2. Bridging the Gap with On-line Analytics
On-line Decision Support for Operations Personnel
– Product quality predictions
– Early process fault detection
Embedded On-line Analytics brings quality information, fault
detection, and abnormal situation knowledge to the operator –
bridging the gap between quality and control.
The PAT Guidelines issued by the FDA emphasized the use of
multivariate analytics as a means of reducing cost, improving
product quality in the pharmaceutical industry.
On-line Data Analytics is targeted for DeltaV v12.
QUALITY CONTROL
3. Information at the Operator Interface
Analytic
Process
Models
Evaluation
process
operation
Process measurements, lab and Truck
analysis over last year
Calculated Feed Composition
Process
measurements
Operator Interface
Predicted End of Batch
Quality
Fault Detection
TT
207
TC
207
TT
206
TC
206
Coolant return
Bioreactor
RSP
AT
205AT
204
FC
203
FC
201
FT
201
Feed
e.g. Glucose
AC
204
Reagent
e.g. Ammonia
FC
202
FT
202
Air
pH
AC
205
Dissolved
Oxygen
Vent
PT
208
PC
208
RSP
Charge
e.g. Media
FT
203 Coolant
supply
IT
209
LT
210
To Harvest
TT
207
TC
207
TC
207
TT
206
TC
206
TC
206
Coolant return
Bioreactor
RSP
AT
205AT
204
FC
203
FC
203
FC
201
FC
201
FT
201
Feed
e.g. Glucose
AC
204
AC
204
Reagent
e.g. Ammonia
FC
202
FC
202
FT
202
Air
pH
AC
205
AC
205
Dissolved
Oxygen
Vent
PT
208
PC
208
PC
208
RSP
Charge
e.g. Media
FT
203 Coolant
supply
IT
209
LT
210
To Harvest
Storage Tank Design
Tank
Design 2
Tank
Design 1
Tank
Design 3
Storage Tank Design
Tank
Design 2
Tank
Design 1
Tank
Design 3
4. 3 Step Monitoring Procedure
1. If either Fault Detection plot
exceeds or approaches the upper
control limit of 1.0, click on that
point in the trend and
-> Select the Parameter in the
lower corner of the screen that
contributed to the fault
2. Evaluate the parameter trends
from process operation standpoint
-> take corrective action if
necessary
3. Inspect impact of fault on quality
prediction plot to find out how
quality may be affected
Note: Use Up arrow to return to the
Analytics Overview.
If a fault is indicted in the analytics overview screen, then selecting the
batch number will bring up the Fault Detection view.
Analytics Overview
Quality Parameter Prediction
ContributionParameter Trend (s)
2
Fault Detection
31
5. Example – Low Hot Oil Flow Rate
When the hot oil
valve is opened,
the flow rate is
much lower
than normal
The lower flow
rate impacts the
time needed for
the mixer to
reach target
temperature –
extending batch
time
6. Example – Low Hot Oil Flow Rate
Fault shows up in
Indicator 2 deviating
above 1.
To find the cause of
the fault, select the
point of maximum
deviation and then
choose the
Contribution Tab or
select the
parameters that
contribute most to
the fault - shown in
the lower corner of
the screen.
7. Example – Low Hot Oil Flow Rate
The trend
confirms that the
media flow rate
is ~ 2 liters/sec
which is much
lower than the
normal flow rate
of 4
liters/second.
8. Example – Low Hot Oil Flow Rate
The
prediction
plot
confirms
that the low
oil flow rate
has no
impact on
the
predicted
product
density
9. Prediction of Product Density
For the Saline
process, the
prediction of
product density
has proven to be
very accurate
even though
variations in the
salt bin level are
a major source of
variation in the
processing
conditions.
10. Example – pH Sensor Drift
O2
Bioreactor
VSD
VSD
TC
41-7
AT
41-4s2
AT
41-4s1
AT
41-2
TT
41-7
AT
41-6
LT
41-14
Glucose
Glutamine
pH
DO
Product Concentration
VSD
VSD
AC
41-4s1
AC
41-4s2
Media
Glucose
Glutamine
VSD
Bicarbonate
AY
41-1
AC
41-1
Splitter
AC
41-2
AY
41-2
Splitter
CO2
Air
Level
Drain
0.002 g/L
7.0 pH
2.0 g/L
2.0 g/L
37
o
C
VSD
Inoculums
VSD
PT
41-3 Vent
MFC
MFC
MFC
PC
41-3
AT
41-15
Coating of the
sensor may
introduce a bias
into the pH
measurement -
resulting in a shift
of the pH
maintained in the
reactor.
May impact cell
growth rate and
product formation
AT
41-1
11. Example – pH Sensor Drift
Fault shows up
as an explained
and unexplained
change –
deviation above
1.
To find the cause
of the fault, select
the point of
maximum
deviation and
then choose the
Contribution Tab.
12. Example – pH Sensor Drift
Drift in the pH
measurement
is reflected in
the pH
measurement
and controller
output.
A trend of the
pH and pH
controller
output can be
obtained by
clicking on
media flow
parameter in
the
contribution
screen.
13. Example – pH Sensor Drift
Impact of the
change in
measurement
bias is show as
an immediate
change in pH.
14. Example – pH Sensor Drift
Longer term the
faulty pH
measurement is
reflected in an
abnormally low
reagent addition
being used to
maintain the
indicated pH.
15. Learning More
A workshop is being offered at Emerson Exchange on
data analytics and field trail at Lubrizol, Rouen. The
schedule for this workshop is:
08-167 Batch Process Analytics (PA) – An In
Depth Update
– Tuesday, Room 206B, 10:00 AM
– Thursday, Room 206A, 11:00 AM
16. Spectral Analyzers
Spectral analyzers may be used at critical
points throughout the process.
– Pharmaceutical - inspection of feedstock,
blend uniformity, granulation, drying and
coating and particle size analysis. Online
QA/QC tool for production.
– Chemical - acid value, adhesive content,
cure, melt index, and polymer processes -
reaction monitoring
– Refinery, petrochemical - fuel production
monitoring
A wide variety of commercial on-line, at-line,
and laboratory spectral analyzers are
available.
Calibration of an NIR analyzer is based on
use of spectral data to develop principal
component analysis(PCA) and projection of
latent structures (PLS) models.
17. Example: NIR Analyzers
Careful development of a set of
calibration samples and their use
in PCA/PLS model development
is the basis for near-infrared
analytical methods.
For purposes of analysis, the
spectral data for a sample
should be saved and accessed
as one set of data e.g. an
array.
3-D plotting of spectral data can
be helpful in screening samples
and in analyzing on-line use of
spectral data.
Off-line PCA/
PLS Model
Development
On-line Quality
Parameter
Prediction
Historian
Array/Data Set
Application
station
NIR Analyzer
Controller
VIM
Interface
3-D Plot of
Spectral Data
18. Learning More
The technical feasibility of providing 3-D plotting and historian
collection of array data has been explore and the value of such
a capability demonstrated in at a field trail conducted on an
absorber and stripper process unit at UT Pickle Research
Center, Austin, TX.
Two presentations on this field trail are scheduled for Emerson
Exchange.
04-132 Process Analysis Using 3D plots
– Tuesday, 3:00:00 PM, Room 207B
– Thursday, 3:15:00 PM, Room 201