FDA investigators found HEPA filters torn in many sections with white powdery materials extruding from the corners. The facility had completed integrity testing on these same filters — and passed them. Equipment surfaces tagged as clean showed visible residue. The Granules India 483 reveals what happens when facility monitoring programs measure compliance on paper rather than conditions on the floor.
On 26 August 2024, FDA investigators at the Granules India Limited facility in Gagillapur Village, Telangana, pulled HEPA filters from randomly selected Air Purification Units — and found them “severely damaged and torn in many sections.” Patches of white powdery material were extruding from the corners of these filters. Dust and wet surfaces coated the filter media. The observation, issued on 6 September 2024, cited a fundamental breakdown: “Building and facilities are not maintained to ensure that products manufactured meet the quality, purity characteristics, identity and strength which they purport.”
What makes this 483 particularly instructive is not the damaged filters themselves. Filters degrade. Seals fail. The critical finding is what happened before the FDA arrived: the facility had completed HEPA filter integrity testing on these same units — and the testing did not identify the damage observed by investigators. A testing programme designed to catch exactly this kind of failure had certified the filters as acceptable. Then an FDA investigator, using visual inspection, found tears visible to the naked eye.
The observation extended beyond HEPA filters. Dust collection systems showed visible build-up and potential cross-contamination between manufacturing areas. Equipment surfaces in contact with product — surfaces tagged as “clean” — showed visible residue. The facility’s environmental monitoring programme, the system designed to catch these conditions, did not correlate with what investigators found on the floor. This was not a single point of failure. It was a systematic disconnect between what the monitoring systems reported and what actually existed in the manufacturing environment.
When a facility’s own integrity testing passes filters that are visibly torn, the problem is not the filter — it is the testing programme, the monitoring architecture, and the assumption that a passing result on paper means acceptable conditions on the floor.
HEPA filter integrity failures (21 CFR 211.42 — Building Design and Construction Features). Section 211.42 requires that buildings used in the manufacture of drug products be of suitable size, construction, and location to facilitate cleaning, maintenance, and proper operations. HEPA filters are not ancillary equipment — they are the primary barrier between the manufacturing environment and particulate contamination. At Granules India, filters were torn in multiple sections. White powdery material was extruding from the corners, indicating that unfiltered air had been passing through compromised media and entering classified manufacturing spaces. The facility’s own integrity testing protocol had been executed on these filters and had not flagged the damage. During the inspection, additional damaged filters were discovered in other modules, suggesting the problem was not isolated to a single unit but represented a systemic gap in the filter monitoring programme.
Dust collection and cross-contamination (21 CFR 211.56 — Sanitation). Section 211.56 requires written procedures for cleaning and maintaining buildings, including provisions to prevent contamination of equipment and drug products. FDA investigators found visible build-up in dust collection systems and indicators of potential cross-contamination between manufacturing areas. Dust collection systems in pharmaceutical manufacturing serve a dual purpose: they remove airborne particulates from the immediate environment, and they prevent product residues from migrating between manufacturing zones. Visible build-up in these systems means both functions were compromised. Material from one manufacturing area could potentially reach another, creating cross-contamination risk that undermines the identity, strength, and purity of every product manufactured in the affected zones.
Equipment cleanliness failures (21 CFR 211.63 — Equipment Design, Size, and Location). Section 211.63 requires that equipment used in the manufacture of drug products be of appropriate design, adequate size, and suitably located to facilitate operations, cleaning, and maintenance. Equipment surfaces in direct contact with product showed visible residue despite being tagged as “clean.” This is a two-part failure: the cleaning process itself did not achieve the required result, and the verification process that cleared the equipment for use did not detect the remaining residue. When equipment tagged as clean carries visible contamination, the cleaning validation programme, the visual inspection protocol, and the status labelling system have all failed simultaneously.
The FDA noted that the facility’s environmental monitoring programme did not correlate with the observed conditions. This is the thread connecting all three findings. Environmental monitoring is supposed to serve as the continuous verification layer — the system that catches what periodic testing misses. At Granules India, that system was generating data that did not reflect the physical reality of the facility.
3
Interconnected CFR sections cited — 211.42, 211.56, 211.63
Multiple
Air Purification Units with damaged HEPA filters found across modules
0%
Correlation between environmental monitoring data and observed facility conditions
The root cause is not inadequate maintenance effort. It is a monitoring architecture where testing programmes, environmental data, and equipment status systems operate independently — each generating passing results while the actual facility conditions drift undetected.
HEPA integrity testing at Granules India followed a defined protocol and produced passing results. But the protocol measured what it was designed to measure — typically aerosol challenge or pressure decay — without accounting for the possibility of gross physical damage. A filter can pass an integrity test at its rated specification while having tears large enough to see. The testing programme was measuring a parameter, not verifying a condition. This distinction matters because it determines what the data actually tells you.
Traditional environmental monitoring programmes collect data at defined intervals — daily viable counts, weekly non-viable particulate sampling, periodic settle plate collection. Between sampling events, conditions can change without detection. At Granules India, the environmental monitoring programme did not correlate with observed conditions because it was designed to sample the environment periodically, not to continuously verify it. A system that checks conditions once a day cannot catch a filter that fails at hour six.
Equipment was tagged as clean despite visible residue. This means someone inspected the equipment, judged it acceptable, and applied a clean status label. Visual inspection is inherently variable — it depends on lighting, angle, inspector training, and cognitive load. More critically, it captures the state of the equipment at a single moment. It cannot account for recontamination after tagging or for residues that become visible only as surfaces dry or environmental conditions change.
The most damaging aspect of the Granules India finding is the gap between what the data systems reported and what existed on the manufacturing floor. Filter integrity: pass. Environmental monitoring: within limits. Equipment status: clean. Actual conditions: torn filters, dust build-up, visible residue. This is not a data quality problem — the data was accurately recorded. It is an architecture problem. The systems generating the data were not connected to real-time physical conditions.
The question is not whether your facility maintenance programme generates passing results. The question is whether your monitoring architecture would detect the conditions the FDA found at Granules India before an investigator walks through the door.
Each comparison below addresses a specific gap documented in the Granules India 483. The digital approach does not add more testing — it connects the testing that already exists so that the correlations the FDA made during inspection are made by the system continuously.
HEPA filter integrity tests run at defined intervals — quarterly, semi-annually, or after maintenance events. Results are recorded in standalone documents. A filter passes its scheduled test, gets documented as compliant, and is not re-evaluated until the next scheduled test or a triggered event. Gross physical damage occurring between test cycles goes undetected.
Particulate monitoring, differential pressure readings, and filter performance data feed into a connected system that correlates environmental conditions with equipment status. A spike in particulate counts in a classified area triggers an immediate investigation workflow — not at the next scheduled test, but at the moment the data diverges from the established baseline. The system connects cause and effect in real time.
An operator or QA reviewer inspects equipment, judges it clean by visual assessment, and attaches a paper status tag. The tag represents a single human judgement at a single point in time. There is no mechanism to verify the status between tagging and the next use. If conditions change — recontamination from airborne particulates, residue becoming visible as surfaces dry — the tag remains unchanged.
Equipment cleaning is logged digitally with timestamp, operator identity, cleaning agent, and parameters. Hold-time limits are enforced automatically — if the interval between cleaning and use exceeds the validated hold time, the system blocks the equipment from production. Environmental data from the surrounding area is linked to the equipment record, so a particulate excursion near cleaned equipment triggers a re-evaluation before the equipment can be used.
HVAC data sits with engineering. Environmental monitoring data sits with QC. Equipment cleaning records sit with production. Dust collection maintenance logs sit with facilities. No single system connects these data streams. The correlation between a failing HEPA filter, rising particulate counts, and equipment contamination is invisible until an investigator connects the dots during an inspection.
All facility data — HVAC performance, environmental monitoring results, equipment status, cleaning records, dust collection maintenance — flows into a single platform. Automated rules identify correlations: if differential pressure across a HEPA filter drops while particulate counts in the downstream area rise, the system flags the relationship, generates an investigation, and places affected equipment on hold. The correlation that the FDA made at Granules India is made by the system continuously.
The three capabilities below directly address the root causes behind the Granules India findings. They work because they eliminate the disconnect between what monitoring systems report and what actually exists on the manufacturing floor.
A modern facility monitoring system must connect HVAC performance data, particulate monitoring, differential pressure readings, and equipment status into a single data layer. When HEPA filter performance degrades, the system must automatically correlate that degradation with environmental monitoring data from downstream classified areas. This is the correlation that was missing at Granules India — and it cannot be achieved when data lives in separate systems maintained by separate departments.
Periodic testing catches problems on a schedule. Real-time monitoring catches them when they occur. The system must continuously evaluate environmental parameters against validated limits, trigger alerts when trends indicate drift before limits are breached, and automatically generate deviation records with pre-populated context. At Granules India, the gap between filter failure and detection was filled by an FDA inspection. In a modern system, that gap is filled by automated detection and escalation workflows.
Equipment status cannot depend solely on a visual inspection and a paper tag. The system must enforce cleaning hold times, link equipment status to environmental conditions in the surrounding area, and prevent equipment from entering production if any connected data stream shows an excursion. If the area around a piece of equipment experiences a particulate event, the equipment's clean status must be automatically re-evaluated — regardless of when it was last inspected.
30
Facilities connected on a single platform — Cipla deployment with 2,500+ concurrent users
100%
21 CFR Part 11 compliance across Piramal's 10+ global facilities and 3 regulatory jurisdictions
20→1
Days for batch review reduced at Valent BioSciences — 2,700 hours saved annually
The objective is not to run more tests. It is to build the data infrastructure that makes every existing test meaningful — connecting HVAC performance, environmental monitoring, equipment status, and cleaning records into a system that correlates data the way an FDA investigator does, but continuously.
The first step is eliminating the silos that allowed the Granules India conditions to persist undetected. HVAC performance data, environmental monitoring results, equipment cleaning records, and dust collection maintenance logs must flow into a unified platform. This does not require replacing existing instruments or monitoring equipment — it requires connecting them. Once the data streams converge, the correlations that the FDA made during inspection become visible to the facility's own quality team in real time.
With connected data, the system can establish performance baselines for every monitored parameter and every correlation between parameters. HEPA filter differential pressure trending against downstream particulate counts. Cleaning effectiveness metrics correlated with environmental conditions at the time of cleaning. Equipment hold-time compliance rates. Automated alerts fire when any parameter or correlation drifts from its baseline — not when it breaches a limit, but when the trend indicates a limit breach is approaching. This is the shift from reactive to predictive monitoring.
The final phase closes the loop between detection and action. When the system identifies a potential facility maintenance issue — a filter performance trend, an environmental excursion, an equipment cleaning anomaly — it does not just generate an alert. It initiates a structured workflow: investigation assignment, root cause analysis template, CAPA tracking, effectiveness verification. Equipment affected by the finding is automatically placed on hold. Production in the affected area is paused pending resolution. The system enforces the response, removing the gap between knowing about a problem and acting on it.
The Granules India 483 did not cite a lack of testing. It cited a lack of correlation between what the tests reported and what existed in the facility. That correlation gap is an architecture problem — and it requires an architecture solution.
The Granules India 483 issued on 6 September 2024 is a case study in the difference between compliance activity and compliance assurance. The facility was not neglecting its monitoring obligations. HEPA integrity tests were being conducted. Environmental monitoring was being performed. Equipment was being inspected and tagged. Every required programme was in place, and every required record was being generated. The problem was that none of these systems were connected to each other, and none of them were connected to the actual physical conditions in the facility.
This is the gap that FDA investigators increasingly focus on during inspections. It is not enough to have a testing programme. The programme must detect the conditions it is designed to detect. It is not enough to have an environmental monitoring programme. The programme must correlate with reality. It is not enough to tag equipment as clean. The equipment must actually be clean, and the system must verify that it remains clean until it enters production.
For quality leaders at multi-site pharmaceutical manufacturers, the Granules India finding is a prompt for a specific question: if FDA investigators walked your facility tomorrow, would your monitoring data match what they observe? If the answer requires qualification — if it depends on which area, which shift, which testing cycle — then the architecture that connects your monitoring systems to your facility conditions has the same gap that the FDA found at Granules India. Closing that gap is not a matter of running more tests. It is a matter of building the data infrastructure that makes every test meaningful.