Sei sulla pagina 1di 17

Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

Dr. D. P. Campbell, B.A., Ph.D., D.P.Campbell@sbe.hw.ac.uk

Drainage Research Group,

The School of the Built Environment

Heriot-Watt University

Edinburgh, EH14 4AS

d.p.campbell@hw.ac.uk, Tel +44 131 451 4618

Title:
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

Abstract
A mathematical model utilising the Method of Characteristics to solve the equations of
continuity and momentum in a finite difference scheme forms a computer based simulation
package (AIRNET). It can simulate conditions in building drainage, waste and vent (DWV)
systems. Recent work extended AIRNET by including the effects of detergents and temperature.
Other work examined the motive force of the water, extending it to include a proportion of the
discharge appearing as droplets falling in the central air core as opposed to assuming it was due
entirely to an annular film. This paper includes these additions and links to previous work on non
dimensional analysis, to yield a more refined AIRNET model which includes the effects of
detergents, temperature and droplet fraction. Significant refinement of the AIRNET model is
described in this paper.

Practical Application
Accurate simulation of detergent dosed waste as described here extends the scope over which
accurate simulations and code guideline comparisons can be made, to include live domestic and
industrial media from modern appliances and practices. This will have a significant impact on
code generation and drainage system design practices, allowing optimisation for reliability,
effectiveness and cost. This paper improves the current level of understanding of the hydraulic
basis for the design requirements imposed on sanitation engineers and code bodies, and
improved an existing computer simulation model in its capacity as a design tool.

1
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

KEYWORDS:
building drainage, remote sensing, fluid flow, stack discharge, modelling

UNITS AND ABREVIATIONS

A area

d partial derivative

d diameter

DC free diameter of wet stack

DRG Drainage Research Group

DWV Drainage, Waste and Ventilation system (in buildings)

f friction factor

FS force due to shear stress

g ratio of specific heats

1 wavelength of laser light

L stack length

Q volumetric flow rate

density

R radius (of stack)

i interface shear stress

u velocity

Vt terminal velocity

Vmax maximum velocity

2
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

Background.
One of the most important roles of a drainage system is the isolation of the habitable space from
the waste material and its associated effects, achieved by the provision of water trap seals at
various points in drainage networks. Design guides aim to produce ventilated drainage networks
which protect these trap seals from unwanted air pressures. These design guides are based on
investigations conducted primarily at BRE from the 1950s onwards, during which it was
assumed that water enters a drainage system as unsteady flow and descends in a stack as an
annulus inducing movement in the central air core, Pink [1]. This air movement manifests as a
suction pressure in the network, which water traps must withstand. Current design guides and
building codes protect trap seals from suction pressures over -375 Nm-2 Wise, [2] in the form of
BS5572[3]. Formative work can be traced back to Wise [7], progressing through a series of site
investigations involving multi-storey buildings by Lillywhite & Wise [4], in which the mechanism
of drainage stack annular down flow was defined.

Recently, this area has become the subject of intensive investigation with a coherent series of
SERC/EPSRC research grants. The first investigated the propagation of air pressure transients in
building drainage vent systems. The equations describing the boundary conditions represented by
common system components were identified by Swaffield & Campbell [6] and incorporated into
an existing mathematical model (AIRNET). The AIRNET model is based on a finite difference
scheme and utilises the method of characteristics as a solution technique to simulate drainage
system operation via the equations that define unsteady partially filled full bore pipe flows and
the boundary conditions represented by water traps and other common system components. In
terms of model algorithm development, the relationship between the suction pressures developed
in the stack and water downflow rates was found to be analogous to that of the operating
characteristics of a fan by Swaffield & Campbell [7] permitting the model to simulate the
interaction between the hydraulic driving force behind air movement and the modifying effects
of the system components themselves. This was later made independent of laboratory based test
systems by Swaffield, Campbell & Jack [8]. The role of water films formed by established
downflows falling past side entries and acting as transient attenuators was also included, as was
the effect of intermittent stack occlusion and subsequent air flow stoppage caused by branch flow
as covered by Wise and Swaffield [9] and Swaffield & Galowin [10]. This investigation was limited
by its application to a single pipe diameter of 100mm. Jack [11] extended the model by applying
the established techniques to pipes in general, overcoming the 100mm-diameter restriction. Site

3
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

investigations using multi-storey buildings and live drainage systems were a feature of this
investigation, as were greater wet and dry stack and horizontal branch lengths, multiple
combinations of user-defined appliance discharges, and higher water flow rates.

Campbell and MacLeod [12,13] reviewed the assumption that airflow induction was due solely to
cold, clean water falling as an annulus and acting as a driving force. A new model was developed
which identified the changes associated with the two main detergent groups: anionic and
cationic; the effect of temperature was also quantified. Campbell [14] developed the work further
and found that a continuous fraction of the stack discharge persisted in the central air core and
accounted for the contribution to airflow conditions made by it. Campbell [15] then applied the
particle tracking velocimetry technique to validate the model. Campbell [16] continued the work
with a dimensional analysis of the factors involved, which provided a potentially useful solution
technique.

This paper extends the work and establishes a general solution for DWV networks, including
boundary conditions, charged with detergent dosed water.

Stack Discharge Analysis


It has been assumed that the velocity profile of the annular flow in a DWV system stack is linear
and that the airflow flow down the vertical stack is as a result of the no-slip condition
between the film and the air. The estimation of the interface velocity of the annular film and
the air core is of great importance as it describes the air-water interaction and driving force
behind air entrainment and thus the pressures that result during stack discharge. It is normal
practice to refer to the annular film attaining a terminal velocity. If a linear profile is
assumed through the film then the interface velocity is given by:
Vi = 2Vt

However there has been no data to support this assumption.

If a falling film is considered having laminar flow the shear stress acting on the film interface is
given by:
dv
d =
dx
This is equal to the weight of the film that is a product of the specific weight, leading to:

4
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

x x2
v= + c1 x + c2
2

The constants of the integration can be obtained by applying the boundary conditions to the
liquid film.
Hence, and by differentiating, the expression for velocity distribution is given by:

dv vmax
= + ( 2 x )
dx 2

The velocity gradient at the air/water interface is zero. Therefore the volumetric flow rate then
becomes:

3
V = vdx =
0
3

The average film velocity is given by dividing V by the film thickness to give:
2
v=
3

The ratio of interface to mean velocity of the film is given by:


v 2
=
vmax 3

Therefore it follows that annular film interface is given by


vi = 1.5vt

This is an estimate for the laminar flow condition. In reality the flow is often turbulent in nature.
This will modify the boundary layer profile which will be thicker and the mean velocity will be
higher. This means that for turbulent flow:
vi < 1.5vt

This analysis makes an assumption that the velocity gradient at the interface of the water film
and air is zero. However if this is not the case then the velocity gradient will be a function of the
air flow rate:
dv
= f (Qa )
dx
So, the velocity gradient will change depending on the airflow however at this stage the above
equation is assumed to be a good approximation.

Airflow analysis

5
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

A characteristic feature of annular flow is the interface between the air core and the film. It is not
a smooth surface, but one with complex wave patterns[17]. As a result of this a single-phase air
core does not exist. The work carried out in this paper supports these findings.

Previous work[12] outlined the classes of detergents covered in this investigation, i.e:
(i) anionic (common in soap powders, tablets and liquids)
(ii) cationic (used as fabric conditioners)
(iii) amphoteric (specialist blenders in de-greasers)
(iv) non-ionic (second most widely used in washing powders)
With further addition of detergents abrupt changes in the physical properties of the solution
occurs at a specific concentration; surface tension, osmotic pressure and turbidity all change
abruptly. The effect increases up to a point (critical micelle concentration (CMC)) and then
remains constant. Since the CMC is reached at detergent concentrations far below those used
domestically or industrially (a factor of 5 is common), the manufacturer recommended dosing
levels were adopted throughout. All detergents are supplied with various builders, perfumes and
carriers with the actual concentration of active ingredients commercially confidential and
reported as lying within a range; i.e. 5-15%. In this case, the median value was used in
calculating the volume required for the tests reported here. The actual detergent concentrations
used (reported here as the weight to weight or w:w ratio) are as follows:
(i) anionic 0.023%
(ii) cationic 0.030%
(iii) amphoteric 0.018%
(iv) non-ionic 0.025%
Two temperatures were used in this work: ambient (18C) and warm (55C), as determined by
measurement in a two-storey 1:1 scale model built in a laboratory, using standard uPVC fittings
and standard fixtures, and mains cold water or 40l discharges of water dispensed at 1 l/s at 60C.
This is the maximum temperature that could be experienced in practice from a operational
industrial dishwasher and this was chosen to maximise the effect.

Friction factor and Core Flow Phases


Three flow zones are known to exist within the air core:
Liquid film with near 100% single phase water or water surfactant flow
Two phase liquid air mixture near film surface

6
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

Like zone 2 with more air and less water

The air movement is a result of the shear stress or friction between the total water movement,
including water droplets, and the air. Within building drainage flow modeling it has been
assumed that two single phases exist and that the air movement is due to the surface of the
annular film.

Wallis & Hewit [18] concluded that for annular flow at low gas and liquid velocities the film
surface is smooth and that there is little droplet entrainment and the friction factor at the film air
core is approximately equal to the pipe roughness. However above a critical value of air and
water movement the friction factor rapidly increases, the tops of the surface waves are sheared
off to form droplets and the density of the core increases. Campbell & MacLeod [14] showed that
this was a simplification, and that a more accurate force balance due to a falling annulus plus a
significant fraction of the discharge in the form of droplets would then be:

Fs = totalDc L

The total shear stress is the sum of the shear stress at the film water interface and the shear stress
of the droplets:

p

4
[
D = Dc L ifilm + drop ]
This gives the total shear stress for the two-phase flow that is the sum of the shear of the droplets
and the annular film interface.

Therefore the total shear can be expressed after substitution as:

p 2C f (Vw Va )
2

=
L Dc

This relationship gives the pressure gradient over the wet stack length, and incorporates the
contribution to the motive force applied by water droplets in the air core and integrates this with
the distributed friction model.

Pressure Zones

7
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

It is recognised that there are pressure zones present in a drainage system. There will be suction
pressure at the base of a dry stack, a localized pressure below a discharging branch above the
curtain and a back pressure at the base of the wet stack. All are critical for determining the entire
pressure profile of the network. Obtaining an accurate profile is necessary to ascertain trap seals
response.

The suction pressure at the top of the wet stack zone 1 is the sum of the vent entry loss and the
dry stack friction loss. This pressure will be dependent upon the venting design that is the vent
piping length, piping material and diameter. It can be calculated from a loss factor for the entry
loss and Darcys equation.
V2
Psuction = (K M 0.5 aVa2 )Entry +
fL 2m

Drysatck

The pressure drop experienced at the branch Zone 2 as a result of the discharge can be calculated
in the same way as the entry loss to the dry stack, except the loss factor K can be obtained from
an equation developed by Jack [11]:

Pj = K m 0.5 aVa2

Where the K loss factor is given by the following equation:

K M = 3.88 X 0.91 6.5


And
15 0 .9 1.4
Free Area Air core AreaU rc Ds
X = 1 +

Air core Area

D
D
StackArea D s B

It should be noted this equation was developed based on the fluid trajectory from the branch to
the stack from scaled drawings. However when comparing the difference in the free area from
the calculated fluid trajectory to that estimated from drawings, the difference was negligible and
it was not necessary to modify the equation as it is a good approximation. As such, the above
equations developed by Jack were used to calculate this pressure drop.

A region of positive air pressure occurs immediately above the base of the wet stack. This arises
as the fluid detaches itself from the pipe bend and a curtain is formed. The entrained airflow has

8
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

to push its way through the curtain and a positive pressure is experienced which then recovers to
atmospheric pressure. This to is calculated from a loss factor K where:
1
Pback = K Vmean
2

Pseudo Friction factor


The pressure zones are as a result of air transients driven by the discharged fluid within the pipe
work. The interaction with the fluid and the air can be described as a friction factor as developed
by the Darcys equation which is expressed as:
4 fLV 2
Pf =
2D
Rearranged where the air velocity is relative to the film velocity gives:

2 Da Ptotal
f =
4 a H w (Va V f )(Va V f )

Where Va is the mean air velocity of the air core and V f is the velocity of the annular water
film at the air water interface. The friction factor equation also has an equivalent height term
H w this height can be defined as: the length at which the annular film does work on the air

core, or the length the air core does work on the annular film.

This length is thus different from the stack height as the stack may be connected to several floor
levels in the building. Also the waste discharge doesnt instantly do significant work on the air
core. Only when annular flow is developed does significant work on the core occur. So an
estimate of the equivalent height could be the distance from the stack base to the point at which
annular flow is developed. The equivalent height is therefore dependent on the geometry of the
junction, the flow rate and the pipe material.

The trajectory of the fluid from a junction could be calculated based on the radius of curvature of
the branch, the discharge flow rate and the friction coefficient of the pipe material. Taking this
analysis further, if the flow from the junction is looked at in more detail when the discharge fluid
element hits the stack wall the fluid splits up and has a high degree of swirl. Some of the
discharge rebounds off the pipe wall and hits the opposite wall at the same angle as the incident
angle. From this approximation the height at which annular flow occurs can be estimated:

9
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

V2
= g cos ( )
R

From this the point the fluid makes contact with the stack wall is given by:
Dp
H1 =
tan ( )

Assuming the rebounded fluid from the wall goes to the other wall at the same incident angle
then the height at which annular flow is developed can be approximated to be:
H annular = 2H1

To obtain this value however the flow rate of the discharge from the junction has to be calculated
using the Chezy coefficient for the pipe material. The equivalent height can now be estimated as:
H equivalent = H Stack H annular

As the water flow rate increases so does the equivalent height as annular flow is reached quicker
and so there is greater height for the film to do work on the air. Enough data is now available to
calculate the pseudo friction factor for the air transient propagation.

Operation of the Boundary Condition and AIRNET Model


A technique was developed by Swaffield & Campbell [6,7] which describes the movement of air in
discharging DWV systems in terms of the equations of continuity and momentum:

u
r +u + = 0
x x t
p u u 4fu u
+ru +r + = 0
x x t 2D

which yield wave speed through the isentropic relationship used by Swaffield & Campbell [5] and
which are linked to air density and pressure. These eventually reduce to two quasi-linear,
hyperbolic partial differential equations L1 and L2:
u 2 c c
r + u + = 0 L1
x 1 c
x t

2 c u u 4fu u
rc +ru +r + = 0 L2
1 x x t 2D

10
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

This expression includes a Colebrook-White - derived friction factor for dry stack and entry
losses, balancing these against the motive force applied by the falling water, which they must
equal and which will also include the detergency, temperature and droplet contribution as defined
by Campbell & Macleod[13]:
surfactant
6 4 7 4 8 6temperature
4 7 48
f h k m
Va g
D j
3 2
k gD w 2 TD
n

Fr Re w 2
Vw H w D w w w D

Equations L1 and L2 can be combined to produce a pair of total differential equations in u, c, x


and t:
du 2 dc 4fu u
+ t = 0
dt 1 dt 2D
x
providing that = uc, which essentially limits the solution to two characteristic lines in the
t
dimensions of distance and time. For the computer-based finite difference solution, conditions at
internal nodes are determined from the pair of characteristic equations
2 4fu u t
uP-uR (cP-cS)+
1 2D 2D

The options may be arbitrarily assigned to flow direction. C+ and C- characteristic equations
are used in determining the air pressure transient propagation, where the C+ and C- equations are
reduced to:
C+
u p = K1 K 2C p

and
C
u p = K 3 + K 4C p

Where u is the gas mean velocity at a section P and c is the gas wave speed velocity. The values
of K1 to K4 are determined by the local conditions within one Dx either side of P at a time step
earlier determined by the Courant criterion.

Suction pressure at base of dry stack


The C characteristic equation is combined with the boundary condition expression for the
suction pressure at the base of the dry stack. Since the pressure depends on the local fluid

11
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

velocity, which is present in the boundary condition expression, then an iterative solution is
required. In this particular instance the bisection method is utilised. The model for suction
pressure at the base of the dry stack has the following procedure.
(a) Upper and lower pressure limits are guessed
Upper pressure limit UP= Atmospheric
Lower pressure limit DP= Suction pressure with Qa=0
(b) Calculate wave pressure air speed:
1
(UP + DN )
ca = 2
a

(c) Calculate U a from characteristic equation


ua = K 3 + K 2Ca

(d) Calculate Dp from boundary condition equation. The general form of the equation was shown
by Campbell & MacLeod [12,13] to be:
c

( (x)) a
2
P = 1 2 Va b d
D


t
surfactant
6 4 7 4 8 6temperature
4 7 48
f h k m

V
Where the function (x) = a Fr Re g D k

j

gD 3
2 2 TD
n

w w
2
Vw H w D w w w D

Specific values of the variables were found by Campbell and MacLeod [12,13].

(e) Compare guessed value with calculated value, if different then set new guess as calculated
value if the same then solution is obtained.

Note that stage (c) for calculation of U a requires the correct time step from the Courant
criterion. This procedure means the air flow at the base of the stack can be calculated based on
the stack water flow rate Vs time profile and the upstream conditions. It can be thought of as the
information from upstream being transferred through the characteristic lines.

Maximum pressure calculation

12
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

When there is zero airflow (when the vent or upper dry stack is blocked) then a new boundary
condition for the suction pressure has to be used. A limit has to be set for the calculation of the
suction pressure to stop as a pressure of infinity would be calculated at low airflows tending to
zero. Equations have been developed by Campbell and MacLeod [12.13] for each surfactant that
calculates this new limit at zero airflow condition.

Active Branch Boundary Condition


To be able to calculate the maximum pressure within the drainage network the pressure drop
across an active branch must be calculated and then added to the value of suction pressure
obtained from the new boundary condition equation. The velocities at the branch of the junction
are:
ua = K 3 + K 4Ca

ub = K1 K 4C B

and
u D = K 3 + K 4CD

Balancing the flows at the junction gives:


uB AB uA AA uD AD = 0

Substituting the characteristic equations for the velocities and arranging gives:
(AB K1 AA K 3 AD K 3) AB K 2CB C A K 4(AA + AD )=0
Also
C A = CD

These can be obtained from:


PB
CB = 1
PB

P o
o
And

PA
CA = 1
PA

P o
o
The pressure loss from the water curtain has been shown to be given by:

13
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

1
Ploss = Kub ub
2
Substituting for ub gives
1
ploss = b K (K1 K 2cB )(K1 K 2cB )
2

The junction pressure model has the following steps:


i. Pressure B is estimated
ii. The wave pressure air speed is calculated
iii. The fluid velocity u B is calculated from the characteristic equation
iv. The fluid velocity is then substituted into the pressure loss equation and calculated
v. The pressure at B can be calculated from PB = PA + Ploss
vi. (AB K1 AA K 3 AD K 3) AB K 2CB C A K 4(AA + AD ) = 0 must be satisfied and then
the loop stopped.

Back pressure at base of wet stack


The solution procedure requires that the C characteristic equation is used. The procedure is as
follows:
i. Upper and lower pressure limits are guessed
ii. Calculate wave speed cc

Pc
cc =
a

iii. Calculate Ua from characteristic equation


u c = K 3 + K 2C c

iv. Calculate Dp from boundary condition equation


v. Calculate Pc = Patm + Dp

Model Performance
The result of these modifications to AIRNET is shown in Figure 1, which shows the output of
the original AIRNET and modified AIRNET models compared to measured data, including the
main detergent classes, at ambient and warm discharge temperatures for a simple two-storey
building constructed as a 1:1 scale replica in the laboratory. The modified AIRNET model
performs significantly better than the original version reported by Campbell & Macleod [13].

14
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

SUGGESTED LOCATION
Figure 1. Comparison of original and modified model output and observed data for a
simple network and various detergents.

Conclusions
The historical development of research in the area of DWV systems has included extensive
research into the hydraulic and pneumatic conditions in the stack and vent pipes, revised recently
by Wise & Swaffield[9]. Historically, the results were interpreted as suggesting that the water in a
stack falls as an annulus which then entrains an airflow causing an associated pressure drop
dependant on the dry stack configuration, although evidence existed which suggested that a
significant proportion of the water was found in the air core. The model proposed by Campbell
and MacLeod [16] builds on the research of many previous workers and current DRG work, and
suggests that the airflow entrainment is shared between the water annulus and the water falling in
the air core. They also showed that detergents in common use today, and temperature, have a
significant effect on the behaviour of a DWV system.

Accurate simulation of detergent dosed waste extends the scope over which accurate simulations
and code guideline comparisons can be made, to include live domestic and industrial media.
This will have a significant impact on code generation and drainage system design practices,
allowing optimisation for reliability, effectiveness and cost, in view of the present drive for water
and materials savings. The work reported in this paper has taken forward the work reported so
far by Campbell & MacLeod [12,13,14] and Campbell [15,16]. It has improved the current level of
understanding of the hydraulic basis for the design requirements imposed on sanitation engineers
and code bodies, and improved an existing computer simulation model in its capacity as a design
tool.

15
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

References
1 Pink BJ, A study of water flow in vertical stacks by means of a probe, BRE current paper,
1973, 36/73
2 Wise AFE, One-pipe plumbing - some recent experiments at the Building Research
Station, 1952, Journal IPHE, Vol 51, pp88-96, London
3 BS 5572:1994, British Standards Institute Publication, Code of practice for sanitary
pipework, ISBN 0 580 10168 1. (Note: current version: BS EN 12056-2:2000, ISBN 0 580
22927 0).
4 Wise AFE, Water, sanitary and waste services for buildings, 1986, 2nd edition, Mitchel,
London
5 Lillywhite MST & Wise AFE, Towards a general method for the design of drainage systems
in large buildings, Journal IPHE, 1969 68(4)
6 Swaffield JA & Campbell, DP, Numerical modelling of air pressure transient propagation in
building drainage vent systems, including the influence of mechanical boundary conditions,
Building & Environment, 1991 Vol 27, no. 4, pp455-67
7 Swaffield JA & Campbell, DP, The simulation of air pressure propagation in building
drainage and vent systems, Building & Environment, 1995 Vol 30, no. 1, pp115-27
8 Swaffield JA, Campbell DP, Jack, LB, The simulation of air pressure propagation in
building drainage and vent systems, CIBW62-94 Conference, 1994 Brighton, England
9 Wise AFE, Swaffield JA, Water, Sanitary and Waste services for Buildings, 4th Edition,
Longman, 1995 London.
10 Swaffield JA, Galowin LS, The engineered design of building drainage systems, Ashgate
Publishing, 1992
11 Jack L.B., An investigation and analysis of the air pressure regime within building drainage
vent systems, Ph.D. Thesis, Heriot-Watt University, Edinburgh, 1994
12 Campbell DP., MacLeod KA, Detergency in Drainage-Waste-Ventilation (DWV) Systems,
D. P. Campbell, K McLeod, Building Serv. Eng. Res. Technol. 2000 21(1) 39-43 (2000)
13 Campbell DP., MacLeod KA, Detergents in Drainage Systems for Buildings, Water
Research, 2000 Vol 18 no.1, pp128-134, ISSN0043-1354
14 Campbell DP., MacLeod KA, Investigation of the causative factors of airflow
entrainment in Building Drainage-Waste-ventilation (DWV) Systems, Building Serv. Eng.
Res. Technol, August 1999; vol. 20/3: pp. 99-104.

16
Developments in Mathematical Simulation of Fluid Flow in Building Drainage Systems

15 Campbell, D.P., 'The Application of Particle Tracking Velocimetry As A Velocity


Measurement Technique', Building Serv. Eng. Res. Technol, Nov 2006; vol. 27/4: pp. 327-
340.
16 Campbell, D.P., 'Surfactant effects on air pressure transients in building drainage systems',
Building & Environment, May 2007; vol. 42/5: pp. 1989-1993.
17 Li, Renardy, "Direct simulation of unsteady axisymmetric core - annular flow with high
viscosity ratio", Journal of Fluid Mechanics (1999), vol. 391, pp. 123-149.
18 G. B. Wallis, 1969 One-dimensional Two-phase Flow, McGraw-Hill Book Company,
rev.2, 2002, ISBN 13-978-007-0679429.

17

Potrebbero piacerti anche