NIR_Blend_Uniformity_Checker

URS & FS — User Requirements and Functional Specification

This document defines the controlled interface, validation behaviour and expected output of NIR_Blend_Uniformity_Checker. The document is intended to reside inside the utility folder together with input.csv and the description file. Expected executable: NIR_Blend_Uniformity_Checker.exe.

Scope

Domain limits and critical parameters

• • Mean API Content: 95.0–105.0%
• • Relative Standard Deviation (RSD): ≤5.0%
• • Maximum single point deviation: ≤10.0%
• • Minimum number of scans/samples: ≥30
• ⚠️ CRITICAL: RSD >5.0% → blend is non-homogeneous! Risk of segregation during tableting.
• NIR_Blend_Uniformity_Checker.exe input.csv output.json → evaluate your data
• BatchNumber,MeanContentPercent,StdDevPercent,NumberOfScans,MaxDeviationPercent,RSD_Percent
• • NIR allows in-line or at-line analysis without sampling (non-destructive).
• • Scanning multiple points (mapping) provides a statistically significant picture of API distribution.
• • Early detection of non-uniformity prevents scrapping the entire tablet batch.
• • Compliance with ASTM E2709 / ASTM E2810 and USP <1033>.
• ⚠️ CRITICAL:
• • RSD ≤5.0% — generally accepted criterion for good homogeneity in dry blends.
• • Mean content must match formulation (Target 100%).
• • Maximum Deviation (Max Dev) shows the worst point in the blender. If high, local segregation exists.
• • Number of scans (N) ≥30 is necessary for representativeness across the blender volume.

Pharmacopoeial and regulatory references found in the description

Production data flow

URS — User Requirements Specification

input.csv contract

BatchNumber,Mean%,StdDev,N,MaxDev,RSD%
BLEND-NIR-2026-001,100.2,1.5,60,4.5,1.5
BLEND-NIR-2026-002,99.8,2.1,45,6.2,2.1
BLEND-FAIL-2026-003,101.0,6.5,30,18.0,6.5

Input validation rules

FS — Functional Specification

Example output.json structure

{
  "utilityId": "nir-blend-uniformity-checker",
  "utilityName": "NIR_Blend_Uniformity_Checker",
  "object": "NIR Blend Uniformity Checker",
  "overallStatus": "PASS|WARNING|FAIL",
  "sourceFile": "input.csv",
  "recordCount": 3,
  "checks": [
    {
      "parameter": "BatchNumber",
      "value": "BLEND-NIR-2026-001",
      "status": "PASS|WARNING|FAIL",
      "message": "Rule-based check result",
      "ruleReference": "VR-001"
    },
    {
      "parameter": "Mean%",
      "value": "100.2",
      "status": "PASS|WARNING|FAIL",
      "message": "Rule-based check result",
      "ruleReference": "VR-002"
    },
    {
      "parameter": "StdDev",
      "value": "1.5",
      "status": "PASS|WARNING|FAIL",
      "message": "Rule-based check result",
      "ruleReference": "VR-003"
    },
    {
      "parameter": "N",
      "value": "60",
      "status": "PASS|WARNING|FAIL",
      "message": "Rule-based check result",
      "ruleReference": "VR-004"
    },
    {
      "parameter": "MaxDev",
      "value": "4.5",
      "status": "PASS|WARNING|FAIL",
      "message": "Rule-based check result",
      "ruleReference": "VR-005"
    },
    {
      "parameter": "RSD%",
      "value": "1.5",
      "status": "PASS|WARNING|FAIL",
      "message": "Rule-based check result",
      "ruleReference": "VR-006"
    }
  ],
  "criticalFindings": [],
  "warnings": [],
  "metadata": {
    "executionMode": "CLI",
    "interface": "CSV to JSON",
    "documentationFile": "NIR_Blend_Uniformity_Checker.documentation.html"
  }
}

Traceability matrix URS → FS → OQ/PQ

OQ/PQ test scenarios

QA/QC, CSV/CSA and change control

• Do not rename columns without updating the validator, test set and documentation.
• Retain input.csv, output.json, executable version, checksum and a copy of the documentation.
• Before production use, perform IQ/OQ/PQ or equivalent risk-based CSV/CSA verification.
• Critical limits shall be approved by the process owner and verified against the registration dossier, pharmacopoeia and local SOPs.
• Any change to rules, thresholds or status interpretation shall go through change control and regression testing.