Capillaries

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Physiological Fluid Mechanics Blood Flow through Capillaries

Medical handbooks often mention that the relationship between the blood ow rate through a capillary and the pressure drop along the capillary is given by Poiseuilles law. This may be correct if the uid viscosity that appears in this law is not the true viscosity of blood, but the apparent viscosity. This apparent viscosity is not only a function of the hematocrit, where one should distinguish between the tube hematocrit and the discharge hematocrit, but strangely enough, also of the diameter of the blood vessel. These puzzling results, discovered rst by Robin F hrus a and Torsten Lindqvist, are discussed in Section 1. This is followed in Section 2 by a description of a uid mechanical model proposed by James Lighthill, Timothy Secomb, Richard Skalak and others to explain the observations of F hrus and Lindqvist. a The discovery that the model of Section 2 gives an accurate description of blood ow in glass capillaries but not of blood ow in capillaries in the human body, was one of the reasons for intensive research in the past 15 years on the structure and function of the glycocalyx, a thin layer covering the endothelial surface of blood vessels. Section 3 describes modications of the model of Section 2, proposed to take into account the presence of this glycocalyx.

From Poiseuille to F hrus a

Blood pressures are reported in mmHg, a custom that goes back to 1828 when the physiologist Jean Poiseuille invented the U-tube mercury manometer to measure pressure in arteries. During the 1830s Poiseuille began his studies of the blood ow in the microcirculation, during which he discovered (in 1835) the plasma-skimming effect: blood ow in the arterioles and venules features a plasma layer at the vessel wall, which is depleted from red blood cells. Poiseuille became best known to uid dynamicists for his exceptionally precise measurements of the ow of water through glass capillaries with diameters in the range from 15 m to 0.6 mm, with the aim of clarifying the laws that govern the ow of blood in the smallest blood vessels. Poiseuille published his results between 1840 and 1843, which included that the volume ow rate Q through a capillary of length and diameter d due to a pressure difference p is given by the relation Q = K(T ) where K(T ) is a function of the temperature T : K(T ) = 1836.7 (1 + 0.033679T + 0.00022099T 2 ) with T measured in degrees Celcius. This relation is now known as Poiseuilles law and is written as 128 p = Q, (1) d4 1 d4 p,

Figure 1: Resistance R, dened as the ratio p/Q, per unit length of vessel as a function of the vessel
diameter d, measured in unbranched vessels of the mesentery. The solid curves are of the form R/ = a dm ; Poiseuilles law states that m = 4. .

the temperature dependence discovered by Poiseuille being that of the uid viscosity . The rst derivations of Poiseuilles law were published in 1860 by Hagenbach and by Helmholtz. Nowadays, it is recognized that Poiseuilles law does not apply to blood ow in arteries and veins (vessels with diameters in the range 100 m to 15 mm), because that ow has a signicant time-varying component and/or because most of these vessels are too short for the ow to become fully-developed (a ow that it is now called Poiseuille ow). But blood ow in arterioles and venules (vessels with diameters in the range 10 m to 100 m) is virtually steady and fully-developed over most of the vessel length; for these vessels, despite the presence of the celldepleted layer at the vessel wall, Poiseuilles law appears to provide an accurate relation between the vessel diameter and the ratio of pressure drop per unit length and volume ow rate, as gure 1 shows. Here a relation of the form (p/Q)/ = a dm is correlated with experimental data; the values for m agree very closely with the m = 4 of Poiseuilles law. However, the real issue here is the value of the coefcient a. In 1928 and 1931 the physiologist Robin F hrus published two articles (the second with his a colleague Torsten Lindqvist) in which he made an ingenious analysis of the data of experiments which were essentially similar to those of Poiseuille. But the uid was blood, not water, and glass capillaries with a smaller diameter could be used: in the range between 3 m and 300 m. This range not only includes the diameters of arterioles and venules but also those of the smallest blood vessels: the capillaries. In the 1928 article F hrus compared the volume concentration HD of red blood cells of the a blood entering and leaving a capillary, the discharge haematocrit, with the instantaneous volume 2

Figure 2: The F hrus effect: the haematocrit of blood owing through narrow capillaries, the tube haemaa
tocrit, is less than that of the blood feeding into and discharging from those capillaries, the discharge haematocrit. The reduction of the tube haematocrit depends on the diameter of the capillaries.

concentration HT of red blood cells of the blood owing inside the capillary, the tube haematocrit. Measurements of the ratio HT /HD as a function of the diameter of the capillary d are shown in gure 2. They illustrate what has become known as the F hrus effect: the tube haematocrit is a less than the discharge haematocrit, a reduction which depends on the diameter of the tube. This remarkable result is explained easily if one recalls the observations of Poiseuille published in 1825, namely that blood ow in the arterioles and venules features a cell-depleted layer at the vessel wall. If the red blood cells have a tendency to gather in the center of a capillary, the uid at the vessel wall is the blood plasma with a relatively low viscosity. The red blood cells in the core region then move relatively fast, so that the mean red cell velocity U0 is higher than the mean velocity of the blood as a whole U , the so-called bulk velocity. The volume ow rate of red blood cell material in the capillary, U0 HT per unit cross-sectional area, must be equal the volume ow rate of red blood cell material that is discharged, U HD per unit cross-sectional area, whence HT U = . HD U0 (2)

a Since U0 > U it is found that HT < HD , as was observed by F hrus. In the 1932 article F hrus and Lindqvist proposed that the relation between pressure drop p a and volume ow rate Q in narrow capillaries is that given by Poiseuiles law, but with a different value of the viscosity: 128 app p = Q, (3) d4 where app is the apparent viscosity. If one measures the pressure drop p and the volume ow 3

Figure 3: The F hrus-Lindqvist effect: the (relative) apparent viscosity of blood owing through narrow a
capillaries is a function of the diameter of the capillaries.

rate Q for given values of the diameter d and length , then the above relation denes the apparent viscosity as p d4 app = . (4) 128 Q A comparison with the viscosity p of pure blood plasma (close to that of water) then denes the relative apparent viscosity as app . (5) rel = p For blood ow in arteries and veins the relative apparent viscosity is an increasing function of the haematocrit, a function that takes a value close to 3 for normal haematocrit values around 0.42. What F hrus and Lindqvist found, see gure 3, is that the (relative) apparent viscosity of blood a owing in narrow glass capillaries depends on the tube diameter: If the tube diameter decreases from values typical of the arterioles and venules (from 100 m to 10 m) down to a value of about 8 m, a typical size of a capillary, the apparent viscosity of blood also decreases. This has become known as the F hrus-Lindqvist effect. Capillaries have diameters in the range 4 m to 10 m, a and F hrus and Lindqvist observed that for the smallest capillaries there is a sharp increase in the a apparent blood viscosity as the tube diameter decreases further below 8 m. For sure, these observations have nothing to do with possible non-Newtonian properties of blood. The reason for the observed behavior at diameters in the range of those of the arterioles and venules is twofold: (i) a reduction of the haematocrit in narrow blood vessels (the F hrus effect), a 4

Figure 4: Comparison between the axial-train model for blood ow in arterioles and venules and measurements in glass capillaries; (a) the F hrus effect; (b) the F hrus-Lindqvist effect. The single-le ow a a model is discussed in Section 6.2.

and more importantly (ii) the presence of a pure blood plasma layer with a reduced viscosity at the vessel wall. This may be illustrated by assuming that the red blood cells are conned to a cylindrical core of diameter 2a, with < 1, surrounded by an annulus of pure blood plasma. If the ow is driven by a pressure gradient dp/dz, the axial velocity distribution in the uid is given by a2 dp a2 dp u(r) = 2 (r/a)2 1 2 , r a, 4c dz 4p dz u(r) = a2 dp 1 (r/a)2 , 4p dz a r a,

where c is the viscosity of the suspension of red blood cells in blood plasma in the core region. The volume ow rate is a4 dp Q= 1 4 (1 p /c ) . 8p dz The relative apparent viscosity is in this case app 1 = 4 (1 / ) p 1 p c and the ratio of tube haematocrit to the discharge haematocrit is HT 1 4 (1 p /c ) = . HD 2 2 (2 p /c ) 5 (7) (6)

Figure 5: Human red blood cells owing in glass tubes of approximate diameters 4.5 m (top), 7.0 m
(middle), and 15 m (bottom). The ow direction is from left to right.

Results of this model, called the axial-train model, are compared with experimental data in gure 4. Here the viscosity ratio was taken as c /p = 3.3 and the width of the cell-depleted layer as = 1.8 m (note that = 1 /a). The reason for the behavior of the apparent viscosity for diameters typical of the capillaries is suggested by the photographs shown in gure 5. The tube diameter in the bottom picture is 15 m; it may be viewed as a visualization of the axial-train model. The pictures for tube diameters 4.5 m and 7 m show that as blood ows through such narrow capillaries it cannot be considered as a continuum. The red blood cells move through the capillary one after the other. It is commonly believed that the equilibrium shape of red blood cells is that of a biconcave disk, about 8 m in diameter and 2 m in thickness. Red blood cells are highly deformable, so that they can pass through capillaries with diameters less than 8 m. However, within the thin layer of blood plasma between the deformed cell and the vessel wall high rates of strain must occur, giving rise to a considerable drag on the cell. The smaller the vessel diameter, the higher this drag, and so the higher the apparent viscosity of blood. In the next section a model for the ow of individual red blood cells through narrow capillaries will be discussed; this is sometimes called the single-le ow model.

Figure 6: (a) Variables describing the geometry of an axisymmetric red blood cell, viewed as a shell moving
in the negative z-direction; (b) stress resultants in an element of an axisymmetric shell.

Squeezing of red blood cells through capillaries (in vitro)

The fundamental ideas for an analysis of the motion of a red blood cell through a narrow capillary were outlined by James Lighthill in 1968, but with an inaccurate model for the mechanics of the cell. His analysis was improved considerably in the 1970s and 1980s, by Richard Skalak (Columbia University) and Timothy Secomb (University of Arizona) in particular. Mechanics of the red blood cell To model its elastic response, the red blood cell is viewed as an axisymmetric membrane, as depicted in gure 6. Let arclength s, measured from the nose of the cell, describe position on the membrane and let be the angle between the normal to the membrane and the central axis. The local membrane curvatures are ks and k ; the membrane tensions are ts and t ; the bending moments are ms and m ; shear force per unit length is qs . Then, the equations for equilibrium of normal stress, tangential stress and bending moments in the membrane are 1 d(rqs ) = ks ts + k t + p, r ds 1 d(rts ) cos = t ks qs , r ds r 1 d(rms ) cos = m + qs , r ds r 7 (8) (9) (10)

respectively. In these expressions p is the pressure and is the viscous shear stress exerted by the uid on the membrane. Note that the radii of curvature are related to (s) by ks = d , ds k = sin . r (11)

The above set of equations needs to be completed by constitutive relations for the behavior of the membrane. The membrane strain is characterized by the principle extension ratios s = ds , ds0 = r , r0 (12)

where the subscript 0 refers to corresponding values of the unstressed shape of a material element. Red cell membranes have a strong resistance to area changes, so a good approximation is r ds = r0 ds0 , from which it follows that s = 1 .

(13)

The principle tensions may be resolved into an isotropic mean tm and a deviatoric component td : ts = tm + td , where td = 1 2 2 t = tm td , 1 2 (14)

(15)

with the shear modulus of the membrane. Finally, the bending moments are assumed to be isotropic and proportional to the change in the total curvature: ms = m = B [(ks + k ) (ks + k )0 ] , (16)

with B the bending modulus. By substituting relations (11)(15) into equations (8)(10), the response of the red cell membrane is nally found to be governed by dr = cos , ds d = ks , ds dks sin cos cos qs = ks + , 2 ds r r B (17) (18) (19)

dqs [ts (2 2 )] sin qs cos = ks ts + +p , ds r r dts (2 2 ) cos = ks qs . ds r

Flow of blood plasma around the red blood cell Following earlier work of Lighthill and T zeren & Skalak, an accurate description of the squeezing o of red blood cells was nally given by Secomb and colleagues in 1986 (reference [5]). The idea is to use an axisymmetric form of lubrication theory over the whole ow domain. The lubrication approximation does not apply near the nose and the rear end of the red blood cell, but the pressure gradient in these regions is small compared to that in the gap between the cell and vessel wall, so that the error in a calculation of the pressure drop based on an application of lubrication theory uniformly along the whole cell is small. In cylindrical coordinates moving with the cell, the ow is steady and the pressure pe (z) in the gap depends only on the axial position z. The axial uid velocity component u(, z), where measures distance from the axis, satises dp = dz u (20)

with boundary conditions, (i) u = 0 on the (yet unknown) cell surface = r(z), and (ii) u = U0 at the vessel wall = a, where U0 is the cell velocity. The solution may be written as 16 dp = 2 U0 dz a r2
1 2 a 2

a2 r 2 4 ln (r/a)

aQ0 / (a2 + r2 ) +

a2 r 2 . ln (r/a)

(21)

Here 2aQ0 is the leakback, the volume ow rate in the gap relative to the cell
a

2aQ0 =
r(z)

u(, z) 2 d.

(22)

Note that in terms of the variable s, dp dp = sin . ds dz The shear stress on the cell surface is given by (r) =
1 4

dp dz

a2 r 2 U0 + 2r . r ln (r/a) r ln (r/a)

(23)

This completes the mathematical model for the motion of a deformable red blood cell through a capillary. The model consists of the ve equations (17)(19), together with the two equations (21) and (23). The seven unknown functions of the variable s are r, , ks , qs , ts , p, and . Conditions that are imposed are that the net axial force on the cell vanishes:
stotal

2
0

(p cos + sin )r d = 0,

(24)

where stotal is the total arclength, that r and vanish on the axis of symmetry and that the surface area and volume of the cell have prescribed values. Some numerically obtained cell shapes are shown in gure 7. Here it was assumed that the cell is unstressed in a spherical shape with the same surface area as the stressed cell. The values of the physical parameters are = 103 kg m1 s1 , = 42 109 N/cm, B = 18 1018 N cm. Numerical difculties made it necessary in the 9

Figure 7: Computed shapes of axisymmetric red blood cells squeezing through glass capillaries. The cell
motion is from right to left. (a) Cell velocity U0 = 0.01 cm/s; displayed numerical values are tube diameters in m. (b) Tube diameter 6 m, displayed numerical values are cell velocities in cm/s.

high cell velocity calculations to neglect bending resistance and membrane shear; this makes the equation for dks /ds redundant and, because dqs /ds = 0, yields an algebraic equation for ks . As a consequence, the high-speed cells show an unrealistic cusp at the rear. Note that the model (the lubrication approximation, in particular) becomes better when the tubes become narrower (as in gure 7(a)).

F hrus effect and F hrus-Lindqvist effect a a The F hrus effect, the reduction of the hematocrit when blood ows through a capillary, may a be calculated from relation (2), U HT = . HD U0 In the model presented above the red blood cell is at rest, while the blood plasma and the vessel wall move in the positive z-direction. The total volume ow rate in this direction is then dened as 2aQ0 . In a frame of reference in which the vessel wall is at rest, the velocity of the red blood cell 10

Figure 8: Variation of the F hrus effect, indicated by the ratio HT /HD , with cell velocity U0 : a

, model including shear elasticity but neglecting bending; , model including both shear and bending elasticity.

is U0 and so the total volume ow rate is 2aQ0 a2 U0 . This means that the mean velocity U is the above relation is given in terms the parameters of the model by U = 2Q0 /a U0 ; hence HT 2Q0 =1 . HD aU0 (25)

The F hrus-Lindqvist effect, the dependency of the (apparent) blood viscosity on vessel dia ameter, may be calculated by reasoning as follows. In the model no hydrodynamic interaction between the red blood cells is taken into account: the cells have equal length and are separated from each other by sections of pure blood plasma of equal length L . The total pressure drop pL over a cell unit of length L is the sum of the pressure drops p over the length containing the cell and the pressure drop pL over the length L containing pure plasma: pL = p + pL . (26)

The volume ow rate through any cross-section of the capillary is a2 (U0 2Q0 /a), so that, by denition, 8app (U0 2Q0 /a)L . (27) pL = a2 The pressure p needs to be determined numerically from the equations given above, but an approximation for pL may be obtained from a simple model in which the ow over the entire length L is considered as fully-developed, i.e. as a Poiseuille ow. Then, by Poiseuilles law pL = 8p (U0 2Q0 /a)(L ) . a2 11 (28)

Figure 9: Variation of the F hrus-Lindqvist effect, indicated by the coefcient KT , with cell velocity U0 . a
Vessel diamter in m is shown on each curve. Theoretical results: , model including shear elasticity but neglecting bending; , model including both shear and bending elasticity. Finite-elements calculations: . Experimental results for different types of cells: , , .

Combining the three expressions (26)(28) yields app = p 1 + a2 p 1 8p (U0 2Q0 /a) L . (29)

If the cell volume is approximated as V a2 then the tube hematocrit is approximately given by HT = V/a2 L /L, and relation (29) may be rewritten as app = p (1 + KT HT ) , KT = a2 p 1. 8p (U0 2Q0 /a) (30)

Results of numerical calculations of HT /HD and KT as a function of the cell velocity and vessel diameter are presented in gures 8 and 9. A comparison of numerical results for a cell velocity U0 = 0.1 cm s1 with empirical correlations is shown in gure 2; there the numerical results are marked single-le ow model.

12

Figure 10: Graph showing ow resistance across microvascular networks as a function of the hematocrit.
The symbols are results from experimental determinations of pressure drop and volume ow rate in 10 microvascular networks. The dashed curve (marked in vitro) is a calculation based on a model similar to that described in section 6.2.

Squeezing of red blood cells through capillaries (in vivo)


The endothelial surface layer

It is fair to say that the model of Section 2 gives an accurate description of the ow of red blood cells in capillaries as observed in in vitro experiments. In the 1990s it became possible, for example in the group of Axel Pries at the Freie Universit t Berlin, to measure the properties of blood a ow in capillaries in vivo. It came as a surprise that there appeared to be a signicant difference between the results of the in vitro and in vivo experiments. This difference is demonstrated in gure 10, which shows data for the blood ow resistance in a capillary network as a function of the haematocrit. The various symbols are in vivo data, and the dashed line shows a calculation based on a model like that presented in Section 2, and so is representative of blood ow in vitro. The conclusion is that the apparent viscosity in vivo at physiological values of the hematocrit (about 45 %) is a factor 3 higher than the apparent viscosity in vitro. Already in 1966, J. H. Lugt had demonstrated that the walls of blood vessels were not smooth, but lined with a very thin layer of small elements protruding into the vessel lumen. After the remarkable discoveries of Priess group, it was immediately realized that this endothelial surface layer, also called the glycocalyx, may be the reason of the high ow resistance observed in in vivo experiments. Since that time the structure, chemical composition and physiological function of the glycocalyx has been a subject of intensive research, but still relatively little is known. An example of the hypothetical structure (proposed around 2000) is shown in gure 11. A comparatively thin (50100 nm) region on the endothelial surface is dominated by molecules (glycoproteins 13

Figure 11: Hypothetical structure of the endothelial surface layer or glycocalyx.

and proteoglycans) bound diectly to the plasmalemma, the glycocalyx in the strict sense. On the glycocalyx proper a much thicker ( 0.5 m) layer is attached which consists of a complex, threedimensional array of soluble plasma components, possibly including a variety of proteins and solubilized glycosaminoglycans. This layer is in a dynamic equilibrium with the owing blood plasma. Therefore, the thickness and composition of the endothelial surface layer depend on the plasma composition and the local hemodynamic conditions. For what concerns the function of the layer, it has become clear it plays a role in inammatory responses, vascular permeability, mechanotransduction, diabetes, atherosclerosis, and other vascular diseases.

Revised model for the ow of blood plasma and the elastic response of the red blood cell Since the glycocalyx appears to be hydrodynamically relevant, it has also been intensively investigated by hydrodynamicists. Especially mentioned should be the work of the groups of Edward Damiano at Boston University, of Timothy Secomb at the University of Arizona (in collaboration with Axel Pries at the Freie Universit t Berlin), and of Sheldon Weinbaum at the City University of a New York (in collaboration with Hans Vink at Maastricht University). Here we briey discuss the revisions proposed by Timothy Secomb and Axel Pries to the model of Section 2. For what concerns the ow of the blood plasma around the cell, the glycocalyx at the vessel wall is represented as a thin poroelastic layer of thickness w, when undisturbed (compare gure 12 with gure 6), in which the axial plasma velocity is described by the Brinkman equation k()u = p + z 14 u , (31)

Figure 12: Conguration assumed in the modied model of section 6.3.

where k() = /kp is the hydraulic resistivity, the ratio of the plasma viscosity and the permeability kp of the glycocalyx. It is assumed that k() = k0 , for > a w, k() = 0, for 0 a w, (32)

so that the small change of the hydraulic resistivity due to compression of the glycocalyx by the red blood cell is neglected. The velocity u in eqn (4) is interpreted as the absolute velocity, rather than the velocity of the interstitial uid relative to the velocity of the solid matrix material (as it should!). Solving eqn (31) analytically now provides modied, and more complex, expressions for the pressure gradient and shear stress (cf. eqns (21) and (23)). Two changes are made to the mathematical description of the elastic response of the cell membrane as given earlier in Section 2. The rst change is that expression (15) for the deviatoric part of the principal stress in the membrane is now modeled as
1 td = 2 2

1 2

1 ms (ks k ) + 2

m , t

(33)

so that the second and third terms on the right-hand side have been added. The second term has been added to represent the effect that bending of (bilayer) membranes are accompanied by in-plan stresses. The third term is a viscous contribution resulting from in-plane shear deformation of the membrane. The second change arises from the presence of the glycocalyx at the endothelial cell wall: eqn (8), expressing the equilibrium of the normal stresses on the membrane, is modied into 1 d(rqs ) = ks ts + k t + p + f, r ds (34)

where the new term f represents an osmotic force acting on the (porous) membrane. This force arises because the absorption of plasma proteins onto the glycocalyx generates a protein concentration gradient between the glycocalyx and the interior of the cell. (It should be mentioned that the presence of this force is still a subject of debate.) Figure 13 presents examples of cell shapes obtained by numerical solution of the modied equations, where it now is assumed that the unstressed cell is a biconcave disk, rather than a 15

Figure 13: Examples of numerical calculated cell shapes in the presence of a glycocalyx. Each panel
corresponds to a different cell velocity,; in each panel the vessel diameter is 6 m.

sphere, and that the pertinent physical parameters have the values = 103 kg m1 s1 , = 60 109 N cm1 , B = 18 1018 Ncm, m = 106 kg m1 s1 , f = 20 105 N/cm2 , w = 0.7 m, k0 = 2 103 N s/cm4 . The red blood cell volume and surface area are taken as 90 m3 and 135 m2 , respectively (values for these latter quantities have not been mentioned in section 6.2, because no information could be found in the original publication [5]). A remarkable result in gure 13 is that for low velocities the cell displaces the glycocalyx almost completely. At larger velocities the cell lifts off from the glycocalyx and a clear layer of blood plasma is found between the red cell membrane and the glycocalyx. These phenomena have indeed been observed in experiments. The power of the modied model is further demonstrated in gure 14. This presents the variation with the red blood cell velocity of the mean width of the gap between the cell and the vessel wall, as calculated with the modied model and with the model of Section 2. The results of the modied model appear to agree very well with experimental observations. Here it might be argued that the choice of the values of the parameters w and k0 may have played a role; nevertheless, these values are quite realistic. 16

Figure 14: Predicted and experimentally observed variation of the mean gap width with red blood cell
velocity. (a) Model results with and without the glycocalyx (solid curves) and experimental data (dots and dashed curve); (b) Model results for three combinations of the parameters w and k0 , chosen to give the same gap width for a red blood cell velocity of 100 m/s (solid curves) and experimental data (dots and dashed curve).

Finally, the inuence of the glycocalyx on the F hrus effect and the F hrus-Lindqvist efa a fect is exemplied in gure 15. Shown are the average gap width (A), the ratio HT /HD of tube hematocrit to discharge hematocrit, i.e. the F hrus effect (B), the variation of the apparent visa cosity, through KT in eqn (30), i.e. the F hrus-Lindqvist effect (C), and the ow resistance for a a discharge hematocrit of 45 % (D). That resistance is dened as R45 = R0 (1 + KT HT ), where R0 is the resistance of pure plasma ow. The glycocalyx is predicted to cause (i) a stronger F hrus effect (lower HT /HD ), (ii) an increased apparent viscosity, and (iii) an increased ow a resistance, as compared to the case without a glycocalyx. The increase in R45 reects the increase in R0 (in the absence of the red blood cells) resulting from the glycocalyx and the proportional effect of the the red blood cells on resistance (higher KT ). Together these more than compensate for the reduction in tube hematocrit HT in the presence of a glycocalyx.

17

Figure 15: Average gap width (A), F hrus effect (B), F hrus-Lindqvist effect (C), and ow resistance for a a
a discharge hematocrit of 45 % (D) as a function of the red blood cell velocity. Dashed curves: predictions without glycocalyx; solid curves: predictions with glycocalyx; dotted parts: regions in which the numerical procedure may have been not sufciently accurate.

References and further reading


Part of Section 1 is based on an interesting article on the life and work of Poiseuille [1] S UTERA , S.P & S KALAK , R. 1993 The history of Poiseuilles law. Annu. Rev. Fluid Mech. 25, 119. An excellent review on the motion of red blood cells in capillaries is [2] S ECOMB , T. W. 2003 Mechanics of red blood cells and blood ow in narrow tubes. In Modeling and Simulation of Capsules and Biological Cells (edt. C. Pozrikidis), pp. 163 196. Chapman & Hall/CRC. Articles relevant to Section 2 are [3] L IGHTHILL , M. J. 1968 Pressure-forcing of tightly tting pellets along uid-lled elastic tubes. J. Fluid Mech. 34, 113134. [4] T OZEREN , H. & S KALAK , R. 1978 The steady ow of closely tting incompressible elastic spheres in a tube. J. Fluid Mech. 87, 116. 18

[5] S ECOMB , T. W., S KALAK , R., O ZKAYA , N. & G RASS , J. F. 1986 Flow of axisymmetric red blood cells in narrow capillaries. J. Fluid Mech. 163, 405423. Articles relevant to Section 3 are [6] P RIES , A. R., S ECOMB , T. W., G ESSNER , T., S PERANDIO , M. B., G ROSS , J. F. & G AE HTGENS , P. 1994 Resistance to blood ow in microvessels in vivo. Circ. Res. 75, 904915. [7] DAMIANO , E. R. 1998 The effect of the endothelial-cell glycocalyx on the motion of red blood cells through capillaries. Microvasc. Res. 55, 7791. [8] S ECOMB , T. W., H SU , R. & P RIES , A. R. 1998 A model of red blood cell motion in glycocalyx-lined capillaries. Am. J. Physiol. Heart Circ. Physiol. 274, H1016H1022. [9] F ENG , J. & W EINBAUM , S. 2000 Lubrication theory in highly compressible porous media: the mechanics of skiing, from red cells to humans. J. Fluid Mech. 422, 281317. [10] S ECOMB , T. W., H SU , R. & P RIES , A. R. 2001 Motion of red blood cells in a capillary with an endothelial surface layer: effect of ow velocity. Am. J. Physiol. Heart Circ. Physiol. 281, H629H636. For an introduction to the structure and physiological relevance of the glycocalyx, see the review articles [11] P RIES , A. R., S ECOMB , T. W. & G AEHTGENS , P. 2000 The endothelial surface layer. P gers Arch.Eur. J. Physiol. 440, 653666. u [12] R EITSMA , S., S LAAF, D. W., V INK , H., VAN Z ANDVOORT, M. A. M. J. & O UDE E G BRINK , M. G. A. 2007 The endothelial glycocalyx: composition, functions and visualization. P gers Arch.Eur. J. Physiol. 454, 345359. u [13] W EINBAUM , S., TARBELL , J. M. & DAMIANO , E. R. 2007 The structure and function of the endothelial glycocalyx layer. Annu. Rev. Biomed. Engng 9, 121167. For an introduction to the mechanics of cells, see [14] B RAY, D. 2001 Cell Movements: From Molecules to Motility. Cambridge University Press, 2nd edition. [15] B OAL , D. 2002 Mechanics of the Cell. Cambridge University Press. [16] K AMM , R. D. 2002 Cellular uid mechanics. Annu. Rev. Fluid Mech. 34, 211232. [17] M OFRAD , M. R. K. & K AMM , R. D., University Press.
EDTS

2006 Cytoskeletal Mechanics. Cambridge

[18] M OFRAD , M. R. K. 2009 Rheology of the cytoskeleton. Annu. Rev. Fluid Mech. 41, 433453. Figure 1 has been taken from reference [1], gures 4 and 5 from reference [2], gures 69 from reference [5], gure 10 from reference [6], gure 11 from reference [11], gures 1215 from reference [10]. 19

You might also like