Drying Mass Balance

For any drying step, the input mass minus the output product mass should equal the moisture removed plus any documented sampling loss. A drift outside the validated band signals one of four things: a leak (gasket, valve), a container loss (residual product in the dryer or transfer line), a sampling-mass mistake (un-weighed samples, lost samples), or an instrument problem (load-cell drift, tare error). Each is a quality event that the balance forces into the open within minutes of cycle end, rather than weeks later in a CPV review.

Mass balance is independent of moisture endpoint and uniformity. A batch can pass LOD and still fail mass balance, indicating a quality concern that endpoint cannot detect. Conversely, mass balance can be in-spec while LOD fails, indicating a drying-kinetics issue rather than a containment issue. The two controls complement each other — together they cover the failure modes that an inspector cares most about.

• Required by 21 CFR 211.103 for yield reconciliation; ICH Q7 §8.13 mirrors the requirement for APIs.
• Validated tolerance typically ±2–5% depending on scale and dryer type.
• Condensate measurement adds direct evidence of solvent removed.
• Mass balance is independent of endpoint — catches problems endpoint cannot.
• Drift over time indicates equipment maintenance need (gasket wear, transfer-line build-up).

For vacuum and tray dryers, condensate is collected in a receiver tank that is weighed empty and full. For fluid bed dryers, condensate is not collected — moisture leaves with the exhaust. The mass balance is then computed by integrating exhaust airflow × (exhaust humidity − inlet humidity) over the cycle time. Most modern fluid bed control systems have this calculation built in and post the result to the batch record automatically.

21 CFR 211.103 requires that 'actual yields and percentages of theoretical yield shall be determined at the conclusion of each appropriate phase of manufacturing, processing, packaging or holding'. Drying is unambiguously such a phase. ICH Q7 §8.13 extends the same requirement to active pharmaceutical ingredient manufacture and is explicit that 'discrepancies should be investigated.' The expected tolerance is product- and process-specific, justified in the validation report; tighter tolerances are easier to set on closed (vacuum, lyo) systems than on open (fluid-bed, tray) systems.

FDA warning letters routinely cite missing or perfunctory yield reconciliation: tolerances set so wide that any imaginable variance fits, or yields outside tolerance simply approved without investigation. Inspectors look for the documented investigation trail when reconciliation fails — the deviation report, the root-cause analysis and the CAPA. A batch record showing repeated near-tolerance excursions without investigation is a classic finding.

• Calibrate all weighing devices before campaign and verify daily.
• Record wet charge and dry product mass in the batch with operator attribution and timestamp.
• For vacuum drying, record condensate receiver weight at end-of-cycle.
• For fluid bed, integrate exhaust humidity if instrumented; otherwise the balance is closed via dryer + transfer-line losses estimated from validation data.
• Record every in-process sample mass on the batch record.
• Investigate any reconciliation outside validated band — the band is set on data, not on tolerance for laziness.
• Tare the discharge container empty before each batch; do not assume tare from previous use.
• Inspect dryer interior for residual product before declaring discharge complete.

• No condensate weight recorded — missing the strongest piece of corroborating evidence.
• Tare not verified — silent bias of several kilograms in large batches.
• Sampling losses not counted — apparent imbalance every batch, dismissed as 'sampling'.
• Single-point weighing of dry product — variation from container residue not captured.
• Reconciliation not tied to release decision — only checked for QA paperwork retrospectively.
• Tolerance set on tradition rather than data — band so wide that real problems hide.
• Manual transcription of weights — typos and reversed digits create apparent imbalances that operators 'correct' to look right.
• Reconciliation expressed as % only — absolute mass discrepancies more easily detected than ratios.

Mass balance tolerance scales inversely with batch size in absolute terms but is usually expressed as a percentage. A 5% tolerance on a 10 kg pilot batch is 0.5 kg — plenty for the precision of a single load cell. The same 5% on a 1000 kg commercial batch is 50 kg, which is enough material to ship and is therefore not an acceptable tolerance for product-quality protection. Most commercial sites tighten the percentage tolerance as scale increases, sometimes to ±1–2%.

Load-cell selection matters. A dryer load cell with 0.5% full-scale accuracy reading at 30% of range can drift the absolute reading by 1.5% of charge. Verification weights traceable to a national standard are needed to detect this. The PM programme should include load-cell verification at fixed frequency and after any service event; the verification record should be in the dryer's PM file, not lost in a spreadsheet.

• API plants — mass balance critical for ICH Q7 yield records, often tied to product cost-of-goods analysis.
• Solid-dose granulation / drying — closes the granulation line balance; feeds into compression-step yield.
• Veterinary medicated articles — yield reconciliation per CVM regulations, mirroring 211.103.
• Lyo biologics — fill mass minus residual moisture vs condenser load is the canonical lyo mass balance.
• Specialty chemical drying — mass balance is the universal data-integrity backbone in fine chemicals.
• Food spray-drying — moisture-in vs moisture-out balance tracks dryer health and energy efficiency.
• Cannabis biomass drying — mass balance + cannabinoid recovery balance underlie regulator-friendly traceability.

The single batch mass balance is one data point; the campaign-wide trend is where mass balance earns its keep. A site that trends mass-balance closure batch-over-batch will see a slow shift — say, from 99% to 98% to 97% closure — well before any individual batch falls out of tolerance. The cause is usually mechanical: gasket wear allowing vapour escape, transfer-line build-up retaining product, or load-cell drift. Caught in CPV, the fix is a PM or a calibration; caught only at a single OOS batch, the fix may be a 6-month root-cause investigation across many batches that all individually 'passed'.

Statistical control limits on mass-balance closure are now standard expectation. Western Electric rules or EWMA against a validated baseline detect drift early. The output of the CPV system is a maintenance signal, not a release decision; an in-control trend approaching the action limit triggers a PM, not a deviation.