CONTINUOUS PROCESS

VERIFICATION
 Topics Covered
1 Core Regulatory Framework 8 CPV Implementation in Real Plants

2 Quality by Design (QbD) 9 Real-Time Release Testing (RTRT)

3 DoE & Design Space 10 Control Strategy Enhancement

4 Stage 3A 11 Quality Risk Management (QRM)

5 Stage 3B Continuous Monitoring 12 GMP Expectations for CPV

6 Process Capability & Statistics

7 Manufacturing Systems &

Operations
 Introduction
Continuous process verification (CPV) is basically the
“always-on” version of process validation: instead of
proving your process works using just a few batches and
then assuming it’s fine forever, you continuously monitor
and evaluate it during routine manufacturing to show it
stays in control over the whole lifecycle.
 Process Validation
Let’s imagine a pharmaceutical Definition (FDA 2011):
company wants to mass-produce Process Validation is the collection
an injectable vial. and evaluation of data, from the
process design stage through
The core question: How do we commercial production, which
know that every batch produced is establishes scientific evidence
safe, effective, and consistent? that a process is capable of
The answer is: By validating the consistently delivering quality
process. products.
 FDA Process Validation Lifecycle
STAGE 2 — PROCESS QUALIFICATION (PPQ)
STAGE 1 — PROCESS DESIGN
Goal: Prove the process works at commercial
Goal: Understand the process
scale.
scientifically.
This stage has 2 parts:
We do:
Part A: Equipment & facility qualification
QbD
IQ → OQ → PQ of equipment
DoE
Part B: PPQ batches
Identify CQAs
Run 3–10 successful commercial-scale
Identify CPPs
batches to demonstrate:
Develop control strategy
Consistency
Process characterization
Stability
Scaling concepts
Reproducibility
 FDA Process Validation Lifecycle
STAGE 3A — Initial CPV STAGE 3B — Ongoing CPV
Goal: Study the true performance of Goal: Ensure the process remains
the process in real commercial consistent over years.
conditions. You compare every batch against:
Stage 3A results
Here, you analyze:
Control limits
First 10–30 commercial batches
Historical behavior
Variability
Tools used:
Capability
SPC charts
CPP–CQA relationships Trends
Inherent process variability (IPV) Drift detection
Analytical vs process variance APQR (Annual Product Quality Review)
Process control performance OOS/OOT investigations
FDA & ICH Together Create CPV
Guidance Contribution Connection to CPV

FDA 2011 PV Guidance Defines Stage 1–2–3 CPV = Stage 3

Defines QTPP, CQAs, CPPs, Design

ICH Q8 Basis of Stage 1 & control strategy
Space

ICH Q9 Defines risk tools Used to select CPPs for CPV

ICH Q10 Quality system & change control Supports long-term CPV (Stage 3B)

ICH Q11 API development & control CPV for API steps

ICH Q12 Manage post-approval changes CPV data = justification for changes
 Quality by Design
Traditional approach = “Quality by QbD = “Quality by DESIGN”
TESTING” Understand the process
Make the product Identify what matters
Test the product Control the process proactively
If it passes, release it Ensure quality is built-in, not tested-in
This is REACTIVE. This is PROACTIVE.

FDA definition:
QbD is a systematic, science-based approach to pharmaceutical
development.
Elements of QbD
1. QTPP — Quality Target Product Profile
QTPP answers: “What should the final drug look like?”
It defines the goals for the product:
Dosage form (injection, tablet, etc.)
Strength
Release profile
Stability requirements
Sterility (for injectables)
Appearance
Dosage accuracy
QTPP is like defining the recipe before cooking.
 Elements of QbD
2. CQA — Critical Quality Attribute
A physical, chemical, biological, or microbiological property
that must be within limits to ensure product quality & safety.
Examples for Injectables:
pH
Assay
Impurities
Osmolality
Particulate matter
Sterility
Endotoxin level
Reconstitution time (for lyophilized products)
 Elements of QbD
3. CMA — Critical Material Attribute
Materials have attributes that affect CQAs.
Examples:
API particle size
Excipient grade
Moisture in raw materials
pH of buffer
Viscosity of polymer solution
API solubility
If a material property affects a CQA → it becomes a CMA.
Elements of QbD
4. CPP — Critical Process Parameter
Process parameters that affect CQAs.
Examples:
Mixing speed
Mixing time
Temperature
pH during reaction
Filter pore size
Homogenizer pressure
Lyophilization shelf temperature
Primary drying time
Compression force (for tablets)
 Elements of QbD
5. Control Strategy
This is the final output of QbD.
A control strategy answers: “How will we ensure consistent product
quality?”
It includes controls on:
CMAs and CPPs
Equipment
In-process tests
PAT tools
Process limits (ranges)
Automation settings and Monitoring frequency and Specifications
 DoE
In drug manufacturing, many variables Every CQA depends on multiple factors.
affect product quality: DoE helps answer:
Formulation factors: Which factors matter most
API particle size How strongly they impact the CQA
Binder amount and Polymer grade How factors interact
Buffer concentration Where failures occur
Process factors: Where the process is stable
Mixing speed What operating ranges are safe
pH
Temperature
Drying time
Compression force
 Design Space
FDA Definition (ICH Q8):
Design Space is the multidimensional combination of input
variables and process parameters that have been
demonstrated to provide assurance of quality.
In simpler words:
Design Space = The safe operating window.
Inside the design space: Can move anywhere, Quality
remains consistent and No risk to CQAs
Outside design space: Quality may fail, Risk increases,
Regulatory re-approval may be needed
 Design Space vs MODR vs PAR
Design Space MODR (Method PAR (Proven Acceptable
Large area Operable Design Range)
Full multivariate Region) One-dimensional
region Used in analytical Based on
Defined by DoE method validation experiments
Requires regulatory Defines region Does NOT give
approval where method regulatory flexibility
You can move freely performs well A tighter subset
inside Less strict than Verified by PPQ
Design Space
Process Analytical Technology
FDA definition:
Process Analytical Technology is a system for designing,
analyzing, and controlling manufacturing processes
through timely measurements of critical quality and
performance attributes.

In simpler words:
PAT gives REAL-TIME EYES AND EARS into the process.
PAT provides continuous measurements of: CQAs, CPPs,
CMAs (indirectly), Intermediate quality
So the process becomes: Predictable, Controllable,
Efficient
 Why PAT?
Traditional QC = Problems:
Make the product Too slow
Take sample Cannot detect issues early
Send to lab Cannot monitor intermediates
Use HPLC/GC/UV Cannot control the process
Wait hours or days Only checks product, not process
Release product later Lots of waste if something goes wrong

PAT advantages: Detect problems immediately, Control process

automatically, Reduce deviations, Reduce out-of-spec batches,
Enable real-time release testing (RTRT), Support continuous
manufacturing, Provide data for CPV (Stage 3)
 PAT Measurement Modes
Inline PAT
Online PAT At-line PAT
The sensor is physically inside the
A side-stream takes Sample taken
process stream.
small sample → sensor manually but analyzed
Example:
measures → returns to right beside the
NIR probe inside blender
process. equipment.
Temperature probe in reactor
Example: Example:
Advantages:
Automated NIR gun used near
True real-time
sampling loop granulator
No sampling error
Raman probe in a Moisture meter
Disadvantages:
bypass line near dryer
Must be sterilizable & robust
 Major PAT Tools
The 14 major PAT tools used in industry are: NIR Spectroscopy (Near-
Infrared), Raman Spectroscopy, UV–Vis Spectroscopy, FBRM
(Focused Beam Reflectance Measurement), TPI (Terahertz Pulse
Imaging), Hyperspectral Imaging (HSI), Mass Spectrometry, Acoustic
sensors, MRT (Microwave Resonance Technology), Eyecon / Lasentec
Imaging Tools, XRF (X-ray fluorescence), OCT (Optical Coherence
Tomography), SFV (Spatial Filter Velocimetry), Chemical Imaging
Tools (NIR-CI, Raman-CI)
 Chemometrics
PAT generates spectra — not direct quality values.
Chemometrics is the math used to convert spectra → predictions.

Key tools:
1. PCA (Principal Component Analysis)
2. PLS (Partial Least Squares Regression)
3. MLR (Multiple Linear Regression)
How PAT connects to RTRT and CPV

PAT → monitors CPPs & CQAs

PAT data → trend analysis in CPV
Stable CPV → proves process is under control
RTRT → release without end-product testing
 Stage 3A
Why needed? But real commercial manufacturing
Stages 1 and 2 (design + PPQ) represent: introduces:
Controlled conditions New operators
Limited batches Different material lots
Ideal operators Seasonal changes
Limited raw material variability Mechanical drift
More batches → more variability

So Stage 3A is like a deep health check of the process under REAL conditions.
What happens in Stage 3A?
Stage 3A consists of 6 major activities:
[Link] how many batches to collect
[Link] real manufacturing data (CPPs, CQAs)
[Link] process variability from analytical variability
[Link] intrinsic process variability (IPV)
[Link] the process using PaCS Index
[Link] the Process Capability & Quality Dashboard (PCQd)
[Link] the Control Strategy
[Link] for Stage 3B
 Stage 3B
Even after Stage 3A creates the “blueprint,” real
manufacturing changes over time:
Machines wear out
New operators join
Raw materials come from new lots
Seasons change humidity
Suppliers change
Facility layout may change
Microbiological environment varies
New equipment may be installed
Therefore: Stage 3B ensures long-term consistent quality. It
is a permanent health-monitoring system for the process.
 What Stage 3B actually does?
Stage 3B focuses on routine monitoring of: Stage 3B is not heavy mathematics like
[Link] Stage 3A.
[Link] Its purpose is:
[Link] Detect signals early
[Link] Confirm process control
[Link]-of-trend behavior (OOT) Support continuous improvement
[Link] and shifts Provide data for lifecycle
[Link] capability over time management (ICH Q12)
[Link] / CAPA effectiveness
[Link] trends
[Link] (Annual Product Quality Review)
The Tools Used in Stage 3B
[Link] Charts (Statistical Process Control)
a.X-bar Chart (Mean Chart)
b.R Chart (Range Chart)
[Link] (I) Chart
[Link] Range (MR) Chart
[Link] Charts (Time Series)
[Link]-of-Trend (OOT) Detection
[Link] Rules (Nelson Rules)
[Link] Capability (Cp, Cpk, Pp, Ppk)
 Monitoring Frequency
Stage 3B monitoring For Example
frequency depends on: Parameter Frequency

Batch frequency
Critical CPP (mixing pH) Every batch
Criticality of the
CQA (Assay) Every batch
parameter
Risk assessment (ICH Q9) Intermediate CQAs Every batch / weekly

Historical variability
Less critical parameters Monthly / quarterly
Volume of production
Yield Every batch

Environmental data Daily–weekly

 What happens if Stage 3B detects a signal?
3. Take action
1. Investigate cause
Corrective action
Batch record
Calibration
Equipment log
Additional tests
Raw material change
Retraining
Operator shift
Modification of CPP limits
2. Classify
4. Improve control strategy
OOT
Stage 3B can feed back into:
Shift
Reopening QbD
Trend
Tightening limits
Spike
Adding PAT
Better sampling plans
Annual Product Quality Review
Under FDA and ICH rules, every product requires:

This is a summary of: It confirms:

CQAs The process stays in control
CPPs No evidence of drift
Trends Continued robustness
Deviations Need for improvements
Complaints APQR is a key element of Stage 3B.
Recalls
OOS/OOT
Stability data
Process Capability & Statistics

This section is extremely important because:

It connects directly to Stage 3A (deep assessment)
It is used heavily in Stage 3B (ongoing CPV)
It forms the math behind capability, control, limits, and
decisions
Every pharma plant uses these tools daily in
QA/Validation/TechOps
 Understanding Variation
Why variation happens?
Every manufacturing
Raw material lots differ
process varies — NOTHING
Operators differ
is constant.
Equipment drift
Example:
Temperature/humidity
Assay might be:
Sampling error
99.1%
Instrument error
99.3%
Statistics exists to separate real variation
98.9%
from noise.
99.0%
Variation is represented by standard
deviation (σ).
Understanding Normal Distribution
Most pharma parameters follow a bell-shaped curve.
This is called a normal distribution.
It has two key components:
Mean (average) - The center of the data.
Standard deviation (σ) - The spread of the data.
Normal distribution has powerful properties:
68% data lies within ±1σ
95% data lies within ±2σ
99.73% data lies within ±3σ
This forms the basis of capability.
 Short-Term vs Long-Term Variation
Short-Term Variation
Variation within a batch (subgroups).
Represents the inherent process noise under controlled conditions.
Used for: Cp, Cpk
Long-Tem Variation
Variation across many batches over time (includes shifts, drifts,
seasonality).
Used for: Pp, Ppk
Key intuition:
Cp/Cpk measure “instant process health.”
Pp/Ppk measure “lifetime process health.”
 Normality Testing and Data Transformation
If data is skewed:
Cp/Cpk assume the data is
Examples:
normally distributed.
Moisture
You must test for normality using:
Particle size
Anderson–Darling Test
Microbial counts
Shapiro–Wilk Test
Impurities
Histogram + Q–Q plot
Use transformations:
If the data is NOT normal:
[Link]-Cox Transformation
Capability indices become
[Link] Transformation
meaningless
[Link] transformation
You must use transformations
Goal: Make the data “normal-ish” so
capability calculations are valid.
 How Capability Fits Into CPV

Stage 3B
Stage 3A
Track capability periodically:
Use capability to:
Every month
Evaluate process performance
Every quarter
Identify weak CQAs
Annually (APQR)
Identify unstable CPPs
If Cpk drops → investigate
Improve control strategy
If Cpk increases → process improving
Benchmark vs historical products
If Cpk becomes <1 → near-failure state
 Batch Manufacturing
Advantages of Batch:
Why batch is common in injectables? High control over each run and Simple to
Easier to ensure sterility validate
Easier to clean equipment between Fits regulatory expectations
batches Easy deviation handling
Easier to validate Disadvantages:
Easier to trace Downtime between batches and Cleaning
Less complex control strategy time (CIP/SIP)
Suits liquid/lyo operations Lower throughput and Large equipment
footprints
More labor and Inventory buildup
 Continuous Manufacturing
Continuous manufacturing means: Why continuous is powerful?
Material flows continuously through Higher throughput and Fewer cleaning cycles
all unit operations, without stops. Real-time control possible and Less variability
Examples: (steady state)
Continuous mixing Smaller equipment footprint and Faster
Continuous granulation changeovers and Less operator involvement
Continuous filtration Disadvantages:
Continuous lyophilization Very complex to design and Requires PAT
(emerging) integration
Continuous compounding of Requires strong control strategies
buffers or solutions Harder to maintain sterility in injectables
High CapEx for advanced equipment
 SUT vs MUT
Multi-Use Technology (MUT): Single-Use Technology (SUT):
Advantages: Advantages:
Reusable and Suitable for large No cleaning validation needed
batches Much faster changeovers
Robust, long-life equipment No risk of cross-contamination
Works with aggressive solvents Small footprint
Disadvantages: Excellent for multi-product facilities
Needs CIP and SIP and Cleaning Disadvantages:
validation is time-consuming Higher consumable cost
Risk of cross-contamination Plastic waste and Limited batch size
Higher downtime and Higher labor Handling fragility and Some solvents
Larger utilities (WFI, steam) incompatible
 Scheduling Basics
Each step has:
Scheduling means planning how Fixed duration and Setup time
different batches and equipment Cleaning time
are used over time. Equipment constraints
Why scheduling is challenging? Manpower requirement
Because a typical injectable process Scheduling determines:
uses SEQUENTIAL steps: How many batches/day
Compounding, Filtration, Filling, How many lyophilizers needed
Lyophilization, Capping, Inspection Whether shift changes support operations
When to clean
When to perform maintenance
Inventory buildup
 Changeover
Changeover = the time required SUT drastically reduces changeover:
to switch between products or No cleaning
batches. No sterilization
Changeover involves: Just replace the bag set
Equipment cleaning MUT takes longer:
SIP / sterilization Cleaning cycle 1–3 hours
Filter replacement Verification
Tubing replacement Documentation
QC sampling Drying
Line clearance Changeover time directly affects: Batch
Documentation scheduling, Throughput, Cost, CPV metrics
 CIP & SIP
CIP (Clean-In-Place) SIP (Steam-In-Place)
Cleaning with detergents/chemicals without Sterilization using pure steam.
dismantling equipment. Used for:
Used for: Tanks, Pipelines, Filters (sometimes) All sterile contact parts
Steps: Pipelines
[Link]-rinse Filters
[Link] wash Holding tanks
[Link] SIP ensures:
[Link] rinse No micro contamination
[Link] Sterility assurance
CIP/SIP cycles must be: Validated, Repeatable and Documented. This is why SUT is
becoming popular: no CIP/SIP → huge time savings.
 Scheduling Basics
Each step has:
Scheduling means planning how Fixed duration and Setup time
different batches and equipment Cleaning time
are used over time. Equipment constraints
Why scheduling is challenging? Manpower requirement
Because a typical injectable process Scheduling determines:
uses SEQUENTIAL steps: How many batches/day
Compounding, Filtration, Filling, How many lyophilizers needed
Lyophilization, Capping, Inspection Whether shift changes support operations
When to clean
When to perform maintenance
Inventory buildup
 CPV Implementation in Real Plants

In real plants, data come from multiple sources, grouped into 4 categories:
[Link] and Sensors (real-time)
[Link] Testing
[Link] or BMR (Batch Manufacturing Records)
[Link] (Quality Management System)
 The 4 Main Digital Systems
SCADA (Supervisory Control And Data Acquisition)
Controls machines + records real-time process data.
Used for: Mixing, Filling lines, Lyophilizers, HVAC/cleanroom systems
Records CPP data every second/minute.
PLC (Programmable Logic Controller)
The actual control hardware that: Runs pumps, Controls motors and
Starts/stops valves. It works under SCADA.
 The 4 Main Digital Systems
LIMS (Laboratory Information Management System)
Stores QC test data: Assay, Impurities, Microbiology, Sterility results
LIMS is the source for CQAs.
MES (Manufacturing Execution System)
Digital BMR.
Tracks: Material movements, Equipment status, Operators, Process
steps, Sampling points
Extremely important for CPV.
How CPV Datasets Are Actually Built

[Link] CQAs & CPPs

[Link] extraction
[Link] cleaning & formatting
[Link] CPV database
[Link] analysis
[Link] CPV reports
[Link] the control strategy if needed
 Real-Time Release Testing (RTRT)
RTRT is the opposite:
Traditional QC: Real-Time Release Testing means releasing
You manufacture the product a batch based on real-time process data,
Take sample NOT waiting for end-product QC testing.
Send to QC lab RTRT proves:
Perform HPLC, GC, particle CQAs were assured during
analysis, etc. manufacturing
Wait hours to days Process control + PAT + models
THEN release product guarantee reproducible quality
This is slow, reactive, and You don’t wait for HPLC results to release the
expensive. batch.
Batch is released based on process controls.
The Technologies That Enable RTRT

PAT (Process Analytical Technology)

Chemometrics & Multivariate Models
Control Strategy
CPV (Stage 3)
Automation (SCADA/MES)
RTRT vs CPV vs PAT vs QbD (Connection Map)

QbD (ICH Q8) → Identify CPPs + CQAs → Design PAT strategy →

Build multivariate models → Implement control loops → Validate
process (PPQ) → Stage 3A: deep variability analysis → Stage 3B:
ongoing CPV → IF stable & predictable → RTRT approved
 Challenges in RTRT
RTRT is powerful BUT hard to implement.
Key challenges:
[Link] validation
[Link] drift over time
[Link] vibration/noise
[Link] fouling
[Link] lifecycle management
[Link] integrity
[Link] approval
[Link] of QA/Operations
[Link] to traditional QC when PAT fails
 Control Strategy Enhancement

Formal Definition (from ICH Q8 & Q10):

A control strategy is a planned set of controls, derived from product and
process understanding, that ensures process performance and product quality.
What Does a Control Strategy Control?

[Link] Controls
[Link] Parameter Controls (CPPs)
[Link]-Process Controls (IPCs)
[Link] Controls / Automation
[Link] Controls
 Feedback vs Feedforward Control

1. Feedback Control
Feedforward Control
React AFTER you see an issue.
Predict & correct BEFORE deviation
Example (Injectables):
happens.
Measure pH
Example:
pH too low → add NaOH
Raw material moisture ↑ → Adjust
pH too high → add HCl
drying time BEFORE drying starts.
You correct output after
detecting deviation.
 How Control Strategy Evolves
Through the Lifecycle
1. Stage 1 (QbD) 1. Stage 3A (Deep CPV) : This is where the strategy is
a. Initial control strategy based on: refined and finalized.
DoE, Risk assessment, Mechanistic a. Stage 3A produces:
understanding i. Final CPP ranges
b. This is a draft control strategy. ii. Final sampling frequency and PAT design
2. Stage 2 (PPQ) iii. Final alarm/alert limits and process capability
a. Control strategy is tested under: assessment and acceptable variability range
b. Commercial scale 2. Stage 3B (Ongoing CPV)
c. Real conditions a. Control strategy is continuously monitored and
d. Full equipment train kept healthy.
e. Weaknesses become visible. b. If drift occurs → update control strategy.
 Control Strategy Pitfalls & Best Practices

[Link] too many CPPs (overcomplicated, impossible to control) → Use risk

assessment + DoE to identify TRUE CPPs.
[Link] limits = specification limits (WRONG — very dangerous) → Control
limits must be tighter than spec limits.
[Link] feedback/feedforward control (all manual adjustments) → Integrate
automation wherever possible.
[Link] PAT in critical process steps → Use PAT especially in: Blending, Drying,
Lyophilization
[Link] updating control strategy after Stage 3A → Stage 3A output MUST
update control strategy.
 What is “Risk”?
Risk is the chance that something will negatively impact product
quality or patient safety. (Risk = Probability × Severity)

QRM (Quality Risk Management) tells you:

What can go wrong
How likely it is
How severe the impact is
How to control it
It is a structured method for these decisions.
 The Big QRM Tools
1. FMEA (Failure Modes and Effects Analysis)
2. Ishikawa (Fishbone Diagram)
3. Fault Tree Analysis (FTA)
4. Risk Ranking and Filtering
5. HACCP (Hazard Analysis and Critical Control Points)
 Best Practices & Pitfalls in QRM
[Link] Practices
[Link] cross-functional teams
[Link] science + statistics
[Link] QRM after Stage 3A
[Link] QRM with CPV dashboards
[Link] risk-based decisions for sampling frequency, monitoring plan
[Link]
[Link] QRM as a paperwork exercise
[Link] scoring of S/O/D without justification
[Link] updating QRM after deviations
[Link] link between QRM and control strategy
[Link] QRM only during development (it must LIVE during CPV)
 GMP Expectations for CPV
GMP requires that: A process must consistently produce quality product
throughout its lifecycle.
CPV is the tool used to prove this consistency.
Regulators expect CPV to:
Continuously monitor the process
Detect drift and variability early
Confirm control strategy effectiveness
Support changes (ICH Q12)
Feed into Annual Product Quality Review (APQR)
Ensure validated state is maintained
THANK YOU