Density of water at 20 °C

    Property · Fixed

    Computed

    CoolProp PropsSI('D','T',293.15,'P',101325,'Water') = 998.2 kg/m3 (IAPWS-95).

  • μw\mu_w0.0010016 Pa·s

    Dynamic viscosity of water at 20 °C

    Property · Fixed

    Computed

    CoolProp PropsSI('V','T',293.15,'P',101325,'Water') = 1.0016e-3 Pa s; rinses at ambient temperature give the lowest Reynolds number (conservative for dispersion).

  • cp,sc_{p,s}4.197 kJ/kg/K

    Specific heat of the dilute cleaning solution (water at 80 °C)

    Property · Fixed

    Computed

    CoolProp PropsSI('C','T',353.15,'P',101325,'Water') = 4.197 kJ/(kg K); the solute (0.5-2 % NaOH) is neglected.

  • zrz_r3.09

    Standard-normal quantile of the rinse end point (residual concentration 1e-2 to 1e-4 of the displaced liquid)

    Scenario · Uniform 2.33 to 3.72

    Declared assumption

    3.09 corresponds to a residual fraction of 1e-3, detected by conductivity.

  • VheelV_{heel}0.2 m³

    Minimum liquid inventory in the CIP tank and pump suction needed to recirculate

    Scenario · Uniform 0.1 to 0.3 m³

    Declared assumption
  • mstm_{st}2,000 kg

    Mass of stainless steel in the cleaned circuit (plates and pipes)

    Scenario · Uniform 1,000 to 3,000 kg

    Declared assumption
  • cp,stc_{p,st}0.5 kJ/kg/K

    Specific heat of austenitic stainless steel

    Scenario · Uniform 0.46 to 0.52 kJ/kg/K

    Declared assumption

    Declared handbook-order value; no registered open source.

  • TwT_w15 °C

    Temperature of the fresh water supply

    Scenario · Uniform 10 to 25 °C

    Declared assumption
  • V˙p\dot V_p20 m³/h

    Product flow rate of the host process

    Scenario · Uniform 10 to 30 m³/h

    Declared assumption

    Same nominal flow as the pasteurization model.

  • ρp\rho_p1,030 kg/m³

    Density of the product (whole milk)

    Scenario · Fixed

    Declared assumption

    Declared value, as in the pasteurization model.

  • trunt_{run}8 h

    Production time between two cleanings of a fouling-limited indirect heat exchanger

    Technology · Uniform 8 to 9 h

    Verified quote
    Show verbatim quote
    indirect heat exchangers have to be cleaned no more than 8–9 hours after start-upBest 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. 401. EUR 29978 EN; JRC118627; doi:10.2760/243911
  • tCIP,typt_{CIP,typ}1.5 h

    Duration of a typical full CIP programme including heating, draining and valve sequencing

    Scenario · Uniform 1 to 2 h

    Declared assumption

    Declared assumption, identical to the cleaning duration assumed in the pasteurization model; replace with the validated programme of the plant.

8Benchmarks

Documented values from open sources used to check the ceilings; no documented value may beat the theoretical limit.
BenchmarkCeilingValueProvenance
CIP water of market-milk dairies, lower end: 50 % of total water (example installation) times the lower BAT level of waste water (0.3 m3/t)Typical0.15 m³/tonneComputed
CIP energy of market-milk dairies, lower end: 20 % of total energy (literature share) times the lower BAT level (0.1 MWh/t)Typical0.02 MWh/tonneComputed
Lowest documented NaOH consumption of market-milk and yoghurt dairies (kg per tonne of processed milk; chemical, not comparable with the water outputs)
Show verbatim quote
Market milk and yoghurt 0.2–10 0.2–5.0Best 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. 400. EUR 29978 EN; JRC118627; doi:10.2760/243911
Best documented0.2 kg/tonneVerified quote
Water consumption after six years of CIP optimisation in an example dairy, as a fraction of the initial consumption (ratio, not comparable with the per-tonne outputs)
Show verbatim quote
consumption of water decreased to 55 % and consumption of detergents decreased to 34 %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. 159. EUR 29978 EN; JRC118627; doi:10.2760/243911
Best available technique55 %Verified quote
Detergent consumption after the same optimisation, as a fraction of the initial consumption (ratio)
Show verbatim quote
consumption of water decreased to 55 % and consumption of detergents decreased to 34 %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. 159. EUR 29978 EN; JRC118627; doi:10.2760/243911
Best available technique34 %Verified quote

9Unavoidable stops

Stops that physics or regulation impose; they set the maximum availability.
  • Cleaning at least once per day of useVerified quote

    Regulatory upper bound of the cleaning interval for milk equipment.

    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
  • Fouling-limited cleaning of indirect heat exchangersVerified quote

    Milk fouling and milkstone reduce heat transfer and flow, so heat exchangers are cleaned after 8–9 h of production; this physical interval is shorter than the regulatory one and sets the cycle used here.

    Show verbatim quote
    indirect heat exchangers have to be cleaned no more than 8–9 hours after start-upBest 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. 401. EUR 29978 EN; JRC118627; doi:10.2760/243911

10Assumptions

  1. All phases circulate at the minimum velocity through a single circuit of equivalent length L and diameter D; spray-cleaned vessels, branches and dead legs are excluded (they lengthen real programmes).
  2. Rinse displacement is plug flow plus Taylor axial dispersion (as in the pasteurization model); a rinse ends when the displaced liquid falls to the residual fraction set by z_r. Desorption from surfaces is neglected.
  3. Alkaline removal follows the two-stage kinetics measured for whey-protein deposits in turbulent flow: a constant thinning rate down to the transition thickness, then a rate faster by the measured factor. The laboratory rate (70 °C, Re 5500) is scaled by a declared acceleration factor for plant conditions.
  4. Solutions are pre-heated in their tanks; heating time is outside the minimum (it can run in parallel with the pre-rinse). Acid and alkali are at the same temperature for the energy estimate.
  5. The required alkaline time scatters lognormally from cycle to cycle; a cycle whose required time exceeds the programmed time fails verification (conductivity, ATP) and is repeated.
  6. OEE of the cleaning operation is defined as the share of the production cycle (run plus cleaning, including repeats) not consumed by cleaning.

11Range of validity

  • Recirculated CIP of closed dairy circuits (plate heat exchangers and piping) fouled with milk-protein deposits; turbulent flow (Re above 10 000 at the minimum velocity).
  • Not for spray-ball cleaning of tanks, evaporators or spray dryers, whose minimum duration is set by film coverage, not by pipe displacement.

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.
  • trunt_{run}Production time between two cleanings of a fouling-limited indirect heat exchanger
    elasticity +0.049Spearman ρ +0.17
  • tacidt_{acid}Minimum circulation time of the acid wash (mineral scale removal)
    elasticity -0.021Spearman ρ -0.58
  • h0,labh_{0,lab}Initial deposit thickness of the laboratory cleaning runs
    elasticity +0.012Spearman ρ —
  • h0h_0Initial thickness of the protein deposit on the hottest surfaces
    elasticity -0.011Spearman ρ -0.19
  • vminv_{min}Minimum mean velocity of the cleaning liquid in the pipes
    elasticity +0.011Spearman ρ —
  • LLEquivalent length of the cleaned circuit (supply and return)
    elasticity -0.011Spearman ρ -0.39
  • tsant_{san}Hot-water sanitisation hold at the end of the programme
    elasticity -0.010Spearman ρ -0.58
  • σln\sigma_{ln}Cycle-to-cycle scatter of the required alkaline time (standard deviation of its logarithm)
    elasticity -0.008Spearman ρ -0.16
  • αk\alpha_kAcceleration of the removal rate at plant conditions (higher velocity and temperature) relative to the laboratory run
    elasticity +0.007Spearman ρ +0.11
  • s1s_1Share of the total cleaning time taken by the first (diffusion–reaction) stage
    elasticity -0.007Spearman ρ —
  • tend,labt_{end,lab}Total cleaning time of the deposit in turbulent flow at 70 °C (Reynolds number 5500)
    elasticity -0.007Spearman ρ —
  • mmExtension of the alkaline phase above the mean required time, as a fraction
    elasticity +0.004Spearman ρ +0.12
  • β\betaRatio of the removal rate of the second (peeling) stage to that of the first
    elasticity +0.002Spearman ρ +0.05
  • hτh_\tauDeposit thickness at which removal switches from diffusion–reaction to peeling
    elasticity -0.001Spearman ρ —
  • zrz_rStandard-normal quantile of the rinse end point (residual concentration 1e-2 to 1e-4 of the displaced liquid)
    elasticity 0.000Spearman ρ 0.00
  • DDInner diameter of the cleaned circuit
    elasticity 0.000Spearman ρ —
  • ρw\rho_wDensity of water at 20 °C
    elasticity 0.000Spearman ρ —
  • μw\mu_wDynamic viscosity of water at 20 °C
    elasticity 0.000Spearman ρ —
  • Output rises with the parameter
  • Output falls with the parameter

13Validation checks

A model is published only when every check passes.
  • Dimensional analysis22 outputs converted to their declared unitsPassed
  • Every cited value verified against its source14 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. Cleaning in place of whey protein fouling - mechanisms of removal (Zhang, van den Berg, Andersen, Portela, Parjikolaei, Bakalis; Journal of Food Engineering 400: 112636) (opens in a new tab). Elsevier (open access, CC BY 4.0); copy in the Delft University of Technology repository, 2025. p. 2. doi:10.1016/j.jfoodeng.2025.112636
  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. 401. EUR 29978 EN; JRC118627; doi:10.2760/243911
  3. fda_pmo_2023
  4. doi:10.1016/j.jfoodeng.2025.112636 — Two-stage removal kinetics of whey-protein deposits; typical CIP velocity, temperature and NaOH concentration.
  5. doi:10.1098/rspa.1954.0130 — Axial dispersion in turbulent pipe flow.
  6. doi:10.1021/ie4033999 — Thermophysical property library (IAPWS-95 water).