Slide 1
Advances in Bioreactor Advances in Bioreactor Modeling and ControlModeling and Control
Greg McMillan, Trish Benton, and Michael BoudreauInterphex – March 17, 2009
http://www.modelingandcontrol.com/http://www.easydeltav.com/controlinsights/index.asp
[File Name or Event]Emerson Confidential27-Jun-01, Slide 2 Slide 2
CoauthorsCoauthorsCoauthors
Greg McMillan - Principal Consultant, CDI Process and Industrial at Emerson Trish Benton – Life Sciences Consultant, Broadley-James CorporationMike Boudreau - Director of Bioreactor Manufacturing and Automation, Broadley-James Corporation
[File Name or Event]Emerson Confidential27-Jun-01, Slide 3 Slide 3
AgendaAgendaAgenda
Mammalian Bioreactor ModelFlexible and Convenient KineticsVirtual Plant ConceptsTypes of Process ResponsesSingle Use Bioreactor (SUB) for Wireless TestsWirelessHART NetworkWireless PID FeaturesWireless SUB Results for pH and Temperature LoopsControl Studies of Wireless PID Control for pHControl Studies of Wireless PID Control for At-Line AnalyzersConclusionsSources for More Info on Modeling and Effect of Sample TimeReferences
[File Name or Event]Emerson Confidential27-Jun-01, Slide 4 Slide 4
Differences between Fungal or Bacterial and Mammalian Bioreactor ModelsDifferences between Fungal or Bacterial and Differences between Fungal or Bacterial and Mammalian Bioreactor ModelsMammalian Bioreactor Models
Kinetics– More than twice as many kinetic terms and parameters– Generalized Michaelis-Menten kinetic parameters– Slower product formation rate and batch cycle timeMass transfer
– Significantly less agitation and bubblesComponents
– Glutamine or glutamate utilization– Lactate and ammonia formationReagents
– Carbon dioxide – Sodium bicarbonateSparge
– Oxygen, carbon dioxide, and inert addition besides airOverlay
– Air, oxygen, carbon dioxide, and inert sweep– No manipulation of overhead pressure for dissolved oxygen control
[File Name or Event]Emerson Confidential27-Jun-01, Slide 5 Slide 5
Mammalian Growth and Product Formation RatesMammalian Growth and Product Formation RatesMammalian Growth and Product Formation Rates
Bioreactor models can handle any user expressions for kinetic rate factors
vTvHvOvbvavsvsvv rrrrrrr ∗∗∗∗∗∗∗= +2max 21µµ
Maximum Specific Growth Rate
(per hr) Growth Rate Factors (0-1)
glucose and glutamine substrates (rvs1) (rvs2), lactic acid (rva), ammonia base (rvb), dissolved oxygen (rvO2), pH (rvH+), and temperature (rvT)
TpHpOpspspp rrrrru ∗∗∗∗∗= +2max 21µ
Maximum Specific Product Formation Rate(g product/g cell per hr)
Product Formation Rate Factors (0-1)glucose and glutamine substrates (rps1) (rps2),
dissolved oxygen (rpO2), pH (rpH+), and temperature (rpT)
[File Name or Event]Emerson Confidential27-Jun-01, Slide 6 Slide 6
Flexible Michaelis-Menten KineticsFlexible MichaelisFlexible Michaelis--Menten KineticsMenten Kinetics
[ ] [ ]jii
i
jii
ji
KXX
KXK
jir21
1
++ ∗=Inhibition parameter Limitation parameter
ConcentrationGrowth or formation rate factor (0 - 1)
Monod Equation
Initialization of kinetic parameters:
If the limitation or inhibition effect is significant the limitation and inhibition parameters are set to 0.1x and 10x, respectivelythe expected set point
If the limitation or inhibition effect is negligible the limitation and inhibition parameters are set to 0 and 100, respectively
[File Name or Event]Emerson Confidential27-Jun-01, Slide 7 Slide 7
Glucose Growth Rate FactorGlucose Growth Rate FactorGlucose Growth Rate Factor
Michaelis-Menten Cell Growth Rate Kinetics
0.0000
0.1000
0.2000
0.3000
0.4000
0.5000
0.6000
0.7000
0.8000
0.9000
1.0000
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Glucose Concentration (g/Liter)
G
luco
se G
row
th R
ate
Fact
or
[File Name or Event]Emerson Confidential27-Jun-01, Slide 8 Slide 8
Convenient pH Model KineticsConvenient pH Model KineticsConvenient pH Model Kinetics
[ ]2maxmin
maxmin
)()()()()(
optpHpHpHpHpHpHpHpHpHpH
vHr
−−−∗−−∗−=+
pHmax = maximum pH for viable cells (8 pH)
pHmin = minimum pH for viable cells (6 pH)
pHopt = optimum pH for viable cell growth (6.8 pH)
[File Name or Event]Emerson Confidential27-Jun-01, Slide 9 Slide 9
pH Growth Rate FactorpH Growth Rate FactorpH Growth Rate Factor
Cardinal pH Model Kinetics
0.0000
0.1000
0.2000
0.3000
0.4000
0.5000
0.6000
0.7000
0.8000
0.9000
1.0000
6.00 6.20 6.40 6.60 6.80 7.00 7.20 7.40 7.60 7.80 8.00
pH
pH
Gro
wth
Rat
e Fa
ctor
[File Name or Event]Emerson Confidential27-Jun-01, Slide 10 Slide 10
Convenient Temperature Model KineticsConvenient Temperature Model KineticsConvenient Temperature Model Kinetics
[ ]])2()()()([)()()(
minmaxminmin
2minmax
TTTTTTTTTTTTTTT
vT optoptoptoptoptr ∗−+∗−−−∗−∗−
−∗−=
Tmax = maximum temperature for viable cells (45 oC)
Tmin = minimum temperature for viable cells (5 oC)
Topt = optimum temperature for product formation (37 oC)
[File Name or Event]Emerson Confidential27-Jun-01, Slide 11 Slide 11
Temperature Growth Rate FactorTemperature Growth Rate FactorTemperature Growth Rate Factor
Cardinal Temperature Model Kinetics
0.0000
0.1000
0.2000
0.3000
0.4000
0.5000
0.6000
0.7000
0.8000
0.9000
1.0000
5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00
Temperature
Te
mpe
ratu
re G
row
th R
ate
Fact
or
[File Name or Event]Emerson Confidential27-Jun-01, Slide 12 Slide 12
Virtual PlantVirtual PlantVirtual Plant
Advanced Control Modules
Process Model
Virtual PlantLaptop or Desktop
or Control System Station
[File Name or Event]Emerson Confidential27-Jun-01, Slide 13 Slide 13
Top Ten Reasons I Use a Virtual Plant Top Ten Reasons I Use a Virtual Plant Top Ten Reasons I Use a Virtual Plant
(10) You can’t freeze, restore, and replay an actual plant batch(9) No separate programs to learn, install, interface, and support(8) No waiting on lab analysis(7) No raw materials(6) No environmental waste(5) Virtual instead of actual problems(4) Batches are done in 14 minutes instead of 14 days(3) Plant can be operated on a tropical beach(2) Last time I checked my wallet I didn’t have $100,000K(1) Actual plant doesn’t fit in our suitcase
[File Name or Event]Emerson Confidential27-Jun-01, Slide 14 Slide 14
Virtual Plant Knowledge SynergyVirtual Plant Knowledge SynergyVirtual Plant Knowledge Synergy
Dynamic Process Model
OnlineData Analytics
Model PredictiveControl
Loop MonitoringAnd Tuning
DCS batch and loopconfiguration, displays,
and historian
Virtual PlantLaptop or DesktopPersonal Computer
OrDCS Application
Station or Controller
Embedded Advanced Control Tools
EmbeddedPAT Tools
Process Knowledge
[File Name or Event]Emerson Confidential27-Jun-01, Slide 15 Slide 15
Self-Regulating ProcessSelfSelf--Regulating ProcessRegulating Process
∆X
∆Y
θp τp
Kp = ∆Y / ∆X(Self-Regulating Process Gain)
0.63∗∆Y
X
Y
ProcessDead Time
Self-Regulating Process Time Constant
Noise Band
New Steady State
Process Output (Y)& Process Input (X)
Time (t)
Response to change in process input with controller in manual
Most continuous processes have a self-regulating response (PV lines out in manual)
[File Name or Event]Emerson Confidential27-Jun-01, Slide 16 Slide 16
Integrating ProcessIntegrating ProcessIntegrating Process
Time (t)θp
Ki = { [ ∆Y2 / ∆t2 ] − [ ∆Y1 / ∆t1 ] } / ∆X(Integrating Process Gain)
∆X
ramp rate is∆Y1 / ∆t1
ramp rate is∆Y2 / ∆t2
X
Y
Process Output (Y)& Process Input (X)
ProcessDead Time
Response to change in process input with controller in manual
To prevent slow rollingoscillations and overshootfrom integral action, the product of the controller gain (Kc) and reset time (Ti) should satisfy the limit:Kc ∗ Ti > 4 / Ki
Most batch processes have an integrating response (PV ramps in manual)
[File Name or Event]Emerson Confidential27-Jun-01, Slide 17 Slide 17
Runaway ProcessRunaway ProcessRunaway ProcessProcess Output (Y)& Process Input (X)
∆X
θp
Kp = ∆Y / ∆X(Runaway Process Gain)
1.72∗∆YX
Y
ProcessDead Time
Runaway Process Time Constant
Time (t)τp’
∆Y
Noise Band
Acceleration
pH and exponential growth phase appear to have a runaway response (PV accelerates in manual)
Response to change in process input with controller in manual
[File Name or Event]Emerson Confidential27-Jun-01, Slide 18 Slide 18
Installation at Broadley JamesInstallation at Broadley JamesInstallation at Broadley James
Hyclone 100 liter Single Use Bioreactor (SUB) Rosemount WirelessHART gateway and transmitters for measurement and control of pH and temperature. (pressure monitored)BioNet lab optimized control system based on DeltaV
[File Name or Event]Emerson Confidential27-Jun-01, Slide 19 Slide 19
WirelessHART Network TopologyWirelessHART Network TopologyWirelessHART Network Topology
Network Manager
Wireless Field Devices– Relatively simple - Obeys Network Manager– All devices are full-function (e.g., must route)
Adapters– Provide access to existing HART-enabled Field
Devices– Fully Documented, well defined requirements
Gateway and Access Points – Allows access to WirelessHART Network from
the Process Automation Network– Gateways can offer multiple Access Points for
increased Bandwidth and Reliability– Caches measurement and control values– Directly Supports WirelessHART Adapters– Seamless access from existing HART
ApplicationsNetwork Manager
– Manages communication bandwidth and routing
– Redundant Network Managers supported – Often embedded in Gateway– Critical to performance of the network
Handheld– Supports direct communication to field device– For security, one hop only communication
[File Name or Event]Emerson Confidential27-Jun-01, Slide 20 Slide 20
WirelessHART FeaturesWirelessHART FeaturesWirelessHART FeaturesWireless transmitters provide nonintrusive replacement and diagnosticsWireless transmitters automatically communicate alerts based on smart diagnostics without interrogation from an automated maintenance systemWireless transmitters eliminate the questions of wiring integrity and terminationWireless transmitters eliminate ground loops that are difficult to track downNetwork manager optimizes routing to maximize reliability and performanceNetwork manager maximizes signal strength and battery life by minimizing the number of hops and preferably using routers and main (line) powered devicesNetwork manager minimizes interference by channel hopping and blacklistingThe standard WirelessHART capability of exception reporting via a resolution setting helps to increase battery lifeWirelessHART control solution, keeps control execution times fast but a new value is communicated as scheduled only if the change in the measurement exceeds the resolution or the elapsed time exceeds the refresh timePIDPLUS and new communication rules can reduce communications by 96%
[File Name or Event]Emerson Confidential27-Jun-01, Slide 21 Slide 21
Traditional and Wireless PID (PIDPLUS)Traditional and Wireless PID (PIDPLUS)Traditional and Wireless PID (PIDPLUS)
PID integral mode is restructured to provide integral action to match the process response in the elapsed time (reset time is set equal to process time constant)PID derivative mode is modified to compute a rate of change over the elapsed time from the last new measurement valuePID reset and rate action are only computed when there is a new valuePID algorithm with enhanced reset and rate action is termed PIDPLUS
[File Name or Event]Emerson Confidential27-Jun-01, Slide 22 Slide 22
Automatically Identified SUB Temperature DynamicsAutomatically Identified SUB Temperature DynamicsAutomatically Identified SUB Temperature Dynamics
[File Name or Event]Emerson Confidential27-Jun-01, Slide 23 Slide 23
Wireless SUB Temperature Loop Test ResultsWireless SUB Temperature Loop Test ResultsWireless SUB Temperature Loop Test Results
[File Name or Event]Emerson Confidential27-Jun-01, Slide 24 Slide 24
Wireless SUB pH Loop Test ResultsWireless SUB pH Loop Test ResultsWireless SUB pH Loop Test Results
[File Name or Event]Emerson Confidential27-Jun-01, Slide 25 Slide 25
Elimination of Ground Noise Spikes by WirelessElimination of Ground Noise Spikes by WirelessElimination of Ground Noise Spikes by Wireless
Wired pH ground noise spike
Temperature compensated wireless pH controlling at 6.9 pH set point
Incredibly tight pH control via 0.001 pH wireless resolution setting still reduced the number of communications by 60%
[File Name or Event]Emerson Confidential27-Jun-01, Slide 26 Slide 26
Control Studies of pH Resolution and Feedforward(Bioreactor batch running 500x real time)Control Studies of pH Resolution and FeedforwardControl Studies of pH Resolution and Feedforward(Bioreactor batch running 500x real time)(Bioreactor batch running 500x real time)
Batch 1 Batch 2 Batch 1 Batch 2
Batch 3 Batch 4 Batch 3 Batch 4
Batches 1 and 2 have 0.00 pH resolution and standard PID
Feedforward Feedforward
Feedforward Feedforward
Batches 3 and 4 have 0.01 pH resolution and standard PID
[File Name or Event]Emerson Confidential27-Jun-01, Slide 27 Slide 27
Control Studies of pH Resolution and Feedforward(Bioreactor batch running 500x real time)Control Studies of pH Resolution and FeedforwardControl Studies of pH Resolution and Feedforward(Bioreactor batch running 500x real time)(Bioreactor batch running 500x real time)
Batch 5 Batch 6 Batch 5 Batch 6
Batch 7 Batch 8 Batch 7 Batch 8
Feedforward Feedforward
Feedforward Feedforward
Batches 5 and 6 have 0.02 pH resolution and standard PID
Batches 7 and 8 have 0.04 pH resolution and standard PID
[File Name or Event]Emerson Confidential27-Jun-01, Slide 28 Slide 28
Control Studies of pH Refresh Time and Feedforward(Bioreactor batch running 500x real time)
Control Studies of pH Refresh Time and FeedforwardControl Studies of pH Refresh Time and Feedforward(Bioreactor batch running 500x real time)(Bioreactor batch running 500x real time)
Batch 9 Batch 10 Batch 9 Batch 10
Batch 11 Batch 12 Batch 11 Batch 12
Feedforward Feedforward
Feedforward Feedforward
Batches 9 and 10 have 30 sec x 500 refresh time and standard PID
Batches 11 and 12 have 30 sec x 500 refresh time and wireless PID
[File Name or Event]Emerson Confidential27-Jun-01, Slide 29 Slide 29
Control Studies of Glucose Sample Time and Feedforward (Bioreactor batch running 1000x real time)Control Studies of Glucose Sample Time and Control Studies of Glucose Sample Time and Feedforward Feedforward (Bioreactor batch running 1000x real time)(Bioreactor batch running 1000x real time)
Continuous FF-NoStandard PID
Continuous FF-YesStandard PID
11 hr Sample FF-NoStandard PID
11 hr Sample FF-YesStandard PID
11 hr Sample FF-NoWireless PID
11 hr Sample FF-YesWireless PID
Batch 1 Batch 2 Batch 3 Batch 4 Batch 5 Batch 6
Glucose Concentration
x1000
Batch 1: Glucose Probe (Continuous - No Delay) + Feed Forward - No + Standard PIDBatch 2: Glucose Probe (Continuous - No Delay) + Feed Forward - Yes + Standard PIDBatch 3: Glucose Analyzer (11 Hr Sample Delay) + Feed Forward - No + Standard PIDBatch 4: Glucose Analyzer (11 Hr Sample Delay) + Feed Forward - Yes + Standard PIDBatch 5: Glucose Analyzer (11 Hr Sample Delay) + Feed Forward - No + Wireless PIDBatch 6: Glucose Analyzer (11 Hr Sample Delay) + Feed Forward - Yes + Wireless PID
[File Name or Event]Emerson Confidential27-Jun-01, Slide 30 Slide 30
Control Studies of Reset Factor & Wireless PID for Real Time Integrating Process (20 sec analyzer sample time)Control Studies of Reset Factor & Wireless PID for Control Studies of Reset Factor & Wireless PID for Real Time Real Time IntegratingIntegrating Process Process (20 sec analyzer sample time)(20 sec analyzer sample time)
Reset Factor = 0.5
Standard PID Standard PID Standard PID
Reset Factor = 1.0 Reset Factor = 2.0
Wireless PID Wireless PID Wireless PID
Reset Factor = 0.5 Reset Factor = 1.0 Reset Factor = 2.0
Improvement in stability is significant for any integrating process with analyzer delay
[File Name or Event]Emerson Confidential27-Jun-01, Slide 31 Slide 31
Control Studies of Lambda Factor & Wireless PID for Real Time Integrating Process (20 sec analyzer sample time)Control Studies of Lambda Factor & Wireless PID for Control Studies of Lambda Factor & Wireless PID for Real Time Real Time IntegratingIntegrating Process Process (20 sec analyzer sample time)(20 sec analyzer sample time)
Lambda Factor = 1.5
Wireless PID Wireless PID Wireless PID
Standard PID Standard PID Standard PID
Lambda Factor = 2.0 Lambda Factor = 2.5
Lambda Factor = 1.5 Lambda Factor = 2.0 Lambda Factor = 2.5
Improvement in stability is significant for any integrating process with analyzer delay
[File Name or Event]Emerson Confidential27-Jun-01, Slide 32 Slide 32
Control Studies of Reset Factor & Wireless PID for Real Time Self-Regulating Process (40 sec analyzer sample time)Control Studies of Reset Factor & Wireless PID for Control Studies of Reset Factor & Wireless PID for Real Time Real Time SelfSelf--RegulatingRegulating Process Process (40 sec analyzer sample time)(40 sec analyzer sample time)
Wireless PID Wireless PID Wireless PID
Standard PID Standard PID Standard PID
Reset Factor = 0.5 Reset Factor = 1.0 Reset Factor = 2.0
Reset Factor = 0.5 Reset Factor = 1.0 Reset Factor = 2.0
Improvement in stability and control is dramatic for any self-regulating process with analyzer delay
[File Name or Event]Emerson Confidential27-Jun-01, Slide 33 Slide 33
Control Studies of Lambda Factor & Wireless PID for Real Time Self-Regulating Process (40 sec analyzer sample time)Control Studies of Lambda Factor & Wireless PID for Control Studies of Lambda Factor & Wireless PID for Real Time Real Time SelfSelf--Regulating Regulating Process Process (40 sec analyzer sample time)(40 sec analyzer sample time)
Wireless PID Wireless PID Wireless PID
Standard PID Standard PID Standard PID
Lambda Factor = 1.5 Lambda Factor = 2.0 Lambda Factor = 2.5
Lambda Factor = 1.5 Lambda Factor = 2.0 Lambda Factor = 2.5
Improvement in stability and control is dramatic for any self-regulating process with analyzer delay
[File Name or Event]Emerson Confidential27-Jun-01, Slide 34 Slide 34
ConclusionsConclusionsConclusionsWireless PID and new communication rules can increase battery lifeWireless pH eliminates spikes form ground noiseWireless PID provides tight control for set point changesFeedforward of ammonia formation rate and oxygen uptake rate (OUR) offers significant improvement. OUR decouples interaction between pH and DO loopsWireless PIDPLUS dramatically improves the control and stability of any self-regulating process with large measurement delay (sample delay). The wireless PID is a technological breakthrough for the use at-line analyzers for control
– The Wireless PIDPLUS set point overshoot is negligible for self-regulating processes with large sample delays if controller gain is less than the inverse of process gain
Wireless PIDPLUS is stable for self-regulating process with large sample delay if controller gain is less than twice the inverse of the process gain
– As the analyzer sample time decreases and approaches the module execution time, it is expected that the wireless PID behaves more like a standard PID
Wireless PIDPLUS significantly reduces the oscillations of integrating processes but the improvement is not as dramatic as for self-regulating processesIntegrating processes are much more sensitive than self-regulating processes to increases in sample time, decreases in reset time, and increases in gainDetuned controllers (large Lambda Factors), makes loops less sensitive to sample time (see Advanced Application Note 005 “Effect of Sample Time ….”)If the controller gain is increased or the wireless resolution setting is made finer, the PIDPLUS can provide tighter control. For a loss of communication, the PIDPLUS offers significantly better performance than a wired traditional PID particularly when rate action and actuator feedback (readback) is used
[File Name or Event]Emerson Confidential27-Jun-01, Slide 35 Slide 35
Top Ten Signs of a WirelessHART AddictionTop Ten Signs of a WirelessHART AddictionTop Ten Signs of a WirelessHART Addiction
(10) You try to use the network manager to schedule the activities of your children(9) You attempt to use RF patterns to explain your last performance review(8) You use so much resource allocation in your network manager, you eat before
you are hungry(7) You propose your wireless device for the “Miss USA” contest(6) You develop performance monitoring indices for your spouse(5) You implement network management on your stock portfolio(4) You carry pictures of your wireless device in your wallet(3) You apply mesh redundancy and call three taxis to make sure you get home
from your party(2) You recommend a survivor show where consultants are placed in a plant with
no staff or budget and are asked to add wireless to increase plant efficiency (1) Your spouse has to lure you to bed by offering “expert options” for scheduling
[File Name or Event]Emerson Confidential27-Jun-01, Slide 36 Slide 36
For More on the Effect of Sample Time on PIDFor More on the Effect of Sample Time on PIDFor More on the Effect of Sample Time on PID
http://www.easydeltav.com/controlinsights/gm/AdvancedApplicationNote005.pdf
[File Name or Event]Emerson Confidential27-Jun-01, Slide 37 Slide 37
For More on Bioprocess Modeling and ControlFor More on Bioprocess Modeling and ControlFor More on Bioprocess Modeling and Control
[File Name or Event]Emerson Confidential27-Jun-01, Slide 38 Slide 38
ReferencesReferencesReferences
1. McMillan, Gregory, et. al., “PAT Tools for Accelerated Process Development and Improvement”, BioProcess International, Process Design Supplement, March, 2008
2. Blevins, Terry, and Beall, James, “Monitoring and Control Tools for Implementing PAT”, Pharmaceutical Technology, Monitoring, Automation , & Control, 2007
3. Boudreau, Michael and McMillan, Gregory, New Directions in Bioprocess Modeling and Control: Maximizing Process Analytical Technology Benefits, Instrumentation, Automations, and Systems (ISA), 2006
4. Boudreau, Michael, McMillan, Gregory, and Wilson, Grant, “Maximizing PAT Benefits from Bioprocess Modeling and Control”, Pharmaceutical Technology Supplement: Information Technology Innovations in the Pharmaceutical Industry, November 2006
5. McMillan, Gregory and Cameron, Robert, Advanced pH Measurement and Control, 3rd edition, ISA, 2005
6. Nixon, Chen, Blevins, and Mok, “Meeting Control Performance over a Wireless Mesh Network”, The 4th Annual IEEE Conference on Automation Science and Engineering (CASE 2008), August 23-26, 2008,, Washington DC, USA.
7. Chen, Nixon, Blevins, Wojsznis, Song, and Mok “Improving PID Control under Wireless Environments”, ISA EXPO2006, Houston, TX
8. Chen, Nixon, Aneweer, Mok, Shepard, Blevins, McMillan “Similarity-based Traffic Reduction to Increase Battery Life in a Wireless Process Control Network”, ISA EXPO2005, Houston, TX