EMS School on Industrial Mathematics

Bedlewo, 11–18 October, 2010

Project: Problem with air bubbles in sausage production process

Report

Joachim Linn, Włodzimierz Bak,

˛ Jacek Domagała, Krystian Kazaniecki, Michał Kawiak, Witold
Kosiński, Andrzej Palczewski, Bogna Pawłowska
Executive Summary
Having considered various possible reasons for bubbles appearance, we chose the most probable hy-
pothesis and investigated it in-depth. We suppose that air bubbles appear at the sausage skin due to
high pressure exerted by intestine on the machine’s steel tube. There is always some air leftover in
the harmonica, but when the pressure is high, it cannot escape. There are three main factors for high
pressure: small radius of the intestine used, its little elasticity and natural irregularities of its surface.
We have provided a quantitative mathematical model to describe pressure as a function of several
adjustable factors.
Introduction
The Objectives
The company producing natural sausages would like to rise the profit of its production by increas-
ing the quality of its product. In this process they have discovered that in a small fraction of their
product air bubbles appear just under the sausages skin. The objective of the project is to find mecha-
nism of bubble formation and propose an improvement which the appearance of bubbles can prevent
altogether.
Due to the company secrets the information available is very limited. Fortunately additional data
can be collected from the sausage filling machine manufacturer (Handtmann) and by performing some
approximate calculations. It should be emphasized that the appearance of air bubbles has a random
character and only about 1% of production has visible bubbles.

The Goals
• recognize the mechanism which produces air bubbles,

• give a quantitative solutions of bubbles formation,

• propose the improvement of sausages quality without increasing the cost of production,

• propose further research on that problem.

Description of the process

In the sausage filling process a vacuum filling machine produced by Handtmann is used (a visuali-
sation of the production process can be seen on a video provided on the Handtmann web page [1]).
Handtmann is a worldwide supplier to the meat processing trade and meat industry. Handtmann man-
ufactures filling and portioning machines for all sizes of business. It offers vacuum filling machine
from small and medium-scale operations to industrial machines. As Handtmann claims ”Product
quality is always first class, irrespective of whether the product is portioned into artificial, collagen or
natural casings”. Nevertheless our customer has some problems with portioning into natural casing
where for a small fraction of production small air bubbles appear.
Our customer is a medium-size company. It produces different type of sausages. The sausage in
problem is a traditional sausage in which homogeneous stuffing is mixed with meat containing large
pieces.
The sausage mass consists of: 30% of homogeneous stuffing with diameter of particle 0.5 mm, in
temperature 8◦ C and 70% of large meat pieces with diameter of particle 8 mm, in temperature 3–4◦ C.
This mass is mixed for 5 minutes (3500 turns/min) in vacuum conditions.
Next the meat mass is left for about 4 hours before putting into filling machine. The filling machine
works also in the vacuum conditions (98% vacuum). The maximal velocity of filling is 50 Liters/min.
The pressure of filling is till 80 bars.
The filling is done with a 22 mm filling tube. The casing which is a natural intestine is put on the
filling tube with a spooling device with a large speed (a natural intestine is 10 m long and it is spooled
in few seconds). The tube has a length of 0.4 m and the wrapped intestine occupies 10–20 cm of the
tube.
 Before filling the intestine are kept for 10 minutes in salty water of unknown temperature. At the
manufacturer of the fillinge machine website there is an information that intestines should be wet for
2 hours in water at 40◦ C.
During the sausage production the intestine is filled with the meat mass with a speed of 0.5 m/s
(the whole 10 m of intestine is filled in 20 sec).
The produced sausage has a diameter of 24–26 mm.
When the air bubbles appear they have a size of 3–5 mm and are of a lens shape. The bubbles
are distributed irregularly over the whole length of the sausage (whole 10 m). On the other hand it
happens very irregularly – sometime just one sausage, sometime several consecutive sausages in a
row. But the overall bad product is of order o 1% of the whole production.

Hypothesis
A number of hypotheses was formulated to explain the formation of air bubbles. Below are those
hypothesis, which explain the most probable mechanism of the process:

• Air bubbles appear in the intestine of a smaller diameter. The main factor in deciding whether
the bubbles will be visible or not is pressure exerted by the intestine on a steel tube of filling
machine.

• Torus-like air bubbles are hidden in the folds of a "harmonica." At the time of the stretch
bending, in accordance with the principle of least action torus-shaped bubble changes to "lens-
shaped" one which is located at random place.

• The pressure exerted by the gut depends in a linear manner from the Young’s modulus. We sus-
pect that through the process of preparation (hydration in salty water) the value of the Yuang’s
modulus can be lowered by two orders of magnitude.

• During the process of dragging an intestine at high speed, air gets between the bowel and the
tube of the filling machine. This process is largely independent of the earlier preparations of
the products, so we do not have a significant influence on the amount of air contained in the
"harmonica".

The main factor which has to be taken into account is the fact that bubbles appear only in 1% of
products. This raises the question why in 99% of cases air bubbles manage to escape through the open
end of the gut?
The most plausible explanation of this last observation is that for the initial radius of the intestine
(less than a certain r , with certain Young’s modulus determined by the preparation process) the thrust
force by which the intestine acts on the steel tube prevents escaping air closed in bubbles through the
open end. For larger intestines the force is too small to trap air bubbles inside the intestine.
On the other hand, if the initial radius of the intestine is the main factor, why the operators of the
filling machine did not report this fact.
The only explanation is the fact the critical value of the radius is in the area where the operator
of the machine cannot fill manually a difference in thrust force. If only 1% of production creates
problems it means that the critical radius is very far away from a mean value of the radius.
Further clarification
To establish the mechanism of formation and transport of air bubbles during the manufacture of
sausage a number of other possibilities has been considered.
First, the following facts have established:
- Meat is prepared at low pressure (150-200 Pa, which corresponds to 1.5% - 2%) of atmospheric
pressure
- It is stored in atmospheric pressure for about 4 hours
- Filling machine works again at low pressure (150-200 Pa).
These facts clearly show that it is reasonable to assume that the volume of air in the meat mass is
so small that we can ignore it.
Then the only plausible assumption would be that the air gets inside with the intestine, when it
moves along the steel tube filling machine.
Another important point is the answer to the question why the bubbles appear in only 1% of cases.
This raises the suspicion that the mechanism responsible for the formation of bubbles is parameter
specific for the intestine but not for the meat mass, chemicals used for cleaning equipment or the
device itself.
It seems that the critical parameter is the size (radius) of an intestine in unstretched condition.
More detailed explanation of the hypothesis:
Membrane (intestine) is stretched at high speed (0.5 m / s) on the tube filling machine, where it
curls up in a "harmonica". Between the tube and the harmonica is a certain volume of air. Initially,
both ends of the bowel are open, but after the start of the filling machine only the rear end (the end of
the remainder of the tube) remains open. If the contact force exerted by the membrane is sufficiently
small, air can easily escape from the space between the membrane and a tube through the open end.
If the force is large enough, some air remains in the space between the membrane and the tube,
causing the occurrence of bubbles in the final product. An important factor influencing the incidence
of bubbles is a velocity of the movement along the tube. If this rate was lower, the air bubbles would
more likely escape through the open end.
When you move the gut on the tube there creates a vacuum between the membrane and tube. This
difference, however, is 125 Pa (corresponding to 1.2% of atmospheric pressure). Given this result, we
found that depression has insignificant effect on the entire process.
Obviously there is also possibility that during the production micro-bubbles are formed and trans-
ported with a layer of transported water (to moisten the steel tubes). Their size is quite comparable to
the width of the water layer, which we can estimate by 10 microns. taking under consideration this
fact, for forming a noticeable-sized bubble, there would have to collect 106 follicles in less than 0.8s
(so a particular point placed of the shell moves along the tube). It is clear that this mechanism also
has a negligible effect on the formation of bubbles.
Therefore the main factor (affecting more than 99%) is the pressure exerted by the gut on the walls
of the device. The pressure that exerts stretched gut on a tube have been calculated and it was found
that it is a determining factor whether bubbles appear or not.
Pressure of the intestine membrane
Table 1 contains data describing the process of sausage production.

parameter value description confidence

r0 10 mm radius of intestine in natural conditions our guess
r1 12 mm radius of filled (stretched) intestine factory value
h 0.4 m length of filling tube factory value
L 10 m length of intestine in natural conditions factory value
l 10–11 m length of filled (stretched) intestine our guess
v 0.5 m/s speed of filling intestine factory value
E 0.01–1 GPa Young’s modulus of intestine our guess
η 200 dynamic viscosity of meat mass our estimate

Table 1: Sausage production data

The pressure equation is the following one

1 E t0
p(r1 ) = 2
[(α2 − 1) + ν(β 2 − 1)], (1)
β 2(1 − ν ) r0
where α = rr01 , β = Ll , ν = 0.5 and t0 = 0.0001 m.
For our model we can assume that α = [1 . . . 1.2] and β = [1 . . . 1.1].
To defend the choice of Young’s modulus for intestine we have colected data available in internet
[2, 3]. These data are in Table 2.

material Young’s modulus in GPa

ABS plastics 2.3
Acrylic 3.2
Collagen 3.7–4
Rubber 0.01–0.1
Skin (human) 0.01
Nylon 2–4
Polycarbonate 2.6
Polyethylene HDPE 0.8
Polyethylene PET 2–2.7
Polypropylene 1.5–2

Table 2: Young’s modulus of different materials

Hence the Young’s modulus for intestine is assumed between the values for skin and collagen.

Derivation of a simple model to estimate the normal pressure of stretched

natural pig intestine on the steel tube of a sausage filling machine
The principle idea of our model is the following: We treat the intestine as a cylindrical elastic mem-
brane. We have to consider moderate geometrically nonlinear effects in longitudinal stretching and
radial extension, with deformation ratios of up to 1.2 (i.e. about 20%). We choose Koiter’s mem-
brane shell model with linear elastic, homogeneous and isotropic material behavior as an appropriate
mechanic model of the intestine membrane structure, cf. [5] and [6].
The starting point of our derivation is the formula for the total elastic energy Wshell of the de-
formed shell, which has roughly the following structure:
Z
2 D
2
Wshell = · change of metric + · change of curvature .
2 2
Here
Et0 Et30
Z= and D=
(1 − ν 2 ) 12(1 − ν 2 )
are the membrane and bending stiffness parameters of the shell, which depend on Young’s modulus
E (units: P a = N/m2 ) and the dimensionless Poisson number ν of the shell material as well as
the initial thickness t0 of the undeformed shell. This model is appropriate for moderate geometrical
nonlinearities, but not too large local strains. It contains fully nonlinear strain measures (i.e. the metric
and curvature tensor of the shell surface), but the material behavior is assumed to be linear. This is
indicated by the appearance of the stiffness constants Z and D which are well known from linear plate
and shell models.

l
t ≈ t0
l0

r0
r

Introducing a typical tube radius of r ≈ 10 mm and the typical thickness of t0 ≈ 0,1 mm, a rescaling
of the energy shows that the membrane part of the energy is proportional to tr0 ≈ 100
1
, while the bend-
t0

3 −6
ing part is proportional to the third power r ≈ 10 of this parameter. Therefore the contributions
of the bending term relative to the membrane term is negligible for the computation of the total energy
of the deformed shell in the special deformation case considered by us.
The initial shape of the tube is a cylinder of the length l0 , radius r0 and thickness t0 . In the
deformed state the tube has different parameters: l > l0 , r > r0 and t ≤ t0 . In our model we have:
l r
∈ [1, 1.2], ∈ [1, 1.2].
l0 r0
As the strains are not too large, we may safely assume — and do so — that the influence of the reduc-
tion of the shell thickness is small (i.e. t ≈ t0 ), for the following reason: For radial and longitudinal
strains in the range given above, the thickness can reduce up to factor 32 in the case of a fully incom-
pressible material (ν = 0.5). If we take the variation of the thickness due to stretching and radial
expansion into account during the derivation of our final pressure formula, the normal pressure will
be reduced by an additional term of order (t0 /r1 )2 ≈ 10−4 , which we can safely omit as a second
order effect.
 Using basic notions from the classical differential geometry of surfaces one can obtain the de-
formation gradient and the change of the metric tensor of the membrane surface corresponding to
the homogeneous deformation of our cylindrical tube. Inserting these ingredients into Koiter’s en-
ergy formula we obtain the following expression for the elastic energy of the deformed cylindrical
membrane:
( 2  2 2 )
r2
 2  2
Et0 l r l
Wshell = 2πr0 l0 · − 1 + 2 − 1 + 2ν −1 −1 .
8(1 − ν 2 ) r02 l0 r02 l02

The normal pressure p(r) to be applied inside the membrane surface or radius r ≥ r0 and length
l ≥ l0 to equilibrate the elastic membrane forces can be calculated by computing the total work which
has to be done during a radial inflation of the membrane at constant stretching:
Z r
1 ∂Wshell
Wshell (r, l) = p(r) · 2πrl dr =⇒ p(r) = · .
r0 2πrl ∂r
r l
In our case the final radius r1 is known. Introducing the stretching ratios α = r0 and β = l0 we can
express the above formulas as follows:
E t0 α  2
(α − 1) + ν(β 2 − 1) .

p(r1 ) = 2
· ·
2(1 − ν ) r1 β

This formula gives a rough estimate of the normal pressure of the stretched and expanded intestine on
the steel-tube.
To obtain the order of magnitude of this pressure we need to estimate the Young’s modulus E
of pig intestine. This material parameter presumably varies between typical values for human skin
(E = 10 MPa) and collagen fibres E = 1 GPa, which differ by two orders of magnitude. An
intermediate value of E ≈ 0.1 GPa might be a reasonable choice for approximations. Typical value
of the Poisson’s ratio for organic intestine material may be close to that of rubber ν = 0.5 (see [4]).
Taking the pre-factor into rt01 ≈ 10−2 account this yields
 
E α 2 1 2 3
p(r1 ) ≈ · α + β − .
150 β 2 2

If we take the final radius r1 ≈ 10 mm as a fixed given quantity, the latter formula may be used to
analyze the normal pressure of the intestine on the steel-tube as a function of the variation of the initial
radius r0 — this is quite hard to estimate — of the intestine and the stretching of the intestine on the
steel tube.
If air bubbles are present between the steel-tube and the intestine, they are probably surrounded by
water. The pressure p(r1 ) can then be used to estimate the pressure inside the air bubbles according
to Laplace formula

pbubble = p(r1 ) + (surface tension) × (mean bubble curvature).

However, a rough estimate for typical (visually observable) bubble sizes shows that the additional
pressure term proportional to the surface tension between air and water is of the order of 100 Pa and
therefore also can be neglected. So we may assume that pbubble ≈ p(r1 ) holds in our estimation of the
pressure of such air bubbles.
 According to our considerations, high pressure increases a probability of appearance of bubbles.
At some critical value (we suppose 1,5–2 bars) this probability substantially increases. But obviously
there is a possibility, that despite really unfavorable conditions the air bubbles will not appear. On
the other hand, there is also a possibility, that despite favorable conditions, air bubbles appear. Our
solution leads to lower the probability as much as possible. But we cannot guarantee that the resulting
calculations will explain the bubbles behavior in every single case.

Tables and graphs describing pressure dependence on α, β and Young’s

Modulus
The following section describes the pressure dependence of the radial stretch ratio α, the longitudinal
stretch ratio β and Young’s modulus. In the tables we can see how the pressure change depends on
the Young’s modulus and α for a fixed parameter β. These dependencies seem to be particularly
important since α is a factor that has a large influence on the pressure, and the Young’s modulus is a
parameter which could be controlled by a suitable process of preparation of intestines. As we expect
air bubbles should appear at high pressure (most likely more than 2 bars), which corresponds to the
lower right corners of the tables.

α\E 10 50 100 500 1000

1 0 0 0 0 0
1,05 0,057 0,285 0,569 2,847 5,694
1,1 0,117 0,583 1,167 5,833 11,667
1,15 0,179 0,896 1,792 8,958 17,917
1,2 0,244 1,222 2,444 12,222 24,444

Table 3: Pressure [in bar] exerted on the membrane at a fixed parameter β = 1

α\E 10 50 100 500 1000

1 0,027 0,136 0,271 1,356 2,712
1,05 0,081 0,407 0,813 4,067 8,135
1,1 0,139 0,691 1,382 6,911 13,823
1,15 0,198 0,989 1,978 9,888 19,775
1,2 0,26 1,3 2,6 12,996 25,992

Table 4: Pressure [in bar] exerted on the membrane at a fixed parameter β = 1, 05

α\E 10 50 100 500 1000

1 0,053 0,265 0,53 2,652 5,303
1,05 0,105 0,524 1,048 5,24 10,48
1,1 0,159 0,795 1,591 7,955 15,909
1,15 0,216 1,079 2,159 10,795 21,591
1,2 0,275 1,376 2,753 13,763 27,525

Table 5: Pressure [in bar] exerted on the membrane at a fixed parameter β = 1, 1

 In the following graph we present the relationship between force (pressure) exerted by the gut
on steel tube and 3 parameters: the Young’s Modulus (E), coefficient of radial stretch (α) and the
longitudinal stretch ratio (β). The Young’s modulus is set at level of 100 MPa. The coefficients α
and β vary between 1 to 2. The right graph shows an enlarged part of the left part (the color changed
to improve readability). The values of α and β vary in the range of 1 to 1.2. We believe that these
values correspond to the actual production conditions, which explains the importance of this part of
the chart.
We suspect that the critical pressure, which may cause the occurrence of bubbles is about 2 bars
(light-blue color on the right graph).
In the model there is a linear dependence on Young’s modulus. If the value measured experimen-
tally proved to be significantly different from the assumed (100 MPa), the pressure read from the chart
would have to be scaled by a constant factor (measured value)/100.
Figure 1: Graph showing pressure dependence of α and β, On the left general graph, on the right the values more specific for our model
 In the next two graphs we show the dependence of pressure exerted by the intestine from the
Young’s modulus (E) (from 10 to 1000 MPa) and the radial stretch ratio α. Longitudinal extention
ratio was set at 1.02 and 1.1 respectively.

Figure 2: p(α, E) with β = 1, 02

Figure 3: p(α, E) with β = 1, 1

We see that if we compare the values of these two graphs for some α and E (especially for
α ≈ 1.2), the difference does not exceed 10%. Thus we see that the growth of α has much more
significant impact on the growth pressure.
References
[1] http://handtmann.com/vf612_video.html - Video clip of sausage production process.

[2] http://en.wikipedia.org/wiki/Young’s_modulus - Some basic information on Young’s Modulus.

[3] http://www.engineeringtoolbox.com/young-modulus-d_417.html - Values of Young’s Modulus

for different materials.

[4] http://en.wikipedia.org/wiki/Poisson’s_ratio - Some basic informations on Poisson’s ratio.

[5] P. Ciarlet –An Introduction to Differential Geometry with Applications to Elasticity, J. Elasticity,
78–79 (2005), 1–215.

[6] A. H. Corneliussen and R.T. Shield – Finite Deformation of Elastic Membranes with Application
to the Stability of an Inflated and Extended Tube, Arch. Rational Mech. Anal., 7 (1961), 274–304.
Appendix
Considered mechanisms of bubbles formation
While trying to find a suitable mechanism many hypotheses have been raised. Most of them we
managed to reject for various reasons, but for the documentation purposes we present them in this
appendix.

1. When mixing the meat mass air gets to it in the form of micro-bubbles, which, after filling join
together and diffuse into the outer layers.
Counterarguments:

• Both mixing and filling takes place at low pressure (150-200 Pa), so the amount of air
trapped in the meat mass is negligible.
• Bubbles are visible immediately after filling, the diffusion cannot last fractions of seconds.
• It would be impossible to determine why the bubbles appear in only 1% of products.

2. Bubbles in fact contain no air, but other gas which is formed in a chemical reaction of meat
mass with the intestine, or diffusing from the inside of meat mass.
Counterarguments:

• Bubbles are visible immediately after filling, the diffusion takes time.
• It would be impossible to determine why the bubbles appear in only 1% of products.
• If intestine would react with a mass of meat, the process of production would probably
not have been approved by the health control.

3. Visible bubbles form by merger of micro-bubbles transported in a layer of water between an

intestine and a steel tube.
Counterarguments:

• Bubbles would have to be comparable to the thickness of layer of water, which we can
estimate by 10 microns.
• there would be necessary to connect about 106 micro-bubbles to form a noticeable bubble.
• It would be impossible to determine why the bubbles appear in only 1% of products.

4. Bubbles appears in intestines, whose radius is bigger then radius of the tube of filling machine.
While filling, some bubbles from outside the intestine are sucked by underpressure created by
movement of the bowel along the tube.
Counterarguments:

• We assume, that a worker keeps his hand on the tube. In this case air bubbles would be
easily squeezed.
• Moreover, on a machine producer’s webpage we saw a special ring-shaped device around
the tube. We know, that the air bubbles problem appears around the world. If this was a
mechanism, it would be already solved. That is why we excluded such possibility.
• Pressure of air trapped under the gut is equal to 99% of atmospheric pressure. The differ-
ence is to small to have a noticeable influence on the process.
 • Taking this result under consideration we cannot explain the mechanism of possible suck-
ing the air bubbles.

5. Bubbles already trapped under intestine are sucked to the prepared sausage by the underpres-
sure.
Counterarguments:

• Pressure of air trapped under the gut is equal to 99% of atmospheric pressure. The differ-
ence is to small to have a noticeable influence on the process.
• In intestines of typical radius, that would be easy to squeeze the bubbles out.
• This effect in fact can constitute a minor effect, but as we estimated bubbles formed by
this effect will have size of 10−8 m = 10nm, which will make them invisible.

6. Bubbles appear if air was not squeezed before stretching the bowel on the tube.
Counterarguments:

• Air appears always between gut and steel tube. This proves, that there must be another
mechanism of forming air bubbles.
• We could not explain why bubbles appear only in 1% percent of products.

7. Bubbles appear because of defect of the filling machine

Counterarguments:

• Bubbles in sausages is a worldwide known problem.

• In this case we would have to assume defect of almost every filling machine.