Saponification Exotherm Control in Continuous Soap Plants via Staged Caustic Solution Addition

Continuous saponification of triglycerides with sodium hydroxide solution is an exothermic liquid–liquid reaction in which each ester bond is cleaved to yield sodium carboxylate and glycerol. The heat release is not uniform across feedstocks because the saponification value, determined according to AOCS Cd 3-25, defines the stoichiometric caustic demand and therefore the total reaction enthalpy per unit mass of oil. For a refined coconut oil with a saponification value of 250 mg KOH/g, the stoichiometric NaOH demand is 178.2 kg per metric ton of oil, equivalent to 356.4 kg of 50 wt% NaOH solution per metric ton; for refined tallow with a saponification value of 190 mg KOH/g, the corresponding demand is 135.4 kg NaOH per metric ton, or 270.8 kg of 50 wt% solution per metric ton. These figures are derived from the molar mass ratio 40.00 g/mol NaOH to 56.11 g/mol KOH, multiplied by the saponification value expressed in grams KOH per kilogram oil. In a continuous plant, the oil feed is metered into a closed recirculating reactor loop at rates typically between 2,000 and 15,000 kg/h; the loop is held at 110–130°C and 1.5–2.5 bar gauge to keep water in the liquid phase while allowing controlled flash drying downstream. When the entire caustic requirement is injected at a single point, the local reaction front can produce a temperature excursion that exceeds the cooling capacity of the recirculation heat exchanger, which in many continuous skid designs is specified for a normal heat removal duty of 0.2–0.5 MW per 10,000 kg/h oil feed. The subsequent temperature rise alters the phase stability of the neat soap phase, accelerates hydrolysis side reactions, and can darken the finished base before any viscosity measurement in the finishing section identifies the deviation. The staged addition of caustic solution therefore operates as both an exotherm control strategy and a reaction completeness control strategy, because the split ratio directly influences free alkali distribution, local temperature rise, and the transition from dispersed oil droplets to a continuous soap phase.

Thermal Runaway Thresholds in Continuous Saponification Reactors Are Feedstock-Specific

The severity of a thermal runaway event in a continuous saponification loop is governed by the saponification value of the oil, the free fatty acid content, the water content of the caustic solution, and the heat transfer coefficient of the recirculation exchanger. Oils with high lauric acid content, such as coconut and palm kernel oils, exhibit saponification values between 240 and 264 mg KOH/g and therefore release more total reaction enthalpy per metric ton than tallow or palm stearin, which typically fall between 190 and 205 mg KOH/g. The local reaction rate increases with temperature, and the apparent activation energy for the two-phase triglyceride–aqueous alkali system has been reported in industrial process design literature to fall between 42 and 55 kJ/mol; this corresponds to a reaction rate that approximately doubles for every 10–12°C increase within the 100–140°C operating window. Because the saponification reactor is intended to run within a narrow thermal window, excursions of more than 5°C above the design setpoint can trigger phase separation in partially saponified high-free-alkali material, particularly when the total fatty matter content exceeds 60 wt%. The margin between a pumpable neat soap and a gel-like phase can be less than 5°C for coconut and palm kernel formulations, while tallow-based systems exhibit a slightly broader margin but also show greater sensitivity to color formation at temperatures above 140°C. The recirculation heat exchanger in a continuous saponification plant typically provides an overall heat transfer coefficient of 350–600 W/m²·K when the tube-side velocity is maintained above 1.8 m/s; below this velocity, the tube-side film coefficient deteriorates, and any calcium soap fouling layer formed from hard water or feedstock impurities further reduces the available heat removal duty. The thermal runaway threshold is therefore not a single reactor temperature but a combined condition of local temperature, local free alkali concentration, and local total soap concentration. A single-point caustic injection can exceed the local threshold even when the bulk loop temperature remains within specification, because the reaction front is not mixed at the molecular scale until several static mixer elements downstream have dissipated the concentration gradient.

Feedstock-specific thresholds also emerge from the fatty acid distribution. Coconut oil and palm kernel oil produce sodium laurate and sodium myristate soaps that thicken rapidly as the water content is reduced; at 65 wt% total fatty matter and 120°C, the apparent viscosity of the neat soap phase is commonly in the range of 300–600 mPa·s, but a local overfeed of 50 wt% caustic can raise the electrolyte concentration and shift the viscosity above 1,200 mPa·s before the recirculation pump can restore homogeneity. Tallow and palm stearin soaps contain higher proportions of sodium palmitate and sodium stearate, which raise the melting point of the soap phase and make the system more susceptible to premature solidification if the loop temperature falls below the design setpoint by more than 5°C. The processing window for continuous saponification is therefore constrained by a low-temperature gel boundary and a high-temperature color and hydrolysis boundary; the two boundaries are separated by as little as 10–15°C for high-lauric feedstocks. Staged caustic addition narrows the local temperature distribution and allows the bulk setpoint to be maintained at the midpoint of the feedstock-specific window rather than at a conservative low setpoint that sacrifices reactor productivity and pumpability.

Regardless of reactor configuration, the staged addition of caustic solution requires a tightly integrated flow-ratio control hierarchy. In a three-stage configuration, the first injection point is located upstream of the first static mixer, the second after the cooler return line, and the third directly before the final high-shear disperser. The split ratio is determined by saponification value, free fatty acid content, and the heat removal capacity of each stage. For coconut oil at a design loop temperature of 120°C, a 35/35/30 wt% split across three stages keeps the local temperature rise per stage between 5 and 8°C; for refined tallow, a 30/35/35 wt% split is often selected to reduce the initial exotherm because tallow has a lower saponification value but a higher melting point and slower initial mass transfer. Each injection quill is fabricated from 316L stainless steel or Alloy 20, with a tip velocity of 3–5 m/s and a pressure drop across the quill of 0.5–1.0 bar to improve dispersion of the aqueous alkali into the oil continuum. Coriolis mass flow meters with a measurement accuracy of ±0.10% of rate are installed on both oil and caustic streams; control valves use equal-percentage trim with digital positioners and a full-scale response time of less than 2 seconds. Static mixer elements downstream of each injection point are sized for a coefficient of variation below 0.05 in caustic concentration, which corresponds to a pressure drop of 0.2–0.4 bar per element at the operating viscosity. The caustic header is maintained at 40–50°C with low-pressure steam tracing to prevent crystallization of 50 wt% NaOH, which has a freezing point of approximately 12°C at that concentration but becomes highly viscous below 30°C.

The split ratios in the table are design-basis values used for continuous skid engineering at a 120°C loop setpoint; they are not universal thermochemical constants and must be recalculated when the feedstock blend, free fatty acid content, or recirculation heat removal capacity changes. The local temperature rise per stage is derived from the sensible heat of the reaction mass and the stage-wise heat removal available from the recirculation cooler. Published data for this specific configuration is limited, so the split ratios are validated during commissioning by thermocouple arrays placed 0.5 m, 1.5 m, and 3.0 m downstream of each injection point.

What Process Interlock Logic Prevents Localized Caustic Overfeed in Continuous Neutralization?

Process interlocks for staged caustic addition are implemented as a safety instrumented function in accordance with IEC 61511-1:2016, typically assigned a safety integrity level of SIL 2 when the consequence of overfeed includes high-pressure relief or loss of containment. The oil flow transmitter and the caustic flow transmitter are wired into a ratio controller with a fail-safe output; if the measured caustic-to-oil mass ratio exceeds the target by more than 2.0%, an alarm is generated, and if it exceeds the target by more than 5.0%, the caustic block valve closes within 2 seconds. A high-high temperature transmitter located immediately downstream of the first static mixer initiates an automatic diversion of partially saponified material to a water-quench rework tank when the measured temperature exceeds the loop setpoint by 10°C. Redundant temperature elements in 1oo2 voting logic reduce spurious trips; the safety system also monitors recirculation pump speed and seal flush pressure because loss of recirculation is the most common cause of localized overheat in continuous soap plants. The free alkali concentration at the reactor outlet is monitored by an online near-infrared analyzer calibrated against ASTM D460-91(2014); the analyzer signal updates every 30 seconds, but the control loop also uses a soft sensor based on inlet flow ratio and temperature to compensate for the analyzer delay. The residual free alkali target for neat soap leaving the saponification loop is 0.05–0.10 wt% NaOH; values above 0.15 wt% trigger automatic divert to rework, while values below 0.03 wt% indicate under-saponification and potential free fatty acid carryover. Interlock setpoints are also applied to the caustic dilution water ratio because a sudden drop in caustic concentration can reduce the apparent exotherm and cause the ratio controller to overfeed concentrated caustic in pursuit of the free alkali target. The dilution water flow meter is interlocked with the caustic flow transmitter so that a loss of dilution water immediately reduces the caustic setpoint to the minimum safe level.

The interlock logic must account for the dead time between the caustic injection point and the first downstream temperature sensor. In a recirculation loop operating at a superficial velocity of 1.5–2.5 m/s, the transport delay from the first quill to the first thermocouple can range from 2 to 8 seconds depending on the distance and the bypass flow. A conventional proportional-integral temperature controller cannot respond quickly enough to a localized overfeed during this dead time, which is why the primary protection is material balance based on the oil and caustic flow meters rather than downstream temperature. The ratio controller uses feedforward compensation from the oil flow transmitter and the recirculation flow meter; the caustic valve position is capped by a high-select block that compares the instantaneous caustic demand to the saponification value and free fatty acid content entered by the operator. If the oil flow transmitter fails or the oil feed pump trips, the caustic block valve closes on a 0 V or 4 mA fail-safe signal within 1.5 seconds. The recirculation pump is provided with an emergency power hold-up of 3–5 seconds to allow the block valve to close without subjecting the static mixer to stagnant caustic pools. The interlock test interval is set at 6 months for the caustic block valve and 12 months for the redundant temperature elements, following the proof test requirements of IEC 61511-1:2016 for the assigned SIL 2 function.

When Feedstock Free Fatty Acid Content Exceeds Design Basis, the Caustic Staging Model Requires Rebaselining

Refined edible oils entering a continuous soap plant normally contain less than 0.10 wt% free fatty acid expressed as oleic acid; in this operating state, the saponification value alone is sufficient to set the caustic demand. When the feedstock blend includes acidulated soapstock, unrefined palm oil, or recovered fat with a free fatty acid content above 2.0 wt%, the first caustic injection stage must be rebaselined because free fatty acid neutralization is instantaneous and releases heat before any appreciable triglyceride saponification. Each 1.0 wt% free fatty acid as oleic acid consumes 1.42 kg NaOH per metric ton of oil, which corresponds to 2.84 kg of 50 wt% NaOH solution per metric ton. If this additional caustic is not accounted for in the first stage, the local temperature in the neutralization zone can exceed the loop setpoint by 12–15°C within a few seconds, producing a free alkali spike that accelerates triglyceride saponification before the cooler can remove the additional exotherm. The design-basis split must therefore shift from a 35/35/30 or 30/35/35 ratio to a first-stage-dominant ratio, typically 50/30/20 or 55/25/20 depending on the measured free fatty acid content and the first-stage heat removal capacity. Feedstock free fatty acid is measured every 2 hours by automatic titration according to AOCS Official Method Ca 5a-40, and the caustic ratio setpoint is updated in the supervisory control system only after the sample is verified by laboratory analysis. The ratio controller is interlocked to reject step changes larger than 5.0 wt% of the incumbent caustic setpoint unless the operator confirms the feedstock change. This prevents an operator from entering a new saponification value that is inconsistent with the actual fat tank and accidentally forcing the first-stage caustic valve into an overfeed condition.

When the free fatty acid content exceeds the design basis by more than 2.0 wt%, the first-stage exotherm is no longer adequately described by the saponification value alone. The neutralization of free fatty acids is a proton-transfer reaction with essentially instantaneous kinetics, while triglyceride saponification is limited by mass transfer between the oil phase and the aqueous caustic phase. The staged addition model therefore separates the caustic demand into two terms: a first-stage term based on free fatty acid content and a subsequent-stage term based on triglyceride saponification value. The first-stage term is calculated as 2.84 kg of 50 wt% NaOH solution per 1.0 wt% free fatty acid per metric ton of oil. The remaining caustic demand is distributed across the later stages according to the original split ratio. For example, a feedstock with 3.0 wt% free fatty acid and a tallow base with a saponification value of 195 mg KOH/g requires 8.52 kg of 50 wt% NaOH solution per metric ton for free fatty acid neutralization in the first stage, followed by approximately 278 kg of 50 wt% solution per metric ton for triglyceride saponification. If the original non-FFA split was 30/35/35, the remaining 278 kg/t is redistributed in that ratio across all three stages, but the first-stage total becomes the FFA term plus 30% of the triglyceride term. This results in a first-stage fraction that can exceed 45 wt% of the total caustic flow, which requires that the first-stage static mixer and injection quill be sized for the higher local heat release. Published data for this specific configuration is limited, but plant commissioning data have shown that failure to rebaseline the first stage when switching from refined tallow to acidulated palm oil can raise the first downstream thermocouple by 10–15°C within 30 seconds.

Whenever the free alkali concentration in the neat soap phase exceeds 0.10 wt%, the electrolyte content and water activity shift the phase inversion boundary enough to increase apparent viscosity from approximately 300–600 mPa·s to more than 1,200–1,500 mPa·s at 120°C in a 65 wt% total fatty matter soap. This rheological change reduces the recirculation pump’s volumetric efficiency and lowers the tube-side velocity below the 1.8 m/s threshold that maintains turbulent heat transfer. When viscosity rises above 1,200 mPa·s, the heat transfer coefficient can fall below 250 W/m²·K, and the resulting increase in loop temperature further accelerates the saponification rate, creating a reinforcing-feedback condition that can rapidly exceed the interlock setpoints. To avoid this, the staged addition control system includes a viscosity override: an inline vibrating-element viscometer in the recirculation loop signals the ratio controller to reduce the first-stage caustic flow by 1.0–2.0 wt% for every 100 mPa·s increase above 1,000 mPa·s, while a dilution water injection valve adds 0.5–1.0 wt% water to moderate the viscosity rise. The recirculation pump is a positive-displacement rotary lobe unit rated for a differential pressure of 8–12 bar and a slip flow below 2% at 1450 rpm; the static mixers downstream of the caustic injection points are designed for a shear rate between 500 and 1,200 s⁻¹ to disperse the aqueous alkali droplets without causing excessive mechanical degradation of the soap gel network. Published data for the exact rheological multivariate response surface in partially saponified high-free-alkali systems is limited, so the safety margin is set conservatively by maintaining the loop operating temperature within ±5°C of the design setpoint and by limiting the free alkali to the specified 0.05–0.10 wt% window.

Viscosity-Controlled Recirculation and Final Free Alkali Verification

After the staged caustic injection loop, the neat soap is transferred to a finishing section where excess water and glycerol are removed by vacuum flash evaporation at 30–50 kPa absolute and 90–110°C; the final free alkali and moisture are controlled to specification before drying. The recirculation pump speed is adjusted by a variable-frequency drive to maintain a loop velocity of 1.5–2.5 m/s, and the inline viscosity signal is used to modulate the pump speed within this band. Final free alkali is verified on a composite sample by potentiometric titration according to ASTM D460-91(2014); the acceptance window is 0.05–0.10 wt% NaOH for neat soap, with a maximum of 0.12 wt% in high-salt formulations intended for frame cooling. The pH of a 1.0 wt% aqueous solution is measured according to ISO 4316:1977 and is maintained between 9.8 and 10.3 for toilet soap base; values outside this range indicate incomplete saponification or caustic carryover. Moisture content is verified by azeotropic distillation or loss on drying at 105°C and is held between 12.0 and 14.0 wt% in the neat soap before finishing; higher moisture reduces viscosity but increases downstream drying load, while lower moisture raises the risk of pump suction cavitation. The process is audited against ISO 22716:2007 for cosmetic manufacturing hygiene, and the caustic handling section complies with EC 1907/2006 REACH exposure scenarios for sodium hydroxide solution.