Caustic to Bauxite Ratio Control in Alumina Digestion

In continuous Bayer alumina digestion, the caustic-to-bauxite ratio (C/B) is not a single laboratory value but a live mass-flow relationship between total available sodium oxide and dry bauxite solids entering the digester series. The control objective is to maintain free caustic concentration after desilication high enough to dissolve gibbsite, boehmite, or diaspore within the designed residence time, while avoiding excess sodium hydroxide that raises thermal load, dissolves additional reactive silica, and increases downstream sodalite scaling in heat recovery equipment. In a typical gibbsitic plant, bauxite slurry with a solids fraction of 0.30–0.45 kg/kg is contacted with spent liquor containing 180–240 g/L total Na₂O as Na₂O equivalent; the resulting C/B at the digester inlet is ordinarily 350–600 kg Na₂O per tonne dry bauxite depending on reactive silica, bauxite moisture, and target alumina-to-caustic ratio. The total caustic concentration in Bayer liquor is reported as Na₂O equivalent because caustic soda is consumed and regenerated as sodium aluminate; the conversion factor from NaOH to Na₂O is 0.775. A change in C/B of 10 kg/t may shift the digestion discharge A/C ratio by 0.01–0.03 units, with direct consequences for red mud settler overflow stability and precipitation yield. The ratio is therefore controlled as a mass-flow cascade rather than as a simple volumetric blend ratio.

What Operational Limits Govern the Caustic-to-Bauxite Ratio in Different Bauxite Mineralogies?

The lower boundary of the caustic-to-bauxite ratio is set by the stoichiometric requirement for sodium aluminate formation plus the irreversible consumption of caustic by reactive silica. For low-temperature gibbsitic bauxite containing less than 2 wt% reactive SiO₂, published refinery operating data place the corrected C/B between 0.35 and 0.50 kg Na₂O per kg dry bauxite, while high-temperature boehmite-diaspore operations with 8–12 wt% reactive SiO₂ routinely operate at 0.65–0.80 kg/kg. Digestion of gibbsite proceeds according to Al(OH)₃ + NaOH → NaAl(OH)₄ at 140–150 °C and 0.35–0.50 MPa, whereas boehmite dissolution requires 240–260 °C and 3.4–4.5 MPa, and diaspore-containing bauxite may require 250–280 °C and 4.5–6.5 MPa. In all cases the ratio cannot be evaluated on total caustic alone; the free caustic remaining after desilication is the effective leaching agent. Reactive silica in kaolinite consumes sodium hydroxide by Al₂Si₂O₅(OH)₄ + 6NaOH → 2NaAl(OH)₄ + 2Na₂SiO₃ + H₂O, and the reaction product sodium silicate subsequently reacts with sodium aluminate to form sodalite-type desilication product. Published mass balances for high-silica bauxite show caustic loss in the range 0.7–1.1 kg Na₂O per kg reactive SiO₂ depending on sodalite retention time and washing efficiency. The upper ratio limit is constrained by heater scaling, evaporation load, and unnecessary dissolution of non-bauxite minerals; free Na₂O concentrations above 250 g/L at digestion discharge are associated with accelerated formation of sodium aluminosilicate scale in flash trains and heat exchangers. Thus the operational window is not a single point but a mineralogy-dependent band, and the control strategy must shift the setpoint when bauxite blends change.

Because reactive silica consumes caustic through kaolinite and sodalite reaction paths, the apparent C/B at the digester feed must be corrected for desilication losses before comparing plants processing different bauxite sources. The correction is commonly expressed as effective caustic-to-bauxite ratio = (total Na₂O input − Na₂O consumed by reactive SiO₂) / dry bauxite mass. Reactive silica is not total silica; only the fraction soluble under digestion conditions contributes to caustic consumption. A bauxite containing 10 wt% total SiO₂ but only 3 wt% reactive SiO₂ will require far less caustic than a bauxite with 6 wt% reactive SiO₂. For this reason, operations with frequent bauxite blend changes install online X-ray fluorescence analyzers on slurry samples and update the ratio setpoint every 30–60 min. Sampling systems that conform to ISO 8557:1985 for bauxite preparation and ISO 6609:1985 for total silicon determination are used to validate the online analyzers. In addition, caustic concentration measurements by automatic titration following ASTM E291-18 provide the required Na₂O concentration input to the ratio controller. The reliability of these analytical inputs is at least as important as the flow-control hardware; a caustic assay bias of 2 g/L Na₂O translates into a C/B error of roughly 10–15 kg/t when the liquor-to-bauxite ratio is 3.0–3.5 m³/t. Without this correction, two plants reporting the same C/B may be operating at completely different effective free-caustic conditions.

When a High-Temperature Digestion Line Drops Below a Corrected C/B of 0.45

In high-temperature plants processing boehmite-diaspore blends, a corrected C/B below 0.45 kg Na₂O per kg dry bauxite creates a processing window narrower than ±5 °C for sustained boehmite extraction. Design calculations for continuous digesters indicate that at 240 °C and 0.45 kg/kg corrected C/B, boehmite extraction can fall from 92–95% to below 80% if the digestion temperature oscillates by more than 5 °C, because the available free caustic after desilication is insufficient to buffer kinetic losses. The undigested boehmite reports to red mud and raises mud solids loading; downstream thickener feed slurry apparent viscosity increases from 30–60 mPa·s to 300–800 mPa·s at 45–55 wt% solids, depending on particle size distribution and residual organic matter. This viscosity rise can reduce rake drive torque margin in high-compression thickeners to less than 10% of rated torque, leading to rake overload alarms and occasional descaling shutdowns. Steam consumption also becomes unstable: a low C/B condition reduces the liquor-to-solid mass ratio and increases slurry apparent viscosity, causing uneven heat transfer in shell-and-tube digesters and local cold spots on the tube side. Published data for this specific configuration is limited, but available heat-transfer design correlations predict that a C/B excursion of 0.05 kg/kg below setpoint can produce a 7–12 °C spread in tube outlet temperatures and a 15–20% increase in differential pressure across the digester discharge manifold in parallel-pass tubular vessels. The control response must therefore include not only ratio correction but also temperature compensation and viscosity-aware slurry pumping interlocks. Operating limits are typically enforced by a DCS ratio controller with a deadband of ±10 kg/t, a high-temperature interlock at 265 °C for boehmite-diaspore service, and a low-density trip on bauxite slurry at 1.15 g/cm³ inferred from a nuclear density gauge.

On the control hardware side, the C/B loop is implemented as a three-element ratio cascade. A Coriolis mass flowmeter on the spent liquor line reports mass flow with a published accuracy of ±0.1 to ±0.5% of rate; a magnetic flowmeter on the bauxite slurry line reports volumetric flow with ±0.5% of rate for velocities above 1 m/s. Dry bauxite mass flow is inferred from slurry volumetric flow and slurry density measured by a nuclear density gauge with an uncertainty of ±0.002 g/cm³. The mass-flow computation is corrected for bauxite moisture by a periodic moisture analyzer conforming to ISO 8557:1985 sample preparation and a gravimetric moisture method with ±0.1 wt% repeatability. The caustic concentration input is provided by an automatic titration station using ASTM E291-18 reagents and calibration buffers; the titration result updates the spent liquor flow setpoint every 5–10 min. The DCS ratio controller multiplies dry bauxite feed by the desired C/B setpoint and divides by the measured Na₂O concentration to produce the required spent liquor mass flow. A deadband of ±10 kg Na₂O per tonne dry bauxite prevents valve hunting when bauxite feed varies by less than 2% of full scale. The output drives a variable-speed centrifugal pump discharging into the predesilication tank; a minimum recirculation line protects the pump below 30% of best-efficiency-point flow. Because bauxite slurry is rheopectic and settles rapidly, slurry lines are designed for minimum velocities of 1.5–2.5 m/s, and loop control is sluggish due to long deadtimes from slurry storage and predesilication. The controller must therefore include feedforward compensation from bauxite moisture and reactive silica analyzer updates; feedback alone is insufficient when a bauxite blend change reaches the digester 30–45 min later.

Caustic-to-Bauxite Ratio, Liquor-to-Bauxite Ratio, and Alumina-to-Caustic Ratio Interdependence

The C/B cannot be managed independently from the liquor-to-bauxite ratio (L/B) and the alumina-to-caustic ratio (A/C). The relationship is defined by C/B = L/B × C_Na₂O, where L/B is cubic metres of digestion liquor per tonne dry bauxite and C_Na₂O is total Na₂O concentration in kg/m³. At a fixed C/B setpoint, increasing L/B requires decreasing caustic concentration, and vice versa. A high L/B of 7–8 m³/t can maintain the same C/B as a low L/B of 3–4 m³/t only if the sodium oxide concentration is reduced substantially, but dilution below 150 g/L Na₂O usually reduces digestion kinetics and affects downstream evaporator steam economy. The A/C ratio at digester discharge is set by the amount of alumina extracted and the residual caustic; in gibbsitic plants the discharge A/C is typically 0.60–0.70, while high-temperature boehmite-diaspore lines may discharge at 0.50–0.60 due to lower solubility of boehmite in spent liquor. These three ratios are linked through the aluminium and sodium mass balance, but they are not interchangeable: C/B is a feed-forward operating variable, A/C is a process state variable, and L/B is a hydraulic throughput variable. A common control error is to maintain C/B by raising L/B while allowing caustic concentration to drift below the kinetic threshold; this generates false security because the total mass of caustic per tonne of bauxite is correct, but the liquor phase is too dilute for the target dissolution rate. The digester then runs with a higher than expected residual alumina content in red mud and a lower than predicted discharge A/C. Conversely, holding L/B low while raising caustic concentration to maintain C/B increases density and viscosity of digestion liquor and can overload the flash train pressure let-down valves. The published operating practice in modern refineries is therefore to maintain L/B within a narrow band and trim caustic concentration independently through evaporation and spent liquor blending.

Excess Caustic in Parallel Digestion Trains Shifts Thermal Load and Scale Formation

At corrected C/B values above 0.70 kg/kg in high-temperature plants, the primary penalty is not alumina extraction but thermal inefficiency and sodium aluminosilicate scaling. Excess free caustic increases the solubility of silica in the digester, and as the slurry is cooled in flash vessels, silica precipitates onto tube walls and flash interstage liners as sodalite-type scale. The scaling rate in flash trains has been observed in published plant trials to rise sharply when flash feed free Na₂O exceeds 230 g/L and the cooling gradient exceeds 90 °C across the first two stages; scale thickness can reach 1–3 mm in 30 days of continuous operation under these conditions. Thermal load also increases because the additional caustic must be heated to digestion temperature and cooled after digestion, and the excess sodium hydroxide leaving digestion must be regenerated through evaporation and causticisation. For a digestion line operating at 250 °C, each additional 50 kg Na₂O per tonne bauxite above the mineralogical requirement adds roughly 70–100 MJ/t bauxite of sensible and reaction heat, depending on heat recovery efficiency from flash steam. In parallel digestion trains, excess caustic can mask uneven bauxite feed distribution: the train receiving the highest bauxite slurry flow may still extract alumina because free caustic is high, while the parallel train operates with a lower liquor-to-bauxite ratio and develops zone-specific scaling. The resulting differential pressure across digester tube banks becomes asymmetric, and the train with higher scaling requires more frequent acid descaling. Industrial practice therefore limits corrected C/B to the upper quartile of the bauxite-specific window and uses a DCS high-caustic interlock at 255 g/L Na₂O in digestion liquor to protect downstream heat recovery. The interlock is typically interlocked with the bauxite feed control and spent liquor dilution water valve to prevent sustained operation above the threshold.

Implementation of a C/B control strategy in an existing refinery requires a structured calibration hierarchy. The bauxite moisture analyzer is validated against gravimetric moisture using ISO 8557:1985 sample preparation; total caustic titration is standardized against ASTM E291-18; and reactive silica is cross-checked by ISO 6609:1985 total silicon determination with a subtractive correction for quartz. Online X-ray fluorescence analyzers are calibrated with fused bead standards prepared according to ISO 12677:2011 for oxide analysis, and particle size distributions of ground bauxite are measured by sieve methods following ISO 2591-1:1988. The ratio controller is only as accurate as the least stable input; published reliability data for nuclear density gauges show drift rates of 0.001–0.003 g/cm³ per month, requiring weekly calibration with process water and slurry samples of known density. Coriolis meter calibration in spent liquor service is verified against a master meter traceable to national flow standards; the acceptance criterion is ±0.2% mass flow deviation over the range 40–120% of maximum flow. When these instruments are combined, the combined C/B measurement uncertainty is usually 2–4% of setpoint under steady feed conditions, but degrades to 5–8% during bauxite blend transitions because of moisture and reactive silica transients. The control system must therefore be tuned for a closed-loop settling time of 15–20 min with minimal overshoot to avoid over-injection of spent liquor during a moisture spike. If the online reactive silica analyzer fails, the C/B loop should be transferred to a fixed conservative setpoint corresponding to the highest reactive silica bauxite currently in the feed blend; continuous operation on a fixed setpoint without blend correction risks either caustic insufficiency or excessive scaling, depending on the direction of the assay bias.

Published data for very high reactive silica ores above 12 wt% and for bauxite containing mixed diaspore-hematite with finely disseminated silica are limited; in such configurations, a fixed C/B setpoint is not sufficient. The control scheme must include continuous or high-frequency reactive silica analysis and a dynamic correction to the C/B setpoint because the caustic consumption term changes with particle size distribution and predesilication residence time. The safe operating boundary is also constrained by refinery-specific materials: spent liquor with sodium oxalate above 3 g/L and organic carbon above 25 g/L can change slurry rheology independently of C/B, and the digestion temperature must be reduced if the flash train cannot handle the additional vapor load from excess caustic flow. In operations where bauxite moisture varies by more than ±3 wt% within a shift, the C/B controller should be supplemented by a dry-solids mass flow meter rather than relying solely on density correction. Avoid combining high free caustic above 240 g/L Na₂O with high organic carbon liquors derived from low-temperature gibbsite plants, because sodium oxalate co-precipitation in heat exchangers is accelerated under those conditions and the resulting scale is more resistant to acid cleaning.