Batch Kettle Saponification Run Control via SAP Value and Caustic Dosing Rate

Batch kettle saponification run control rests on the stoichiometric bridge between the saponification value of the fatty charge and the mass of sodium hydroxide delivered to the vessel. The saponification value, determined according to ISO 3657:2013 or AOCS Cd 3-25, reports the mass of potassium hydroxide in milligrams required to saponify one gram of oil or fat. For a refined, bleached, deodorized palm kernel oil lot with a saponification value of 245 mg KOH/g, the theoretical sodium hydroxide requirement at 100% purity is calculated as 245 × 0.713 = 174.7 kg NaOH per 1,000 kg oil. In a full-scale batch kettle the actual weighment is derived from the inverse of the NaOH assay: a 98.0% w/w sodium hydroxide feedstock would require 178.3 kg per metric ton for the same lot. The conversion factor 0.713 arises from the molar mass ratio of sodium hydroxide to potassium hydroxide (40.00 g/mol ÷ 56.11 g/mol). Because the saponification value is lot-specific and can shift by 5–15 mg KOH/g between tropical oil shipments and by 10–25 mg KOH/g when tallow is substituted for palm stearin, control systems that fix caustic flow solely on batch volume without adjusting for the certificate of analysis create either residual unsaponified matter or excessive free alkali at the endpoint.

In a typical 10,000 kg oil charge, the caustic dosing rate is not applied as a single instantaneous addition but as a staged mass flow profile integrated over 45–90 min. The profile is generated by the distributed control system from the target lye excess, the assay-corrected stoichiometric demand, and a phase-dependent dosing maximum. Early in the run, only 20–30% of the total caustic may be introduced at 4.0–6.0 kg/min per metric ton of oil to avoid forming a persistent invert emulsion. The control loop uses a positive-displacement or Coriolis mass flow meter with a measurement uncertainty of ±0.1% of reading, a temperature-compensated density input for the caustic solution, and a bottom-mounted PT100 resistance thermometer conforming to IEC 60751. Kettle pressure, agitator motor current, and vapour-duct temperature are interlocked with the caustic feed valve. If the agitator current exceeds the phase-specific trip point—commonly configured at 85–90% of motor nameplate for a close-clearance scraped-wall anchor—the dosing pump is ramped back by 30–50% until the torque value returns below the alarm band.

What Limits Caustic Dosing Rate During the Paste Phase in Palm Kernel Oil Kettle Saponification?

The limiting variable during the first third of the reaction is not stoichiometric availability but local mass transfer. When 50% w/w sodium hydroxide solution at 20°C with a density of approximately 1.53 g/cm³ contacts the oil phase at 75–80°C, the interfacial reaction produces a coherent sodium soap film. This film is insoluble in the oil and traps unreacted caustic droplets. In palm kernel oil, which has a high lauric acid content and a saponification value of 240–256 mg KOH/g, the early soap phase can raise the apparent viscosity of the reacting mass to 120–180 Pa·s at a shear rate of 0.1 s⁻¹, although published data for this specific configuration is limited and batch-to-batch variation of 30–40 Pa·s is observed on production kettles. If the caustic dosing rate exceeds the point at which the soap film can be sheared into the bulk oil, unmixed sodium hydroxide accumulates at the surface. The resulting localized saponification creates a crust that adheres to the upper wall of the kettle. Once formed, this crust is not reincorporated; it degrades heat transfer and leaves the batch with a visible speck count after finishing.

The upper caustic feed rate is therefore derived from agitator torque, not from pump capability. A helical ribbon or close-clearance anchor impeller with a tip speed of 0.8–1.5 m/s is typically operated at 12–20 rpm in a 10 m³ hemispherical-bottom vessel. During the paste phase, the caustic dosing rate is normally held between 2.5 and 4.0 kg/min per metric ton of oil. At 4.5 kg/min per metric ton, the reaction exotherm can outpace jacket heat removal. The jacket heat-transfer coefficient drops from an initial value of 450–600 W/(m²·K) in the mixed liquid phase to below 150–200 W/(m²·K) as the paste thickens. The decline is caused by the formation of a stagnant soap layer on the jacket wall and by reduced radial mixing at the wall boundary. This is why a dosing profile that is acceptable in the later fluid phase will fail during the paste phase.

On a 15 m³ batch kettle processing 12,000 kg of refined bleached deodorized coconut oil, the initial charge temperature is set at 75°C and the caustic solution at 48–50% w/w is preheated to 35°C before injection through a dip pipe at the kettle centre. The first 20% of the total caustic charge is added at 6.0 kg/min for the first 10 min, after which the control system reduces the rate to 3.2 kg/min because the agitator current rises from 68 A to 82 A against a full-load rating of 95 A. The paste phase lasts approximately 25–35 min and is recognized by a plateau in the reaction temperature, a rise in torque, and an increase in the opacity of a sampled drop from translucent to pearlescent. After the paste phase breaks, the batch becomes fluid and the remaining caustic is charged at 5.0–5.5 kg/min per metric ton without exceeding the 95°C ceiling. Wet alkalinity samples are drawn every 15 min from the steam-jacketed sample port; free alkali, expressed as Na₂O, moves from 0.8% at mid-run to an endpoint of 0.08–0.12% on a neat soap basis. The endpoint is accepted only when two consecutive samples separated by 10 min show a free alkali change of less than 0.03% Na₂O, a condition prescribed by the plant quality plan to prevent downstream neutralization with fatty acid from overshooting the target.

The table represents stoichiometric values only; the actual batch addition includes an intentional excess of 0.05–0.15% free NaOH by mass in the neat soap to ensure complete saponification. A feedstock delivery with a saponification value near the upper end of the range, such as coconut oil at 264 mg KOH/g, increases the caustic demand by 9.95 kg NaOH per 1,000 kg oil relative to the lower end. If the batch weighment error on the oil charge is ±50 kg and the SAP analytical repeatability under ISO 3657:2013 is ±1.0 mg KOH/g, the combined dosing uncertainty for a 10,000 kg charge can reach ±10–12 kg NaOH. This uncertainty is material at the endpoint; therefore the control system includes a trim step: after 95% of the stoichiometric caustic is added, the remaining 5% is controlled by an in-line near-infrared probe calibrated against the wet alkali titration method.

SAP-Based Stoichiometric Compensation in High Free Fatty Acid Tallow Charging

Tallow feedstocks with a free fatty acid content above 1.0% w/w as oleic acid require a caustic correction that is separate from the triglyceride saponification value. The acid value, determined by AOCS Ca 5a-40 or equivalent titration, reports the free fatty acid concentration as milligrams of KOH per gram of fat. A tallow charge with an acid value of 4.0 mg KOH/g consumes 2.85 kg NaOH per 1,000 kg fat at 100% purity before any triglyceride saponification begins. At 50% w/w caustic, this corresponds to 5.70 kg of solution per metric ton. The neutralization reaction is faster than triglyceride saponification and therefore occurs at the oil-caustic interface during the first addition. When the dosing rate is set only from the saponification value and the free fatty acid correction is omitted, the initial caustic charge is depleted by neutralization, leaving an apparent induction period. The batch then falls behind the temperature ramp and the control logic incorrectly interprets the reduced exotherm as a need for more heat. This is a common failure mode on tallow-fed kettle lines where the raw material source switches between mechanically separated tallow and rendered tallow without a new certificate of analysis being entered into the recipe.

The equation implemented in the batch recipe is total NaOH mass at 100% purity = oil mass in metric tons × 0.713 × (SAP in mg KOH/g + acid value in mg KOH/g). The assay correction divides this value by the sodium hydroxide mass fraction. For a 10,000 kg tallow charge with a saponification value of 198 mg KOH/g and an acid value of 4.0 mg KOH/g, the 100% sodium hydroxide requirement is 10 × 0.713 × 202 = 1,440.3 kg. Using 98% w/w NaOH, the actual weighment becomes 1,469.7 kg. The acid value contribution is small in absolute terms but is concentrated in the first 5–10% of the caustic feed. If the initial dose is injected at 5.0 kg/min per metric ton, the neutralization front consumes the caustic so rapidly that the local pH remains below the saponification initiation condition. This produces a delayed viscosity surge and can result in a batch with a non-uniform soap crystal size distribution because the subsequent saponification occurs in an already partially neutralized emulsion.

Where the feedstock is a blend of palm stearin and coconut oil intended for a semi-boiled washer soap, the arithmetic mean saponification value may be inaccurate if the constituent oils saponify at different rates. A 70:30 palm stearin:coconut oil blend has a calculated saponification value of 0.7 × 202 + 0.3 × 255 = 217.9 mg KOH/g if the palm stearin lot is 202 mg KOH/g and the coconut oil lot is 255 mg KOH/g. The blend demands 155.4 kg NaOH per metric ton at 100% purity. However, the coconut oil component reacts faster due to its shorter-chain fatty acid composition. During the first 20 min of dosing, the caustic demand is disproportionately driven by the coconut fraction. The dosing system therefore overrides the linear average and applies a front-loaded profile of 60–65% of the total caustic in the first 30 min, followed by a tapered profile for the remaining 35–40%. This profile was developed on a production line using a 12 m³ steam-jacketed kettle with an internal helical ribbon agitator and a bottom discharge plough. The observed temperature rise was 12°C over 30 min, and the batch did not show a separate paste phase after the initial coconut-rich invert emulsion broke.

When Sodium Hydroxide Purity Drops Below 97% by Mass in Batch Dosing Manifolds

Industrial 50% w/w sodium hydroxide solution is specified by hydroxide mass fraction rather than total alkalinity. A membrane-grade product may carry 49.5–50.5% w/w NaOH and 0.1–0.3% w/w sodium carbonate. Lower assay solution arises from tank dilution or carbon dioxide absorption during storage. The recipe compensation uses the following expression: corrected solution mass = stoichiometric NaOH mass ÷ hydroxide mass fraction as NaOH. If the certificate of analysis reports 48.3% w/w NaOH, the effective NaOH fraction is 0.483, not 0.500. A batch requiring 1,440.3 kg NaOH at 100% purity would require 2,982.0 kg of the 48.3% solution, compared to 2,880.6 kg if the solution were exactly 50.0%. The 101.4 kg difference is sufficient to shift final free alkali by several tenths of a percent if ignored. Although sodium carbonate contributes to total alkalinity, it does not saponify triglycerides at a rate that can be credited in a 90 min kettle cycle below 100°C; it may neutralize free fatty acids but cannot be substituted for hydroxide in the saponification reaction under these conditions.

The low-purity condition also changes the density and viscosity of the caustic feed. A 48.3% w/w NaOH solution at 20°C has a density near 1.51 g/cm³, while a 50.0% w/w solution has a density near 1.53 g/cm³. If the dosing control uses a volumetric flow meter without density correction, a 4.0 L/min setting delivers 6.04 kg/min of the 48.3% solution instead of 6.10 kg/min of the 50.0% solution. The density-related deficit is small compared to the assay deficit: over a 60 min feed window, the volumetric error is 3.6 kg of solution, whereas the assay error for the same batch can exceed 100 kg. Nevertheless, the Coriolis mass flow meter corrects for both, and the recipe manager locks the caustic feed from the certificate of analysis. If the caustic storage tank is recharged mid-batch, the software rejects a new lot unless the operator enters the updated assay and density. This interlocks against mixing of two caustic concentrations in the day tank.

Alkali addition at the nitrogen-blanketed feed nozzle is terminated when the in-line guided-wave radar level sensor confirms that the fill volume has dropped below 2% of the day tank capacity. The final 5% of caustic is not delivered at the same rate as the early bulk charge; instead, the control algorithm divides the remaining demand into three equal increments delivered at 1.0 kg/min per metric ton, each separated by a 10 min equilibration period under agitation. During this trim phase, the batch is sampled through a recessed sample point that is flushed with steam after each draw. The free alkali result is entered into the batch historian and the system computes an adjusted final increment using the equation: adjusted mass = remaining theoretical mass − 2.0 × batch mass × (sampled free alkali − target free alkali) / 100. This equation, calibrated on a 12,000 kg neat soap mass, reduces the final free alkali variance from ±0.08% Na₂O under single-point control to ±0.02% Na₂O across 120 consecutive batches. The improvement is statistically meaningful because downstream vacuum drying is sensitive to free alkali excursions above 0.15% Na₂O, which increase soap hydrolysis and odour development during storage.

Torque Rise Profiles Cannot Replace Wet Alkali Titration During Final Adjustment

Agitator torque is an excellent mid-run phase indicator but it is not a direct measurement of saponification completion. In a 10 m³ kettle, torque rises through the paste phase, falls after phase inversion, and then plateaus. A stable torque plateau at 78–80% of the motor full-load value can coexist with free alkali values from 0.02% to 0.25% Na₂O because the rheology of the neat soap is dominated by the crystalline network rather than by the residual alkali concentration. A batch discharged at 0.02% free Na₂O appears identical in viscosity to a batch at 0.15% free Na₂O when measured with a Brookfield viscometer at 70°C. The absence of a torque signal for free alkali means the endpoint must be confirmed by wet alkalinity titration according to ISO 685:1975 or by an in-line spectroscopic method calibrated against the same reference method.

Raman spectroscopy has been deployed on some kettle lines to measure free alkali in the neat soap at 85°C. The method uses a sapphire-windowed probe inserted into a recirculation loop. Partial least-squares calibration models built with 120 samples from 12 separate production campaigns achieved a root mean square error of prediction of 0.018% Na₂O under isothermal conditions. The calibration fails if the moisture content deviates more than 3% absolute from the training set mean, because water bands influence the baseline. For this reason, the Raman signal is used as a trim signal only after the batch moisture has been brought below 30% by vacuum flashing. On lines without vacuum flashing, the final control action is a laboratory titration. The caustic feed valve is closed after the penultimate bulk addition, and the batch is held at 85°C for 15 min before sampling. The final caustic addition is limited to 0.3 kg/min per metric ton, and the control system enforces a maximum final addition of 0.5% of the total stoichiometric mass to prevent overshoot.