Chemical Engineering Science, 1971, Vol. 26, pp. 963-967. Pergamon Press. Printed in Great Britain.

Oxygen transfer to blood flowing in round tubes including Bohr and

Haldane effects
JOHN CARL FAIR
and
MICHAEL H. WEISSMAN
Carnegie-Mellon University, Pittsburgh, Penna. 152 13, U.S.A.

(First received 8 October 1969; accepted 26 October 1970)

Abstract- An iterative numerical solution for the coupled oxygen and carbon dioxide convective
diffusion problems for blood flow in a large gas permeable tube is given. By comparison with the prior
work of Weissman and Mockros, it is shown that the Bohr and Haldane effects are of secondary
importance in these problems.

INTRODUCTION [2]. The interrelationship of 0, storage in the

THE NEED to carry on circulation and respira- blood and amount of CO2 in the blood (Bohr
tion artificially during certain surgical operations effect) and of CO, storage to the amount of 0,
has promoted the development of heart-lung in the blood (Haldane effect) which, where not
machines. Removal of metabolically produced included previously, are treated here.
carbon dioxide and replacement of consumed
oxygen is accomplished in the oxygenator MATHEMATICAL MODEL
section of such a machine. Transfer of sufficient The physical model of a single tube in which
amounts of gas has been accomplished in screen, oxygenation is occurring is shown in Fig. 1.
disc and bubble type oxygenators but with the Assumptions used in the mathematical descrip-
necessity of a direct blood to gas interface. Such tion of this situation include: a uniform circulator
a direct interface has been shown to cause cross-section tube; axial symmetry; a fully
damage to the blood which increases in extent developed velocity profile ,uniform and parabolic
as time of oxygenator usage increases [ 11. limits may be taken); red blood cells present
To reduce damage due to artificial oxygena- as storage sites for carbon dioxide and oxygen
tion, a thin, gas permeable membrane can be and uniformly distributed throughout the plasma;
placed between the gas and liquid blood. Thus, chemical kinetics are fast enough to make
an oxygenator could consist of alternate layers transport of gases diffusion limited in the blood
of blood and gas separated by thin membrane (rather than reaction limited); wall permeability
sheets or of blood flowing in capillaries formed so exceeds permeability in blood that wall
by the membrane and surrounded by gas. The thickness may be neglected, (in practice silicone
work reported in this paper represents an ex- rubber tubes, having an 0, permeability some
tension of the analysis of Weissman and Mockros, 20 times greater than blood, are used); in the
of the oxygenation of blood flowing in round, gas absence of chemical reaction some gross
permeable, membrane-tubes [2], whose diameters diffusivity exists for whole blood treated as a
varied from 635 to 6350 CL. homogeneous fluid: and the oxygen rich atmos-
The process is assumed to be limited by the phere surrounding the tube remains constant in
diffusion of the gases in plasma and resistance composition.
to diffusion of gases through the membrane Mass conservation equations may be written
wall is considered to be negligible. These for both oxygen and carbon dioxide. These
assumptions were found valid in previous work equations include diffusion, convection and
963
 J. C. FAIR and M. H. WEISSMAN

storage of gas in chemical form. The equations

have the form:

;[&(R$$] = (1-R’) (g+g). (3)

where For the uniform of flat velocity profile the non-

c(r, z) = concentration of dissolved gas in dimensional equation is:
plasma
D = gross diffusivity of gas in whole blood
~[-&(~)]=;(g+~). (4)
s = storage term
V = blood velocity in the axial direction.
The equations for 0, and CO, mass conserva-
Since convective transport is much larger than tion must be solved numerically due to the form
diffusion in the axial direction, the equation of the storage terms.
may be simplified:
TREATMENT OF GAS STORAGE TERM
The functional relationship between the
amount of 0, stored in the blood and the
amount dissolved in the plasma is expressed
The three necessary boundary conditions are: by the oxyhemoglobin dissociation curve. A
similar curve exists for carbon dioxide. Each of
c(r,O) =cg @ z=o these curves is parametrically influenced by the
amount of the other gas present in the plasma.
c(a,z) =c,@ r=u For example, the oxyhemoglobin dissociation
curve position is a function of the carbon
g (0, z) = 0 @ r = 0. dioxide content of the plasma.
The mathematical description of the functional
The presence of the storage term in this equation relationship between blood gas storage and
makes the equation non-linear since s is de- dissolved plasma gas content used in this paper
pendent on c in a complicated, non-linear was reported by Grodins[3]. It consists of a
manner. mixture of chemically derived equations and
Equation (2) may be put into non-dimensional curve fits to the data of Boothby[4]. The
form, for convenience of solution, through equations are:
introduction of the dimensionless variables:
s co2 = BHCOS +0*375 [O-2 - CHbo2]
R = r/a,

c = c/co,
co*- K~cozPcoz
0.01 PO, 1 >
_ o. 14

z = (7?D/2Q)z,
1. CO, STPD
+ fho8’coz. (5)
where Q = volume flow rate, = VV&, 1. blood
1. O2 STPD
s = s/co. 302 = KaoZ’o. + CHbOz , blood (6)

The resulting non-dimensional equation for a

C HbOs= (0.2) [ 1 - e+1po*]2 ” pas,‘,‘,” (7)
parabolic velocity profile
964
 Oxygen transfer to blood flowing in round tubes

Sl = 044921 pH -0*10098 (PH)~ sets of boundary and initial conditions:

=49mmHg
+ O-006815 (PH)~ -0.454, (mm Hg)-’ (8) Set 1: Initial conditions pPC00
02 =54mmHg

pH = 9-LogC,+ PC02=lmmHg
KcYCO*PC()*
(9) At tube wall
IP 02 = 715 mm Hg
CH+ = K’ (10) Set 2: Initial conditions P =49mmHg
[-sco2 - K%O.PC02
1 pco*
=32mmHg
02

where Eq. (5) is the CO, buffer equation includ- P co2 =lmmHg
At tube wall
ing the Haldane effect, Eq. (6) is the oxyhemo- P 02 = 715 mm Hg
globin dissociation curve including the Bohr P =49mmHg
effect, and Eqs. (7) and (8) are empirically fit. Set 3: Initial conditions pCQ
02 = 43.5 mm Hg
SC02 and so2 are the total O2 and CO, content of
the blood. The required constants are: P cot =lmmHg
At tube wall
(yco. = O-510 (1. STPD/atm. - 1. of blood), solu- P 02 =715mmHg.
bility coefficient of COz gas in blood
cu,, = O-024 (1. STPD/atm. - 1. of blood), solu- The conditions on Pco2 were chosen so that the
bility coefficient of 0, gas in blood high value (49 mm Hg) was representative of
K = O-00132 (atm/mmHg), conversion factor “average” venous blood and the low value
BHCO, = O-5470 (1. of CO, at STPP/l. of blood), (1 mm Hg) was close to zero. A value of zero
the standard bicarbonate content of blood could not be used directly for reasons of con-
K’ = 794 (nmoles/l.), the dissociation constant vergence of the numerical methods. The wall
for carbonic acid. PO2of 7 15 mm Hg was taken to coincide with the
Partial pressures of the two gases in the plasma value reported earlier. In set 1, the initial value
are related to molar concentrations by the of PO2 was chosen so that the same pressure
equations: gradient as previously reported existed. Since
the oxyhemoglobin dissociation curve for human
PO2= 6.9 x lo5 Coz (11) blood is different from that of cattle blood,
(which was used in the earlier work) the oxygen
P co2 = 3.4 x 104cco* saturation resulting was 84.6 per cent, compared
to 75 per cent for cattle blood at the same PO,.
where P is in mmHg and C is in moles/l. However, the changes in saturation as a function
Since Eq. (5) is transcendental in Sco2, an of Z for a given pressure gradient should not
iterative scheme was used (as suggested by depend on the type of blood[2]. The values of
Grodins) for the calculation of blood gas the initial PO, in sets 2 and 3 were chosen to
content[3]. This scheme utilized Eqs. (5-10) give initial saturation values of 50 and 75 per
except that Eqs. (5) and (6) were modified so cent respectively.
as to yield only chemically stored, rather than A parabolic velocity profile was used for all
total, gas content. three solutions. The parabolic profile is reason-
Eleven radial points were used in the numeri- able for capillary tubes which are large compared
cal model. This has been found in similar solu- with the size of a red cell and in which the shear
tions to give sufficient accuracy [2]. rates are greater than about 50 to 100 set-‘[5,6].
Figure 3 shows the mixing-cup PO* and tte
increase in mixing-cup oxygen saturation for
RESULTS AND DISCUSSION the three sets of data. The results are seen to be
Solutions were obtained for the following systematic in that the blood starting at lowest
965
 J. C. FAIR and M. H. WEISSMAN

the old. The similarity of the mixing cup satura-

tion increases has been predicted by Weissman
and Mockros using a qualitative argument to
show that saturation increases should not be
sensitive to shifts in the oxyhemoglobin dissocia-
tive curve, if the PO2gradient remains fixed. The
divergence of the PO, curves for old and new
solutions, however, is a result of the difference
in the oxyhemoglobin dissociation curves
used in obtaining these solutions. These results
indicate that the Bohr effect is not of primary
importance in blood oxygenation in a large tube.
The CO, results for the data of set 3, seen in
Fig. 3, showing mixing-cup pC02, CO, content
and pH, represent an improvement in accuracy
over the oversimplified Graetz solution used
earlier (also shown in Fig. 3). The new solution
predicts a greater rate of CO, loss than did the
old, under the same pressure gradient. Results
for sets 1 and 2 are not shown since they are
nearly identical to those of set 3. The fact that
the results are similar for all three cases shows
that the Haldane effect is only of secondary
importance, in CO, removal in a large tube. The
Z data show that when mixing-cup pCOZ is reduced
from 49 to 35 mm Hg, the more oxygenated
Fig. 2.
blood (data set 1) contains only 0.8 vol. % CO,
less than the less oxygenated (data set 2) blood,
out of a total CO, content of about 50 vol. %.
The pH is seen to rise as CO, is eliminated
(from 7.36 to 7+49), in contrast with the constant
I, pH used previously. The absolute level of the
pH is very slightly higher than the “normal”
human range for arterial blood (7.37-7.43) and
apparently the equations given by Grodins[3]
require an initial pC0, slightly higher than
49 mm Hg to bring the pH into this range.
Fig. 3.
Acknowledgment-This work was supported in part by
Grant No. HE 12714, from the National Heart Institute, at
PO, takes on oxygen most rapidly. Also appearing Carnegie-Mellon University.
in Fig. 2 is the Weissman and Mockros solution
for cattle blood utilizing the same initial PO2
NOTATION
gradient as was taken in data set 1. It can be
seen that there is close coincidence between tube radius
these two solutions in so far as the saturation BHCO: standard bicarbonate content of blood
increase is concerned, but the mixing cup PO2 dimensional concentration
for the new solution rises more rapidly than for : dimensionless concentration
966
 Oxygen transfer to blood flowing in round tubes

Cl0 concentration at r = a stored gas

co initial concentration dimensionless stored gas
D diffusivity in blood total gas content
K conversion factor empirical function
K’ dissociation constant for H&O, blood velocity in z direction
Q flow rate axial coordinate
radial coordinate dimensionless axial coordinate
R’ dimensionless radial coordinate solubility coefficient

REFERENCES
[l] GALETTI P. M. and BRECKER G. A., Heart-Lung Bypass: Principles and Techniques of Extracorporeal Circulation.
Grune and Stratten, New York 1962.
[2] WEISSMAN M. H. and MOCKROS L. F., J. Engng Mech. Div., Proc. Am. Sot. Civil Engrs 1967 93EM6 225.
131 GRODINS F. S. BUELL J. and BART A. J.. J. at&. Phvsiol. 1967 22 260.
[4] BOOTHBY W. M. et al., Handbook ofRespira&y D&a in Aviation, Washington, D.C., Office of Scientific Research
and Development and the National Research Council 1944.
[5] BUGLIARELLO G. and HAYDEN J. W., Trans. Sot. Rheol. 1963 7 209.
[6] MERRILL E. W. and PELLETIER G. H., J. appl. Physiol. 196723 178.

RCsume-On donne une solution numerique iterative des probltmes de diffusion par convection de
I’oxygene et de l’acide carbonique, pour un ecoulement sanguin dans un grand tube permeable aux gaz.
La comparaison avec les travaux anterieurs de Weissman et Mockros, pennet de montrer que les
effets de Bohr et Haldane sont d’importance secondaire dans ces problemes.

Zusammenfassung-Es wird eine iterative numerische Losung fur die gekoppelten Sauerstoff-und
Kohlendioxyd-Konvektivdiffusionsprobleme fur die Blutstriimung in einem grossen, gasdurchllssigen
Schlauch dargelegt. Gegeniiber den friiheren Arbeiten von Weissman und Mockros wird gezeigt,
dass die Bohr und Haldane Effekte in diesen Problemen von sekundarer Bedeutung sind.

967