Professional Documents
Culture Documents
Estimation of Pressure
Loss Coefficients of Pipe
Fittings: An Undergraduate
Project
KUMAR PERUMAL,1 RAJAMOHAN GANESAN2
1
INTRODUCTION
Pipe ttings such as valve, bend, tee, elbow, contraction and
expansion are integral part any piping system found in chemical
and allied industrial processes. These are mainly used to control
the ow rate and change the direction of ow, which causes energy
loss in addition to that caused by the uid ow through straight
pipes. Flow of uids in a piping system is accompanied by both
skin and form friction, resulting in pressure or energy loss. Skin
friction, which is responsible for pressure loss in straight pipe ow,
is the friction between the pipe wall and the uid and also between
the uid layers. Whereas, form friction is caused by pipe ttings as
the uid is subjected to sudden velocity and direction changes.
Reliable pressure loss coefcients for various pipe ttings are
needed to calculate the additional energy loss and determine the
correct pump size [1]. The classic reference for such data is the
Chemical Engineers handbook [2]. But this data are limited to only
single phase ow of Newtonian uids. Industrial ows are often
complex involving either multi phases (i.e., solidliquid, gas
liquidsolid) or non-Newtonian uids. The pressure loss
180
coefcients for such uids and ows are not readily available in
hand books. So there are several attempts by researchers to
determine these coefcients. In one of the earliest works, Griskey
and Green [3] determined pressure loss coefcient data for dilatant
uids. Turian et al. [4] provided loss coefcients for turbulent ows
of concentrated non-Newtonian slurries and Telis-Romero et al. [5]
presented the data for laminar ow of pseudo plastic uids.
The summary of the published literature is given in Table 1.
A careful examination of the literature reveals that most of the
previous work is experimental (E) in nature, which is expensive
compared to the numerical simulation (CFD) studies. The cost of
experimental studies further increases when it involves sophisticated instruments such as Electrical Capacitance Tomography,
Wire Mesh Sensor Tomography. Because of the low cost, use of
CFD for the study of uid mechanics, heat and mass transfer of
various chemical processes has increased signicantly in the last
decade or so.
Computational Fluid Dynamic (CFD) simulations predict
ow variables such as velocity, pressure ect, by solving the
mathematical equations describing the relationship between the
ow variables. So the accuracy of the simulation results depends
on how well the mathematical model or equations captures the
ow physics. The accuracy of the mathematical model is evaluated
by comparing the simulation results with the experimental results.
This process is called validation in CFD parlance. The
Table 1
181
Reference
Pipe fitting
Fluid
Remarks
908 bend
Airwater
908 bend
Airwater
Return bend
Airsolid
908 bend
Single phase
908 bend
Airwater
Oil in water
measured emulsion
Coalair
Deshpande and
Barigou [13]
Gasliquid
Deshpande and
Barigou [14]
Gasliquid foam
Xanthan gum
Venkatasubramanian
et al. [12]
Spedding and
Bernard [17]
Spedding et al. [18]
Fester and Slatter
[19]
Liu and Dian [20]
Abdulkadir et al. [21]
Water, glycerol
and CMC
908 Elbow
Gasliquid
908 Elbow
Globe valve
Wateroil air
Water, glycerol
and CMC
Coal water slurry
Airsilicone oil
Diaphragm valve
CMC
Oilwater
Oilwater
Tee junction
Airwater
Elbow
Gassolid
908 bend
Airwater
908 bend
Gasliquid
Gasliquid
182
Momentum conservation:
@
r~
u r ruu~ rp r t
@t
CFD MODELING
mt @e
e2
p
rC1 Se C2 r
s e @xj
k ne
k2
e
Governing Equations
The mass and momentum conservation equations are expressed as
follows:
Mass conservation:
@r
r r~
u 0
@t
Figure 2
183
loss coefcient, which accounts for the form friction. The Fanning
friction factor is dened as [41]:
f
DPs D
2rV2 L
Where DPs is the pressure loss caused by the straight pipe section
of length L and Kf is dened as follows:
Kf
Boundary Conditions
Water at ambient temperature (300K) was used as the working
uid. Simulations were carried out by specifying velocity at the
inlet of the horizontal pipeline. Turbulent intensity, I and the
hydraulic diameter, Dh were specied for an initial guess of
turbulent quantities (k and e). The turbulent intensity was
estimated for each case based on the formula I 0.16(Re)1/8
and was about 3% for all the cases. Outow boundary condition
was used at the outlet boundary.
2DPf
rV2
Where DPf is the pressure loss caused by the pipe tting. A plot
between DPf and rV2/2 results in a straight line passing through
origin with Kf as the slope, which is the average value of the
pressure loss coefcient for the given ow condition. But, for
accurate determination of pressure loss, knowledge of loss
coefcient as a function of Reyolds number is essential. Figure 4
shows the comparison between loss coefcients from simulation
and published literature.
Hooper [42] developed an empirical two k method, which
correlates the loss coefcient with the Reynolds number and the
diameter of the tting through the following equation.
K1
1
10
K1 1
Kf
D
Re
The K1 and K1 values for the 90 degree bend are taken as
800 and 0.25 respectively and D is the diameter of the tting in
inches. It should be noted that this method is applicable only for
single-phase ow through pipettings. Several authors [15,22]
have modied this equation for the ow of complex uids and
ows and estimated the K1 and K1 values for both laminar and
turbulent regimes. Csizmadia and Hos [43] used experiments and
CFD modelling to determine the loss coefcient of Bingham and
Power law uids for ow though diffuser and elbows. It can be
observed that the agreement between the K values is reasonably
good, particularly for Re <100000. Beyond this Re value, the
empirical Kf values become constant, whereas the simulated value
of [43] is lower than their own experimental value and of Miller
[44]. They claimed that this under prediction by simulation maybe
due to the assumption of hydraulically smooth pipe. In spite of the
similar assumption, the realizable ke turbulence model, (with
standard wall functions) used in the present work is able to predict
6
g
rg
Where P is the static pressure of ow and r is the uid density,
while the subscripts indicate points 1 and 2, respectively. The term
F accounts for the friction losses, which include losses in the
straight pipe section (i.e. skin friction) and from pipe ttings (i.e.
form friction) in the system. These can be formulated as
X
X 2fV2L
D
X KfV2
2
Where f is the Fanning friction factor, which accounts for the skin
friction loss, V is the average velocity of the uid, L is the pipe
length and D is the pipe diameter. Kf is the dimensionless pressure
184
Figure 5
CONCLUSIONS
The usefulness of CFD simulation for the estimation of pressure
loss coefcients of pipe ttings has been demonstrated through a
senior undergraduate project. The results are conclusive that a
validated CFD model is a cheap alternative for simulating
complex industrial ows and hence the determination of loss
coefcients. It is strongly believed that contour plots will help
students in the visualisation of uid ow phenomena such as
boundary layer separation.
REFERENCES
[1] B. S. Massey, Mechanics of uids, 2nd ed., Van Nostrand Reinhold,
London, UK, 1970.
[2] R. H. Perry and D. W. Green, Perrys chemical engineers hand book,
7th ed., McGraw-Hill, New York, 1997.
[3] R. G. Griskey and R. G. Green, Flow of dilatants (shear-thickening)
uids, AIChE J 17 (1971), 725728.
[4] R. M. Turian, T. W. Ma, F. L. G. Hsu, M. D. J. Sung, and G. W.
Plackmann, Flow of concentrated slurries: 2. Friction losses in bends,
ttings, valves and venturi meters, Int J Multiphas Flow 24 (1998),
243269.
[5] J. Telis-Romero, M. A. Polizelli, A. L. Gabas, and V. R. N. Telis,
Friction losses in valves and ttings for viscoplastic uids, Can J
Chem Eng 83 (2005), 181187.
[6] G. C. Gardner and P. H. Neller, Phase distributions ow of an air?
water mixture round bends and past obstructions, Proc Inst Mech
Engrs 184 (1969), 93101.
185
BIOGRAPHIES
Kumar Perumal obtained his PhD degree from
the University Institute of Chemical Technology
(UICT), Mumbai, India under the guidance of
Padmabhushan Prof. J.B. Joshi. Kumar is
currently an associate professor and associate
Dean for Teaching and Learning in the Faculty of
Engineering and Science, Curtin University,
Sarawak Campus, Malaysia. His research interests include Computational Fluid Dynamics
(CFD), Fluids and Thermal Engineering, Process Intensication Studies, and Engineering Education Research.