Solubility assay sensitivity is 100%. If your assay catches only X% of problem compounds, discount the savings by that fraction.
Microfluidics damage is partly driven by problem compounds. The model assumes pre-filtering eliminates ~50% of fluidics-damage risk. The residual reflects wear, contamination, and operator error.
Catastrophic failures waste an average of half a chip. In the conservative scenario only the pre-failure batch is re-run; in the aggressive scenario the full chip batch is invalidated.
Instrument downtime, project delays, and protein supply are not priced. These factors typically strengthen the savings case further.