Академический Документы
Профессиональный Документы
Культура Документы
Business Challenge
! -! -! Reduction of dimentionality ! Conformity check Real time release testing Trend analysis Root cause analysis
Solution
! ! ! !
Topics
! Introduction: MVDA in the context of pharmaceutical production ! Case studies for MVDA I
! Process monitoring of a granulation process in pharmaceutical production
Umetrics
! ! ! ! ! ! ! ! ! Part of ~1Billion conglomerate The market leader in software for multivariate analysis (MVDA) & Design of Experiments (DOE) 25+ years in the market Off line analysis tools On-Line process monitoring and fault detection 700+ companies, 7,000+ users Pharmaceutical, Biotech, Chemical, Food, Semiconductors and more Worldwide Presence with MKS Close collaboration with universities in USA, Sweden, UK and Canada
Copyright 2013 OSIsoft, LLC.
QRA:
Quality Risk Assessment
MVDA
QFD
Quality Function Deployment
! !
DOE is a knowledge building tool for process development MVDA is used both for process understanding and process monitoring
Copyright 2013 OSIsoft, LLC.
The information is found in the correlation pattern - not in the individual variables!
Copyright 2013 OSIsoft, LLC.
Process Sensors
! !
Require MVDA methods to visualise and extract reliable information from raw data MVDA handles noise, missing data, correlation and visualize in graphs
10
11
tPS[1]
-1
-2
12
MSPC
Observation Level ! Example of a drying step
PO_WST3433_EXJADE_Drying_V01.M3:3 Predicted Scores [comp. 1]
+3 Std.Dev t[1] (Avg) -3 Std.Dev tPS[1] (Batch S0058_A_854826)
Control limits
tPS[1]
-1
-2
-3
13
-2 -3
t[1]
-4 -5 -6 -7 -8 -9
Model Data Ciclo - Oct 2010 v5 - batch -100 level (scores).M2 0 (PCA-X) t[Comp. 1]/t[Comp. 2]
20
10 614 615 625 633 624 612 616 639 635 634 618 631 620 626 641 628 638 627
632 621
630
t[2]
-10
637
623
617
-20 -80 -70 -60 -50 -40 -30 -20 -10 0 t[1] R2X[1] = 0.873728 R2X[2] = 0.0564179 Ellipse: Hotelling T2 (0.95) 10 20 30 40 50 60 70 80
14
15
Reduction of Dimensionality
ObsID(Obs ID ($PhaseID)) Mixer Power rate of change precss variable 0.01 * Mixer torqute process variable 0.1 * Mixer speed process variable 0.1 * Product temperature process variable Mixer power process variavle (electrical)
13
12
11
10
320
330
340
350
360
370 Num
380
390
400
410
420
!
Copyright 2013 OSIsoft, LLC.
16
PO_WST10332_EXJADE_GRAN_Steintraining - batch level (scores).M1 (PCA-X), All Batches, PS-Complementory tPS[Comp. 1]/tPS[Comp. 2]
Batch Level
-40
tPS[2]
-60 -80 -100 -120 -140 -160 -180 -20 -10 0 10 20 30 40 50 tPS[1] R2X[1] = 0.402627 R2X[2] = 0.341738 Ellipse: Hotelling T2PS (0.95) 60 70 80 90 100 110 S0028-B_85 120
Observation Level
1
Missing
-1
Product Temperature
Mixer Torque
Chopper Speed
Mixer Speed
-2
Flow liquid feed
Aims: -! Conformity check -! Real time release testing - Trend analysis - Root cause analysis
Var ID (Primary)
+3 Std.Dev XVar(Liquid feed pump speed) (Aligned) (Avg) -3 Std.Dev XVarPS(Liquid feed pump speed) (Batch S0007_B_854825)
120
100
80
60
40
20
-20
Bowl Temperature
Bowl Pressure
Mixer power
50
17
! Variables
18
19
Granulation Process
Data Setup Input: Initial conditions X(t)
Variable initial conditions :
!Amount of granulation liquid !Granulation time
1.! 2.! 3.! 4.! 5.! 6.! 7.! 8.! 9.! Power consumption Power consumption rate Torque Product Temperature Mixer Speed Chopper Speed Water addition rate Flow Liquid Speed Liquid Speed Pump
time
X0
Y
Copyright 2013 OSIsoft, LLC.
20
Example Granulation
Results of DoE investigation
6 4
S0015_B
2
S0011-B S0011-A
S0012-B S0012-A
t[2]
-2
-4
10
11
12
"! Process variability are reflected by the red lines "! Clustering of DoE batches can be visualised "! Common cause vs. special cause variation
Copyright 2013 OSIsoft, LLC.
21
Loadings
Identification of process parameters contributing to process variability
Loadings of Granulation Models
0.5 0.4 0.3 0.2 0.1
Mixer Power Liquid Added Mixer Torque Power (calcula Bowl Bowl Temperature Pressure Product Temperature
p[2]
Flow Liquid Feed Liquid Pump Speed Chopper Speed Mixer Speed
0.0 0.1 0.2 p[1]
R2X[1] = 0.401621 R2X[2] = 0.318985
SIMCA-P+ 11 - 04.01.2010 18:07:36
0.3
0.4
0.5
0.6
"! Which parameters are most influential? "! How do the variables correlate to each other?
Copyright 2013 OSIsoft, LLC.
22
Last campaign This campaign Last batch after liquid feed tube change
t[2]
10 S0064_A_854826
S0080_A_854826
-10
-20
-10 t[1]
10
20
23
24
25
-0.38843 * XVar(Glucose (mg/L)) - 0.38843 -0.0251397 * XVar(Titer (mg/L)) - 0.0251397 -0.150825 * XVar(Spec Feed (mL/cells)) - 0.150825
All the variables for each batch are summarized into one quantity (carrier of the information) process signature
Perfusion Start
10 5
tPS[1]
3 sd from ref.
-2
t[2]
-4
0
-6
-8
-10
-12 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Day 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
SIMCA-P+ 12.0.1 - 2011-03-18 12:30:07 (UTC+1)
42 43 44 45
27
3 sd from ref.
2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0
tPS[1]
-2
-4
-6
-8
Score Contribution
Viability (%)
Weight (kg)
-10
-12 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Day
Var ID (Primary)
SIMCA-P+ 12.0.1 - 2011-04-29 18:06:16 (UTC+1)
pH
28
MVDA learning
! Generated useful process knowledge
! Enhanced process understanding supported by data ! Improved process consistency
non-g
tPS[2]
R2X[1] = Ellipse:
0.430057 Hotelling
R2X[2] T2 (0.95)
0.145144
29
Petter Moree
petter.moree@umetrics.com Director, Global Product Manager Umetrics
30
Brought to you by