Perfusion System Controller Strategies during an ECMO Support

International Journal on Soft Computing (IJSC) Vol. 5, No. 3, November 2014
DOI: 10.5121/ijsc.2014.5302 13
PERFUSION SYSTEM CONTROLLER STRATEGIES
DURING AN ECMO SUPPORT
M.Dhinakaran 1
and S.Abraham Lincon2
1, 2
Department of Instrumentation Engineering, Annamalai University,
Annamalai Nagar, Tamil Nadu, India
ABSTRACT
In this work modelling and control of Perfusion system is presented. The Perfusion system simultaneously
controls the partial pressures during Extra Corporeal Membrane Oxygenation (ECMO) support. The
main Problem in ECMO system is exchange of Blood Gases in the Artificial Lung (Oxygenator).It is a
highly Nonlinear Process comprising time-varying parameters, and varying time delays, it is currently
being controlled manually by trained Perfusionist. The new control strategy implemented here has a
feedback linearization routine with time-delay compensation for the Partial pressuresof Oxygen and
Carbon dioxide. The controllers were tuned robustly and tested in simulations with a detailed artificial
Lung (Oxygenator) model in Cardiopulmonary bypass conditions. This Automatic control strategy is
proposed to improve the patient’s safety by fast control reference tracking and good disturbance rejection
under varying conditions.
KEYWORDS
Cardio Pulmonary Bypass (CPB), Blood Gas Analyser (BGA), Extra Corporeal Membrane Oxygenator
(ECMO)
1. INTRODUCTION
ECMO support has been established as a routine treatment used for the patients whose Heart or
Lungs is not working properly and is most often used for Bypass procedures, and Heart
transplantations. The ECMO procedure is similar to a Heart-Lung bypass used during open-Heart
surgery. Extracorporeal Membrane Oxygenation can take over the function of the child's Heart
and / or Lungs for a limited time until the child recovers from the initial cause of the failure [4].
Figure 1 shows a new born having respiratory failure is placed in ECMO setup in ICU. The most
frequent use for ECMO has been with new born infants with lack of a fully functioning
respiratory system or other birth defect,although it may also be helpful in selected cases of severe
heart failure in adults and children. During CPB the functions of the Heart and Lungs are taken
over by the ECMO setup there by guaranteeing a bloodless, motionless operating field. In Extra
Corporeal Membrane Oxygenator the blood is circulated from the venous to the arterial side of
the patient’s vascular system by a pump in the machine. Additionally an Oxygenator takes over
the function of the Lungs by the addition of fresh Oxygen (O2) to and the removal of Carbon
dioxide (CO2) from blood. This Gas exchange is achieved by a gas flow that is in close contact to
the Blood pumped through the Oxygenator. This Gas mixture and its partial gas pressures initiate
a diffusion process in the Oxygenator, which in turn determines the amount of Oxygen and
Carbon dioxide on the arterial side of the patient. The Gas flow and the mixture of Gas that flows

14
through the Oxygenator are currently still being manually adjusted to have correct partial gas
pressures in the arterial blood [5].
Figure 1. ECMO Setup
This manual control Process can lead to errors, which may cause tissue or nervous cell damage.
An automatic blood gas control strategy is suggested to improve the patient’s safety by fast
control reference tracking of reasonable set points and a good disturbance rejection, which means
keeping the blood gas values in physiological ranges during disturbances. In addition to that, the
automatic control is suggested to remove workload from the perfusionist staff, and may thereby
help to increase the amount of correct decisions in situations of extreme workload.Figure 2 shows
the circuit diagram of an ECMO setup.
Figure 2.Complete Circuit Diagram of ECMO

15
2. ECMO MODELLING
As the exchange and transport of oxygen and carbon dioxide are complex biological processes
Comprising nonlinearities, time delays, uncertainties, and time-varying parameters, detailed
system knowledge is needed for the development of a suitable feedback control [1]. For that
reason a model of the blood gas transfer process was developed as follows. The system subject to
control was divided into three subsystems: the gas mixer; the Oxygenator, and the Blood Gas
analyzer (BGA) as shown in thefigure 3.
Figure 3. Gas Blender, Oxygenator, and BGA diagram
2.1. Gas Blender Model Design
As the gas concentrations are normally set mechanically with flow tubes, an electronic gas mixer
was designed based on electronic flow dosage modules.Figure 4 shows the Gas blender,
Oxygenator, and Blood gas analyser model diagram. The static output of the gas mixer is given
by
QG = Qo2 + QN2 + Qco2 (1)
QO2 = QG * FiO2 (2)
QCO2 = QG * FiCO2 (3)
QN2 = QG (1-FiO2-FiCO2) (4)
The combination of the two gases is known as inspired Input Oxygen (FiO2).QG is nothing but
total gas flow of the cardiopulmonary Blood Gases. QG is measured by blood Gassensor.QCO2is
nothing but total flow of carbon dioxide.QN2 is nothing but total flow of atmospheric air.T(T)is
a transport delay due to the distance from gas blender to oxygenator and applies to the Oxygen
fraction in the gas.The time-delay depends on gas flow and the tubing system. Total time delay
for the Gas mixer process based on the equation [1] - [2].

16

() =

.

,

, ()
Where ,

is the square diameter of the Gas Blender,and ,is the length of the Gas Blender
to Membrane Oxygenator,is the Total blood flow in the Membrane Oxygenator.
2.2. Membrane Oxygenator Model
The Membrane Oxygenator replicates the functionality of the Human Lungs in that it exposes the
blood to regulate amount of O2 and CO2. Gas entering on top inside the fibres while the blood is
flowing first down through a Heat Exchanger and then up through the Gas Exchanger. Gas
Exchange between Gas and blood phase is a very complicated process involving position
dependent quantities and partially nonlinear mechanisms which include Gas transport and
diffusion in the Blood or the diffusion across the Membrane. [3]Mass Balance Equation for Gas
Stored in the Blood Compartment in membrane Oxygenator.
V
d[c]
dt
= T![C],$ − [C],'() + Dp,-.( − p) + Ri (6)
Where V is the Volume of the compartment mixing chamber in an Oxygenator, [2]3is the
Concentration of the Components,[4]3 is the Diffusion Capacity over the Membrane Oxygenator,
[53]is the internal and external Partial Pressure of the Membrane Oxygenator, 6i is the external
disturbance in membrane Oxygenator cross bonding to chemical reaction .7is the Total Blood
flow through in the Membrane Oxygenator.
2.3.Blood Gas Analysis Model Design
Blood gases are only sampled a few times a day in order to check blood gas status. This clinical
blood gas analyser’s use up a few millilitre bloods for each sample, so these devices cannot be
used for continuous blood monitoring on a frequent basis (e.g. every second). For our setup we
chose the CDI 500 blood gas analyser from Terumo which offers online measurement of pH,
pCO2, pO2, potassium, temperature and etc. The BGA was modelled according to measurements
with first order differential equations with its dominant. The modelled BGA time-delay is
T8
T) =
0.7853
b((0.01 T! + b(@)
d(,ABC
D
l(,ABC (8)
Where d(,ABC is the diameter of the Blood Gas analyzer tube , l(,ABC is the length of the of the
Blood Gas analyzer tube,7
is the time delay for the Blood Gas Analyzer, 7
is the Conversion
factor for Blood Gas Analyser, 7 is the total blood flow through the Blood Gas Analyser .

17
Figure 4. Gas blender, Oxygenator, and Blood Gas Analyser modelDiagram
2.4. Time Delay
Total time delay of the blood flow can be calculated by time delay between the Gas blender to
Membrane oxygenator Td1(Tg) and time delay between membrane oxygenator to Blood Gas
analysis Td2(Tg).
TT) = TT) + TDq) (9)
2.5. Input/OutputLinearization of O2
The process is linearized with an analytical input/output linearization for the oxygenation process
with input FiO2 and output pO2. Direct disturbance to the states can be measured with high
dynamics in case of the blood-flow which are measured via the blood-gas analyser the
disturbance dynamics are expected to be much slower than the dynamics of the blood gas
analyser. So the input oxygen from the Gas blender to membrane Oxygenator as follows.
u.I = JiOD ∗ p!MN(10)
The input and output of the oxygen can be linearized with the help of barometric pressure and
percentage of inspired input Oxygen depends on the patient respiration.
2.6. Nonlinear State Space Process Model
Process model depends on the Gas blender, membrane oxygenator, blood gas analysis and so on.
Finally we have to derive the mathematical model for Gas transfer in an oxygenator. So the strong
mathematical knowledge should be needed to design the membrane oxygenator model.[3]
Initially the model design and control by single input single output (SISO) Processes the may be
defined
XP(t) = A x(t) + Bu(t)(11)
Y(t) = C x(t) + D u(t) (12)

18
Where x(t) is the state vector; A is the system matrix;B is the input matrix; C is the output matrix
is zero matrix; y(t) is the output vector ; u(t) is the input control matrix ;[1]
The continuous time invariant state space model for an membrane oxygenator to control the blood
gases with the help of controller .The controller approach is based on the mathematical nonlinear
model of the gas transfer process in membrane oxygenator. Input control matrix of the single
input as followsas
BV
u = W0 0 0 0 0 0 0 0 0 0 0 0
q
v
Y (u)(13)
Input vector u = pO2 = FiO2 * 760 mm Hg for barometric pressures[3]. The state vector may be
defined as
X = [xxDxZx[xx]x^x_x`x@xxDxZ]V
u (14)
The full process model of the three inputs, two output of the system with input composed of FiO2,
Gas flow and Blood flow the output vector is based on pressure of O2and CO2
YV
= [pODpCOD]V
(15)
3. CONTROLLERDESIGN
The pCO2 process is much less critical in control terms; compared with the pCO2 process need
not Linearization of the carbon dioxide for pCO2 controller design.pO2 process is more
complicated in control terms. An input/output state linearization routine with delay-time
compensation and an external linear gain scheduled controller had to be developed to handle the
strong static and dynamic nonlinearity and the time delay at input and output. Like the pO2
controller was gain-scheduled in dependence of blood flow through the oxygenator.[1] The PID-
pO2 and the PID-pCO2 controllers, were discretized at the sampling time of the blood gas
analysis, Ts, BGA = 6 s.
3.1. PID Controller
The Proportional-Integral-Derivative (PID) method is a kind of feedback controller which is
generally based on the error (e) between desired set point and measured process value. Three
parameters must be designed in the PID Controller and each parameter has an effect on the error.
The transfer function of the PID controller is writtenas
b(S) =
U(S)
E(s)
= kh +
K
S
+ K (16)
4. DISCUSSI ON
By using a detailed model of the gas blender, the Oxygenator, and the Blood gas analysis, a
feedback control could be developed for the simultaneous control of Oxygen and Carbon dioxide
partial pressures. As the control input for pO2, the oxygen fraction in the gas flow FiO2 to the
oxygenator was used. As the control input for pCO2, the total gas flow qg to the Oxygenator was

19
used. The direct coupling of qg to the FiO2 control input is similar to a disturbance at pCO2
controller gas flow input changes and can be seen as a disadvantage. The advantages of this
strategy are that an additional CO2 gas supply is saved and that this technique is commonly used
in most cardiovascular Heart surgery centres. The pO2 process is more complicated in control
terms. An input/output state linearization routine with delay-time compensation and an external
linear gain scheduled controller had to be developed to handle the strong static and dynamic
nonlinearity and the time delay at input and output. In contrast to that, a PI controller was
developed for the O2 process, as it shows an output delay time only and much less static and
dynamic nonlinearities. The controllers were tested in simulations over the whole operating range
with the model and incorporated uncertainty and showed a good performance and robust stability
5. SIMULATION RESULTS
Figure 5. A. Oxygen linearisation I/O Response
Figure 6. Fraction of Oxygen (FiO2) Control Response
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
0
0.005
0.01
0.015
0.02
0.025
0.03
Time (sec)
P
r
e
s
s
u
r
e
o
f
O
x
y
g
e
n
(
m
m
H
g
)
OxygenInput / Output LinearizationModel Output
5 10 15 20 25
0.024
0.026
0.028
0.03
0.032
0.034
Blood Flow (littre/sec)
F
ra
c
tio
n
o
f
O
x
y
g
e
n
(F
iO
2
)
PI Controller Responses For Inspired Oxygen(FiO2)

20
Figure 7. State Space output for pressure of oxygen
Figure 8. State Space output for pressure of carbon dioxide
Figure 9. PID Controller Responses
6. CONCLUSION
The proposed model of the blood-gas controller for Oxygen and Carbon dioxide exchange during
ECMO support shows fast reference tracking and good disturbance rejection results, while
process coupling is kept low. This is possible because of the Oxygen process I/O linearisation and
time-delay compensation. Linear process models for the two-input- two output plant were
0 20 40 60 80 100 120 140 160 180
0
5
10
15
20
25
30
Time (sec)
P
re
s
s
ur
e
of
O
xyg
en
(
m
m
H
g
)
SS Model Output
0 20 40 60 80 100 120 140 160 180
0
5
10
15
20
25
30
35
Time (sec)
P
re
s
s
u
r
e
o
f
C
a
r
b
o
n
-
d
i
o
x
i
d
e
(
m
m
H
g
)
SS Model Output
0 1 2 3 4 5 6 7 8 9 10
0
10
20
30
40
50
60
Time (Sec)
B
loo
d
G
a
se
s
(m
m
H
g
)
Closed loop PID Controller Responses
pO2
pCO2

21
obtained for the linearised system without time-delay over the complete operating range during
surgeries. Uncertainties in the resulting transfer functions, as well as time delay uncertainties
were modeled in a multiplicative output uncertainty bound. Only one linear controller was applied
in nonlinear simulation with a validated Oxygenator model and showed very good results
However, lumping uncertainties of various sources at the system output in a multiplicative
uncertainty bound is suggested to yield conservative controller result.
REFERENCES
[1] Misgeld BJE, Werner J, and Hexamer M. Nonlinear robust blood gas control by state linearisation for
the cardiopulmonary bypass. Control Eng Pract 2008; 16:884–95.
[2] Hexamer M., Werner (2003). A Mathematical model for the gas transfer in an oxygenator IAFC
conference on modelling and control in biomedical systems (pp.409-414).Australia: Melbourne.
[3] Hexamer M, Misgeld B, Prenger-Berninghoff A, et al. Automatic control of the extra- corporal
bypass: system analysis, modelling and evaluation of different control modes. BiomedTech 2004;
49:316–21.Kay PH, Munsch CM. Techniques in Extracorporeal Circulation. London: Arnold, 2004.
[4] Robust and self-tuning blood flowcontrol during extracorporealcirculation in the presence of
systemparameter uncertaintiesB. J. E. Misgeld J. Werner M. HexamerMedical Biological
Engineering Computing 2005, Vol. 43, PP 589-599
[5] Hammon J Wi . Extracorporeal Circulation: Perfusion SystemCohn Lh, ed. Cardiac Surgery in the
Adult. New York: McGraw-Hill, 2008:350-370..
[6] Cardiopulmonary Anatomy PhysiologyEssentials for Respiratory Care, 4th EditionTerry Des
JardinsCOPYRIGHT © 2002 by Delmar, a division of ThomsonLearning
[7] Introduction to Extracorporeal Circulation Maria Helena L. Souza Decio O. Elias
[8] A.N.K.Nasir, M.A. Ahmad and M.F.Rahmat, Performance comparison between LQR and PID
Controller for an inverted pendulum system International Conference on Power Control and
Optimization, Chiang May, Thailand, 18-20, July 2008.
[9] Mahmoud, M.S.: ‘Switched delay-dependent control policy for water quality systems’, IET Control
Theory Appl., 2009, 3, (12), pp. 1599–1610.
[10] Zhang, H., Feng, G.: ‘Stability analysis and H1 controller design of discrete-time fuzzy large-scale
systems based on piecewise Lyapunov functions’, IEEE Trans. Syst. Man Cybern., B, 2008, 38,
(5),pp. 1390–1401. M. Drummond, “Technological Evolution of Membrane Oxygenators,” April,
2005, RevistaBrasileira de Cirurgia Cardiovascular, Vol. 20, pp. 432-467.
[11] Allen J, Fisher AC, Gaylor JD, Razieh AR. Development of a digital adaptive control system for PO2
regulation in a membrane oxygenator. J Biomed Eng 1992; 14:404–11.
[12] Sastry S. Nonlinear Systems. New York: Springer-Verlag, 1999.
[13] Scott I. Mars’, Robert H. Bartlett’, Janice M. Jenkins3, Pierre Ababa. Controller design for
extracorporeal life support (1996) 18th Annual International Conference of the IEEE Engineering in
Medicine and Biology Society, Amsterdam 1996 .1733-1735.
[14] N. Chauveau, W. Vanneurs, Armurier, R. Barthelemy (1992) Closed loop control of blood gases
during Surgical extracorporeal circulation IEEE 377-399.
[15] M. Korari and E. Zafriou. Robust Process Control. Prentice-Hall International, Englewood Cliffs, NJ,
1989.
[16] Cooper JB, Newbower RS, Kitz RJ. An analysis of major errors and equipment failures in anesthesia
management: consideration for prevention and detection. Anaesthesiology 1984; 60:34–42.
[17] S. Marlow, J. D. S. Gaylor, P. H. Mook, R H. Wildevuur, and IC. M. Taylor, “A pO2Regulation
system for membrane oxygenators,” ASAIO Truns., vol. 27, pp.299-303, 1981.
[18] De Jong DS, Van Dalen A, Nauta J. Introduction of an automatic oxygen regulation device to be used
in combination with bubble oxygenators during cardiopulmonary bypass. Extract from the 19th
International Conference of the American Society of Extra Corporeal. Technology 1981.

22
AUTHORS
M. Dhinakaran obtained his B.E degree in Electrical and Electronics Engineering in the year 2000 from
the Madras University and further received M.E. degree in Instrumentation Control Engineering in 2003
from the Thapar University, Punjab. Presently he is working as Assistant professor in the Department of
Instrumentation Engineering Annamalai University. His areas of research are Bio Medical Instrumentation,
Process Control.
S. Abraham Lincon obtained his B.E degree in Electronics Instrumentation Engineering in the year 1984
and received M.E. degree in Power System Engineering in 1987 and another M.E degree in Process control
and Instrumentation Engineering in 2000. He Completed his Ph.D.in the Field of Process Control in
2008from the Annamalai University, Chidambaram. Presently he is working as a Professor of
Instrumentation Engineering in Annamalai University. His areas of research are Process Control,
FaultDetection and Diagnosis and Multivariable Control.

Perfusion System Controller Strategies during an ECMO Support

More Related Content

What's hot (20)

Similar to Perfusion System Controller Strategies during an ECMO Support (20)

Recently uploaded (20)

Perfusion System Controller Strategies during an ECMO Support