Respiratory Physiology & Neurobiology 163 (2008) 214–221

Contents lists available at ScienceDirect

Respiratory Physiology & Neurobiology

journal homepage: www.elsevier.com/locate/resphysiol

The mechanics of airway closure

Matthias Heil a,∗ , Andrew L. Hazel a , Jaclyn A. Smith b
a School of Mathematics, University of Manchester, Oxford Road, Manchester M13 9PL, UK
b School of Translational Medicine, University of Manchester, University Hospital of South Manchester, Southmoor Road, Manchester M23 9LT, UK

a r t i c l e i n f o a b s t r a c t

Article history: We describe how surface-tension-driven instabilities of the lung’s liquid lining may lead to pulmonary
Accepted 7 May 2008 airway closure via the formation of liquid bridges that occlude the airway lumen. Using simple theoretical
models, we demonstrate that this process may occur via a purely fluid-mechanical “film collapse” or
Keywords: through a coupled, fluid-elastic “compliant collapse” mechanism. Both mechanisms can lead to airway
Pulmonary airway closure closure in times comparable with the breathing cycle, suggesting that surface tension is the primary
Liquid lining
mechanical effect responsible for the closure observed in peripheral regions of the human lungs. We
Surface tension
conclude by discussing the influence of additional effects not included in the simple models, such as
gravity, the presence of pulmonary surfactant, respiratory flow and wall motion, the airways’ geometry,
and the mechanical structure of the airway walls.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction Experimental evidence that airway closure occurs in healthy

human lungs was first described by Burger and Macklem (1968)
The pulmonary airways are lined with a thin liquid film extend- whose elegant studies provided evidence of small-airway clo-
ing from the trachea down to the alveoli. The film provides sure during expiration, when the lung’s residual volume was
protection against infection and oxidation, and is believed to facil- approached. Animal studies of airway closure by Hughes et al.
itate the transfer of gases between the pulmonary circulation and (1970) found histological evidence that during lung deflation, the
the inspired air at the level of the alveoli. The surface tension that site of airway closure was confined to the very small airways
acts at the film’s air–liquid interface has a significant effect on the (≈0.4–0.6 mm in diameter).
lung’s mechanical behaviour and is responsible for 50–60% of the In healthy lungs, airway closure occurs only in the small air-
lung’s elastic recoil. Pulmonary surfactant reduces the surface ten- ways, towards the end of expiration when the airway diameters
sion of the film and helps the fluid to spread over the airway surface. are smallest and, by virtue of mass conservation, the liquid film is
The liquid lining usually forms a thin, relatively uniform layer on at its thickest. The liquid blockages normally rupture during the
the airway wall but it is possible for the airway to become blocked early stages of inspiration. The development of more permanent
by the liquid, resulting in airway closure. It is the main aim of this occlusions or the occurrence of airway closure in the larger air-
paper to describe the various mechanisms involved in this process. ways is either a consequence of certain pathological conditions
Throughout this paper we shall distinguish airway closure (e.g. an increase in the volume of fluid in pulmonary oedema, or
(which involves the formation of a liquid blockage) from the purely an increase in the surface tension in neonatal respiratory distress
structural collapse of the airway walls in response to external or syndrome); or the result of deliberate medical interventions, such
internal forces. The latter can occur, e.g. during expiratory flow as the installation of a liquid plug in the larger airways during sur-
limitation, discussed in the article by Bertram (2008), and tends factant replacement therapy; see e.g. Espinosa and Kamm (1998)
to leave the airway lumen open and thus available for respiration. and the article by Grotberg et al. (2008). The tendency for airway
We shall demonstrate, however, that the structural collapse of the closure to occur is also increased during sleep, general anaesthesia
airway wall can aid the formation of liquid blockages, and further- and in obesity (Appelberg et al., 2007; Rothen et al., 1998).
more, that the formation of an occlusion can itself cause a structural The principal mechanical effect responsible for airway closure is
collapse. the surface tension acting at the interface between the liquid lining
and the air in the lumen. Fig. 1 shows a sketch of the cross-section
through an idealised, straight, circular airway of radius a, lined with
∗ Corresponding author. Tel.: +44 161 275 5808.
a thin, static film of uniform thickness h0  a. The surface tension
E-mail address: [email protected] (M. Heil).
acting at the curved interface induces a jump in pressure across the

1569-9048/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.resp.2008.05.013
 M. Heil et al. / Respiratory Physiology & Neurobiology 163 (2008) 214–221 215

interface, and the pressure in the fluid is given by the Young–Laplace ics: if the airway wall is compressed, the curvature of the air–liquid
equation interface will increase, causing an increase in the compressive load
on the wall (via the final term in Eq. (2)), which will further increase
pfluid = pint − , (1)
the compression. Kamm and Schroter (1989) referred to this pro-
where pint is the air pressure in the lumen,  is the surface tension cess as “compliant collapse” and we shall discuss this mechanism
and  is the mean curvature of the air–liquid interface. For the uni- in Section 3. The propensity of an airway to become occluded by
form, axisymmetric liquid film shown in Fig. 1 the mean curvature this mechanism is largely determined by the ratio of the surface-
is constant and given by  = 1/(a − h0 ), i.e. it is inversely related to tension-induced pressure jump over the air–liquid interface to the
the radius of the air–liquid interface. wall stiffness. Since the curvature of the air–liquid interface is great-
Eq. (1) shows that spatial variations in the interfacial curvature est in the smallest airways, they are most likely to be susceptible to
generate pressure gradients in the fluid. Hence, if the uniform liq- “compliant collapse”. In emphysema, destruction of parenchymal
uid lining is disturbed, the induced pressure gradients drive flows tissue and reduced tethering decreases the airway stiffness and may
that will redistribute the fluid and this process may lead to airway allow the occurrence of such collapse in larger airways.
closure via a mechanism that Kamm and Schroter (1989) termed We note that the competition between surface tension and wall
“film collapse”. We shall discuss this mechanism in Section 2. stiffness can also be cast in energetic terms. The equilibrium config-
Another mechanical property that contributes to airway closure urations of the liquid-lined airway are characterised by a minimum
is the elastic nature of the airway walls, which deform in response of the total energy, comprising the surface energy (given by the
to the pressures exerted on them. Denoting the external (pleural) product of surface tension and the area of the air–liquid interface)
pressure acting on the airway by pext , the net (compressive) load and the strain energy (the elastic energy stored in the deformed air-
experienced by the airway wall in Fig. 1 is way walls). The compression of the airway wall reduces the former
but increases the latter. The airway will therefore tend to collapse if
pwall = pext − pfluid = pext − pint + . (2)
the axisymmetric, axially uniform configuration sketched in Fig. 1
This compressive load is resisted by the wall’s internal stiffness has a higher total energy than a corresponding collapsed/occluded
(characterised here by its Young’s modulus E and the wall thickness configuration. In that case, the system will evolve towards a new
hw ) and by the structural support provided by the parenchymal equilibrium configuration of lower total energy, i.e. the collapsed
tethering, represented here by distributed elastic springs of stiff- state.
ness k. In this paper, we describe the details of both the “film collapse”
Eq. (2) shows that the structural collapse of the airway wall may and “compliant collapse” mechanisms and their interaction. We
be caused by an increase in the pleural pressure, for example under begin with the purely fluid-mechanical “film collapse”, which tends
conditions of pneumothorax and/or pleural effusion, and/or by a to occur when the surface tension is much smaller than the wall
reduction in the internal pressure, e.g. during expiratory flow lim- stiffness, before turning our attention to the more complex fluid-
itation. The last term in Eq. (2) indicates that it is also possible for elastic “compliant collapse”, which occurs when the surface tension
an airway to collapse if the compressive load generated by the liq- is relatively high. In both cases we include the minimum number
uid lining becomes sufficiently large. Kamm and Schroter (1989) of physical effects required to elucidate the mechanism. In partic-
observed that the latter provides a mechanism for airway closure ular, we neglect the effects of gravity, the presence of surfactant,
via a destabilising feedback between the fluid and solid mechan- the curvature of the airways and the presence of airway bifurca-
tions. In addition, the models of the lining liquid and airway wall
are kept deliberately simple. The influence of the neglected effects
is discussed in Section 4, in which we also consider some remaining
open questions.

2. “Film collapse”: liquid plug formation in rigid

axisymmetric vessels

In larger airways the surface tension is small, relative to the

airway stiffness, hence the surface-tension-induced compression
does not produce significant wall deformations and the airway
wall is essentially rigid. The redistribution of the liquid from a thin
lining film into an occluding liquid bridge can then occur via a clas-
sical fluid-mechanical instability, known as the Rayleigh–Plateau
instability which is also responsible for the surface-tension-driven
breakup of liquid jets into individual droplets (Plateau, 1873;
Rayleigh, 1879)
To explain the origin of the instability, Fig. 2 illustrates how the
fluid in a uniform liquid film that lines an axisymmetric airway
responds to an axisymmetric perturbation of its thickness from
the initial value h0 to h(z). The mean curvature, , of the axisym-
metric air–liquid interface is given by the sum of its two principal
curvatures, i.e. the curvatures along the length of the airway (the
principal axial curvature) and in its cross-section (the principal cir-
Fig. 1. A simple model of an elastic airway (radius a, wall thickness hw and Young’s cumferential curvature); for small-amplitude perturbations to the
modulus E), supported by external tethering (representing the parenchyma, and film thickness we have  ≈ d2 h/dz 2 + 1/(a − h); see, e.g., Gauglitz
modelled as distributed springs with spring stiffness k), and lined with a thin liquid
film of uniform thickness h0 . The surface tension at the air–liquid interface is  and
and Radke (1988). The principal axial curvature, d2 h/dz 2 , is nega-
the airway is subjected to an external (pleural) pressure pext . The pressure in the tive where the film is thickest and positive were the film is thinnest.
lumen is pint . The principal circumferential curvature, 1/(a − h), is always posi-
216 M. Heil et al. / Respiratory Physiology & Neurobiology 163 (2008) 214–221

Fig. 2. Illustration of the axisymmetric Rayleigh–Plateau instability. Sinusoidal perturbations to an initially uniform film of thickness h0 create pressure variations that drive
flows within the fluid. (a) and (b) If the wavelength of the perturbation L, is less than the initial circumference of the air–liquid interface, the fluid will return to the axially
uniform state. (c) and (d) Perturbations with a wavelength in excess of the initial circumference of the air–liquid interface will generate flows that cause the perturbation to
grow in amplitude.

tive and greatest where the film is thickest. From Eq. (1) it follows would continue to drive a flow. Fig. 4 shows the system’s possible
that the axial curvature generates a pressure distribution that axisymmetric equilibrium configurations: a cylindrical air–liquid
attempts to return the fluid in the liquid lining to an axially uni- interface, bounding an axially uniform film (Fig. 4a); an unduloid-
form state, with a high pressure in the thick lobes and low pressure shaped interface, bounding an annular collar of fluid (Fig. 4b); and
in the depleted regions. Conversely, the circumferential curvature occluding liquid bridges (Fig. 4 c,d) in which the fluid is bounded
is destabilising in the sense that it tends to drive yet more fluid into by two spherical caps that meet the vessel walls at a certain contact
the regions where the film thickness is already elevated. The rela- angle, . In the lung the fluid in the liquid lining wets the airway
tive importance of these two competing effects depends on the axial walls, implying a zero contact angle.
wavelength, L, of the perturbation. For short-wavelength perturba- The volume of fluid required to form the “minimal” liquid bridge
tions the (stabilising) variations in the axial curvature dominate, shown in Fig. 4(c) is given by
and as a result the fluid pressure increases in the regions where
the film is thickest, generating flows that return the fluid to its axi- 2a3 2 sin3  + 3 cos2  − 2
Vmin = .
ally uniform state, as shown in Fig. 2(a). Conversely, if the axial 3 cos3 
wavelength of the perturbation is sufficiently large, the destabilis- We note that this provides a lower bound on the volume of fluid
ing effect of the circumferential curvature dominates, leading to a required to form an occlusion in an axisymmetric airway because
reduction in the fluid pressure in the thickest regions, driving more it is unlikely that a “minimal” liquid bridge (in which the oppos-
fluid into these regions, and enhancing the non-uniformity of the ing air–liquid interfaces just touch on the airway’s centreline) can
film further. This scenario is illustrated in Fig. 2(c). be formed via a continuous redistribution of the fluid initially
A formal linear stability analysis (e.g. Goren, 1962) shows that contained in an axially uniform film. The actual minimal film thick-
the liquid lining is unstable to sinusoidal perturbations, along the
length of the airway, whose wavelength L exceeds the circumfer-
ence of the undeformed air–liquid interface so that L > 2(a − h0 ).
This threshold may also be derived from energetic considerations
by determining the configurations that minimise the interfacial
energy (or, equivalently, the area of the air–liquid interface), sub-
ject to the constraint that the volume of fluid contained in the
liquid lining remains constant; see e.g. Everett and Haynes (1972).
Within the energy-based framework, the competition between sta-
bilising/destabilising axial/circumferential effects is related to the
fact that undulations of the air–liquid interface along the length of
the airway, tend to increase the interfacial area (as shown in the
longitudinal sections sketched in Fig. 2), while the reduction in the
interface radius in the thick-film regions tends to decrease it.
A similar argument may be employed to demonstrate that the
uniform liquid film is stable to perturbations in the transverse plane
(Hammond, 1983). This is illustrated in Fig. 3 which shows that the
variations in the curvature of the air–liquid interface, induced by a
perturbation to its shape, generate flows that return the interface
to an axisymmetric configuration.
Once a suitable initial perturbation to the air–liquid inter-
face has initiated the axisymmetric Rayleigh–Plateau instability,
the fluid in the liquid lining continues to redistribute until the
system reaches a new equilibrium configuration with lower inter-
Fig. 3. Sketch illustrating the stability of the liquid lining with respect to non-
facial energy. In this configuration the air–liquid interface must
axisymmetric perturbations. Non-axisymmetric perturbations to the film thickness
again be a surface of constant mean curvature—if it were not, the generate surface-tension-induced circumferential pressure gradients that drive the
pressure gradients induced by non-uniformities in the curvature fluid back towards the axisymmetric configuration.
 M. Heil et al. / Respiratory Physiology & Neurobiology 163 (2008) 214–221 217

Fig. 4. Possible equilibrium configurations for the liquid lining in an axisymmetric airway: (a) a uniform liquid film; (b) an unduloid; (c) a minimal liquid bridge; (d) a
finite-thickness liquid bridge. In all four cases, the air–liquid interface is a surface of constant mean curvature. In (b–d)  is the contact angle between the air–liquid interface
and the airway wall.

ness, hmin , required for the formation of an occluding liquid bridge surface-tension-induced pressure  exerted by the liquid film; the
can only be determined by following the system’s evolution after pressure in the lumen was set to zero, pint = 0. (Note that the results
the onset of the initial instability to determine whether it evolves in Fig. 5 were actually calculated with an elastic airway wall, but
towards a non-occluding unduloid configuration (as in Fig. 4b) or the wall deformation remains small because the wall is relatively
towards a finite-thickness liquid bridge (as in Fig. 4d). stiff.)
Numerical simulations (e.g. by Gauglitz and Radke, 1988; Fig. 5(a) shows the early stages of the system’s evolution dur-
Johnson et al., 1991) and experiments (e.g. by Cassidy et al., 1999) ing which the perturbation to the air–liquid interface, and hence
indicate that a film thickness of at least 12% of the vessel’s radius is its curvature, is approximately sinusoidal. During this stage the
required for an occluding liquid bridge to form. perturbation to the film thickness grows exponentially, so that
The temporal development of the instability in an axisymmetric |h(z, t) − h0 | ∼ exp(t). The growth rate  is given by
tube is shown in Fig. 5 for a case in which h0 < hmin so that the sys-
tem evolves towards a non-occluding unduloid. Fig. 5 presents axial
sections through the liquid-lined airway at three different times.  h 3 n2  
Also shown is the total pressure exerted along the length of the air-  0 a2 2
= −n , (3)
way wall, comprising the constant external pressure, pext , and the 3 a a − h0 (a − h0 )
2

Fig. 5. Evolution of the axisymmetric fluid-elastic instability: The figures on the left show the evolution of (one half of) the growing axisymmetric lobe; the figures on the
right show the total inward pressure (comprising the surface-tension-induced fluid pressure , and the external pressure, pext ) acting on the airway wall. As the lobe grows,
the fluid exerts a large compressive load on the wall. The inset in (c) shows details of the induced wall deformation and the thin “neck region” that bounds the main lobe.
Modified from White and Heil (2005).
218 M. Heil et al. / Respiratory Physiology & Neurobiology 163 (2008) 214–221

where n = 2a/L is the wavenumber of the axisymmetric perturba-

tion; see e.g. White and Heil (2005). The above expression reflects
the fact that the instability is driven by surface tension (the growth
rate increases with ) and resisted by viscous effects (the growth
rate decreases with an increase in the fluid’s viscosity, , and with
a reduction in the film thickness, h0 ).
As fluid is drawn into the growing lobe, the air–liquid interface
approaches a shape of constant mean curvature and as a result the
pressure in the lobe becomes approximately uniform; see Fig. 5(b).
During the final stages of the fluid’s redistribution, a thin “neck
region” , shown in the inset in Fig. 5(c), tends to develop at the end of
the growing lobe. This region offers a large viscous resistance to the
flows that continue to slowly drain fluid from the remaining thin- Fig. 6. Cross-sections through an axially uniform, collapsed airway loaded by the
film region into the main lobe as it evolves towards the unduloid- surface-tension-induced fluid pressure and the external pressure. After Rosenzweig
and Jensen (2001).
shaped collar, its ultimate equilibrium configuration.
For films whose initial thickness exceeds hmin , the initial stages
of their evolution is similar to that shown in Fig. 5. However, once way wall to buckle non-axisymmetrically. The fluid in the liquid
the minimum radius of the air–liquid interface is reduced below lining redistributes into isolated pockets formed by the lobes. Com-
a certain threshold, the evolution accelerates dramatically as the pared to the axisymmetric configuration shown in Fig. 6(a), the
interface contracts to form an occlusion. buckled states have a much lower total energy because the non-
The overall timescales required for an occlusion to form are very axisymmetric collapse allows a significant reduction in the area of
sensitive to the initial film thickness, h0 . However, the final stages of the air–liquid interface, while the bending-dominated wall defor-
the “film collapse” proceed extremely rapidly, allowing the devel- mation only requires relatively small strain energies.
opment of an occluding liquid bridge within a single breathing In the case shown in Fig. 6(d), the non-axisymmetric collapse
cycle; see e.g. Johnson et al. (1991). results in a massive reduction in the airway lumen although a small
central air core remains open and thus available for respiration. For
larger values of the surface tension and/or greater volumes of fluid
3. “Compliant collapse”: fluid-elastic instabilities in the liquid lining, the non-axisymmetric collapse can also lead to
complete airway closure.
3.1. Axisymmetric wall deformations Axially non-uniform equilibrium configurations were calculated
by Heil (1999a,b). Fig. 7 shows a sequence of static equilibrium con-
The effect of wall elasticity on the dynamics of the figurations of an airway that is occluded by a liquid bridge and
Rayleigh–Plateau instability was first investigated by Halpern and subjected to different values of the external (pleural) pressure, pext .
Grotberg (1992, 1993). They showed that the reduction in fluid The volume of fluid contained in the liquid bridge is just suffi-
pressure in the region of the growing lobe, shown in the plots cient to form a minimal liquid bridge, such as the one shown in
on the right of Fig. 5, can lead to a significant compression of Fig. 4(c), when the airway is in its undeformed, axisymmetric state.
the airway wall. The wall compression reduces the radius of the As the airway collapses, the liquid bridge spreads out and subjects
air–liquid interface even further and thus facilitates the formation an increasing fraction of the airway’s length to a compressive load.
of an occluding liquid bridge. The (relatively small) wall deforma-
tion induced by the fluid pressure can be seen in the inset in Fig. 5(c).
Halpern and Grotberg (1992, 1993) also considered cases in
which the axisymmetric wall deformations are much larger than
those shown in Fig. 5. However, such cases are unlikely to be
observed in practice because compressed cylindrical tubes tend
to buckle non-axisymmetrically when the compression exceeds a
certain threshold; see e.g. Yamaki (1984). This buckling instabil-
ity arises because the bending stiffness of thin-walled elastic tubes
is much smaller than their extensional stiffness. Beyond a certain
degree of compression it is more energetically favourable for the
tube wall to buckle non-axisymmetrically than to undergo further
axisymmetric deformations. Observations of occluded airways ex
vivo show that the formation of occlusions does indeed tend to
be accompanied by the non-axisymmetric collapse of the airway
walls; see e.g. Hughes et al. (1970).

3.2. Non-axisymmetric wall deformations

If the wall deforms non-axisymmetrically, a wider class of

Fig. 7. (a–c) Sequence of static equilibrium shapes of a non-axisymmetrically buck-
equilibrium states are possible than for purely axisymmetric defor- ling airway, occluded by a liquid bridge. The volume of fluid is sufficient to form
mation. Hill et al. (1997) and Rosenzweig and Jensen (2001) an occluding liquid bridge in the axisymmetric airway. Only one half of the (sym-
computed the non-axisymmetric equilibrium shapes of axially metrically collapsed) airway and one of the two air–liquid interfaces are shown.
uniform, liquid-lined tubes. Fig. 6 shows typical results from Configurations (a) and (b) are statically unstable. The stable configuration shown
in (c) (and again in (d)) has three distinct regions: (I) the undeformed airway; (II)
Rosenzweig and Jensen’s (2001) study. The combination of the
a transition region which contains the air–liquid interface; (III) a liquid-filled, uni-
large pressure jump over the highly curved air–liquid interface formly collapsed region in which large parts of the airway wall are in opposite wall
and the external pressure causes the initially axisymmetric air- contact. From Heil (1999b).
 M. Heil et al. / Respiratory Physiology & Neurobiology 163 (2008) 214–221 219

Fig. 8. A minimal liquid bridge occluding a non-axisymmetrically buckled airway.

The liquid bridge contains only about a tenth of the volume of fluid required to
Fig. 9. Evolution of a non-axisymmetric fluid-elastic instability. (a) The ini-
occlude the airway in its axisymmetric state. Only one half of the (symmetrically
tial, axisymmetric, axially uniform state; (b) the evolving axisymmetric
collapsed) airway and one of the two air–liquid interfaces are shown. The small inset
Rayleigh–Plateau instability: fluid drains into the axisymmetric lobe and creates a
shows the overall shape of (one half of) the occluded airway. From Heil (1999a).
strong compressive load on the airway wall; (c) the early stages of non-axisymmetric
collapse: the airway wall is beginning to buckle; (d) late stages of non-axisymmetric
collapse, just before the radius of the air–liquid interface is reduced to zero. Only
Furthermore, the curvature of the air–liquid interface increases
one half of the (symmetrically collapsed) airway is shown and a sixth of the liquid-
strongly with the degree of collapse, leading to a rapid increase lined airway is cut away to show thickness of the liquid lining. From Hazel and Heil
in the compressive load on the wall. The combination of these two (2005).
effects is strongly destabilising and the two intermediate configu-
rations shown in Fig. 7(a) and (b) are, in fact, statically unstable. This
indicates that if an elastic airway were occluded by an axisymmet- evolution towards a non-occluded equilibrium state may cause
ric liquid bridge it would collapse further into a much more strongly the formation of occluding liquid bridges from fluid volumes that
deformed configuration, such as the one shown in Fig. 7(c) where are too small to form an axisymmetric liquid bridge. Moreover,
the occurrence of opposite-wall-contact over a significant fraction the mechanism does not require the initial compressive load to
of the occluded segment generates sufficiently strong elastic restor- exceed the buckling pressure of the tube. The process is illus-
ing forces to balance the large compressive load generated by the trated in Fig. 9. Fig. 9(a) shows the axisymmetric airway, lined with
liquid bridge. a uniform liquid film which is unstable to the primary, axisym-
Heil (1999a) also demonstrated that non-axisymmetric airway metric Rayleigh–Plateau instability. Following the onset of this
collapse allows the formation of occluding liquid bridges with far instability, fluid drains into an axisymmetric lobe, as shown in
less fluid than would be required to occlude an axisymmetric air- Fig. 9(b), causing an increase in interfacial curvature and a reduc-
way. Fig. 8 shows an example of a non-axisymmetric minimal liquid tion of the fluid pressure, which increases the compressive load
bridge formed with a volume of fluid that is about a tenth of Vmin . on the airway wall. When the compressive load exceeds a certain
The airway is held in a strongly collapsed configuration by the threshold, the airway begins to buckle non-axisymmetrically, as
pressure jump over the two highly curved air–liquid interfaces shown in Fig. 9(c). The resulting non-axisymmetric distortion of
that bound the liquid bridge. The plot shows only one half of the the air–liquid interface generates circumferential pressure gradi-
symmetrically collapsed vessel and one of the two menisci. The ents which set up draining flows. These act to return the air–liquid
minimum liquid bridge configuration is characterised by the two interface to an axisymmetric shape, and also oppose the wall’s
opposite air–liquid interfaces just touching in the symmetry plane. non-axisymmetric deformation (cf. Fig. 3). However, the total com-
We note that the airway collapse is restricted to the immediate pressive load on the airway wall increases very rapidly as the radius
vicinity of the liquid bridge—away from the occlusion the airway of the air–liquid interface decreases, causing the non-axisymmetric
rapidly reopens to its undeformed axisymmetric shape. collapse to accelerate dramatically towards a complete occlusion;
see Fig. 9(d).
In both the two- and three-dimensional simulations, it was
3.3. Dynamic non-axisymmetric collapse
established that the timescales required for the non-axisymmetric
collapse to occur are comparable to the duration of the breathing
As in the axisymmetric case, the mere existence of occluding
cycle, as is the case for axisymmetric collapse.
liquid bridges that contain a relatively small volume of fluid does
not imply that these can be realised via a continuous evolution
from an initial state in which the axisymmetric airway is lined by 4. Discussion
a thin, uniform layer of fluid. In two-dimensional simulations, Heil
and White (2002) demonstrated that the axially uniform collapsed The studies reviewed above demonstrate that surface-tension-
states considered by Hill et al. (1997) and Rosenzweig and Jensen driven instabilities of the airway’s liquid lining can lead to airway
(2001) may be realised, provided that the initial compressive load closure via the formation of occluding liquid bridges. The insta-
exceeds the critical buckling pressure of a two-dimensional elastic bilities are driven by the tendency of the fluid in the liquid lining
ring. to redistribute such that the area of the air–liquid interface (or,
In three-dimensional simulations, White and Heil (2005) and equivalently, its surface energy) is minimised. Wall elasticity facil-
Hazel and Heil (2005) demonstrated that the occurrence of itates airway closure through the “compliant collapse” mechanism
non-axisymmetric instabilities during the system’s axisymmetric and in the smaller airways, where the pressure jump over the
220 M. Heil et al. / Respiratory Physiology & Neurobiology 163 (2008) 214–221

highly curved air–liquid interface can become comparable to the 4.3. Airway bifurcations
wall stiffness, non-axisymmetric airway collapse allows the for-
mation of occluding liquid bridges with relatively small volumes In vivo, the pulmonary airways are slightly curved and the dis-
of fluid. Moreover, the studies established that increasing the sur- tance between successive bifurcations is only a few tube diameters.
face tension, decreasing the wall stiffness, and/or increasing the Although short, such tube lengths are sufficient for the initial
initial film thickness caused a more rapid development of the Rayleigh–Plateau instability to develop. Furthermore, Figs. 7–9
instability. show that airway closure tends to be a strongly localised phe-
The studies described deliberately employed rather idealised nomenon. The presence of bifurcations at the end of the airway
models of the pulmonary airways and their liquid lining in order segments, which was not taken into account in any of the studies
to focus on what we believe to be the most important mechanisms reviewed here, is therefore unlikely to have a strong effect on the
involved in the airway closure process. The inclusion of additional airway closure process.
physical effects and/or the use of more realistic models for the fluid
or solid mechanics, are likely to affect the precise details of the
system’s evolution but will not alter the fundamental instability 4.4. Centreline curvature
mechanisms. We now briefly discuss how the airway closure pro-
cess is affected by various mechanisms that we have ignored thus Jensen (1997) studied the effect of slight airway curvature on
far. the distribution of fluid in the liquid lining but he only considered
cases in which the volume of fluid was too small for occluding liq-
uid bridges to form. The airway’s centreline curvature was shown
4.1. Wall mechanics
to cause the fluid to redistribute along the circumference to form a
thicker layer on the “outside” wall (i.e. away from the centre of cen-
The precise details of the wall mechanics appear to be irrel-
treline curvature), resulting in a distribution of fluid similar to that
evant for most aspects of the “compliant collapse”. For instance,
achieved by transverse gravity. By analogy, therefore, it is possible
both Halpern and Grotberg (1992) and White (2003) axisymmetric
that the Rayleigh–Plateau instability may be suppressed in small
studies yield qualitatively similar results even though rather differ-
curved airways, but we are not aware of any studies that assess how
ent wall models were employed. One feature of the instability that
the airway’s centreline curvature affects the initial stages of the air-
can be strongly affected by the specific wall model is the azimuthal
way closure process; its later stages are unlikely to be affected by
wavenumber of the pattern that develops once the airway has buck-
slight variations in the airway’s initial shape as by then the process
led non-axisymmetrically. The three-dimensional models used by
is driven by the rapid increase in the curvature of the air–liquid
White and Heil (2005) and Hazel and Heil (2005) did not include
interface which is located at a significant distance from the airway
any representation of the parenchymal tethering and assumed the
walls.
airway wall to be a thin-walled shell structure of uniform thickness.
The prediction of a three-lobed collapse pattern, therefore, only
applies to airways with weak external support and relatively thin
4.5. The composition of the liquid lining
walls. Observations of occluded airways tend to show collapse with
much larger azimuthal wavenumbers, consistent with theoretical
The studies reviewed here treated the liquid lining as a thin
investigations of the effects of external tethering (Wang et al., 1983)
layer of homogeneous, Newtonian fluid. The exact composition and
and the complex multi-layer structure of an actual airway wall
distribution of the liquid lining is still a matter of some debate
(Wiggs et al., 1997). In addition, the buckled configurations of actual
(Dorrington and Young, 2001). It is generally accepted that the
airways are not perfectly symmetric—a consequence of inevitable
lining consists of an aqueous phase of relatively low viscosity imme-
spatial non-uniformities in the parenchyma and the airway wall
diately adjacent to the airway wall and a surfactant layer at the
which may be exacerbated by diseases such as emphysema
air–liquid interface (Geiser and Bastian, 2003). In the conducting
and asthma.
airways, an upper mucus layer is present, but does not normally
extend to the small airways in healthy lungs (Williams et al., 2006).
4.2. Gravity Moreover, the active transport of fluid by beating cilia in the con-
ducting airways, discussed in the article by Blake et al. (2008) in this
If gravity acts in a direction transverse to the airway’s centreline, special issue, leads to a totally different fluid-mechanical behaviour
fluid drains towards the “bottom” of the airway and the film thins in these regions.
(or even ruptures) at the “top”, resulting in a strongly non-uniform The aqueous layer is continuous in the central airways and
distribution of fluid around the airway circumference. The sur- the combined height of the layer and the underlying non-uniform
face tension forces that drive the formation of an occluding liquid topography, caused by variations in thickness of the epithelial
bridge must therefore compete with the gravitational forces that cells, is approximately constant (Geiser and Bastian, 2003). In
drive the fluid towards the “bottom” of the airway. Estimates of the vitro experiments indicate that the epithelial cells can absorb
Bond number (a measure of the relative importance of gravitational and secrete liquid in order to control the thickness of the layer
and surface-tension forces) by Kamm and Schroter (1989) suggest (Boucher, 2003). This active control of the layer thickness pre-
that gravity has only a minor effect on the liquid film dynamics sumably explains why airways do not exhibit the “dry regions”
in the small airways and Dutrieue et al. (2005) found that airway predicted by theoretical models of the system’s equilibrium con-
closure in healthy subjects (where airway closure only occurs in figurations (e.g. Fig. 4), except in diseases such as cystic fibrosis
these airways) is not affected by microgravity conditions, justify- where airway fluid regulation is defective (Boucher, 2007). The
ing the neglect of gravity in the studies presented here. In contrast, effect of such active control during airway collapse does not
Duclaux et al. (2006) found that at sufficiently large values of the appear to have been investigated theoretically. A particularly inter-
Bond number the Rayleigh–Plateau instability can be suppressed esting set of questions concerns the relative timescales of the
in rigid tubes, which suggests that in the larger airways trans- control mechanism and collapse process. Can the control mech-
verse gravity may suppress airway closure by the “film collapse” anism act over sufficiently fast timescales to suppress airway
mechanism. closure?
 M. Heil et al. / Respiratory Physiology & Neurobiology 163 (2008) 214–221 221

4.6. Surfactant Cassidy, K.J., Halpern, D., Ressler, B.G., Grotberg, J.B., 1999. Surfactant effects in model
airway closure experiments. Journal of Applied Physiology 87, 415–427.
Dorrington, K.L., Young, J.D., 2001. Editorial III: Development of the concept of a liquid
Airway closure is driven by surface tension and so the pres- pulmonary alveolar lining layer. British Journal of Anaesthesia 85, 614–617.
ence of surfactant has a strong effect on the dynamics of the Duclaux, V., Clanet, C., Quéré, D., 2006. The effects of gravity on the capillary insta-
process. An increase in surfactant concentration primarily leads to bility in tubes. Journal of Fluid Mechanics 556, 217–226.
Dutrieue, B., Verbanck, S., Darquenne, C., Prisk, G.K., 2005. Airway closure in micro-
a net reduction in surface tension which increases the timescale gravity. Respiratory Physiology & Neurobiology 148, 97–111.
for the development of the initial Rayleigh–Plateau instability (cf. Espinosa, F.F., Kamm, R.D., 1998. Meniscus formation during tracheal instillation of
Eq. (3)), and reduces the surface-tension-induced compression of surfactant. Journal of Applied Physiology 85, 266–272.
Everett, D.H., Haynes, J.M., 1972. Model studies of capillary condensation. I. Cylindri-
the airway wall (cf. Eq. (2)), thus reducing the airway’s propen- cal pore model with zero contact angle. Journal of Colloid and Interface Science
sity to collapse. Additionally, surfactant is advected by the fluid 28, 125–137.
flow, leading to a higher surfactant concentration and, therefore, Gauglitz, G.A., Radke, C.J., 1988. An extended evolution equation for liquid film
breakup in cylindrical capillaries. Chemical Engineering Science 43, 1457–1465.
a lower surface tension in the regions where the film thickness
Geiser, M., Bastian, S., 2003. Surface-lining layer of airways in cystic fibrosis mice.
increases. This reduces the growth rate of the Rayleigh–Plateau American Journal of Physiology: Lung Cellular and Molecular Physiology 285,
instability yet further but cannot suppress it altogether (Campana L1277–L1285.
et al., 2004; Campana and Saita, 2006; Cassidy et al., 1999; Halpern Goren, S.L., 1962. The instability of an annular thread of fluid. Journal of Fluid
Mechanics 12, 309–319.
and Grotberg, 1993; Otis et al., 1993). Grotberg, J.B., Fujioka, H., Halpern, D., 2008. Liquid and surfactant delivery into
pulmonary airways. Respiratory Physiology & Neurobiology.
4.7. Respiratory flow and wall motion Halpern, D., Grotberg, J.B., 1992. Fluid-elastic instabilities of liquid-lined flexible
tubes. Journal of Fluid Mechanics 244, 615–632.
Halpern, D., Grotberg, J.B., 1993. Surfactant effects on fluid-elastic instabilities of
Estimates of the timescales required for the formation of occlud- liquid-lined flexible tubes: A model of airway closure. ASME Journal of Biome-
ing liquid bridges by the mechanisms discussed above indicate chanical Engineering 115, 271–277.
Halpern, D., Grotberg, J.B., 2003. Nonlinear saturation of the Rayleigh instability
that the process is sufficiently rapid to be largely unaffected by in a liquid-lined tube due to oscillatory flow. Journal of Fluid Mechanics 492,
the respiratory motion of the airway walls; see also Otis et al. 251–270.
(1993). However, a recent study by Halpern and Grotberg (2003) Hammond, P.S., 1983. Non-linear adjustment of a thin annular film of viscous fluid
surrounding a thread of another within a circular cylindrical pipe. Journal of
shows that, at least in certain parameter regimes, the periodic shear Fluid Mechanics 137, 363–384.
stresses generated by the respiratory flows may interact with the Hazel, A.L., Heil, M., 2005. Surface-tension-induced buckling of liquid-lined elastic
Rayleigh–Plateau instability and stabilise the air–liquid interface. tubes—a model for pulmonary airway closure. Proceedings of the Royal Society
461, 1847–1868.
Heil, M., 1999a. Minimal liquid bridges in non-axisymmetrically buckled elastic
5. Summary tubes. Journal of Fluid Mechanics 380, 309–337.
Heil, M., 1999b. Airway closure: liquid bridges in strongly buckled elastic tubes.
ASME Journal of Biomechanical Engineering 121, 487–493.
We have described the mechanics of airway closure—the for-
Heil, M., White, J.P., 2002. Airway closure: Surface-tension-driven non-axisymmetric
mation of occluding liquid bridges via surface-tension-driven instabilities of liquid-lined elastic rings. Journal of Fluid Mechanics 462, 79–109.
instabilities of the lung’s liquid lining. Airway closure may occur as Hill, M.J., Wilson, T.A., Lambert, R.K., 1997. Effect of surface tension and intraluminal
a consequence of a purely fluid-mechanical “film collapse”, see Sec- fluid on mechanics of small airways. Journal of Applied Physiology 82, 233–239.
Hughes, J.M.B., Rosenzweig, D.Y., Kivitz, P.B., 1970. Site of airway closure in excised
tion 2, or alternatively a coupled fluid-elastic “compliant collapse”, dog lungs: Histologic demonstration. Journal of Applied Physiology 29, 340–344.
see Section 3. Although in principle these mechanisms operate Jensen, O.E., 1997. The thin liquid lining of a weakly curved cylindrical tube. Journal
throughout the lung, they are most likely to occur in the small air- of Fluid Mechanics 331, 373–403.
Johnson, M., Kamm, R.D., Ho, L.W., Shapiro, A., Pedley, T.J., 1991. The non-linear
ways where the curvature of the air–liquid interface (and hence growth of surface-tension-driven instabilities of a thin annular film. Journal of
the surface-tension-induced pressure change between the air in Fluid Mechanics 233, 141–156.
the lumen and the liquid lining the walls) is highest. Theoretical Kamm, R.D., Schroter, R.C., 1989. Is airway closure caused by a liquid film instability?
Respiration Physiology 75, 141–156.
models predict that both mechanisms can lead to airway closure in Otis, D.R., Johnson, M., Pedley, T.J., Kamm, R.D., 1993. Role of pulmonary surfac-
times comparable with the breathing cycle, suggesting that surface tant in airway closure: a computational study. Journal of Applied Physiology 75,
tension is indeed the primary mechanical effect responsible for the 1323–1333.
Plateau, J., 1873. Statique expérimentale et théorique des liquides soumis aux seules
observed closure in peripheral regions of the human lungs. forces moéculaires. Volume Tome Second Paris: Gauthier–Villars.
Rayleigh, L., 1879. On the capillary phenomena of jets. Proceedings of the Royal
References Society of London 29, 71–97.
Rosenzweig, J., Jensen, O.E., 2001. Capillary-elastic instabilities of liquid-lined air-
ways. ASME Journal of Biomechanical Engineering 124, 650–655.
Appelberg, J., Pavlenko, T., Bergman, H., Rothen, H.U., Hedenstierna, G., 2007. Lung
Rothen, H.U., Sporre, B., Engberg, G., Wegenius, G., Hedenstierna, G., 1998. Airway
aeration during sleep. Chest 131, 122–129.
closure, atelectasis and gas exchange during general anaesthesia. British Journal
Bertram, C.D., 2008. Oscillating flow in collapsed tubes. Respiratory Physiology &
of Anaesthesia 81, 681–686.
Neurobiology 163, 256–265.
Wang, C.Y., Watson, L.T., Kamat, M.P., 1983. Buckling, postbuckling and the flow
Blake, J.R., Smith, D.J., Gaffney, E., 2008. Modelling muco-ciliary transport. Respira-
through a tethered elastic cylinder under external pressure. ASME Journal of
tory Physiology & Neurobiology 163, 178–188.
Applied Mechanics 50, 13–18.
Boucher, R.C., 2003. Regulation of airway surface liquid volume by human airway
White, J.P., 2003. Airway closure: Surface-tension-driven non-axisymmetric insta-
epithelia. Pflügers Archiv. European Journal of Physiology 445, 495–498.
bilities of liquid-lined elastic tubes. PhD thesis. University of Manchester.
Boucher, R.C., 2007. Cystic Fibrosis: A disease of vulnerability to airway surface
White, J.P., Heil, M., 2005. Three-dimensional instabilities of liquid-lined elastic
dehydration. Trends in Molecular Medicine 13, 231–240.
tubes—A lubrication theory model. Physics of Fluids 17, 031506.
Burger, E.J., Macklem, P., 1968. Airway closure: Demonstration by breathing 100%
Wiggs, B.R., Hrousis, C.A., Drazen, J.M., Kamm, R.D., 1997. On the mechanism of
O2 at low lung volumes and by N2 washout. Journal of Applied Physiology 25,
mucosal folding in normal and asthmatic airways. Journal of Applied Physiology
139–148.
83, 1814–1821.
Campana, D.M., Di Paolo, J., Saita, F.A., 2004. A 2-D model of Rayleigh instability in
Williams, O.W., Sharafkhaneh, A., Kim, V., Dickey, B.F., Evans, C.M., 2006. Airway
capillary tubes—Surfactant effects. International Journal of Multiphase Flow 30,
mucus: From production to secretion. American Journal of Respiratory Cell and
431–454.
Molecular Biology 34, 527–536.
Campana, D.M., Saita, F.A., 2006. Numerical analysis of the Rayleigh instability
Yamaki, N., 1984. Elastic stability of circular cylindrical shells. North-Holland, Ams-
in capillary tubes: The influence of surfactant solubility. Physics of Fluids 18,
terdam.
022104.