12/16/2018

1

McMaster University

Continuous manufacturing & robust process control for safe pharmaceutical manufacturing

Qinglin Su, Claire Liu, Marcial Gonzales, Rex Reklaitis, Zoltan K. NAGY

Davidson School of Chemical Engineering

Purdue University, West Lafayette, IN

Deember 12, 2018

Purdue Process Safety and Assurance Center (P2SAC)

2

Why do we need to use advanced control techniques in the pharmaceutical industries?

Lack of real-time measurements

Lack of models

Lack of actuators

Quality assurance

Process validation

Continuous verification

Safety

Large variations in product pipeline

Downward pressure by healthcare providers

Growth in emerging markets

12/16/2018

2

3

Integrated

Intensified

Monitored

Controlled

Safer

Advanced control enables a new generation of integrated,

intensified & intelligent manufacturing systems with

drastically improved flexibility, predictability, stability, controllability & safety

Advanced control enables a new generation of integrated,

intensified & intelligent manufacturing systems with

drastically improved flexibility, predictability, stability, controllability & safety

Nagy&Braatz, Handbook of Ind. Cryst., 2013; Nagy&Braatz, Annu. Rev. Chem. Biomol. Eng., 2012

Holistic systems view of Integrated & Continuous Manufacturing of Pharmaceuticals

4

Develop a methodology for integrated Design-Control-Safety (DCS) platform for pharmaceutical processes

Safety by control: Fault tolerant control (FTC) for safe plant operation

risk-based control system design and validation

Safety by design: Improve inherent process performance and safety via adoption of continuous manufacturing of pharmaceuticals

Overall Project Objectives

12/16/2018

3

5

Quality-by-Control (QbC) paradigm

6

Past and Present Target Processes for Implementation

Continuous crystallization MSMPR (cascade of MSMPRs)

– Advantages: equipment in place, easier use of PAT

– Disadvantage: similar to batch processes (large residence time distribution, low heat transfer area/volume ratio)

Plug flow crystallizers (PFC) – different technologies– Advantages: faster, smaller volume, higher product consistency

– Disadvantage: need for new equipment, difficulty of using PAT

Continuous tablet manufacturing

12/16/2018

4

7

Continuous Crystallization

Two main technologies:

MSMPR (cascade of MSMPRs) – stirred tank or dynamic baffled– Advantages: equipment in place, easier use of PAT

– Disadvantage: similar to batch processes (large residence time distribution, low heat transfer area/volume ratio)

Plug flow crystallizers (PFC) – different technologies– Advantages: faster, smaller volume, higher product consistency

– Disadvantage: need for new equipment, difficulty of using PAT

Typical problems:– Optimal operation

– Control of CQAs

– Startup

– Fouling

8

Cascade MSMPR system

12/16/2018

5

9

NMPC – attainable regions for control for safety

100 200 300 400 500 600 700 8000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Mean size [m]

Yie

ld [-

]

12

3

Nucleation ControlGrowth ControlAntisolvent ControlTemperature ControlGlobal Control 0 100 200 300 400 500

100

300

500

Siz

e [

m]

0 100 200 300 400 5000.5

0.75

1

Yie

ld [-

]

(a) 21 1 3Set-point:

Yield

Size

0 100 200 300 400 500

0

5

10

Sta

ge 1 a

ntis

olven

tadd

ition

rate

(F1) [m

l/min

]

0 100 200 300 400 500

0

5

10

Sta

ge 2 a

ntisolvent

add

ition ra

te (F

2) [m

l/min]

Time [min]

(b)

F1

F2

Point Yield Size

1 Medium Medium

2 High Small

3 Low Large

Attainable region depends on control approach

Depending on safety constraints change of control structure needed

Robust fault-tolerant control

Safer control

10

Continuous plug flow crystallization systems

12/16/2018

6

11

Encrustation is a process by which solute (API) deposits on to the crystallizer surface.

Encrust reduces heat transfer, and subsequently product quality and yield.

Causes non-steady state process and leads to blockage.

Fouling or encrustation

12

On-Off Fault-tolerant Feedback Control

• The Encrust and CSD controller assures state of control.

• The CSD feedback controller detects the upper and lower bounds and trigger the point of collection (blue).

• Example of level-1 control, in which an active process control system is used to monitor CQAs in real-time.

• Process will work indefinitely due to QbC approach despite fault due to encrustation

12/16/2018

7

13

Purdue Continuous Tableting Pilot Plant

Hierarchical Control System Development and Analysis

14

Pharmaceutical continuous manufacturing pilot plant at Purdue

Feeders: Schenck AccuRate AP‐300, DP‐4Blenders: Gericke GCM 250Roller Compactor: Alexanderwerk WP120Tablet Press: Natoli BLP‐16, Natoli NP‐400

12/16/2018

8

15

Layer 2

•Use of dynamic models for prediction•Model Predictive Control, Adaptive Control, Gain Scheduling, etc.•Integrated across multiple unit operations

Layer 1

•Based on PAT to measure CQAs and closed loop control (PFC)•PID based more advanced control schemes•Single or multiple unit operations

Layer 0

•Built‐in control systems by manufacturers at unit operation level•Simple PID control•Single unit operation

Hierarchical control architecture

R1 Medium

R0 Low

R2 High

Risk levels & failure mode

Sensor accuracy, precision, sampling time Sensor location Control architecture reconfiguration Control method Controller tuning⁞

Evaluation & regulatory filling

Good

Excellent

Fail

T2P: Time to productM2P: Magnitude to productMRI:  Morari’s resilience index

Continuous improvement 

e.g. Good Manufacturing Practices (GMP) guidelines

Manufacturing processes are clearly defined and controlled. All critical processes are validated to ensure consistency and compliance with specifications.

Manufacturing processes are controlled, and any changes to the process are evaluated. Changes that have an impact on the quality of the drug are validated as necessary.

⁞

Formulation optimization Process reconfiguration ⁞

Risk-based systematic design and analysis of feedback control strategies for continuous manufacturing processes

[1] Su Q, Moreno M, Giridhar A, Reklaitis GV, Nagy ZK. Journal of Pharmaceutical Innovation. 2017;12: 327-346.

16

Level 1 PID control scheme for Natoli Tablet Press

Cascade control: Tablet weight master loop (by main compression force), dosing slave loop (by dosing level)

Production rate by turret speed

12/16/2018

9

17

Level 2 MPC control scheme for Natoli Tablet Press

A linear MPC control of 2 by 2 with an additional constraint variable of main compression force was implemented in DeltaV platform, controlling the tablet weight and production rate by manipulating the fill depth and turret speed.

18

Natoli BLP‐16

Mettler Toledo ME4001E 0 500 1000 1500 2000 2500

Time (sec)

0

100

200

300

400

500

Main comp. force

Kawakita model

Dell Laptop

0 100 200 300 400 500

Tablet weight (mg)

0

2

4

6

8

10

12

AT4-2%AT4-10%MP

c=0.2499

Formulation characterization: compressibility

Data reconciliation ‐ the power to combine knowledge, speed, accuracy, and precision of sensors

Tablets

0 500 1000 1500 2000 2500

Time (sec)

0

100

200

300

400

500

Sotax Auto Test 4

* Feb 19, 2018, API% = 10.0, MCC% = 89.8%, SiO2% = 0.2

Changes in material properties ‐> compressibility

12/16/2018

10

19

Comparison of Level 1 PID and Level 2 MPC control performance (with Level 2 DR)

June 15, 2018, API% =10, MCC% =89.8, SiO2% = 0.2

Data reconciliation reinitialized

Sotax AT4

20

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

Alwaysundercontrol

3 No patient effectProcess performance decreasing (e.g., divert of non-conforming product)

Low 1 occurrence during the continuous production in 1to 2 years (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.

12/16/2018

11

21

45 < R227 <R1 <451< R0 < 27

Severity

1None

3Low

5Medium

7High

9Very high

Probability

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

Sensor calibration error

Raw material propertyvariation

System interaction

Sensor noise

Sluggish Level 0 or machine control

Bulk densitychanges by shear

Level 0 control risk analysis in tablet press

22

45 < R227 <R1 <451< R0 < 27

Severity

1None

3Low

5Medium

7High

9Very high

Probability

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

SensorCalibrationerror

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

12/16/2018

12

23

45 < R227 <R1 <451< R0 < 27

Severity

1None

3Low

5Medium

7High

9Very high

Probability

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

SensorCalibrationerror

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

24

Conclusions

Improved safety via better control and use of CM

Continuous pharmaceutical manufacturing enhances safety of manufacturing processes

Developed systematic hierarchical control structure

Quality-by-control (QbC) - new paradigm for risk-based control system development for safer plant operation

Exemplified for continuous crystallization & tablet manufacturing

24