XXXI IAHR CONGRESS 4923

UNDERESTIMATION OF THE FRICTION FACTOR

IN A PIPELINE AND ITS INFLUENCE IN
AN ACCIDENT CAUSED BY WATERHAMMER

GABRIEL SOTO-CORTES 1 and DARIO GUAYCOCHEA-GUGLIELMI 2

1
Professor Researcher, Department of Energy, Universidad Autónoma Metropolitana
Azcapotzalco, Av. San Pablo 180, 02200 Azcapotzalco, Mexico City, Mexico
(Tel/Fax: +5255-53189065, e-mail: gsc@correo.azc.uam.mx )
2
Professor Researcher, Department of Energy, Universidad Autónoma Metropolitana
Azcapotzalco, Av. San Pablo 180, 02200 Azcapotzalco, Mexico City, Mexico
(Tel: +5255-53189003, Fax: +5255-53944831, e-mail: sacbi@correo.azc.uam.mx )

Abstract
The Nuevo Teapa-Venta de Carpio oil pipeline in Mexico registers head losses which
cannot be predicted by conventional formulas, or attributed to pipe age or inlay deposits. An
analysis, based on field measurements and supported on experimental and numerical concepts,
using internal flows criteria and optimization numerical fitting procedures, shows that the
friction factor is significantly influenced by dynamic effects. The underestimation of the
friction coefficient originates in turn an underestimation of the service pressure in the line.
The transients theory is applied to investigate an accident recently registered between two
pumping stations in the mentioned oil line. A model, supplied with field data, was used to
simulate this phenomenon. A comparison is made considering two hypotheses: 1) friction
factor calculated following conventional formulas (Coolebrook and White) and 2) friction
factor calibrated considering dynamic effects. The results show that high steady flow pressure
combined with the overpressure due to the transient, approaches the pipe breaking point in
certain sections. Although the case study refers to oil transportation, similar works were
performed for aqueducts, as some references show.

Keywords: Oil pipelines; Head losses; Dynamic effects; Waterhammer

1. INTRODUCTION
Dynamic effects caused by alignment changes in pipelines, produce consequences such as
vortexes, separation, secondary flows and changes in the velocity profiles, which in turn have
considerable influence in energy losses. This issue was reported by Idelchik (1994) and Miller
(1985); their works show that non-axial speed components produce alterations in the flow
pattern and increase energy dissipation. Carmona et al (2002) performed studies about similar
effects in a long aqueduct (diameter: 2.1 m and length: 42.6 km) with important alignment
changes. After an analysis based upon field data and numerical modeling, these authors report
evidence of a dynamic mechanism influencing energy dissipation. This effect on the friction
factor is sometimes even more important that roughness increase due to pipe aging. Similar
4924 September 11~16, 2005, Seoul, Korea

phenomena have been reported by El-Emam et al. (1997), Idelchik (1994), García &
Tzátchkov (1991) and Weiderhold (1966).
The Nuevo Teapa-Venta de Carpio oil pipeline in Mexico is one of the longest ones in the
country, with 480 km. Due to an irregular topography and to the necessity of transporting oil
from coastal areas to the central plateau, the pipeline has continuous vertical and horizontal
curves. Fig. 1 shows the line profile, in which it is possible to see the great altitude rise. The
oil pipeline consists of a system of two parallel conduits of 24 and 30 inches respectively
(0.5938 m and 0.7462 m), designed according to API standards. With five intermediate
pumping stations, the combined transportation capacity reaches about 550,000 barrels daily.
Table 1 shows the relevant stations data; the average pumping power is 30,000 HP.

2. DYNAMIC EFFECTS AND THE FRICTION COEFICIENT

Table 2 shows some field data and hydraulic calculations, regarding the operation for both
conduits (24 and 30 inches) in steady flow, for the section limited by pumping stations 2 and 3.
The formulas quoted in the table are:
Coolebrok & White formula
1
f = 2
(1)
  2.51 ε  
− 2 log + 0.269 
  Re f D 
Darcy Weisbach formula
π 2 gD 5
fm = hf (2)
8 LQ 2
In equations (1) and (2) Re is the Reynolds number, f and f m the friction coefficient,
ε the pipe absolute roughness, D the pipe diameter, Q the flow rate, L the length, h f
the friction losses and g acceleration due to gravity.
The operative routine for the pipeline is scheduled weekly using an empirical formula for
the calculation of the head losses. Its results are very similar to those ones obtained by the
Coolebrook and White or the Swamee Jain formulas.
The comparison between the theoretical and real friction factors is shown in the last line of
Table 2. The difference is important and can be explained as a result of a dynamic
phenomenon, associated to alignment changes and not to roughness increase or inlay deposits,
because the program for maintenance and internal inspection is efficient; paraffin scales are
removed every six months. If roughness variation is assumed as the only cause for this
difference, resultant values are not credible, as the same table shows. Moreover, as it is shown
in Table 2, dynamic variations in the friction coefficient are more important for partial flows
rates below the design flow rate. The authors reported a preliminary study, including the
analysis of field data along the whole line (Soto and Guaycochea, 2004).
 XXXI IAHR CONGRESS 4925

2.1 TRANSIENT EVENT TO ANALYSE

In the present paper, only the sector limited by Pumping Stations 2 and 3 is considered,
because the objective is to analyse the influence of the friction factor underestimation in a
recent accident occurred between those stations.
In fact, five pumps stopped at Station 3, once their engines ran out of gas. As a consequence,
pressure between those two stations increased and pumps at Station 2 were stopped, either
manually or automatically, causing the line to come out of service. The overpressure caused a
collapse in the 30 inches pipe, in the form of a longitudinal opening (1.51 × 0.45 m), at
kilometre 280.352 (Fig. 1). As a result, a considerable volume of crude oil was spilled and a
fire started, being controlled three hours later. Several factors coincided to provoke the
accident, other than the shortcoming in the system operation. The protection valves at the inlet
of Station 3 did not act effectively to relief the overpressure. On the other hand the pipe
thickness reduction due to corrosion, affected the breaking point of the material; in fact, the
30 inches conduit, in which the failure was registered, was constructed between 1979 and
1981 and started operations in this last year; the 24 inches conduit was initially designed as a
gas pipeline and started operations in 1994. However, another factor, that is the purpose of
this discussion, is the underestimation of the normal pressure acting in this particular section
of the pipeline. The real operative pressure (considering dynamic effects, Fig. 2), combined
with the overpressure due to the transient (Fig. 3), reached a value high enough to produce the
collapse; it is shown in Fig. 2 that for this condition the steady flow pressure in the section
close to Station 2 is higher than the maximum operation pressure; represented by the curve
called MAOP; this limit is used by the people in charge to schedule the pipeline operation. It
is important to point out that two other major accidents happened since the pipeline is
operating in sites which are very close to the presently referred accident.

3. UNSTEADY FLUID FLOW MODEL

Governing equations for transients flow are as follows:
The continuity equation
∂H a 2 ∂V
+ =0 (3)
∂t g ∂x
where a is the pressure wave speed, V the average velocity in the pipe, parallel to the x-
axis and H the hydraulic grade line or HGL.
From the Navier-Stokes equations neglecting the convective acceleration terms, since flow
perpendicular to pipe walls is approximately zero, flow can be considered one-dimensional,
for which the momentum equation reduces to:
∂V ∂H fV V
+γ + =0 (4)
∂t ∂x 2D
where f is the Darcy-Weisbach friction coefficient, D the inside diameter of the pipe, V
the velocity of fluid and γ the specific weight of the fluid.
Eqs. (3) and (4) are solved using the Method of Characteristic (MOC). For this problem,
specific gravity = 0.877, pressure wave speed = 968 m/s, vapor pressure = -0.098 MPa.
4926 September 11~16, 2005, Seoul, Korea

3.1 BOUNDARY AND INITIAL CONDITIONS

The hydraulic transient analyses consider:
• Sudden power failure in station 3.
• The power failure is assumed to be instantaneous and the check valve is allowed to
close without delay.
• Reverse spin not allowed.
• Pump bypass followed by a shut down in station 2.
As the system of two parallel conduits works as independent lines, the 30 inches conduit
was modeled according equation (3) and (4) considering field data (Table 2) for a flow rate of
198 x 103 bbl/d (0.3647 m3/s) for the following two hypothesis:
• Condition 1: f1 = 0.02549, no dynamic effects considered.
• Condition 2: f m = 0.08137, dynamic effects considered

4. RESULTS AND CONCLUSIONS

Friction coefficient calculated following conventional formulas, f1 , and that one calibrated
considering dynamic effects, f m , show important differences, specially for partial flow rates
(Table 2). If suction head at station 3 must remain invariant, for the transient analysis of this
work, f m f1 implies an underestimation of the dissipative phenomena, which require a
significant increment in the pump head at station 2. Fig. 2 compares both conditions for
steady state flow. It is important to say that this high steady flow pressure is not considered
for the oil pipeline operative routine. Additionally, the overpressure due to the transient
(section 2.1) is shown in Fig. 3; maximum head gets closer to MAOP line. From Fig. 3 there
are not doubts about a risk zone without protection equipment susceptible to collapse.
According to the accident conditions, the positive overpressure wave initially traveled from
station 3 to station 2, it could explain the breaking point position. For the exposed reasons, the
accident could have happened independently of the thickness reduction due to corrosion.

ACKNOWLEDGEMENTS
The financial support that made this work possible was provided by PEMEX Refinación
and Universidad Autónoma Metropolitana Azcapotzalco. The authors wish to acknowledge,
especially, the hold of Mario Ortega Pérez, General Superintendent, Coordinación de
Operaciones y Ductos, PEMEX Refinación. Also to Ricardo Cepeda, scholarship holder, for
his collaboration.

REFERENCES
Carmona-Paredes, R.B., Ortiz-Nuñez, L.A. and Sánchez-Huerta, A. (2002), “Efectos
dinámicos y temporales en la reducción de la capacidad de conducción en grandes
acueductos –estudio con medidas en prototipo”, Revista Ingeniería Hidráulica en México,
Vol. 17(2), pp.5-16 (in Spanish).
Chaudry, M.C. (1979), Applied hydraulic transients, Van Nostrand Reinhold Co, New York.
 XXXI IAHR CONGRESS 4927

El-Emam, N., Gad, F., Nafey, A. and Zoghaid, N. (1997), “New friction factor derived from
study of Egyptian gas-field pipelines”, Oil & Gas Journal, Vol. 10, pp. 72-78.
García, M.C. and Tzátchkov, V.G.. (1992), “Variación del coeficiente de cortante en tuberías
de agua potable en operación”, Proc. XII Congreso Nacional de Hidráulica, Puerto Vallarta,
México (in Spanish).
Idelchik, I. (1994), Handbook of hydraulic resistance, 3rd. edition, CRC Press, Inc., Boca
Raton, Florida.
Miller, D. (1985), Internal flow systems, BHRA, Fluid engineering series, vol. 5.
Soto, G. and Guaycochea, D. (2004), “Evidencia de efectos dinámicos en grandes oleoductos”,
Proc. XXI Latino-American Hydraulic Congress, San Pedro, Brasil (in Spanish).
Weiderhold, (1966). “Discussion: Factor influencing flow in large conduits”, Journal of
Hydraulics Division, ASCE, (HY4), pp. 168-218.
4928 September 11~16, 2005, Seoul, Korea

Fig. 1 Nuevo Teapa-Venta de Carpio oil pipeline profile and site of accident

Table 1. Pumping stations

Nuevo Teapa Station 1 Station 2 Station 3 Station 4 Station 5 San Mart?

Kilometer 0 156.30 267.78 307.15 340.80 351.79 480.45

altitude over the sea
level (m) 34 45 221 776 1371 1825 2250

Installed power (HP) 29000 30 000 30 000 35 000 20 000 35 000 35 000

Table 2. Field data and friction factor calculations for the sector between stations 2 and 3
(kilometres 267.78 to 307.15)
24 " Conduit (internal 30 " Conduit (internal
_diameter: D = 0.5938 m) _diameter: D = 0.7462 m) Remarks
Event 1 2 3 4 1 2 3 4
pressure at St. 2 p2 (kPa) 6668 6080 6668 6472 6668 6080 6668 6472 measured
pressure at St. 3 p3 (kPa) 588 686 686 883 588 686 686 883 measured
head loss hf 151.94 72.13 140.54 94.93 151.94 72.13 140.54 94.93 (z2+p2/γ) - (z3+p3/γ)
flow rate Q (103 barrels/day) 105.29 141.21 156.64 164.87 198.21 294.67 293.27 342.62 measured
flow rate Q (m3/s) 0.19373 0.25983 0.28822 0.30336 0.36471 0.54219 0.53962 0.63042
flow speed V (m/s) 0.70 0.94 1.04 1.10 0.83 1.24 1.23 1.44
Reynolds number Re 29672 39795 44143 46462 44450 66081 65767 76835 calculated
relative roughness ε/D 0.00168 0.00168 0.00168 0.00168 0.00134 0.00134 0.00134 0.00134
friction factor f1 0.0277 0.0267 0.0264 0.0262 0.0255 0.0243 0.0244 0.0240 calculated with form.(1)
friction factor fm 0.0918 0.0242 0.0384 0.0234 0.0812 0.0174 0.0343 0.0170 calculated with form.(2)
due relative roughness ε/D 0.08168 0.00076 0.00926 0.00067 0.065 0.0006 0.00737 0.00054 obtained from form.(2)
due absolute roughness ε 0.0485 0.00045 0.0055 0.0004 0.0485 0.00045 0.0055 0.0004 _with f = fm
relation fm/f1 3.3118 0.9085 1.4565 0.8933 3.1878 0.7166 1.4090 0.7081

General data: Oil specific gravity s: 0.877, Oil specific weight γ : 8.6 kN/m3, Oil kinematic viscosity ν : 1.4 × 10-5 m2/s, Sector

length L : 39,380 m, Static head (st. 2 altitude z2 – st. 3 altitude z3 ): H s = - 555 m, Pipe absolute roughness ε : 0.001 m
 XXXI IAHR CONGRESS 4929

1400

Flow 198 103 bbl/d

1200
M.A.O.P

1000
HGL Condition 2

HGL Condition 1
800

600

Pipe elevation
400

KM
200
265 270 275 280 285 290 295 300 305 310

Station 2 Station 3

Fig. 2 Increasing of pump head due to dynamic effects. Condition 1: f = 0.02549, no dynamic
effects considered. Condition 2: f = 0.08137, dynamic effects considered. Maximum
operation pressure (MAOP) for a new pipe according to ASME B31.8.

MAOP
Max. head condition 2
Max. head condition 1

HGL Condition 1

Risk zone Pipe elevation

Flow 198 103 bbl/d

Station 2 280+352 Station 3

Fig. 3 Maximum transient head and risk zone. Condition 1: f = 0.02549, no dynamic effects
considered. Condition 2: f = 0.08137, dynamic effects considered.