DECO THEORY - BUBBLE MODELS

This page describes principles and theories about bubble generation and bubble growth in the scuba divers body and
about the effect of bubble formation on decompression and decompression sickness (DCS, bends) in scuba diving.
Whereas classical (neo-)Haldane theories are mainly empirical and only take dissolved gas into account, bubble
theories intend to give a physical explanation of the effects of bubbles on decompression. Bubble theories take
dissolved and free gas into account. Especially the Varying Permeability Model (VPM) and Reduced Gradient
Bubble Model (RGBM) give good explanation.

History

In classic decompression theory according to Haldane and successors a certain amount of supersaturation of the
divers tissue with dissolved inert gas is allowed. The divers tissue is divided in a number of hypothetical tissue
compartments. A certain limit (M-value) is associated with each compartment to supersaturation levels of
dissolved inert gas in the compartment (tissue tension). This theory suggests efficient decompression by pulling
the diver as close to the surface as possible with constraint that in all tissue compartments the supersaturated
tissue tension remains within the limits. By pulling the diver as close to the surface the pressure gradient
between the supersaturated tissue tension and the pulmonary (or arterial) gas is maximized. This enhances the
elimination of the excess gas in the tissue. This theory is mainly empirical and based on experiment. At the
moment most diving tables and computers are based on this theory.
Since the early days, diving has become more sophisticated by diving deeper and longer, the use of other
breathing mixtures, etc. Some tech divers have made their own adaptations to the decompression schedules
by inserting depression stops at greater depth (’deep stops’, sometimes called ’Pyle stops’ after Richard Pyle).
These divers report feeling better when using these deep stops. This suggests that classic decompression theory
fails in some situations and cannot be extrapolated to every diving situation. In order to gain insight in
the principles of decompression, forming of bubbles during decompression has been studied for the last three
decades. This has resulted in new theories like the Varying Permeability Model (VPM) by Yount et al. and
the Reduced Gradient Bubble Model (RGBM). Bubble theories do not only take into account the dissolved gas
(like the Haldane models), but also the free gas in the divers body. In this chapter we will have a look at some
features of bubble theory. Lots of mathematics will be presented. The most important equations however, will
be highlighted.

Bubbles and surface tension

Consider a small air bubble in a glass of water. For the moment we neglect the solubility of the air in water. The
small amount of air within the bubble is surrounded by a surface. The surface consists of water molecules which
are unbound to one side. An unbound molecule represents more energy than a molecule which is completely
surrounded by other water molecules. A surface tension γ is associated with this surface between air and water.
The surface tension is the amount of energy per unit of surface area and is expressed in J/m2 or N/m.

1
Figure 1: In equilibrium the internal pressure in the bubble is equal to the sum of the ambient pressure and the
skin pressure due to the surface tension

A system will always try to minimize energy. Surface tension tends to minimise the bubble’s surface. Hence, a
bubble tends to collapse. However, collapsing a bubble decreases its volume. This will increase the gas pressure
in the bubble (Boyle’s law), until equilibrium is established: the internal pressure compensates the surface
tension. The internal pressure due to the ambient pressure and surface tension is given by the Laplace equation:

2γ
Pin = Pamb + Psurf = Pamb + (1)
r
r Radius of the bubble in m
γ Surface tension in joule/m2 of N/m. The surface tension of water at 273 K is 0.073 N/m.
Pin Pressure inside the bubble in N/m2 =10−5 bar
Pamb Ambient pressure in N/m2 =10−5 bar
Psurf Pressure due to the surface tension in N/m2 =10−5 bar
From this equation we learn that the smaller the bubble, the higher the pressure inside. You can experience the
radius dependency of the pressure by trying to blow a balloon (bubble principles perfectly apply to a balloon
up to the point where the balloon explodes). To get the first blow of air into the balloon (small radius) is a hell
of a job, whereas it becomes easier if the balloon becomes larger.

Bubbles and diffusion

When we have a bottle of beer things get a bit more complicated (usually the opposite holds, but when we
look at the bubbles it might be). Bubbles in beer contain Carbon Dioxide. There is also Carbon Dioxide in
solution in the beer. Carbon Dioxide can diffuse from the solution into the bubble or vice versa, depending on
the partial pressure of the Carbon Dioxide in solution and in the bubble. If we assume that the bubble consist
of only Carbon Dioxide, the Carbon Dioxide pressure in the bubble is given by equation 1 and depends on the
radius of the bubble.
We define the partial pressure of the Carbon Dioxide in solution in the beer to be Pt (if we regard the bottle of
beer as a primitive model for a diver, we could call it ’tissue tension’). If the bottle is closed, the partial pressure
of the Carbon Dioxide in solution Pt is in equilibrium with the ambient pressure Pamb , which is the pressure in
the bottle. If we assume there is only Carbon Dioxide gas in the (closed) beer bottle, the beer is saturated with
Carbon Dioxide and Pt will be equal to Pamb (we can neglect hydrostatic pressure). The pressure in the bubble
Pin will be higher than Pt due to the surface tension. Gas from within the bubble will diffuse into solution and
the bubble will collapse. So every bubble will collapse eventually due to this gradient Pin − Pt . This is why in
a closed bottle of beer there are no bubbles and there is no foam. However, if we open the bottle things will be
different. The ambient pressure will drop, whereas the value of Pt remains the same, at least for the moment.
In this case Pt is larger than Pamb : the beer is supersaturated with Carbon Dioxide.

2
Given an ambient pressure Pamb and the partial pressure Pt of the Carbon Dioxide in solution, there is a critical
bubble radius rmin at which the pressure inside the bubble Pi n equals Pt . The critical radius can be found by
substituting Pin by Pt in equation 1:

2γ
rmin = (2)
Pt − Pamb

For bubbles which size exceeds this critical size the pressure Pin in the bubble is smaller than the partial pressure
Pt of the Carbon Dioxide in solution. Carbon Dioxide will diffuse from solution into the bubble. The bubble will
grow. For bubbles smaller than the critical size, the opposite holds: gas from the bubble diffuses into solution
and the bubble shrinks until it collapses completely. Bubbles at the critical size are in equilibrium, though it is
an unstable equilibrium. This is depicted in Fig. 2.

Figure 2: Bubble growth dependens on the radius

So every bubble with a radius larger than rmin will start to grow. When we look at our opened bottle of beer
we see bubbles becoming visible and heading for the surface, where they form foam. If you scrutinize a bubble
you’ll see that it grows during ascent. Its diameter might have doubled or tripled when it arrives at the surface.
You might think this is due to Boyle’s law. However it takes an ascent of several meters for a bubble to double
its diameter. The growth of the bubble is due to the diffusion described above.
As an example, we can calculate critical radii for Spa Barisart Soda (6.4-8.0 g/l Carbon Dioxide). The pressure
in the bottle specified by Spa is shown in next table (dependent on temperature). The partial pressure Pt of
the Carbon Dioxide in solution is roughly that value. If we open the bottle the ambient pressure Pamb drops to
1 bar, whereas the partial pressure Pt remains at the high value. Using equation 2 we can calculate the critical
radius rmin .

3
 Temperature (◦ C) Pressure (bar = 105 P a) rmin (µm)
15 3 0.73
20 3.75 0.53
25 4.5 0.42
30 5.3 0.34
35 6 0.29
40 7 0.24

The Varying Permeability Model

According to previous chapter, in a supersaturated situation any bubble exceeding a critical size rmin will grow
(and will disappear by floating to the surface) and any bubble smaller than this size will collapse. In a normal
non-supersaturated situation, rmin approaches infinity. Any bubble will collapse. So we do not expect any
bubbles around after a while. You might expect that if no initial bubbles are around, there is no bubble to
grow on supersaturating the liquid. The tensile strength of water is estimated on 1000 atm, making immense
supersaturations possible, before bubbles (voids) are created. If no initial bubbles would be present in the water
making up the diver, a diver could easily dive to a kilometer depth and pop up to the surface without any
problems. In practice, this is not the case. Bubbles form on modest decompression as low as 1 atm. Here comes
in the Varying Permeability Model (VPM). The VPM was initially defined by Yount et al. [2] in order to give
a quantitative explanation on the formation of bubbles in decompressed gelatin [1] (as model for divers tissue).
Later on, they showed this model can be used to calculate dive tables as well [3], [4]. In next paragraphs we
will have a look at the gelatin theory. Later on we will apply the theory to diving.

Figure 3: Skins of varying permeability are the base of the VPM

The gelatin experiments

Experiments on gelatin have been performed, by David Yount and other researchers [1]. The advantage of
gelatin over water is that any bubble appearing during decompression gets trapped and won’t flow to the
surface. In this way they can be observed and counted. Yount applied the rudimentary pressure of Figure 4 to
gelatin samples: Gelatin samples were made at ambient pressure Pamb=P0 of 1 atm. The samples were rapidly
compressed in a 100

4
 Figure 4: Pressure schedule applied to the gelatin samples by Yount

According to the VPM, in aqeous media like water and gelatin stable gaseous cavities are present. They are
called nuclei. Radii range from a few 1/100 m up to around 1 m. Any nucleus in water larger will flow to the
surface and disappear. Whereas an ordinary bubble with these radii would collapse under normal conditions
(no supersaturation), these nuclei appear to be exceptionally stable and have a long life. Yount proposed this
stability is due to an elastic skin made up of surfactant, as shown schematically in Figure 3. Surfactant consists
of (hydrophobic) surface active molecules, which are aligned. During the compression stage, these skins are
permeable for gas up to a pressure of around 8 atm. Diffusion through the skin takes place. The pressure Pin
of the gas in the nucleus is equal to the dissolved gas tension Pt in the surrounding liquid. Above this pressure,
the skin becomes impermeable. Upon decompressing (reducing the ambient pressure) the skins are regarded
permeable. The skin gives rise to a ’surface compression’ Γ which opposes the regular surface tension γ of the
water/air surface, as shown in Figure 5:

2Γ 2γ
Pin + = Pamb + (3)
r r

Figure 5: Pressures acting on the surface of the bubble

The skin tension Γ is not constant but ranges from 0 to a maximum γc , which is called the ’crumbling com-
pression’. The idea is that small variations of the size of the nucleus can be supported by varying the distance
between the molecules in the skin. This gives rise to varying Γ. This situation is described by equation 3 and
is referred to as the small-scale situation. In this equilibrium situation and in the permeable region, due to

5
diffusion the internal pressure Pi n is equal to the tension Pt . In the samples (no hydrostatic pressure, 100%
Nitrogen) Pt equals Pamb. So Pin = Pt = Pamb . In this situation Γ equals γ, according to equation 3.
Upon compressing and decompressing, variation of the size of the nucleus becomes to large to be supported
by varying distances between molecules. Surfactant molecules have to be expelled from or taken up into the
skin in order to compensate for the area decrease resp. increase of the nucleus. This is schematically shown
in Figure 6. The skin is surrounded by an amount of surfactant, which is not part of the skin. This amount
acts as a reservoir, taking up or supplying surfactant molecules from or to the skin. The reservoir molecules are
not aligned and cannot support a pressure gradient. Γ takes its crumbling value γc in this large-scale situation.
Yount proposes two derivations of the VPM [2]: one from a thermodynamic point of view and one from a
mechanical point of view.

Figure 6: The large-scale situation: variation in the size of the nucleus result in expelling molecules from the
skin

In the original sample there is a initial distribution of nuclei with radii distributed according to some function
f (r0 ). (The ’0’ in r0 refers to the initial situation). On applying the pressure schedule, it is assumed that all
nuclei with a radius larger than some minimal initial radius r0min will grow into bubbles. The number of bubbles
N that occur is given by the integration of f (r0 ) from r0min to infinity.
Z ∞
N= f (r0 )dr0 (4)
r0min

Applying this theory to a diver, it might be assumed that the severity of Decompression Sickness (DCS) might
be related to this number of bubbles, which occur after decompression. Hence, r0min becomes an indication
for the severity of DCS.
It is assumed that no nuclei are extinguished or created during application of the pressure schedule. Furthermore
it is assumed that the ordering of nuclei is preserved: if one nucleus is larger than an other one, this is still true
after a pressure change (ordering hypothesis). At the end of the pressure schedule there is a new distribution
of radii g(rf ) and a new radius rfmin above which all nuclei will grow into bubbles. Note: a nucleus with radius
r0min ends up as a nucleus with radius rfmin after application of the presure schedule. The aim of next VPM
calculations is
1. To define the allowed number of bubbles by defining r0min
2. To find a relation between rf and r0 (and hence between rfmin and r0min )
3. To find the relation based on rfmin which governs the bubble formation on decompression.
4. To calculate the resulting restricting relations for the pressure schedule, given the value of r0min and hence,
the number of resulting bubbles after application of the pressure schedule.

6
Thermodynamic equilibrium

From a thermodynamic point of view the left-hand side of equation 3 represents the skin pressure PS :

2γc
PS = Pin + (5)
r
Γ has been replaced by the large-scale value γc . Similarly the right hand term of equation 3 represents the
reservoir pressure PR :

2γ
PR = Pamb + (6)
r
In the large-scale situation transport of surfactant is not described by setting PR equal to PS but by the
requirement that the electrochemical potential in the skin and reservoir are equal. The electrochemical potential
ξ is given by

ξ = µ + kT ln (ρ) + pv + Ze ψ (7)
ξ Electrochemical potential
µ Pure chemical potential
k Bolzmann contant
T Absolute temperature in K
ρ Molecular concentration or number density
p Static pressure
v Active volume occupied by one surfactant molecule
Ze Effective charge of one surfactant molecule
ψ Electrostatic potential
In the reservoir we have

ξR = µR + kT ln (ρR ) + PR v + (Ze ψ)R (8)

and in the skin we have

ξS = µS + kT ln (ρS ) + PS v + (Ze ψ)S (9)

Requiring ξR is equal to ξS and substituting PS and PR by the values in equation 5 resp 6 results in:

2γc 2γ
Pin + − β = Pamb + (10)
r r
in which
 
1 ρR
β= kT ln + (µR − µS ) + (Ze ψ)R − (Ze ψ)S (11)
v ρS

Equations 10 and 11 can be used to calculate the changes in radii after applying a pressure step. We have a
look what happens when applying the pressure schedule of Figure 4 to the sample. At the beginning of the
pressure schedule Pamb = P0 . The pressure of the gas in the nucleus is

Pin = Pt0 = Pamb = P0 (12)

Just before compression equation 10 is:

2γc 2γ
P0 + − β 0 = P0 + (13)
r0 r0

After the pressure rise to the pressure P ∗ where the skins becomes impermeable equation 10 reads:

7
 2γc 2γ
P0 + − β∗ = P ∗ + ∗ (14)
r∗ r
Substracting equation 14 from 13, assuming β0 = β ∗ and rewriting it a bit result in:
 
1 1
2(γc − γ) ∗ − = P ∗ − P0 (15)
r r0

We continue to compress rapidly from P ∗ to Pm . Since the skin is not permeable now, the pressure in the
nucleus varies with its radius according to Boyle’s law (P V = P 4/3πr3 = constant):

r∗ 3
Pin = Pt∗ (16)
r3
In this equation r∗ is the radius of the nucleus at the beginning of the impermeable process. Pt∗ is the corre-
sponding dissolved gas tension, which is equal to P0 . So after the compression to Pm we have:

3
r∗

2γc 2γ
P0 + − β m = Pm + (17)
rm rm rm

Assuming βm = β ∗ = β0 , subtracting equation 17 from 14 and rewriting a bit we end up with:

" 3 #
r∗
  
1 1
2 (γc − γ) − ∗ = Pm − P ∗ + P0 1 − (18)
rm r rm

We now have relations (equation 15 and 18) between the radius rm of the nucleus after compressing and the
radius r0 prior to compression. The saturation phase (Pamb = Ps ) that follows saturates the liquid so that
finally the dissolved gas tension Pts = Ps . We might expect that the radius of the nucleus increases to its
original value (see the the note in the ’Mechanical Equilibrium’ section). However, this has not been observed.
So we assume

rs = rm (19)

In fact the radius restores quite slowly, but for the moment equation 19 holds. Prior and after decompressing
(which is fully permeable) equation 10 reads:

2γc 2γ
Ps + − β s = Ps + (20)
rs rs
resp.

2γc 2γ
Ps + − β f = Pf + (21)
rf rf

Assuming βf = βs (not equal to βm , subtracting equation 21 from 20 and rewriting a bit result in:
 
1 1
2 (γc − γ) − = Pf − Ps (22)
rf rs

So we now have a relation between all radii of the nucleus during the entire profile. A nucleus with radius r0
ends up as a nucleus with radius rf through a number of stages (r∗ , rm , rs ) defined by the relations 15, 18, 19
and 22.
We now define the criterion for bubble formation, which is given by the Laplace equation 1:

2γ
Pin − Pamb = Ps − Pf ≥ (23)
rf min

8
There is no reference to Γ or γc in this equation. We assume the skin of the nucleus to be permeable. So the
skin does not restrict bubble formation: gas simply flows through the skin and forms a gas shell outside the
skin. If the skin should not be permeable as has been proposed by others some tearing strenght or tearing
tension Γ = −γT is introduced. The bubble forming equation becomes:

2 (γ + γT )
Pin − Pamb = Ps − Pf ≥ (24)
rf min

By combining equation 15, 18, 19, 22 and 23 we find the VPM equations.
For the ever-permeable region Pm ≤ P ∗ :

γc − γ γ
Pss min = 2γ min
+ Pcrush (25)
r0 γc γc

For the permeable-impermeable-permeable situation Pm > P ∗ we find:

"  ∗ 3 #
min γc − γ r γ
Pss = 2γ min + Pm − P0 (26)
r0 γc rm γc

In these equations we have the crushing pressure

Pcrush = (Pamb − Pt )max = Pm − P0 (27)

and the supersaturation pressure

Pss min = (Pt − Pamb )max = Ps − Pf (28)

Equation 26 can be written as:

γc − γ γ γ 1
Pss min = 2γ + (P ∗ − P0 ) + ∗
∼
(Pm − P ) (29)
r0 min γc γc γc 1 + rB

The parameters r∼ and B are defined as:

r∗
r∼ = r∗ (30)
rm

2 (γc − γ)
B=   (31)
r∗ rm
P0 rm +1+ r∗

Mechanical equilibrium

The other way the VPM is derived is by looking from a mechanical point of view. Changes in nuclear radius
can be calculated by the equation proposed by Love, which reads (in the VPM form, [2]):

∂r
2(Γ − γ) = ∂Pin − ∂Pamb (32)
r2
In the ’permeable’ region of the VPM, Pin remains constant and equal to Pt . Here ∂Pin is 0. For large-scale
variations in the ’permeable’ region of VPM equation 32 reads

∂r
2(γc − γ) = ∂Pamb (33)
r2

9
In the ’impermeable’ region, Pin is given by equation 16. Differentiated it reads:

∗3
 
∗r
∂Pamb = − 3Pt ∂r (34)
r4

For the large-scale variations in the impermeable region, equation 32 reads

∗3
 
∗r ∂r
2 (γc − γ) + 3Pt 2 = ∂Pamb (35)
r r2

Together with the Laplace equation 23 and the assumption of equation 19, equation 33 and 32 can be used
(by integrating) to derive the VPM equations 25 and 26: equation 15 and 22 can be obtained by integrating
equation 33, equation 18 can be obtained by integrating equation 35. The derivation is given in [2].
Note: During the compression phase ∂Pin is zero in equation 32(24). Assuming the pressure schedule takes
place in the permeable region, integration of 32 from Pamb = P0 to Pm results in:
 
1 1
2 (γc − γ) − = Pm − P0 (36)
rm r0

During the saturation phase that follows ∂Pamb is zero in equation 32. Integration of 32 from Pi n = P0 to Ps
results in:
 
1 1
2 (γc − γ) − = P0 − Ps (37)
rs rm

Adding equation 37 to 36 results in rs = r0 , assuming Pm = Ps (in fact in the non-permeable situation or

a situation in which Pm 6= Ps we could derive the same result, though it takes some more derivation). This
suggests that the nucleus is fully restored to the original size during saturation. The effect of the crushing is
lost in this situation. However, this is in sharp disagreement with experiment. Further indication is given by
special cases in which a pressure spike is present at the start of the schedule (Figure 4), so that Pm < Ps . In
this cases the bubble count only depend on Pss and Pcrush and not on Ps ! Hence, the assumption defined by
equation 19 is made: rs = rm .

Equilibrium considerations

In this section we will consider some implications from the equilibria discussed above.
Rewriting equations 13, 14, 17, 20 and 21 gives us:

2 (γc − γ)
β0 = (38)
r0

2 (γc − γ)
β∗ = − (P ∗ − P0 ) (39)
r∗

3 !
r∗

2 (γc − γ)
βm = − Pm − P0 (40)
rm rm

2 (γc − γ)
βs = (41)
rs

2 (γc − γ)
βf = − (Pf − Ps ) (42)
rf

(We recall our assumptions that β0 = βm = β ∗ and βs = βf .) According to equation 11 β is independent of

radius. However, according to equation 38 β0 appears to be a function of r0 . Assuming that β0 is constant for
all nuclei at Pamb = P0 we obtain an remarkable prediction that γc increases with increasing r0 :

10
 r0 β0
γc = γ + (43)
2
Another consideration stems from mechanical equilibrium: Small scale equilibrium is given by equation 3:

2Γ 2γ
Pin + = Pamb + (44)
r r
All properties of the skin and the reservoir are incorporated in the small scale skin compression Γ. The equation
can be obtained from 10 by setting:

2Γ 2γc
= −β (45)
r r
Substituting the β values of equation 38 to 42 in 45 results at the respective ambient pressure values Pamb = P0 ,
P ∗ , Pm , Ps and Pf in:

Γ0 = γ (46)

γc − γ ∗
Γ∗ = γc − r (47)
r0

γc − γ
Γm = γc − rm (48)
r0

Γs = γ (49)

Γf =? (50)

A plausible small scale/mechanical equilibrium criterion for bubble formation is that Γf is less than or equal to
zero. This results in:

2 (γc − γ) 2γc
≥ (51)
rs rf

Substituting this in equation 22 results in the Laplace equation 23 as used for the thermodynamic derivation.
Equation 46 to 50 shows that during the compression Γ increases. During saturation Γ ’relaxes’ to its value
prior to compression γ, keeping rm constant. During decompression Γ drops to 0, the point at which bubble
formation just starts.

Consequences of the VPM relations

Plotting Pss vs. Pcrush

min min
Most conveniently, equation 25 is plotted as Pss vs. Pcrush . In these plots, Pss − Pcrush pairs resulting in
the same number of bubbles (and hence, the same DCS morbidity) form straight lines.

Diver vs. gelatin

The VPM originally was developed to quantitatively explain bubble formation in gelatin during decompression
[2]. The ultimate goal was to gain understanding of decompression sickness. To apply VPM to a diving situation
it first was suggested that decompression sickness (DCS) symptoms were related to the number of bubbles. Say,
severe symptoms occur at a number NDCS of bubbles in some tissue. Given the radial distribution f (r0 ),
equation 4 defines a r0min . If all nuclei with a radius equal or larger than this radius grow into bubbles, we
end up with NDCS bubbles (and some bad DCS). Given a dive to some depth resulting in an ambient gradient
min
Pcrush , equation 25 gives the maximum allowed gradient Pss resulting in the NDCS bubbles

11
The VPM relations

The VPM relations 25 and 26-28 define the maximum allowed gradient between the ambient pressure and the
tissue tension. In other words: it defines the minimum allowed ambient pressure Pamb , given the tissue tension
Pt . In a diving situation it defines the depth the diver is allowed to ascend given the tissue tension. The relation
should be applied to each tissue compartment of the diver.
The initial compression (defining Pcrush ) is important for Pssmin. During this stage nuclei are crushed to a
smaller size, making them less active in bubble formation. The secret lies in equation/assumption 19, which
states that no regeneration of the bubble size takes place during saturation. It implies that a descent during a
dive should be as quick as possible, the deepest part of the dive should be at the start of the dive and deeper
dives should precede shallower dives in a repetitive dive situation. These facts have been empirically found
during a century of decompression research.

Not a 100% Nitrogen saturation dive

The derivation of the VPM assumed 100% Nitrogen and fully saturated gelatin. If we apply the equations to
a non-saturating diving situation in which the Nitrogen fraction is less than 100% (for example air, containing
79% Nitrogen), the VPM equations 25 and 26-28 can be regarded as a conservative restriction to the dive profile.

Applying VPM to diving

In this section we will apply the VPM to a diving situation and describe a method to generate diving tables.
Whereas the VPM theory of previous sections applies to a special situation of fully saturated gelatin in a
100% Nitrogen atmosphere, situations during diving are different. The assumption that the severity of DCS
is proportional to the absolute number of bubbles leads to very safe diving tables, not covering all of the
conditions of modern diving tables and often leading to unacceptable long decompression periods. The VPM
was reformulated, as described in this section, to fit it with conventional diving tables. Conventional diving
tables were regarded as valid measurements. We will follow the derivation of Yount [3]. During this derivation
we assume only one inert gas. Later on we will place remarks on using more inert gasses (Trimix, etc). Another
assumption is the dive takes place in the ’permeable’ region of the VPM.

The reformulated VPM

The derivation of the theory below is based on a number of more or less ad hoc assumptions. The most
important assumptions concern the relationship between decompression symptoms and the amount of free gas
(bubbles) in the divers tissue:
1. There is an amount of bubbles Nsaf e which can be tolerated by the divers body, independent of all
circumstances (like tissue tension, degree of supersaturation, etc). The initial critical radius corresponding
to this number is r0min (equation 4).
2. The actual number of bubbles Nactual may be higher than Nsaf e as long as the total volume V of all free
gas always remains below a critical value Vcrit . This is called the ’critical-volume hypothesis’. A initial
radius r0new smaller than r0min is associated with this number.
3. The volume of free gas V inflates at a rate proportional to Pss (Nactual − Nsaf e ), where Pss is the saturation
Pt − Pamb .
The first of these assumption agrees with physiological studies, which state that the lungs are able to continue
functioning as a trap for venous bubbles to a certain degree. From this assumption can be deduced that the rate
at which the body can dissipate free gas by exchange in the lungs is proportional to both the supersaturation
pressure Pss and Nsaf e .
The assumption defined by equation 19 is fine tuned according to observations: the radius rm slowly regenerates
during saturation instead of remaining unchanged, as stated by equation 19. The regeneration is exponential,
governed by a regeneration time constant τR :

 tR 
rs (tR ) = rm + (r0 − rm ) 1 − e τR (52)

12
 rs Nuclear radius just prior to ascent and decompression (m)
rm Nuclear radius after compression by Pcrush (m)
r0 Nuclear radius before descent (m)
tR Regeneration period: time from start of dive up to start of ascent and decompression
(min)
R Regeneration time constant (min)
If we wait long enough the crushed nucleus will end up with its initial radius prior to compression r0 .
In contrast with other decompression models, VPM takes the effect of other gasses (water vapor, Oxygen,
Carbon Dioxide) into account in calculating the tissue tension:

Pt total = Pinert gasses + Pother gasses (53)

Pt total Tissue tension
Pinert gasses Sum of the partial pressures of the dissolved inert gasses
Pother gasses Pressure due to water vapor, Oxygen and Carbon Dioxide. Yount specifies a nearly
constant value of 102 mm Hg (corresponding to 0.136 bar) for inspired partial Oxygen
pressures up to 2 atm [5]
The supersaturation is now defined as:

Pss = Pt + Pamb (54)

The reformulated VPM now consist of the following steps:

1. Specify the parameters defining the VPM: surface tension γ, the crumbling compression γc , the minimum
initial radius r0min , the regeneration time constant τR and a composite parameter λ. The latter is related
to the critical volume Vcrit . The parameters are the same for each compartment.
2. Calculate the initial allowed supersaturation that is just sufficient to probe r0min and that results in Nsaf e
bubbles. The equation for this is:

min γ γc − γ
Pss =2 tR (55)
γc rs

In fact, this is an enhanced equation 25, taking nuclear regeneration into account. In this equation the
regenerated radius rs (tR ) is given by equation 52. Since the VPM parameters are the same for each tissue
compartments, this initial allowed supersaturation gradient will be the same for each compartment.
min
3. Calculate a decompression profile, using this Pss . The total decompression time defined by the profile
is tD .
new
4. Calculate a new allowed supersaturation gradient Pss using:

new 1h
1/2 i
Pss = b + b2 − 4c (56)
2

where

min λγ
b = Pss + (57)
γc tD + k1

 2
γ λPcrush
c= (58)
γc tD + k1

In these equation is k = ln(2)/τ , where τ is the half-time of the tissue compartment. This result in a larger
new
allowed supersaturation gradient Pss . Of course, this step is repeated for each tissue compartment.
new min
5. Perform a number of iteration of step 3-4, until tD and Pss converge. Of course, occurrences of Pss
new
are now substituted by Pss .

13
 new
After the iterations we end up with a more severe decompression profile and a Pss corresponding to a new
new min
initial critical radius r0 , which is smaller than r0 . This new radius results in a larger number of bubbles
Nactual and a maximum volume of free gas approaching Vcrit .

More inert gasses

In some (tech) diving situations, other gas mixtures are used consisting of more than one inert gas (for example
Trimix, containing Oxygen, Nitrogen and Helium). In next remarks we assume Helium and Nitrogen to be the
inert gases.
1. For each gas, the VPM parameters should be specified. For each tissue compartment a half-time for each
gas should be specified.
2. For each gas, the allowed supersaturation gradient should be calculated using the method in previous
section. In this case the supersaturation gradient for Helium is Pss He and for Nitrogen is Pss N
3. If Pt He and Pt N are the Helium and Nitrogen tissue tensions, the total tissue tension is given by:

Pt total = Pt He + Pt N + Pother gasses (59)

4. The allowed supersaturation gradient is given by the weighted average:

Pt He Pss He+Pt N Pss N
Pss total = Pt total − Pamb = (60)
Pt He + Pt N

Derivation

min
In this section we will derive the new VPM equations 56-58. The allowed supersaturation gradient Pss as
given by equation 25, 26 and 55 can be applied to diving as a safe-ascent criterion. Whereas they can be derived
new
directly from VPM, the derivation of Pss in equation 56-58 involves a number of ad hoc assumptions.
Assumption 1: The total volume of free gas in the divers body should never exceed a critical volume value
Vcrit at any time t (not during the dive, nor thereafter).
Assumption 2: The rate at which the free gas inflates is proportional to Pss (t)(Nactual − Nsaf e ). In this
equation P ss(t) = Pt (t) − Pamb (t).
Assumption 1 and 2 result in the decompression criterion:

Z t
Pss (t) (Nactual − Nsaf e ) dt ≤ αVcrit (61)
0

In this equation is α a proportionality constant. This criterion should hold for any t. To minimise the decom-
pression time tD , the ≤ sign is replaced by the = sign.
Assumption 3: The actual number of bubbles Nactual and the number of bubbles always allowed Nsaf e are
determined by the initial decompression stop and remain constant thereafter. The decompression criterion now
reads:

Z tmax
αVcrit = (Nactual − Nsaf e ) Pss (t)dt (62)
0

In this equation tmax is the value of t at which the integral reaches the maximum value.
new
Assumption 4: The decompression profile is chosen so that Pss (t) remains at constant value Pss during the
ascent period tD and decays exponentially to zero thereafter (at the surface). This is in agreement with As-
new
sumption 3: Pss (t) is always positive and never exceeds its initial value Pss . This initial value is the maximum
value defining Nactual . The latter remains constant thereafter. The exponential decay to zero is a conserva-
tive approximation: according to Yount&Lally [5] humans are ’inherently unsaturated’ when equilibrated at
atmospheric pressure by about 54 mm Hg (0.072 bar). Eventually, Pss (t) will become negative by this amount.
Due to Assumption 4 and the exponential decay, the integral of equation 62 reaches it maximum value in the
limit as tmax approaches ∞. The criterion for decompression now becomes:

14
 Z tD Z ∞
new new −k(t−tD )
αVcrit = (Nactual − Nsaf e ) Pss (t)dt + Pss e (63)
0 tD

In this equation is k = ln(2)/τ where τ is the tissue compartment halftime.

Assumption 5: The distribution of nuclei in humans is not known. An decaying exponential relation is
assumed, observed in vitro:

β0 Sr new
0
Nactual = N0 e− 2kT (64)

β0 Sr min
0
Nsaf e = N0 e− 2kT (65)
β0 VPN Constant
N0 Normalisation constant
S Constant area, occupied by one surfactant molecule in situ
k Bolzmann contant
T Absolute temperature in K
The decompression criterion can be rewritten:

new αVcrit
Pss = (66)
(Nactual − Nsaf e ) tD + k1

where
 β Srnew β0 Sr min

0 0 0
− 2kT − 2kT
(Nactual − Nsaf e ) = N0 e −e (67)

Assumption 6: The exponential arguments in equation are small enough so that they can be expanded.
According to [3] this approximation is in some question, since the model parameters are not fixed nor well
known. The true distribution is unknown. According to this assumption equation 67 becomes:

N0 β0 Sr0min r0new
 
(Nactual − Nsaf e ) ≈ 1− (68)
2kT r0min

Substituting 68 in 66 and rewriting this a bit results in:

  
N0 β0 S min 1 1 1 new 1
r − min tD + Pss − αVcrit new = 0 (69)
2kT 0 r0new r0 k r0

The radii r0new and r0min can now be replaced using the VPM equations (rewriting 38-42 and 60):

γc − γ
r0min = 2 (70)
β0
 
1
min
γc Pss − Pcrush γγc
= (71)
r0min 2γ (γc − γ)
 
1
new
γc Pss − Pcrush γγc
= (72)
r0new 2γ (γc − γ)

Pcrush is here Pm − P0 . The equations 70-72 apply to the permeable region. Applying them to the impermeable
−1
region results in an acceptable error of only 3% for values of Pcrush below 10 bar. Substituting r0min , r0min
new −1
and r0 in the relation 69 by the equations given in 70-72 results in:
   
N0 (γc − γ) S new min

new 1 new γ
Pss − Pss Pss tD + − αVcrit Pss − Pcrush ≈0 (73)
kT k γc

15
Rewriting this leads to the quadratic equation:

new 2 new
aPss − bPss +c=0 (74)

a=1 (75)

min λγ
b = Pss + (76)
γc tD + k1

 2
γ λPcrush
c= (77)
γc tD + k1

Equation 56-58 is the solution of equation 74-77, where:

αVcrit γc kT
λ= (78)
γN0 (γc − γ) S

Parameter values

The Yount article [3] reports the following parameter values:

Parameter Value
γ 17.9 dyn/cm = 0.179 N/m
γc 257 dyn/cm = 2.57 N/m
τR 20160 min
r0min 80 µm
λ 7500 fsw min = 250 bar min

Some adaptation to VPM

In the Yount/Maiken/Baker article [4] VPM is applied to reverse dive profiles (Pss < Pcrush ). Some adaptations
are made to the VPM.
First, the descent is not assumed to be instantaneous but takes place at a certain rate. During descent gas is
loaded into the tissue compartments, leading to a smaller Pcrush value than on instantaneous descent (no gas
loading). This effects the faster tissues more than the slower ones. Using the Schreiner equation one can derive
a new, more general version of equation 55 for compartment j:

new 2γ (γc − γ) γ
Pss j = min
+ ∆j (79)
γc r0 j γc

The effects of nuclear regeneration have not been taken into account in this equation. In this equation the set
of effective crushing pressures ∆j is given by

QN 2 Rc
1 − e−kj tc


∆j = Pcrush (1 − QN 2 ) + (80)
kj

In this equation is QN 2 the Nitrogen fraction in the breathing gas mixture and Rc is the crushing change rate of
the partial Nitrogen pressure. In the case of rapid descent, where in the limit tc approaches 0, Rc tc approaches
Pcrush and ∆j goes to P crush. This results in the original equation 25.
In the Yount/Maiken/Baker article [4] equations 56-58 has been replaced by:

new 1h
1/2 i
Pss = b + b2 − 4c (81)
2
where

16


min λγ Ptjdive − Pm tD
b = Pss j +
 −   (82)
1
γc tD + kj 2 tD + k1j

2 min dive


Pss j Ptj − Pm tD

γ λPcrush
c= − (83)
tD + k1j
 
γc 2 tD + 1 kj

In this equation Ptjdive denotes the set of compartment tissue tensions. The last terms have been added to b
and c, compared to equations 56-58. These terms become zero for saturated, not-metabolizing systems, where
Ptjdive is Pt ≈ Pm .

The Reduced Gradient Bubble Model

Bruce Wienke extended the VPM by incorporating repetitive diving, including multi day diving [6]. This
resulted in the Reduced Gradient Bubble Model (RGBM). In this section we follow the derivation given in [6].
Following equation 63, the critical volume hypothesis states for J repetitive dives becomes:

J
X Z tD Z ts j 
new new −kt
(Nactual − Nsaf e ) Pss dt + Pss e dt < αVcrit (84)
j=1 0 0

J
X  Z ts j 
new new −kt
(Nactual − Nsaf e ) Pss tD + Pss e dt < αVcrit (85)
j=1 0

In this equation tsj is the time of the surface interval after the j th dive. In [6] Wienke uses G (of Gradient)
new
instead of Pss . Hence, we followed this ’Yount notation’ so far, we keep up doing so. After the last dive no
more dives are made. Hence tsJ goes to infinity. Rewriting equation 84 (introducing ∆N = Nactual − Nsaf e )
results in:

J     
X 1 1 1
new
∆N Pss tD + − e−ktsj new
+ ∆N Pss tDJ + < αVcrit (86)
j=1
k k k

We now define Gj :
For j = 1:

new
∆N G1 = ∆N Pss (87)

For j = 2..J:
   
1 new 1 new 1 −ktD(j−1)
∆N Gj tDj + = ∆N Pss tDj + − ∆N Pss e (88)
k k k

Using equations 87 and 88, equation 86 now reads:

J   
X 1
∆Gj tDj + < αVcrit (89)
j=1
k

One important property of G is:

new
Gj ≤ Pss (90)

17
Comparing this result to equation 63, equation 90 treats the dives as if they were independent dives. However,
to account for the influence of the previous dives, reduced gradients are used for subsequent dives. The reduced
gradients can be written as:

new
Gj = ξj Pss (91)

with

0 ≤ ξj ≤ 1 (92)

We will look at the factors that influence ξ. However, [6] only shows the resulting equations, not the derivation.

Regeneration

During the surface intervals bubble sizes regenerate.

∆N tcum
j−1 cum
ηjreg = = e−τR tj−1 (93)
∆N

Reverse diving profiles

ηjexc = (94)

Repetitive dives

ηjrep = (95)

Gmin
   
(∆N )max cum
ξj = ηjreg ηjexc ηjrep = 1 − 1 − new e −τM tj−1
e−τR tj−1 (96)
(∆N )j Pss

References
[1] David E. Yount, Richard H. Strauss, ’Bubble formation in gelatin: A model for decompression sickness’,
Journal of Applied Physics, Vol. 47, No. 11, November 1976, p5081-5089
[2] David E. Yount, ’Skins of varying permeability: a stabilization mechanism for gas cavitation nuclei’, Journal
of the Acoustical Society of America 65 (6), june 1979, p1429-1439
[3] David E. Yount, D.C. Hoffman, ’On the use of a bubble formation model to calculate diving tables’, Aviation,
Space and environmental medicine, feb. 1986, p149-156
[4] David E. Yount, Eric B. Maiken, Erik C. Baker, ’Implications of the Varying Permeability Model for Reverse
Dive Profiles’, presented at the Reverse Dive Profiles Workshop October 29 and 30, 1999 Smithsonian
Institution, Washington DC. See http:www.phys.hawaii.edu/ dey .
[5] David E. Yount, D. A. Lally, ’On the use of oxygen to facilitate decompression’, Aviation, Space and
environmental medicine, 1980, 51 page 544-550.
[6] Bruce R. Wienke, ’Abyss/reduced gradient bubble model: algorithm, bases, reductions, and coupling to
ZHL critical parameters’.

18