Question 8 of 9 Intermediate Scenario
A downstream purification step loses recovery only at production scale. How would you investigate?
Short answer
A purification recovery loss appearing only at manufacturing scale may originate in loading, fluid dynamics, residence time, hold conditions, materials or analytical accounting. Localize the step before redesigning the process.
Why
Scale-up changes bed height, column packing, superficial velocity, pressure, mixing and product exposure to interfaces. Yield calculated from concentration alone is misleading unless volumes and all fractions are measured. Product loss may be adsorption, aggregation, degradation or simply mass-balance error.
How to work it through
- Build a complete mass balance: feed, flow-through, washes, eluate, strip, filter or hold-up and samples with assay uncertainty.
- Compare binding capacity and breakthrough, pressure traces, residence time, conductivity/pH, filter fouling, hold times and small-scale representativeness.
- Test material-contact adsorption and analytical recovery.
- Make focused changes under change control with quality-attribute assessments.
An example
At pilot scale, an affinity column elutes 85% recovery. Production scale reports 62%. Product quantification in the flow-through reveals early breakthrough from incorrect bed packing, while total recovery reconciles when all fractions are included.
Where the answer stops holding
Assay interference by buffers can give false apparent losses. Improved yield can worsen purity or aggregation. A successful investigation checks critical quality attributes, not yield alone.
A strong answer covers
- A purification recovery loss appearing only at manufacturing scale may originate in loading, fluid dynamics, residence time, hold conditions, materials or analytical accounting
- Localize the step before redesigning the process
- Build a complete mass balance: feed, flow-through, washes, eluate, strip, filter or hold-up and samples with assay uncertainty
Common mistakes
- Scaling volumes linearly without considering residence time
- Ignoring filter dead volume
- Altering validated operation before documenting impact
The interviewer might ask next
How would you distinguish irreversible adsorption from aggregation in a filter train?