Remote evaluation of a THF production asset requires reading capacity not from a single nameplate line item but from a mass-balance-consistent set of equipment datasheets, process flow diagrams, utility logs, quality records, and historian trends. The phrase THF Mill & Plant Scale: How to Read Capacity Without Visiting defines an audit approach in which plant throughput is reconstructed from published process data, column dimensions, reboiler duties, batch cycle times, and storage turnover. Tetrahydrofuran is produced mainly by dehydration of 1,4-butanediol in the presence of an acid catalyst. The crude product forms a minimum-boiling azeotrope with water at approximately 94.6 wt% THF and 63.4 °C at 1 atm. This azeotrope dominates downstream separation design and makes distillation equipment a reliable capacity indicator.

Unit conversions are necessary before comparing vendor datasheets prepared in different regional units. A rating expressed in MM lb/year converts to kt/a by the factor 0.454; a plant rated at 100 MM lb/year corresponds to 45.4 kt/a. Volume-based ratings should be corrected by THF density, approximately 889 kg/m³ at 20 °C. A rating of 50,000 m³/year therefore corresponds to approximately 44.5 kt/a. Without such normalization, a remote audit can misread plant scale by the difference between mass and volume throughput.

Two process configurations are common: liquid-phase dehydration using a homogeneous acid catalyst and vapor-phase dehydration over a solid acid bed. In both, remote capacity analysis starts with the limiting reactor volume or catalyst bed volume and proceeds through the distillation train. A documented feed flow meter range must be corrected for conversion, selectivity, and annual operating hours. Published continuous THF train data often use an 8,000 h/year nameplate basis; actual capacity can be lower if catalyst regeneration or peroxide-safeguard shutdowns reduce uptime.

Limitations of nameplate throughput in THF dehydration trains

Nameplate throughput figures in kt/a are not self-validating. A plant rated at 45 kt/a may be hydraulically limited in the azeotrope column if the reboiler duty is insufficient to deliver the required boil-up at high reflux ratio. Capacity should be read from three coupled parameters: reactor feed rate, conversion across the dehydration reactor, and recoverable THF after water removal. A remote audit equation is Q = F × X × S × τ, where Q is annual THF output, F is the limiting BDO feed flow, X is BDO conversion, S is THF selectivity, and τ is effective operating hours. If F is 8,500 kg/h, X is 0.96, S is 0.98, and τ is 7,900 h/year, the resulting Q is approximately 63.2 kt/a. Discrepancies between this estimate and nameplate figures usually indicate a downstream purification bottleneck, a peroxide inhibitor limitation, or a utility constraint.

When a PFD or datasheet lists a fixed-bed vapor-phase reactor, the catalyst bed volume can be converted into a feed rate by multiplying volume by the design liquid hourly space velocity. A catalyst bed of 18 m³ with an LHSV of 0.4 h⁻¹ corresponds to a feed rate of 7.2 m³/h. Using BDO density near 1,000 kg/m³ at process temperature, this is roughly 7.2 t/h. Liquid-phase batch reactors are read differently: the working volume, jacket heat-transfer area, and condenser capacity set the cycle time. A glass-lined reactor with a nominal volume of 20 m³, working volume of 14 m³, and condenser duty of 2.1 MW can remove the exothermic heat of dehydration only up to the corresponding heat load; if reaction exotherm plus reflux load exceeds condenser capacity, cycle time lengthens. Remote capacity estimation should therefore treat condenser rating as a harder upper bound than reactor volume.

How are distillation column diameters converted into THF recovery capacity?

Distillation capacity in THF purification is governed by the azeotrope column and the downstream anhydrous THF column. For a packed column, the internal diameter determines cross-sectional area; vapor velocity at flooding is related to the capacity factor. A column with internal diameter 1.8 m has cross-sectional area 2.54 m². At a superficial vapor velocity of 1.2 m/s, volumetric vapor flow is 3.05 m³/s. If vapor density is 2.1 kg/m³, mass flow is 6.4 kg/s, or 23.0 t/h. At a reflux ratio of 2.5, net distillate flow would be roughly 6.6 t/h. This is a scale-calculation example, not a universal design value; packing type, liquid distributor fouling, and pressure drop constraints reduce achievable capacity.

If the reboiler duty is listed at 2.6 MW and the mixed THF-water latent heat is approximately 420 kJ/kg, the boil-up rate is 22.3 t/h. At a reflux ratio of 2.5, net distillate rate is approximately 6.4 t/h, aligning with the column diameter estimate. Remote capacity audits should require agreement between column hydraulic capacity and reboiler heat input. A PFD that shows a 2.6 MW reboiler but a column diameter of only 1.0 m indicates either a partial vapor load from a second source or an error in documentation.

Column differential pressure provides online evidence of capacity utilization. A packed azeotrope column operating at roughly 80% of flood may show a pressure drop of approximately 2.3 kPa/m for high-performance structured packing under total reflux, but published values vary with packing type and liquid load. A sustained pressure drop above 4.0 kPa/m usually indicates flooding or distributor malfunction, not additional capacity. Plant logs showing repeated high differential pressure at fixed reboiler duty indicate that the column cannot accept more feed without stability loss.

Batch systems require heat removal and solvent recovery checks

In batch THF processes, remote capacity is derived from isolated product volume, cycle time, and recovery yield. A batch record indicating 14 m³ of dry THF isolated per 12 h cycle corresponds to 1.17 m³/h THF production before purity adjustments. If the same train includes a water-THF recovery column that processes 18 m³ of aqueous distillate per batch, the THF losses can be cross-checked by the column mass balance. Batch-to-batch variance in the water content of the distillate, visible in Karl Fischer data, is a stronger predictor of capacity drift than reactor charge weight alone. A rise in water content above the azeotropic value without a pressure change indicates entrainment or flooding, not a true capacity increase.

Vendor datasheets frequently list jacket heat-transfer area. In a batch THF dehydration system, production rate can be inferred from the heat-removal equation Q = U × A × LMTD. A reactor with a jacket area of 32 m², an overall heat-transfer coefficient of 350 W/m²·K, and a temperature driving force of 40 K removes approximately 448 kW. If the reaction exotherm plus agitator power exceeds 448 kW, the batch cycle is heat-removal-limited. This is a frequent bottleneck in scaled-up THF batches. Remote auditors compare this heat-removal capacity with the condenser duty and the documented batch time.

THF peroxide formation imposes a capacity boundary that is often undocumented in PFDs. Uninhibited THF stored with oxygen ingress can accumulate peroxides; stabilizers such as butylated hydroxytoluene are often specified at 100–300 ppm. Remote audits should treat stabilizer consumption rate and storage turnover as operational evidence of actual throughput. A storage tank with 200 m³ usable volume and a documented turnover of 3 days implies an average daily withdrawal of 66.7 m³/day. Published data for inhibitor depletion rates under high-humidity storage conditions is limited, so the storage-turnover method should not be used as a sole capacity indicator.

When reboiler duty is not listed, utility loads and storage turnover apply

Condenser heat load can be estimated from cooling water or chilled water return data. The heat removal rate is m × Cp × ΔT. A cooling water stream of 120 m³/h with a return-temperature rise of 18 K removes approximately 2.51 MW; this can be compared with the distillation reboiler duty if the condenser is dedicated. Remote auditors should verify the flowmeter tag location and the cooling water network configuration. In some THF plants, the same cooling water header serves the BDO feed preheater, the vent condenser, and the product cooler; using total header temperature rise will overstate column capacity by 15–30%. Published data for specific THF plant utility networks is limited; capacity readings should therefore be bracketed as a range rather than a single value.

Equipment tags and vessel mechanical datasheets permit further remote verification. A distillation column tag with a design pressure of 0.5 MPa and design temperature of 120 °C in THF service is typically not capacity-limiting; the hydraulic internals and reboiler are. A reactor tag showing an emergency relief device sized for 15 t/h vapor release is not a direct throughput indicator but suggests the maximum credible vapor source used in relief calculations. Batch records with time-stamped valve operations from a distributed control system historian can provide cycle times without an on-site visit. The DCS trend of reactor temperature, condenser outlet temperature, and column differential pressure is often more informative than the nominal PFD capacity block.

Process historians provide capacity evidence that static datasheets cannot. A continuous dehydration reactor with a feed flow trend showing stable 7,900 kg/h over 30 days but periodic dips to 5,500 kg/h for catalyst regeneration yields an average output different from nameplate. The cumulative feed total over 90 days divided by 90 days gives a more reliable operating rate. Column differential pressure, reflux flow, and bottom temperature trends allow mass balance verification across the azeotrope column. If the reflux flowmeter is calibrated per ISO 5167-1:2022, the recorded volumetric flow can be corrected for density and temperature. Published data for specific DCS tag naming in THF plants is limited, but typical distributed control systems record flow, temperature, and pressure at 1 min intervals, sufficient for remote capacity reconstruction.

Catalyst deactivation in BDO dehydration reduces conversion and capacity. Remote review of laboratory reports showing BDO conversion falling from 96% to 88% over 12 months implies a throughput loss of roughly 8% at constant feed. If feed rate is raised to compensate, side-product formation may increase and overload the lights column. The capacity estimate must therefore be tied to catalyst age or cumulative feed per kilogram of catalyst, not only to the initial nameplate.

Capacity definitions should distinguish crude THF, unstabilized THF, stabilized THF, and polymer-grade THF. A remote capacity figure for a dehydration train may include recycled THF from the solvent recovery column, which inflates total throughput but not net make-up. Mass balance should subtract recovered THF inventory from gross column feed. For integrated BDO-THF plants, the recycle ratio can exceed 0.5; without this subtraction, remote auditors can overestimate net capacity by the recycle fraction.

Selected standards and documentation codes for remote auditing

Certification to ISO 9001:2015 does not quantify output, but clause 8.5.1 requires controlled process conditions, and audit records may contain the intended throughput range. Under REACH, the registered tonnage band for a THF producer can provide an upper capacity boundary; however, the registered tonnage band is often wider than actual plant output.

Remote capacity audits have explicit boundaries. They cannot detect undersized reactor agitation at high viscosity, localized hot spots in fixed-bed catalyst tubes, or leak paths that bypass the distillation column unless the DCS historian includes the relevant pressure and composition signals. THF is incompatible with strong oxidizers and can form explosive peroxides; storage and transport capacity claims that omit inhibitor type and concentration are incomplete. Published data for specific deactivation kinetics of acid catalysts in impure BDO feed is limited. Where process data is absent, the capacity reading remains a bracketed estimate derived from hydraulic and heat-transfer constraints rather than a guaranteed output.