real time release in continuous solid dose manufacturingreal time release in continuous solid dose...
TRANSCRIPT
Real Time Release in Continuous Solid Dose Manufacturing:Systematic Characterization of Material Properties, and Optimal Design of Sensing and Control Methods
Data reconciliation in robust fault-tolerant control for continuous direct compaction
Session 5 Safety in the Pharmaceutical Industry and Manufacturing
Content
▪ Introduction▪ Continuous tablet manufacturing▪ Data reconciliation for tablet weight▪ Process control development▪ Risk analysis & assessment▪ Conclusions
▪ Introduction
[1] http://berkeleysciencereview.com/importance-uncertainty/[2] Khaleghi B, Khamis A, Karray FO, Razavi SN. Multisensor data fusion: A review of the state-of-the-art. Information Fusion.
2013; 14: 28-44.[3] https://blogs.salford.ac.uk/careers-employabilty/2017/01/20/day-6-decision-making-made-easier/
SafetyQuality
Control
https://www.truesocialmetrics.com/blog/three-things-everyone-should-learn-about-analytics
min𝑧
𝑧 − 𝑧𝑚𝑇W−1 𝑧 − 𝑧𝑚
subject to 𝑧 ⊆ [𝑦, 𝑥, 𝑢, 𝜃]
𝑓 𝑦, 𝑥, 𝑢, 𝜃 = 0
𝑔(𝑦, 𝑥, 𝑢, 𝜃) ≤ 0
Data reconciliation principle
𝑧𝑚 vector of measured process variables
𝑧 vector of reconciled measurement variables
W weighting matrix
y process output variables
x process state variables
u process input variables
𝜃 process model parameters
f (·) process equality equations
g (·) process inequality equations
Su Q, Bommireddy Y, Gonzalez M, Reklaitis GV, Nagy ZK. Data reconciliation in the Quality-by-Design (QbD) implementation of pharmaceutical continuous tablet manufacturing. 2018, in preparation.
Data reconciliation method and application
Valle ECd, Kalid RdA, Secchi AR, Kiperstok A. Collection of benchmark test problems for data reconciliation and gross error detection and identification. Computers and Chemical Engineering. 2018; 111: 134-148.
Data reconciliation in industrial applications
Rafiee A, Behrouzshad F. Data reconciliation with application to a natural gas processing plant. Journal of Natural Gas
Science and Engineering. 2016; 31: 538-545.
The equality constraints for the total mass balance of the plant
Based on the measured data, the mass flow of output streams is greater than the mass flow of the raw feed to the plant which leads to an imbalance of about 7%.
Data reconciliation with gross error test indicates there is no gross error in the measurements.
Data reconciliation in industrial application
Weiss GH, Romagnoli JA, Islam KA. Data reconciliation-An industrial case study. Computers and Chemical
Engineering. 1996; 20(12): 1441-1449.
Based on simplified mass and energy balances, the overall heat transfer coefficient, estimated using reconciled measurement data could be used to determine better regeneration cycle time of the reactor.
Pharmaceutical continuous manufacturing
crystallization
Secondary manufacturing
UK EPSRC Center for Continuous Manufacturing and Crystallisation (CMAC)
▪ Continuous tablet manufacturing
Pfizer, the Portable Continuous Miniature and Modular (PCMM) design, 2016 Facility of the Year
Awards
Vertex, FDA approved in July 2015Orkambi (lumacaftor/ivacaftor)
Janssen, FDA approved in April 2016darunavir
▪ In the last decade, the FDA’s Quality by Design (QbD) and Process Analytical Technologies (PAT) initiatives have been supporting the development of continuous manufacturing based on scientific process understanding.
▪ The pharmaceutical continuous manufacturing has recently been moving from conceptual designs to pilot/production processes, specifically in secondary manufacturing.
[1] http://blogs.fda.gov/fdavoice/index.php/2016/04/continuous-manufacturing-has-a-strong-impact-on-drug-quality/
Pharmaceutical continuous manufacturing (PCM) is identified as an emerging technology.
CMA CQA CPP
Quality by Design (QbD)
CQA monitoring
CQA control
Process Analytical Technology (PAT)
Risk Assessment
RiskEvaluation
Risk management
▪ Controllability▪ Stability▪ Robustness▪ Control design▪ Level 1 PID loops▪ Level 2 MPC▪ Evaluation
▪ Risk scoring▪ Risk mapping▪ Risk assessing▪ Risk planning
▪ Design space: A state-space model
The Continuous manufacturing pilot plant in Purdue University
▪ API composition▪ Mixing uniformity▪ Powder flowrate
Rotation speed
Tablet hardness
Tensile strength
Dissolution rate
Tablets mass flowrate
Tablet weight
Main compression force
Pharmaceutical continuous manufacturing pilot plant at Purdue University
Process disturbance or failure
Schenck AccuRate PureFeed® AP-300 loss-in-weight
feeders
“Rat hole”
Sensor measurement fault
Innopharma EyeConTM for particle size analyzer
‘’fouling’’
Sensor measurement uncertainty
Mass flowrate =Δ Gain in mass
ΔTime
Sensor measurement uncertainty
Tablet weight =Mass flowrate
Turret speed ∗ Station No
Main Comp. kN
Tablet production rate, kg/hour
Averaged tablet weight, mg
Turret, RPM
Dosing, mm
Sotax Auto Test 4
* Feb 26, 2018, API% = 10.0, MCC% = 89.8%, SiO2% = 0.2
Continuous tableting process: no quality control
Natoli BLP-16
Tablets
Sotax Auto Test 4
▪ The built-in Level 0 control in Natoli press BLP16 can only maintain critical process parameters (CPPs) of dosing position, turret speed. With newer version of NP-400, additional control of main compression force, also one of the CPPs, is possible.
▪ However, no critical quality attributes (CQAs), e.g., the tablet weight, hardness, is real-time monitored and controlled.
▪ Variations in material properties, e.g., particle size, bulk density, or the interactions between dosing and turret speed, can affect the die filling and thus the tablet weight, main compression force, hardness, etc.
▪ Turret speed▪ Dosing position▪ Feeder speed
Die filling Pre-comp. Main-comp.
▪ Pre-compression force▪ Turret speed
▪ Main-compression force▪ Turret speed▪ Punch displacement
▪ Powder bulk density▪ Powder flowability
▪ Powder bulk density▪ Powder/tablet weight
▪ Powder compressibility▪ Powder/tablet weight
▪ Main compression force is one of the critical process parameters (CPPs) that determines the final drug properties, e.g., hardness, dissolution rate, etc.
▪ Tablet weight is a critical quality attribute (CQA) concerning about the potency of the drug, but also a good process indicator of tableting process operation in die filling and compressing. Furthermore, tablet weight will also reflect the changes in material properties. The interlink between main compression force and tablet weight play a critical role in tablet press process control and quality assurance.
▪ The lock symbol here also means that if you use the relationship between tablet weight and main compression force to reconciliate the measurement of tablet weight, it would not create any process control stability issue, since the turret speed and dosing position are not included in the data reconciliation.
▪ Data reconciliation for tablet weight
Natoli BLP-16
Mettler Toledo ME4001E
Main comp. force
Kawakita model
Dell Laptop
Formulation characterization: compressibility
Data reconciliation, the power to combine knowledge, speed, accuracy, and precision
Tablets
Sotax Auto Test 4
* Feb 19, 2018, API% = 10.0, MCC% = 89.8%, SiO2% = 0.2
Changes in material properties -> compressibility
𝜌𝐶 = 0.2499
𝜌𝐶 = 0.2499
𝜌𝐶 = 0.2577
𝜌𝐶 = 0.2607
Tablet press data reconciliation with open-loop operation
Tablet press data reconciliation with closed-loop operation
▪ No process dynamic information was removed or filtered out, as the data reconciliation using tablet weight and main compression force was not interacting with the Level 2 MPC process control loop.
▪ The use of reconciliated tablet weight measurement in process control loop should be possible, improving the process control performance.
▪ NOTE: At this stage, the Level 2 MPC process
control and the data reconciliation were running independently.
Feb 5, 2018, API% = 10.0, MCC% = 89.8%, SiO2% = 0.2
▪ No process dynamic information was removed or filtered out, as the data reconciliation using tablet weight and main compression force was not interacting with the Level 2 MPC process control loop.
▪ The use of reconciliated tablet weight measurement in process control loop should be possible, improving the process control performance.
Tablet press data reconciliation with closed-loop operation
▪ NOTE: At this stage, the Level 2 MPC process
control and the data reconciliation were running independently.
Feb 5, 2018, API% = 10.0, MCC% = 89.8%, SiO2% = 0.2
Data reconciliation under different tuning strategies
Slow update with good initial value of 𝜌𝐶
Slow update with rough initial value of 𝜌𝐶
Fast update with rough initial value of 𝜌𝐶
0.2600
0.2600
0.2600
Level 1 control scheme for Natoli Tablet Press
▪ Process control development
Level 2 control scheme for Natoli Tablet Press
Feb 26, 2018, API% =10, MCC% =89.8, SiO2% = 0.2
▪ Open loop operation for the tablet press to reach steady state
▪ At the mean time, the model in the Data Reconciliation was also automatically calibrated and verified with Auto Test (AT4) measurement of tablet weight
▪ The Level 1 control showed promising performance with the reconciliated measurements of tablet weight and tablet production rate.
▪ Disturbance due to the longer duration of AT4 gate opening to take tablet samples was introduced, the Level 1 control was capable to bring the process back to normal after the abnormal disturbance.
▪ The interaction between dosing and turret speed under Level 1 control can also be observed at time 1600s, when the reduced turret speed causing variations in the dosing, main compression force, and turret speed.
Integration data reconciliation with close-loop operation (1)
Open loop Level 1 close loop
Integration data reconciliation with close-loop operation (2)
Open loop Level 1 close loop
Feb 27, 2018, API% =10, MCC% =89.8, SiO2% = 0.2
▪ Set-point changes for both tablet weight and production rate were taken, the Level 1 control also showed promising control performance.
Open loop Level 1 closed loop
AT4 samplingBucket changingTurret speed changing
Feb 27, 2018, API% =10, MCC% =89.8, SiO2% = 0.2
Data reconciliation to reject process error and filter process noise
Feb 27, 2018, API% =10, MCC% =89.8, SiO2% = 0.2
𝜌𝐶 = 0.2575
𝜌𝐶 = 0.2610
Data reconciliation to update the model parameter: critical density
▪ The initial value of critical density was obtained from previous run with raw materials from the same tanks.
Adaptive moving window time to collect 100 tablets: turret 13 RPM
Adaptive moving window time to collect 100 tablets: turret 20 RPM
Operation panel for tablet press Level 1 control
Comparison of Level 0 and Level 1 control
Level 0 control
Level 1 control
Mar 12, 2018, API% =10, MCC% =89.8, SiO2% = 0.2
▪ Relative density 𝜌𝑐 updated by data reconciliation▪ Both updated from initial value of 0.2575 to 0.2645▪ Auto calibration of the tablet weight measurement▪ Almost consistent material properties of the two runs
Comparison of Level 1 and Level 2 control performance
Better control performance was obtained with the Level 2 MPC control for the process set point change to the tablet weight while maintaining the same tablet flow. Unknown disturbance occurred to the tablet weight and main compression force, as highlighted in red, could be due to the feeding of the powder into die. After its removal, the system was back to normal and under control.
Level 1 Level 2
04/09/2018, API 10%, MCC 89.8%, SiO2 0.2%
Rate Severity Probability Controllability
1 No effect None < 1 occurrence during the continuous production in 5 years (e.g., control instrument failure)
Very rare All the applicablemeasurements &controls are in place
Always undercontrol
3 No patient effectProcess performance decreasing (e.g., divert of non-conformingproduct)
Low 1 occurrence during the continuous production in 1 to 2years (e.g., calibration errors)
Rare Most of the applicablemeasurements & controls are in place
High credibility of under control
5 No patient effectProcess performance decreasing(e.g., large amount of non-conforming product, process shut-down)
Medium < 1 occurrence during a single continuous production campaign (e.g., variance in raw material properties)
Sometimes Some of the applicablemeasurement &controls are in place
Moderate credibility of under control
7 Potential patient effect (e.g., large variance in critical quality attributes)
High > 1 occurrence during a single continuous production campaign (e.g., raw material reloading)
Frequent/No information
Few of the applicable measurements & controls are in place
Remote credibility of under control
9 Significant patient effect (e.g., large variance in target product quality profile, loss of clinicalperformance)
Very high
At any time during the continuous operation (e.g., machinery vibration)
High frequent None of the applicablemeasurements & controls are in place
No control
[1] Potter C. PQLI application of science- and risk-based approaches (ICH Q8, Q9, and Q10) to existing products. Journal of Pharmaceutical Innovation. 2009;4(1):4-23.
▪ Risk analysis and assessment
A risk mapping, from the perspective of process control, will be designed based on scores of severity and probability of risks during continuous manufacturing, upon which the proposed real-time release control strategies will be evaluated to give corresponding controllability scores for each risk.
45 < R227 <R1 <451< R0 < 27
Severity
1None
3Low
5Medium
7High
9Very high
Pro
bab
ility
9High frequent
7Frequent
5Sometimes
3Rare
1Very rare
1 Always under control
3 High credibility of under control
5 Moderate credibility of under control
7 Remote credibility of under control
9 No control
Machinery vibration
Raw material propertyvariation
System interaction
Level 0 control risk analysis in tablet press
Sensor noise
Sluggish Level 0 or machine control
Bulk densitychanges by shear
45 < R227 <R1 <451< R0 < 27
Severity
1None
3Low
5Medium
7High
9Very high
Pro
bab
ility
9High frequent
7Frequent
5Sometimes
3Rare
1Very rare
1 Always under control
3 High credibility of under control
5 Moderate credibility of under control
7 Remote credibility of under control
9 No control
Machinery vibration
Raw material propertyvariation
System interaction
Sensor noise
Sluggish Level 0 or machine control
Level 1 control risk analysis in tablet press
Bulk densitychanges by shear
45 < R227 <R1 <451< R0 < 27
Severity
1None
3Low
5Medium
7High
9Very high
Pro
bab
ility
9High frequent
7Frequent
5Sometimes
3Rare
1Very rare
1 Always under control
3 High credibility of under control
5 Moderate credibility of under control
7 Remote credibility of under control
9 No control
Machinery vibration
Raw material propertyvariation
System interaction
Sensor noise
Sluggish Level 0 or machine control
Level 2 control risk analysis in tablet press
Bulk densitychanges by shear
▪ Conclusions and future work
▪ The importance of data reconciliation in reducing data imperfection was demonstrated for the real-time tablet weight measurement in the pharmaceutical continuous manufacturing.
▪ The current data reconciliation strategy is implemented to a subsystem of the tablet press based on the Kawakita model for main compression force and tablet weight.
▪ The future work would consider the implementation of data reconciliation to the feeding-blending system, as well as to the plant-wide system to reconcile more critical process parameters and critical quality attributes.
Acknowledgement
Prof. G.V. ReklaitisProf. Z.K. Nagy
M. Moreno J. Liu S. Ganesh
Prof. M. Gonzalez
Y. Bommireddy Dr. A. Giridhar Dr. D. Pai