PAT Particle Size Analyzer (PSA)
Application for API Milling
Manufacturing Process
ISPE Great Lakes Chapter 2011 Meeting
Eric J. Correa
Sr. PAT Process Engineer
Abbott- North Chicago
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
1
© 2011 Abbott Laboratories
1
Agenda
• PAT Background
• PAT Case Study
• Insitec® Particle Size Analyzer
• PAT PSA Current State and Setup
• Online Milling Engineering Study
• Milling PAT Implementation
• Closed Loop Feedback Control Results
• Summary and Next Steps
• Acknowledgements
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
2
© 2011 Abbott Laboratories
2
What is PAT?
Diagnosis
• Process analytical technology (PAT) has been
defined by FDA as a mechanism to design,
analyze, and control pharmaceutical manufacturing
processes through the measurement of Critical
Process Parameters (CPP).
• PAT is process analysis in real-time. Real-time
analytical measurements can replace off-line time
consuming chemical analyses and reduce process
cycle time.
• PAT - An Innovative Way to Achieve Continuous
Process Improvements and Quality by Design.
• PAT tools and technology includes; Near Infrared
(NIR) spectrophotometers, Raman, laser
diffraction for particle size analysis, NIR chemical
imaging, etc.
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
3
© 2011 Abbott Laboratories
3
PAT
Monitoring
Understanding
Control
PAT Case Study: API Milling Operation
•
Typical API Pharmaceutical Process:
Reaction
Crystallization
Drying
Milling
• API PSD acceptance range: D50 within [30, 80] microns
• Non-critical, no final testing on the composite sample.
• API milling in-process test target range: D50 within [50, 65]
microns
• Two mills, fixed feed rate RPM, blower speed, and screen size.
• Initial milling speed @ 3600 RPM, mill ~ 1-5 kg, PSD testing (offline).
• Adjust milling speed to reach the target range, mill remainder lot.
• Offline method
• Wet method using Malvern MasterSizer® 2000.
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
4
© 2011 Abbott Laboratories
4
Insitec® Online Particle Size Analyzer
• Laser diffraction technology (0.1,
1000 um).
• First principles measurement,
model independent analysis.
• Wide range of mass
concentration
– Low mass concentration – Beer’s
Law.
– High mass concentration –
Multiple scattering correction.
li
ed
r
e
t
t
Sca
• Easily integrated into the process.
ght
Incident light
Refracted light
Scatt
ered
light
Illustrations from Malvern Instruments
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
5
© 2011 Abbott Laboratories
5
How Does it Work?
• A laser diffraction instrument does not measure particle size (PS).
• Laser diffraction measures light scattering, which is dependent of PS.
• Laser beam hits the sample and generates light scattering.
• Scattering signal is collected by the detectors (32 in total).
• Scattering pattern is generated.
Illustrations from Malvern Instruments
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
6
© 2011 Abbott Laboratories
6
How to Generate the Size Distribution
Illustrations from Malvern Instruments
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
7
© 2011 Abbott Laboratories
7
Mill/Insitec® Current Setup at Abbott Facility
OSI-PI
Controls
Nitrogen
Wireless
TCP/IP Network
Software/Server/Interface
Insitec PSA
Sample In
RS232
Wall
Interface
RS5422
Return
Control PC
Allen Bradley PLC
Wireless
TCP/IP Network
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
8
© 2011 Abbott Laboratories
8
PSA Operational Sequence and Logistics
Initial
Start-Up
2
3
PLC
HMI
Analyzer perform
background check
Operator collects
background
Data Tag
Operator enter
Batch ID
Batch ID
Enable
Mill RPM
Measurement
Active
Mill Starts
Start Malvern's
Auto Shut
Down
D10, D50, D90
4
D10, D50, D90
Auto Shut Down
Low Scattering
Mill Stops
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
OSI-PI
Operator Enters
D50 & Control
parameters
PLC setups
parameters
Start-Up
1
Backgrd
Ref.
Analyzer
9
© 2011 Abbott Laboratories
9
API Milling Operation – Current State
• API PSD acceptance range: D50 within [30, 80] microns
• Non-critical, no final testing on the composite sample.
• API On-line milling
• On-line milling using PSA-PAT.
• Define initial parameters (i.e. Initial milling speed, desired D50, etc).
• Milling speed via feedback control.
• PAT PSD Process Monitoring
• Support via remote desktop connection.
• PSD data available for viewing in OSI-PI.
• PSD data track and trending.
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
10
© 2011 Abbott Laboratories
10
Engineering Study
• To establish the feasibility of process monitoring.
• To demonstrate the correlation between online Insitec® and offline
Mastersizer® particle size results.
• To demonstrate feasibility of feedback control during the milling
operation.
• To study and optimize the following parameters:
–
–
–
–
–
Flow Titration
Flute Optimization
System Sensitivity
Process Control Parameters
Feedback Control
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
11
© 2011 Abbott Laboratories
11
ES Outcome- Feed Back Control Parameters
Parameter
Setting
Venturi Eductor flow rate (SCFM)
2.5
D50 set point (microns)
58
Target mill speed at startup (RPM)
The nitrogen flow rate needed for proper
aspirating a representative sample stream from
the main process stream
3600
Particle size cycle time (sec)
10 ~ 20
(15)
Mill speed adjustment step size (RPM)
50 ~ 100
(50)
Mill speed adjustment limit (RPM)
+/- 2600
D50 dead band range (microns)
+/- 2
Particle size deviation alarm (microns)
+/- 20
D50 rolling average window size (sec)
30
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
12
© 2011 Abbott Laboratories
Comments
12
Target particle size value
Start up mill speed target, same as the current
milling starting RPM
Frequency of changes to the mill speed if the
rolling average D50 value is outside the dead
band range
Value of change to the mill speed during every
cycle time if D50 is outside of the dead band range
Maximum value of change to the mill speed,
allowed by PLC
The range around the rolling average D50 set
point, within which no change is made to the mill
speed
The range around the rolling average D50 set
point, outside which Malvern Insitec® issues an
alarm signal. D50 acceptable range is 30 – 80
microns.
The window size of the time over which the
average is reported
Offline-Online Correlation
Mastersizer Offline D50 (um)
80
75
70
65
y = 1.0023x + 1.4477
R2 = 0.9417
60
55
50
50
55
60
65
70
75
Insitec Online D50 (um )
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
13
Offline Sample 1
Offline Sample 2
Offline Average
Linear (Offline Average)
© 2011 Abbott Laboratories
13
80
Engineering Study Conclusions
•
On-line particle size analyzer can monitor the milling operation
continuously.
•
On-line results have good correlation to the results from the off-line
method.
•
Feedback loop using Insitec® provides consistent and well controlled
particle size distribution throughout the whole milling operation.
•
Analyzer needs to be integrated into the milling system for regular
implementation.
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
14
© 2011 Abbott Laboratories
14
Implementation Strategy & Outcome
Engineering
Study
System
Qualification
PAT
Implementation
Strategy
Manufacturing
Implementation
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
15
© 2011 Abbott Laboratories
Communication
Established Correlation to offline method
Established
Critical Parameters
Defined
Analyzer & Mill
HMI PLC, Software,
OSI-PI
PAT Operation Procedure
PV/PQ
Manufacturing Direction
Training
Monitor and Control
Track and Trend
15
Qualification Approach
Equipment
& APC
OSI-PI
Malvern Analyzer
IOQ
APC Controls & Interface IOQ
Engineering
Study
URS
PAT OSI-PI OPC Configuration
API PSA Tags
Malvern PSA
Software
API
Process
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
16
PAT Software Operational Qualification (SOQ)
PAT Process Integration
© 2011 Abbott Laboratories
16
API Process Impact
• In-Process (offline testing) sample was removed.
• Mill RPM via feedback control.
• Continuous Monitoring of D50 PSD.
• Reduction in cycle time for milling stage.
• Integration of PSA into the Mill.
• Integration of PSA into the API process.
• Modification of current HMI PLC program and APC interface.
• PAT BOP to incorporate PAT system along with personnel training.
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
17
© 2011 Abbott Laboratories
17
Closed Loop Feedback Control
Transmission
Dv90
Dv50
Process Upset
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
18
© 2011 Abbott Laboratories
18
End of a Batch
Closed Loop Feedback Control
• Step 1: D50 SP = 58 ± 2 m, mill speed startup SP = 2600 RPM
– Took 5 minutes to reach dead band
– Offline sample D50 = 61 m
• Step 2: D50 SP = 58 ± 2 m, mill speed startup SP = 3800 RPM
• Step 3: D50 SP = 50 ± 2 m
• Step 4: D50 SP = 58 ± 2 m
80
4000
4500
75
3500
75
70
3000
70
65
2500
2000
55
1500
50
1000
45
500
D50 (um)
60
Mill Speed (RPM)
D50 (um)
SP=58um
SP=50um
SP=58um
4000
65
3500
60
3000
55
50
2500
45
40
23:11:02
40
0
22:56:38 22:58:05 22:59:31 23:00:58 23:02:24 23:03:50 23:05:17 23:06:43 23:08:10
23:13:55
23:16:48
23:19:41
Avg D50
UCL
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
19
LCL
23:25:26
23:28:19
2000
23:31:12
Time
Time
D50
23:22:34
UDB
LDB
Target
© 2011 Abbott Laboratories
D50
MILL RPM
19
Avg D50
UDB
LDB
Target
MILL RPM
Mill Speed (RPM)
80
D50 Feedback Control Using PAT for 2009
Campaign
Initial D50
Constant RPM
D50 Adjustment
Feedback
Controlled D50
Mill Speed Adjustment
D50 Rolling Average
D50 Particle Size (real-time value)
Mill Speed
D50 Set Point 57m | Mill Start-UP 3600RPM
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
20
© 2011 Abbott Laboratories
20
On-Line Milling PSA Benefits
• Integrated milling system leads to real time particle size
distribution monitoring, control, and process understanding.
• Removing In-Process testing leads to streamlined operation and
reduction in cycle time.
• Data integration to OSI-PI leads to real time tracking and trending
and improved trouble shooting capability.
Previous Process
In-Process Sample
RPM Manual Adjustment
Offline testing
Limited PSD Data
Archived QA Data
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
21
© 2011 Abbott Laboratories
Current Process
No In-Process Sample
RPM Feedback Control
On-line testing
Continuous PSD Data
Data Historian OSI-PI
21
Summary and Next Steps
•
PAT initiative encourages the pharmaceutical industry to embrace new
analytical approaches with the aim of transforming process development and
manufacture.
•
PAT On-Line Particle Size Analyzer (PSA) for Fitz-Mill was implemented during
2009 API manufacturing campaign successfully and it represents the first OnLine PAT PSA project completed and used in Abbott commercial production.
•
PAT PSA has replaced the In-Process sample and reduced the cycle time
during the milling staged as well as improved the process in overall.
•
API Manufacturing will continue to use the PAT PSA for future campaigns.
•
Currently evaluating other APIs and Drug Products (DP) pharmaceutical
processes in order to determine PAT PSA feasibility.
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
22
© 2011 Abbott Laboratories
22
Last Word: Vendor-Customer Partnership
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
23
© 2011 Abbott Laboratories
23
Acknowledgements
•
Abbott
•
•
Adrian Del Moral Ballesteros, Elvin Negron, Osvaldo Rodriguez,
Amarillys de Jesus, Amarillys Carde, Angel Roman, Maricely Colon,
Yamilda Cruz, Ana Cocero, Xenia Guerrero, Jose Restituyo, Li Sun,
Jay Paterson, Cynthia Curty, Dan Crandall, Yahui Hu, Eric Nielsen,
William Kloud, Michael Hoffman, Radames Garcia, Kenneth Wierzba,
Radames Garcia.
Malvern Instruments
•
•
Alon Vaisman, Boris Makovoz, Jeff Denigris, Gerard Michel.
Fitzpatrick
•
Joe Ream
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
24
© 2011 Abbott Laboratories
24
ISPE Great Lakes Chapter
2011 Meeting
GDS_100671_v1
25
© 2011 Abbott Laboratories
25