My work at Grifols involved managing the site’s established CPV environment and that ultimately became the basis of my Master’s project in Industrial and Systems Engineering. I evaluated the existing monitoring model, mapped how manufacturing data moved from production through analysis and reporting, and developed risk-based approaches for strengthening and scaling the program.
The work reinforced a principle that has continued to shape my approach to manufacturing:
The Manufacturing Environment
The Emeryville site manufactured bulk immunodiagnostic products, including recombinant protein antigens produced through biological manufacturing processes.
The established CPV program supported monitoring across a broad product portfolio using a custom, qualified Manufacturing Information Management System (MIMS).
Simplified representation for illustrative purposes
From Manufacturing Data to Process Understanding
The monitoring system centralized manufacturing and QC data so performance could be evaluated across batches and over time. The analysis methods that I used for routine monitoring, analysis, and reporting included:
Simplified representation for illustrative purposes
A signal should trigger technical evaluation, not an automatic conclusion.
Designing the Monitoring Strategy
One of the central questions was:
Monitoring every possible manufacturing parameter is neither practical nor useful. I developed a risk-based framework that began with process understanding and criticality, then translated those inputs into a practical monitoring strategy.
Simplified representation for illustrative purposes
I prioritized parameters based on process risk, criticality, manufacturing knowledge, and the value of the information being generated. The framework also considered:
- how data would be collected and verified
- appropriate statistical methods
- control-limit development
- sampling frequency
- reporting cadence
- ownership and review responsibilities
- how unexpected trends would be investigated
- how CPV findings would feed back into process understanding, process control, validation, and improvement
A Key Technical Judgment: Signal vs. Noise
A mature CPV program should detect meaningful changes without creating unnecessary investigations.
One value differs. Review context before drawing a conclusion.
A directional pattern can be more informative than a single point.
A shift may warrant checking what changed between batches.
Widening variation can matter even without a shift in the average.
Synthetic patterns shown for illustration — not statistical conclusions about actual processes.
Not every rule violation represents a meaningful process issue. Overreacting to expected variation can consume significant resources and can even introduce unnecessary process adjustments. The more useful question is:
Evaluation should consider statistical behavior together with:
This combination of statistical evidence and engineering judgment helps distinguish routine variation from a signal that deserves deeper technical review.
Manufacturing Changes Matter
Historical data cannot always be treated as one homogeneous population. One of the more significant challenges that I encountered was understanding how manufacturing changes affected historical process performance. Facility transfers, equipment changes, utilities, scale, or other process change points can shift process behavior even when product requirements continue to be met.
A stable process can shift following a meaningful manufacturing change. Control limits and historical reference populations should be evaluated accordingly.
Control limits need technical context. When manufacturing conditions change, the historical data set used to characterize process behavior may no longer represent current performance.
The approach I recommended was to document significant process change points, identify representative reference batches with appropriate SME input, and reassess control limits when changes could materially affect process performance. Through continuous monitoring and analysis of change effects and control limits, I ensured that the right signals were flagged, evaluated, and addressed with SMEs.
Making CPV Operational
CPV is more than a set of charts and statistical analyses. It’s a structured framework for turning manufacturing data into process understanding: Data collection → verification → analysis and review → interpretation → response or escalation → reporting → continuous improvement and feedback.
Final yield / immediate process output
Recent batch performance and parameter review
Yield, quality attributes, targets, and broader performance
Trends, open actions, and sustained process behavior
Simplified representation for illustrative purposes
One of the practical constraints that I identified was the time required for data entry, verification, and batch closure before complete analysis could occur. Based on that constraint, I implemented the following improvements:
Faster Data Flow
Reduce delays between manufacturing and analysis.
Real-Time Monitoring
Enable earlier identification of potentially meaningful signals.
Electronic Integration
Reduce transcription and duplicate verification where appropriate.
Change-Point Traceability
Connect manufacturing changes directly to performance history.
Focused Monitoring
Periodically remove or reduce monitoring of parameters that no longer provide sufficient value based on stability, capability, risk, and process knowledge.
Scaling the Framework
My Master’s project extended the site experience at Grifols into two risk-based models:
New Program Approach
A structured framework for another manufacturing site to establish its own CPV capability.
Expanded Service Approach
A model for leveraging an established monitoring capability, infrastructure, experience, and systems across additional manufacturing sites.
NEW PROGRAM
- Purpose
- Scope
- Requirements
- Monitoring Plan
- System Configuration
- Routine Monitoring
- Continuous Improvement
EXPANDED MODEL
- Existing CPV Framework
- Site-Specific Requirements
- Product / Parameter Configuration
- Local Processes & Training
- Shared Monitoring Capability
- Local Ownership / Continuous Improvement
Developed frameworks and recommendations. My work at Grifols was at the Emeryville site only.
Results / Deliverables
Structured CPV Framework
Developed a risk-based approach for defining CPV scope, requirements, monitoring strategy, statistical evaluation, and continuous improvement.
Process-Monitoring Architecture
Mapped the interactions among manufacturing data, the monitoring system, analysis, reporting, stakeholders, and supporting quality systems.
Scalable Operating Model
Developed approaches for both establishing a new site-level CPV capability and extending an existing monitoring framework to additional manufacturing sites.
Improvement Roadmap
Identified opportunities involving real-time monitoring, electronic data integration, control-limit governance, parameter rationalization, change-point management, and faster access to actionable process information.
What I Took Away
Data needs context.
A chart can identify a signal. Understanding whether the signal matters requires process knowledge.
Monitor what matters.
A sustainable CPV program prioritizes parameters based on criticality, risk, and information value rather than attempting to monitor everything equally.
Manufacturing changes reset assumptions.
Equipment, facilities, scale, materials, and process changes can alter historical performance. Monitoring strategies and reference populations have to evolve with the process.
CPV should create a feedback loop.
The strongest monitoring programs do more than demonstrate continued control. They feed learning back into process understanding, investigations, validation strategy, and continuous improvement.
My work at Grifols formed the basis of my graduate project, “Developing the Continued Process Verification Program at Grifols Diagnostic Solutions Inc.”
The project evaluated the existing CPV operating model and developed risk-based frameworks and recommendations for program implementation, expansion, and continuous improvement.
Good process monitoring turns manufacturing data into better technical questions and better technical questions lead to better manufacturing decisions.