Yield-Adjusted Batch Size

Yield-Adjusted Batch Size is the planned batch size, re-scaled before any dispense weighing begins, when the full nominal size on the MMR cannot or should not be made for this run. Five legitimate triggers: (1) input shortage — only enough of a critical raw material on hand for a 70% run; (2) vessel-capacity ceiling — the requested batch would exceed the mixer's working volume because a new excipient is lower density; (3) sub-potent active that, after PF + LOD + SBF recalculation, would push the counter-balance excipient negative at full size but lands positive at 60% size; (4) equipment downtime — one of two parallel mixers is offline so the batch is halved to fit the remaining equipment's schedule; (5) deliberate scale-down for a clinical, stability, or validation lot.

It is not the actual yield (that's measured at the end), not the theoretical yield (that's the nominal MMR target), and not a yield-loss cover (that's overage). It is the new top-of-batch denominator that every other adjusted target then derives from. The scale factor = adjusted size ÷ nominal size, and every formula row — active, excipient, counter-balance, processing aid — is multiplied by the scale factor before any per-lot correction (PF, LOD, SBF) is applied.

Scale-adjustment is allowed only inside the validated batch-size range declared on the MMR. That range is the product of process-validation work under 21 CFR 211.100 + Annex 15 — typically expressed as a min/max kg (or L) pair and sometimes with discrete validated sizes (e.g. 100 kg, 250 kg, 500 kg only). Inside the range, scale-adjustment is a routine release decision with a two-person e-sig. Outside the range — even by 1% — it is a process change requiring change control, re-qualification, and (depending on regime) regulatory notification.

Scale factor = adjusted batch size ÷ nominal batch size. Every formula row is multiplied by the scale factor before any per-lot correction is applied. The order of operations matters: scale first, then PF / LOD / SBF / counter-balance, then dispense.

A yield-adjusted release is a deliberate decision, not a kiosk override. It is documented in the BMR with: the trigger (input shortage / equipment / vessel ceiling / sub-potent active / planned scale-down), the requested size, the scale factor, the validated range from the MMR, confirmation the requested size is inside the range, the scaled formula table, the IPC sample-mass re-check (still inside the validated IPC range), the fill-volume re-check, the mixing-vessel fill re-check, and a two-person e-signature under 21 CFR 11.50 + 11.70.

• Preparer (production supervisor or planner) — proposes the scale, documents the trigger, attaches the scaled formula table.
• Reviewer (independent QA or designated production reviewer) — confirms validated-range coverage, IPC / fill-volume / vessel-fill re-checks are inside their respective validated windows, e-signs the release.
• Both signatures are Part 11 e-sigs (link to user, link to record, link to timestamp, link to meaning — 'authorised yield-adjusted release at 180 kg').
• The scaled MMR is snapshotted into work_orders.mmr_snapshot at release; dispense, IPC, fill, and reconciliation all run against the snapshot, never against live MMR.

1. Recalculate first, then scale — operator (or upstream system) applies PF / LOD / SBF to the nominal MMR figures, then scales the recalculated numbers; counter-balance derivation breaks; total batch mass drifts; multi-batch accumulation produces a systematic under- or over-dose pattern that surfaces months later in yield reconciliation trending.
2. Scale outside validated range without change control — input shortage forces a run at 120 kg against a 150–250 kg validated range; production decides 'we'll just adjust the time / speed / fill' without re-qualification; the resulting batch has unvalidated process parameters, and even if CQAs pass, the batch is a deviation under §211.100.
3. Scale without IPC re-check — IPC sample mass is typically 50–250 g taken at a specific point in the process; at a different batch size, the same absolute sample mass represents a different fraction of the batch, which can fall outside the validated IPC range. Recurring 483 area when IPC procedures don't carry the scale-sensitivity check.
4. Shared-vessel scale-up violation — plant decides to scale up from 180 kg to 220 kg to use spare capacity but the mixer's working-volume ceiling is 200 kg at the formulation's density; mixing is incomplete; granulation fails.
5. Partial-scale e-sig missed — production supervisor changes scale at WO release but the reviewer e-sig is captured against the original (nominal) scaled formula table; the BMR carries a mismatch between authorising e-sig and what was actually dispensed.
6. Scale-adjustment used to mask yield-loss trend — repeated 'scale-adjustment for input shortage' across batches actually masks a yield-loss trend in upstream operations; the unscaled batches would show the yield-loss visibly, the scaled batches close on the new (lower) theoretical, and the trend stays invisible until quarterly yield-reconciliation review surfaces it.

• Yield-adjusted batch coverage % — fraction of WO releases at non-nominal scale that ran through the full YA release contract (two-person e-sig + validated-range confirmation + IPC re-check) (target 100%).
• In-validated-range scale % — fraction of YA batches whose requested size fell inside the validated range (target 100%; outside-range cases must run through change control, not YA release).
• Scale-first compliance % — fraction of WO snapshots whose scaling was applied before per-lot recalculation (target 100%; checked by comparing snapshot order in the audit trail).
• IPC re-check compliance % — fraction of YA batches where the IPC sample-mass re-check is captured in the BMR (target 100%).
• Fill-volume re-check compliance % — fraction of YA liquids batches where the fill-volume re-check is captured in the BMR (target 100%).
• Vessel-fill re-check compliance % — fraction of YA batches where the mixing-vessel fill at the scaled size has been confirmed against the validated tip-speed window (target 100%).
• YA two-person e-sig compliance % — fraction of YA WO releases with both preparer + reviewer e-sigs under Part 11 §11.50 + §11.70 (target 100%).

1. MMR schema: validated_batch_size_min, validated_batch_size_max, and validated_discrete_sizes[] are first-class fields on the master record; the MMR-approval e-sig flow requires these to be populated.
2. WO planning: the planner can request any size, and V5 immediately surfaces validated-range coverage — green if inside the range / inside a discrete validated size; amber with a 'change control required' callout if outside.
3. WO release gate: a non-nominal scale requires the YA release contract — trigger reason picker (input shortage / equipment / vessel ceiling / sub-potent active / planned scale-down / other-with-justification), scaled-formula table preview, IPC re-check, fill-volume re-check, vessel-fill re-check, and two Part 11 e-sigs (preparer + reviewer, two distinct users).
4. Snapshot order enforcement: scaled MMR is computed first and locked into work_orders.mmr_snapshot before any per-lot correction is applied; the dispense engine runs PF / LOD / SBF / counter-balance against the snapshot, never against live MMR.
5. Kiosk hard-block: kiosk operators cannot scale unilaterally — there is no scale field on the dispense screen. Scale is a release-time decision, full stop.
6. Counter-balance integration: counter-balance derivation runs against the scaled formula, not the nominal formula; the negative-counter-balance check fires against the scaled figure, not the nominal one.
7. IPC integration: in-process tests run with sample-mass + timing recomputed for the scaled size; IPC acceptance criteria hold; out-of-range sample mass blocks the IPC.
8. BMR rendering: the BMR header shows nominal MMR size, requested scaled size, scale factor, trigger reason, validated range, and the two YA e-sigs alongside the standard MMR-snapshot e-sig.
9. Yield reconciliation: closes against the scaled theoretical, not the nominal theoretical; trending separates scaled vs nominal yield trajectories so an upstream yield-loss trend masked by scaling becomes visible quickly.
10. Quarterly review: ICH Q10 §3.2.5 PQS management review surfaces the YA-batch report — total count, distribution across triggers, validated-range coverage, and trending — so management sees scale-adjustment patterns and either re-qualifies a wider range or addresses the upstream supply trigger.