In Bayer refineries processing low-grade bauxite ores where the reactive silica (RxSiO₂) content exceeds 6 wt% on a dry basis, the specific caustic soda consumption—measured as kilograms of Na₂O per tonne of calcined alumina—frequently surpasses 70 kg/t and can approach 120 kg/t when the bauxite contains abundant kaolinite and moderate boehmite. The dominant irreversible sink for NaOH is the formation of sodium aluminosilicate hydrate phases, commonly grouped under the term desilication product (DSP), with stoichiometries approximating Na₆[Al₆Si₆O₂₄]·2NaOH·xH₂O for sodalite-type structures in the absence of lime. Continuous digestion circuits, particularly high-temperature tube digesters operating at 240–280°C and 35–65 bar, afford a degree of spatial control over the temperature–time trajectory that batch autoclaves cannot replicate. This control directly influences the rate and extent of quartz dissolution, the phase evolution of DSP, and the partitioning of alumina and caustic into process losses. The discussion that follows examines critical mechanisms, equipment-specific experience, and threshold process windows that determine whether continuous digestion mitigates or exacerbates caustic loss when treating silica-rich, low-grade bauxites.
While batch digestion imposes a single temperature plateau after a finite heat-up ramp, a plug-flow tube digester subjects the slurry to a precisely managed temperature gradient—typically an initial preheating zone from 90°C to 220°C in 4–7 minutes, a hold zone at the target digestion temperature for 15–25 minutes, and a flash-cooling stage. The desilication reaction between dissolved silicate species and sodium aluminate liquor exhibits an activation energy of approximately 85–100 kJ/mol for DSP formation in the absence of lime; thus, the extent of caustic consumption is extremely sensitive to the time integral of temperature. In continuous systems, the residence time distribution (RTD) deviates from ideal plug flow due to pipeline roughness, bends, and steam injection turbulence, yielding a RTD variance quantified by the axial dispersion coefficient. Published tracer studies on industrial-scale tube digesters (inside diameter 0.18 m, total heated length 950 m) indicate a Peclet number (Pe = uL/Dₐₓ) of 80–150, corresponding to an RTD standard deviation of 10–14% of the mean residence time. This narrow dispersion ensures that 95% of the slurry volume experiences the digestion temperature within a ±3°C band, whereas batch autoclaves frequently exhibit temperature gradients of ±8°C within the vessel due to incomplete agitation. The consequence for low-grade bauxite processing is a more uniform conversion of reactive silica to DSP, which can prevent localized under-digestion zones where undissolved silica later precipitates as scale in downstream heat exchangers—a loss mechanism that is often underestimated in caustic consumption accounting. However, the same tight RTD also means that the average kinetic driving force for silica precipitation remains near the maximum throughout the hold zone, potentially increasing the total DSP yield if the liquor’s dissolved silica concentration exceeds the equilibrium solubility at the hold temperature too quickly. Therefore, the net caustic consumption is dictated by the interplay between pre-desilication efficiency, the timing of lime addition, and the position of the silica concentration peak along the tube length. Measurements in refineries processing boehmitic bauxite with 8% RxSiO₂ have shown that shifting the point of lime slurry injection from the tube inlet to a location 10% downstream of the preheater exit can reduce caustic loss by 2.8 kg Na₂O/t Al₂O₃ because it retards DSP precipitation until after the bulk of alumina has been extracted, thereby lowering the effective silica supersaturation during the initial dissolution phase. These data, obtained from plant trials using segmented sampling ports and on-line FTIR-liquor analyzers, underscore the importance of spatial control unique to continuous flow digestion.
| Bauxite Mineralogy | RxSiO₂ (wt% dry) | Digestion Temperature (°C) | Mean Residence Time (min) | Lime Dosage (% CaO dry bauxite) | Specific Na₂O Consumption (kg/t Al₂O₃) | Caustic Loss Determination Standard |
|---|---|---|---|---|---|---|
| Gibbsitic, high kaolinite | 8.2 | 145 | 18 | 3.8 | 78 | ASTM D3872-05(2019) / mass balance |
| Boehmitic, moderate quartz | 7.4 | 250 | 22 | 5.2 | 98 | ASTM D3872-05(2019) |
| Boehmitic–diasporic mixed | 9.1 | 268 | 28 | 7.0 | 115 | ISO 18894:2006 liquor balance |
| Diasporic, high quartz | 11.5 | 275 | 35 | 8.3 | 132 | ISO 18894:2006 |
Lime slurry injection at the tube digester preheating stage is practiced in the majority of continuous circuits treating low-grade bauxite because the conversion of reactive silica to a calcium silicate hydrate phase (Ca₃Si₂O₇·3H₂O) rather than sodium aluminosilicate DSP consumes one mole of CaO per mole of SiO₂ while sparing NaOH. The stoichiometric lime requirement calculated from ISO 6606:2008 reactive silica determination is typically increased by an excess factor of 1.2–1.5 to account for the formation of calcium carbonate from residual organic carbon and to promote a distinct crystalline phase that resists re-dissolution. On a 1200 m tube digester processing boehmitic bauxite at a feed rate of 65 t/h dry solids, the addition of lime at 6.3 wt% CaO on dry bauxite via an in-line high-shear mixer (Netzsch M-Ovas type, tip speed 40 m/s) has been correlated with a caustic consumption reduction from 105 kg/t to 93 kg/t alumina when the mixer is located precisely 18 m after the last preheat steam sparger. This positioning ensures that the lime is dispersed into the liquor when the dissolved silica concentration has already reached a local maximum of 3.8 g/L SiO₂, as verified by automated XRF slurry analyzers. Deviating the injection location by as little as 5 m upstream results in a caustic penalty of +3.5 kg Na₂O/t due to premature calcium consumption by alumina-rich liquor forming calcium aluminate hydrates that later revert and re-release caustic in the flash tanks. The operational window is therefore constrained by both the silica dissolution profile and the need to avoid unreacted CaO surviving into the settlers where it causes blinding of filter cloths; this requires that the filter cake free CaO content, measured by ASTM C25-19 rapid sugar test, remains below 0.5 wt%.
Bauxite feeds with reactive silica exceeding 10 wt% on a dry basis and a boehmite content above 15% impose a condition where the lime dosage required to suppress DSP-based caustic consumption would exceed the solubility limit of Ca(OH)₂ in concentrated caustic liquor at digestion temperature—roughly 8–12 g CaO/L in 200 g/L Na₂O solutions at 260°C. Adding lime beyond this limit leads to undispersed solids accumulating in the tube digester bends, which is documented as a primary cause of localized overheating and catastrophic tube rupture in refinery operations. In one North American refinery, processing a blend of 60% Jamaican high-silica bauxite with 12.8% RxSiO₂, an attempt to maintain lime addition at 9.5% CaO resulted in a measured pipeline scale layer of cancrinite-type DSP with incorporated excess CaO reaching 4 mm thickness after 180 days of continuous operation, causing the pressure differential across the digester to increase from 2.1 bar to 6.8 bar. This scale not only reduces the effective cross-section, distorting the RTD and expanding the residence time tail, but also permanently traps caustic in the form of hydroxycancrinite, raising total Na₂O loss by 11 kg/t alumina above the pre-scale baseline. The refinery was forced to implement an alternating lime injection strategy, where for 72-hour cycles the lime dosage was dropped to 6.0% while the digestion temperature was raised by 5°C to 265°C to cope with the temporarily higher caustic consumption through greater boehmite extraction—a practice that, in turn, raised the risk of boehmite recrystallization in the flash tanks. These operational constraints illustrate the delicate equilibrium that continuous digestion must maintain, where the maximum tolerable lime rate is not a fixed stoichiometric figure but a function of real-time pressure drop monitoring across the digester tube, correlated with on-line ultrasonic thickness measurements per ASTM E797/E797M-21.
Continuous tube digesters designed with a length exceeding 800 m and internal diameter of 0.2 m at slurry Reynolds numbers of 5 × 10⁴–1 × 10⁵ achieve near-plug flow, which is beneficial for avoiding short-circuiting of unreacted silica, but it simultaneously creates a condition where the liquor’s alumina concentration reaches supersaturation with respect to boehmite earlier in the digestion cycle than in a well-mixed batch autoclave. When the slurry exits the final heating zone and enters an adiabatic flash vessel, the rapid pressure drop from 50 bar to atmospheric pressure causes a temperature quench of 105–110°C in under 3 seconds, creating a massive supersaturation driving force for boehmite precipitation. The resultant fine boehmite particles (0.2–1.5 µm as measured by dynamic light scattering per ISO 22412:2017) have a high specific surface area and occlude caustic within their agglomerates, with thermogravimetric analysis indicating that such occluded Na₂O can account for 2–5% of the total soda loss when the flash exit temperature is below 108°C. In continuous operation, the extent of this re-crystallization is strongly influenced by the steepness of the temperature drop, which is dictated by the flash vessel nozzle design and the back-pressure setpoint. Refineries using a two-stage flash with intermediate pressure hold at 12 bar have demonstrated a caustic loss reduction of 4.2 kg Na₂O/t relative to single-stage atmospheric flash, because the intermediate depressurization allows a more controlled release of supersaturation and reduces the instantaneous nucleation rate. This effect becomes critical when processing low-grade bauxites where the pregnant liquor A/C ratio (alumina-to-caustic) is already depressed to 0.58–0.62; any additional boehmite loss further lowers productivity, indirectly raising the caustic consumption per tonne of recovered alumina. Plant data from a refinery in the Urals, operating a 700 m tube digester on diasporic bauxite, indicated that replacing a single flash nozzle with a multi-hole distributor at the first flash vessel inlet reduced caustic occlusion loss from 8.1% to 5.6% of total Na₂O input, as measured by a ASTM C114-18 soda balance over a 12-month campaign.
| Parameter | Acceptable Limit | Test Method | Impact on Digestion Caustic Efficiency |
|---|---|---|---|
| Total alkalinity (as NaOH) | 50 ± 0.5% | ISO 3195:2020, ASTM D3872-05(2019) | Lower concentration demands higher volumetric feed, altering RTD |
| Chloride (Cl⁻) | < 100 mg/kg | ISO 3195:2020 Annex B | Chloride enters DSP structure as sodalite, increasing Na₂O loss per unit SiO₂ |
| Sulfate (SO₄²⁻) | < 200 mg/kg | ASTM D516-16 | Sulfate stabilizes noselite phase DSP, enhancing caustic incorporation |
| Iron (Fe) | < 10 mg/kg | ISO 3195:2020 | Iron precipitates as Fe₂O₃ scale on heat exchanger, impairing heat transfer and requiring higher steam input that alters RTD |
| Carbonate (as Na₂CO₃) | < 0.3% | ISO 3195:2020 | Carbonate promotes cancrinate DSP with higher caustic payload per unit mass |
Continuous digestion facilities commissioned after 2010 increasingly integrate on-line caustic concentration monitoring using toroidal conductivity sensors (Emerson Rosemount 1056 model) with temperature compensation to 25°C reference, installed in the recirculating spent liquor line upstream of the digester feed pump. The measurement is correlated to grab samples analyzed per ASTM D3872-05(2019) and used as the primary input for a model-predictive controller that adjusts the bauxite feeding rate when the exit pregnant liquor A/C ratio deviates from the setpoint by more than 0.02 units. In a tube digester operating at 260°C on a boehmitic feedstock, a drop in the measured A/C from 0.64 to 0.61, if sustained for 20 minutes, signals either a slump in available alumina or an acceleration of DSP formation, both of which increase the specific caustic consumption. The controller responds by reducing the bauxite slurry feed rate by 5–8% while holding steam injection constant, effectively increasing the liquor-to-solid ratio and decreasing the driving force for silica precipitation. Field validation using ISO 21285:2019 reactive silica determination on quenched slurry samples collected at the tube exit showed that this intervention reduces the DSP-bound Na₂O by 1.9–2.4 kg/t over the following 1.5 hours of operation, although at the cost of a 2.3% drop in hourly alumina output. The permissible latency between sensor detection and feed rate adjustment is constrained by the digester’s transportation delay, which for a tube of 1200 m length at a linear velocity of 1.2 m/s is approximately 16.7 minutes; hence, the controller must incorporate a Smith predictor algorithm to maintain process stability, without which the system oscillates and creates alternating periods of over- and under-causticization that each contribute to elevated lime consumption and DSP formation. Operational records from an Australian refinery confirm that implementing such a predictive control loop reduced the standard deviation of daily caustic consumption from ±4.2 kg/t to ±1.7 kg/t over a 6-month period, with the improvement attributed almost entirely to the minimization of transient excursions outside the defined temperature-A/C window.
Low-grade bauxites often contain significant free quartz (2–8 wt%), which is not measured as part of the reactive silica fraction by standard ISO 6606:2008 but dissolves at rates that become appreciable above 200°C in concentrated caustic liquors. In a tube digester held at 250°C or higher, the dissolution of quartz releases monomeric silicic acid at a rate of approximately 0.08–0.12 g SiO₂ per m² of quartz surface per hour in 200 g/L Na₂O solutions, based on rotating disc electrode experiments reported in TMS Light Metals proceedings. Even though the total dissolved silica from quartz may represent only 10–15% of that from kaolinite, its release occurs continuously along the tube length and into the post-digestion flash circuit, where the lower temperature drives immediate DSP precipitation on heat exchanger surfaces. This post-digestion scaling does not consume caustic during the digestion step itself, but it represents a latent caustic sink because the scale must be periodically removed by mechanical reaming or acid washing, during which the occluded NaOH is neutralized and lost. For a tube digester processing bauxite with 5% quartz at a throughput of 80 t/h, the annual cumulative Na₂O loss attributable solely to quartz-derived DSP scale has been estimated at 180–240 tonnes of Na₂O per year, equivalent to an additional 2.2–2.9 kg/t alumina when amortized over production. Continuous digestion aggravates this mechanism because the unidirectional flow continuously feeds fresh quartz surfaces into regions of high temperature, whereas a batch autoclave reaches a terminal quartz particle size that limits further dissolution. To mitigate this, some tube digester configurations adopt a split-circuit approach: the first 300 m of the tube is lined with ceramic sleeves and operated at a reduced temperature of 195°C to dissolve kaolinite while minimizing quartz solubility, after which the slurry is heated rapidly to 270°C in the remaining 600 m. This staged temperature profile, monitored by multiple internal thermowells, has been shown to reduce quartz-derived scale thickness by 40% versus a uniform high-temperature profile, as quantified by ultrasonic thickness gauging per ASTM E797/E797M-21 during annual shutdown inspections. The penalty, however, is a slightly lower boehmite extraction efficiency (–1.8%) due to the shortened time at the maximum temperature, which indirectly increases caustic consumption per unit of alumina produced, a classic trade-off that defines the processing window for silica-rich low-grade bauxites in continuous circuits.
The tendency of caustic soda to co-precipitate with aluminium hydroxide during the precipitation stage—commonly termed “soda loss to hydrate”—is highly sensitive to the total organic carbon content that accumulates in closed Bayer circuits processing low-grade bauxites. Continuous digestion, by enabling a tighter control of the slurry’s exit temperature and flash conditions, allows for a more consistent decomposition of humic and fulvic acids into lower-molecular-weight organics, which has a direct bearing on the final supersaturation and agglomeration kinetics in the precipitators. When a refinery switches from a batch autoclave train to a tube digester, the steady-state organic carbon concentration in the spent liquor adjusts over a period of 6–9 months because the tube digester’s greater surface-area-to-volume ratio accelerates oxidative degradation reactions at 250–270°C. Plant data from a Chinese refinery treating guangxi bauxite with 9% RxSiO₂ demonstrated that, after a 10-month equilibration, the total organic carbon dropped from 23 g/L to 18 g/L, and the corresponding soda incorporation into the alumina trihydrate product, measured by photometric sodium determination per ISO 8038:2011 on calcined hydrate, fell from 0.35% Na₂O to 0.28% Na₂O. Although this improvement is not a direct caustic “consumption” in the digestion sense, it reduces the overall Na₂O requirement for the refinery, making the continuous digestion circuit’s net caustic efficiency superior by an estimated 3.5 kg Na₂O/t alumina beyond reductions attributable solely to silica reactions. The long transient underscores the need for plant-scale mass balance trials extending over multiple bauxite shipments before the full caustic consumption benefit of continuous digestion on low-grade materials can be confirmed.