23

Minor Loss Coefficients

for Storm Drain Modeling with SWMM

William H. Frost

In the 1990s, as one of the updates to version 4.0 of SWMM, an EXTRAN

routine was included which allowed the modeler to include input parameters
on the C1 card for calculating minor losses in flow transitions. The ability to
model these types of losses explicitly, and to track them as part of the input
file, is an important addition to EXTRAN's capabilities. Particularly in
cases where there is no flow data for the storm drain network being
modeled, it is important to try to represent all the physical processes that can
have an effect on flows and water quality.
The goal of this chapter is to review commonly used loss coefficients to
determine their applicability for modeling storm drains in EXTRAN. The
literature review found that many of the coefficients in use were derived
from experimental data unrelated to the type of flows found in storm drains.
There [Link] more information for pressure flow scenarios than for free
surface flow. There was also more information for transitions in pipes than
for transitions through junctions such as manholes.
In reviewing the original experimental research used to derive these
coefficients, the chapter identifies several methods of estimating them which
should be more applicable to storm drain modeling. The chapter also
recommends that researchers conduct new experimental studies with the
goal of developing loss coefficients that can be used for better estimates of
junction losses in pipes with free surface flow.
Frost, W.H. 2006. "Minor Loss Coefficients for Storm Drain Modeling with SWMM." Journal of Water
Management Modeling R225-23. doi: 10.14796/JWMM.R225-23.
© CHI 2006 [Link] ISSN: 2292-6062 (Formerly in Intelligent Modeling of Urban Water
Systems. ISBN: 0-9736716-2-9)

517
518 Minor Loss Coefficients

23.1 Introduction
In the 1990s, as one of the updates to version 4.0 of SWMM, an EXTRAN
routine was included which allowed the modeler to include input parameters
on the C1 card for calculating minor losses in flow transitions. The
variables are KENTR, KEXIT, and KOTHER, which represent the
coefficients for entrance, exit, and other losses (e.g. expansion, contraction,
bends, or valves). The minor loss equation in EXTRAN takes the form of:

v2
hL = K (23.1)
2g
where:
hL = the head loss
K = a loss coefficient
v = the velocity in the conduit, and
g = the gravitational constant

The ability to model these types of losses explicitly, and to track them as
part of the input file, is an important addition to EXTRAN's capabilities.
Particularly in cases where there is no flow data for the storm drain network
being modeled, it is important to try to represent all the physical processes
that can have an effect on flows and water quality. Minor losses (head
losses from flows through junctions and bends) can be significant. UWRRC
(1992) said that
In long conduits where L/D >> 1000 these local losses are
usually very small in comparison to the friction losses
and…can be neglected. However, if the channel or
conduit is very short and/or there are a number of
manholes, changes in direction, junctions, or changes in
pipe size, then the sum of these losses can exceed the
friction loss.

23.1.1 Objective

The goal of this chapter is to develop loss coefficients for use in EXTRAN
to model minor losses for storm drain design and analysis based on original
experimental research. The method of finding loss coefficients should be
Minor Loss Coefficients 519

simple to use, and where simplifications are made from the research, they
should result in only minor variations from the more accurate calculations.

23.1.2 Modeling versus Design

There are differences in the process of head loss calculations between

modeling and design. First, the modeler may not take into account every
pipe and junction, but instead folds several short pipes into one longer
equivalent pipe to improve model stability. A designer, on the other hand,
must calculate the HGL for every part of the network to be built. Similarly,
a SWMM model typically will ignore inlets and small branches, lumping
them into RUNOFF calculations, while inlet and branches are a key part of
storm drain design.
A second difference is in the type of flow. Designers typically work with
a single peak flow value and design a closed storm drain system for full flow
conditions. In a modeling situation, particularly with continuous models, a
wide range of flows may occur. A conduit may operate in open channel, full
flow, or pressure flow condition. Unlike the design situation, it is not
possible for a modeler to predict in advance what flow regime will occur.
For this reason, the head loss equation and coefficients should take into
account the flow regime.

23.1.3 Conditions to be Modeled

Table 23.1 shows the conditions to be modeled, described in the EXTRAN

manual:

Table 23.1 Minor losses modeled in Extran.

Variable Condition Description

KENTR Entrance flows entering a conduit from a larger waterbody
KEXIT Exit flows exiting a conduit to a larger waterbody
KOTHER Expansion transition from a smaller to larger conduit
Contraction transition from a larger to smaller conduit
Bends change in flow direction
Valves gate valve or other flow control device
Inlet flows entering the conduit from an inlet
520 Minor Loss Coefficients

Some of these losses are unlikely to be included in a model. Individual

inlets and their losses, for example, are frequently assumed in the RUNOFF
block and not modeled explicitly. Some may only be modeled occasionally.
Most minor losses in a closed storm drain system will take place in a
junction, usually a manhole structure, where the flow enters through one or
more pipes and exits through a single conduit. This situation is more
complex than either the entrance and exit situation where flow is entering or
leaving a pipe network or a simple expansion or contraction in pipe size.
The key to modeling losses in a storm drain network is understanding
and developing computations for losses in junctions. These, in turn, will
need to be modified so that they can be entered into EXTRAN as entrance,
exit, and other losses for each conduit.
While there is some research on junction losses, most has been done for
pressure flow situations. There has been much less study done on the
conditions of flow in pipes flowing partially full, under free surface
conditions. As shown in Table 23.2, these are the situations that are most
frequently modeled for storm drain analysis.

Table 23.2 Types of minor losses to be found in storm drain models.

Frequently Occasionally Rarely

Type of Loss Modeled Modeled Modeled

Pipes (Full or Partially Full)

Entrance X
Exit X
Expansion and Contraction X
Inlet on branch X
Curves or bends X
Outfall X
Junctions (Full or Partially Full)
Flow through junction X
Bend within junction X
Junction with lateral X
Junction with inlet X
Channels
Expansion and Contractions X
Curves or bends X X
Culvert entrance X
Culvert exit X
Outfall X
Minor Loss Coefficients 521

23.2 Literature Review

The literature review had two goals. The first was to review readily
available design criteria, textbooks, handbooks, and manuals to identify
commonly accepted engineering practice. These included the following list.
Journal articles were reviewed to identify methods and criteria for
calculating minor losses, but these were not found to be in general use in
engineering practice.

Design Criteria:
Highway Drainage Manual (MSHA 1981)
Virginia Drainage Manual (VDOT 1988)
Urban Storm Drainage Criteria Manual (UDFCD 1969)

Handbooks
Handbook of Hydraulics (King and Brater, 1963)
Standard Handbook for Civil Engineers (Merritt, 1983)
Design of Small Dams (USBR, 1977)
Modern Sewer Design (AISI, 1980)
Design and construction of urban stormwater management systems
(UWRRC 1992)
Handbook of Applied Hydraulics (Davis, 1954)

Software Manuals
HEC-RAS
HYDRAIN
HEC-22

Textbooks
Chow, 1959
Henderson, 1966
French, 1985
Daugherty, Franzini, and Finnemore, 1985
Streeter and Wylie, 1975
Linsley and Franzini, 1975

Several problems were found with the methods and data used in the
design criteria manuals, software manuals, and handbooks listed above:
522 Minor Loss Coefficients

1. There are few commonly accepted values or calculation methods.

No two sources approached every situation in the same way.
2. Within a single reference, there may be two or more methods
presented to calculate one type of loss. Frequently these methods
do not give the same result.
3. Many of the design criteria do not have explicitly sourced ties to
research verifying the data, so there is no way to check its validity.
4. Some of the criteria and data which can be tied to original research
have been extrapolated to design situations that may be
inappropriate.
5. While there is research and a commonly used standard for pressure
flow in pipes and junctions, culvert entrance and exit losses, and
open channel flow in channels, there remains almost no easily used
research results on losses in junctions.
6. There is no standard method or coefficients for modeling losses
through junctions for either pressure flow or open channel flow.
The attempt to identify commonly used loss coefficients led in turn to the
second goal of the literature review: further examination of the references in
these works, sometimes going back to the original experimental research
underlying many of the methods.
As recently as 1994, one researcher commented, “A literature search
showed that junction loss data for the range of flow conditions required in
the design were either non-existing or so uncertain that they were useless.”
(Serre et al., 1994)
Lack of good junction data is a significant issue, although there were a
number of studies found. In the last 50 years, many researchers have
investigated different aspects of the problem. The most thorough
experimentation remains that of Sangster et al (1958). Their results,
however, were expressed as design charts for determining pressure losses,
which are not directly useful for SWMM modeling requirements.

23.3 Loss Coefficients and Calculation Methods

One finding of this research was that there are a number of ways to calculate
minor losses. Most are based on a factor applied to some variation of the
velocity head at the junction of two pipes. These variations include the
difference between upstream and downstream velocity heads:
Minor Loss Coefficients 523

2 2
Vu - Vd
(23.2)
2g
the upstream head alone

2
Vu
(23.3)
2g
or the downstream head alone

2
Vd
23.4)
2g

In at least one case (King and Brater, 1963), tables of loss coefficients are
keyed to the velocity head in the smaller pipe: upstream for expansions and
downstream for contractions.
Several researchers (such as Sangster, et al, 1958) have made a
distinction between calculation of head (energy) losses and calculation of
pressure (HGL) losses. A pressure coefficient applied to the velocity
through a junction outlet will provide direct calculations of the hydraulic
grade line (HGL) in the upstream pipes, taking into account differences in
velocity and water surface elevation changes caused by different pipe
diameters. These coefficients are not suitable for EXTRAN, which
calculates the HGL internally. What is desired is an energy loss coefficient,
which describes only the head loss through the junction, without reference to
different pipe sizes.
It is relatively straightforward to convert pressure change coefficients to
energy change coefficients. Marsalek (1985) derived an equation for
circular pipes from definitions of coefficients and junction geometry to make
this conversion:
4
D 
K p = K e + 1 −  u  (23.5)
 Dd 
Conversion of the pressure change coefficient to an energy change
coefficient is done with a minor manipulation of this equation:
4
D 
K e = K p +  u  − 1 (23.6)
 Dd 
524 Minor Loss Coefficients

For pressure flow, it is also relatively straightforward to convert

coefficients applied to upstream velocity head to downstream loss
coefficients. In this case, the relative size of the pipes is used, along with
continuity, to show

4
K u  Du 
≈  (23.7)
K d  Dd 

These conversions will be used against published loss coefficients and

research findings throughout this chapter to normalize all the data to a single
format. The most convenient to use will be a single coefficient applied to
the velocity head exiting the junction. Since EXTRAN applies losses to
conduits, not junctions this approach could be used by applying all the
junction loss coefficients as a single factor on the downstream conduit.

23.4 Coefficients for Pressure Flow in Pipes

Minor loss calculations and coefficients for pressure flow are reported in
many sources – several can be traced back to the same original research.
Much of this research dates back to the early 1900s, when estimates of
losses in water distribution systems motivated the studies. These are the
coefficients most often shown in the design criteria and handbooks.

23.4.1 Entrance Losses

Entrance losses deal with the situation where water enters a submerged pipe
from a reservoir. For this situation, the equation relating losses to
downstream velocity is used:
2
Vd
hL = K (23.8)
2g

Entrance loss coefficients are based on the shape of the entrance pipe, as
shown in Table 23.2 (Task Force, 1965). These, or similar coefficients, have
been widely reproduced (Merritt, 1983, Daugherty et al 1985, King and
Brater, 1963). The equation where a free flowing stream enters a pipe is
Minor Loss Coefficients 525

similar to that of a culvert entrance. The coefficients for this situation are
given in a later section.

Table 23.3 Entrance loss coefficients for pressure flow .

Entrance Type K
Inward Projecting 0.80
Sharp-Cornered 0.50
Slightly Rounded 0.25
Bell-Mouthed 0.05

23.4.2 Exit Losses

Exit losses occur when the pipe system discharges into a lake or other
receiving water which is large enough so it is essentially still with respect to
the discharge. In this case, the entire velocity head is dissipated, and the loss
coefficient is 1.0, applied to the velocity of the upstream pipe. The situation
where a pipe discharges to a free-flowing stream will be considered in the
section on free-surface flow.

23.4.3 Transitions

Transitions consist of expansions and contractions in the size of the pipe.

Estimating losses for transitions in pressure flow was one of the earliest fluid
mechanics problems.

Expansion
The theoretical value for head loss through a sudden enlargement can be
derived from the energy and momentum equations:

(V1 − V2 )2
hL = (23.9)
2g

Expressed as a coefficient on the downstream velocity, this becomes

(Vd )2
hL = K (23.10)
2g
where:
526 Minor Loss Coefficients

2
 D 
2

K =  d  − 1 (23.11)
 Du  

One of the most widely used procedures for calculating losses in pressure
flow sudden expansion transitions is described in King and Brater (1963),
whose tables have been reproduced in several texts and design manuals
including HEC-22, Linsley and Frazini (1975) and UWRRC (1992),
showing the loss based on velocity in the upstream pipe.
The tables were created by Horace King (King and Brater, 1963) using
an equation for these types of losses derived from experimental results by
Archer (1913).
(V1 − V2 )1.92
hL = 1.1 (23.12)
2g

The tables give head loss, hL, and K based on the ratio of upstream and
downstream pipe diameters. The data can’t be applied directly in EXTRAN
because the exponent of the velocity component is not 2.0.
As an alternative, the theoretical equation gives results reasonably close
to the tables derived from Archer’s equation. For the likely range of storm
drain transitions (Dd/Du < 2.0) the theoretical curve underestimates Archer’s
values for K by more than 10 percent at velocities under 4 fps and
overestimates the values for velocities by more than 10% at velocities over
15 fps. If the modeler anticipates high velocity flows, reducing the
theoretical value of K by a factor of 0.9 gives a reasonable approximation to
the tabular values.

Contraction
The theoretical value for head loss in sudden contractions has also been
determined in the early 1900s by analysis to be

2
 1  V 2
hL =  − 1 d (23.13)
 Cc  2g

which results in a loss coefficient of

Minor Loss Coefficients 527
2
 1 
K =  − 1 (23.14)
 Cc 

The contraction coefficient Cc has been experimentally determined by

several researchers beginning in the late 1800s. An equation for it was
derived and given in King and Brater (1963). The results match the
experimental data fairly closely for higher velocities.

0.0418 (23.15)
C c = 0.582 +
 D 
1.1 − d 
 Du 

King derived similar tables for head loss factors for contraction as those
for expansion by smoothing the graphed results of several researchers’ data,
with the loss coefficient varying based on both the ratio of pipe sizes and
flow velocity. These, like the tables for expansions, have been widely
reproduced, including in AISI (1980) and UWRRC (1992).
Fluid mechanics texts and handbooks are another source of these factors.
Daugherty (1985), Merritt (1983) and Linsley and Frazini (1955) provide
information which can be used to derive head loss factors or contraction
coefficients based on the ratio of upstream to downstream diameters or
cross-sectional area. These match reasonably closely withKing and Brater’s
table for V = 20 cfs, as shown in Table 23.4. For pressure flow in storm
drains, theKing and Brater values are recommended.

Table 23.4 Head loss coefficients for contraction in pressure flow.

Du/Dd 1.1 1.2 1.25 1.4 1.6 1.8 2.0 2.2 2.5 3.0
Daugherty .06 .15 .22 .28 .33 .36
Linsley .10 .30 .40
Merritt .06 .10 .22 .27 .32 .34 .38
King .05 .09 .18 .25 .31 .33 .35 .37 .39

Bends
Similarly to closed conduit transitions, losses in bends have been studied
extensively. Most of the researchers have focused on small pipes with bends
of 90°, suitable for calculating water distribution or plumbing losses. Losses
528 Minor Loss Coefficients

were found to vary depending on the ratio of the bend radius to the pipe
diameter, R/d. Two sources of original research were identified: Crane
(1965) and Anderson and Straub (1948).
A few researchers have worked with these data to apply them to large
pressure conduits, among them King and Brater (1963), USBR (1977), and
Creager and Justin (1950). USBR (1977) and Creager and Justin (1950)
have both analyzed available research and have come up with substantially
the same results. The King and USBR data area both based on the
Anderson/Straub (1948) results.

Table 23.5 Loss coefficients for 90° bends.

Anderson/ King /
R/d Crane Straub Brater USBR
1 .50 .23 .23 .23
2 .30 .14 .13 .13
4 .25 .09 .08 .09
6 .15 .08 .08 .07
8 .15 .08 .08 .07

Smaller bend angles have proportionately smaller loss coefficients.

Again, these have been reported frequently with good agreement in a
number of handbooks, including King and Brater (1963), USBR (1977), and
Creager and Justin (1950). King and Brater (1963) reported two sets of
these factors, one by Fuller, which was an approximation, and one by
USBR. The USBR factors, derived from a study of large conduits, are
probably the most suitable for use in calculating storm drain losses.
Hinds (1928) cited in Merritt (1983), recommended adjusting the losses
for other bend angles with the following equation. Results are shown in the
last column of Table 23.6.

0.5
α 
K =  (23.16)
 90 

This equation matches the USBR curve closely for angles between 45o
and 90o but will overestimate losses for more shallow angles.
Minor Loss Coefficients 529

Table 23.6: Loss coefficient reduction factors for other bend angles.

Angle USBR Fuller Creager and (a/90)0.5

Justin
22.5 .42 .50 .45 .50
45 .70 .75 .70 .71
60 .83 .85 .82
90 1.00 1.00 1.00 1.00

23.5 Coefficients for Free Surface Flow in Open

Channels
23.5.1 Transitions in Size or Shape

Two equations for entrance and exit losses were widely reported:

for flow contraction:

2 2
V − Vd
hL = 0.1 u (23.17)
2g

for flow expansion:

2 2
V − Vd
hL = 0.2 u (23.18)
2g

These were cited in ASCE (1961), UDFCD (1969), and AISI (1980),
where they were attributed to Hinds (1928). They were also cited in French
(1985), attributed to Chow (1959).
For rectangular channel transitions, the recommended angle of the
transition section is 12.5 degrees, cited in ASCE (1961), Daugherty (1985),
and French (1985).
All of these references eventually can be traced back to the work of
Hinds (1928), which was a study of transitions in trapezoidal and rectangular
open channels. They are based on well-designed transitions in flumes,
which are smoother than typical storm drain transitions.
530 Minor Loss Coefficients

Loss coefficients for rectangular open channels were subsequently

developed by Chow which vary depending on the shape of the transition
structure as shown in Table 23.7. These appear to be the best researched
coefficients for transitions in man-made channels.
All of the coefficients in this table are based on the differential in
velocity head and thus can’t be applied directly to the downstream flow
only. The equation can be normalized using the cross-sectional area for each
section, with the continuity equation used to equate Q at both sections:
2 2
Vu − Vd
hL = K (23.19)
2g

A 2 V 2
hL = K  d 2 − 1 d (23.20)
 Au  2g

Table 23.7 Transition loss coefficients for rectangular open channels.

Type of transition Inlet Outlet

Warped 0.10 0.20
Cylinder-quadrant 0.15 0.25
Simplified straight line 0.20 0.30
Straight line 0.30 0.50
Square-ended 0.30+ 0.75

23.5.2 Curves or Bends

Unlike the situation for calculating transition losses, there do not appear to
be any agreed upon equations and coefficients for bend losses in open
channel flow. Handbooks and design criteria use different methods, usually
without attributing any research or studies to back them up. Where there is
concurrence, it us usually because the coefficients for pressure flow have
been used.
There have been several studies of bend losses in open channels,
however. The most thorough analysis, and the best source of energy loss
coefficients is a study by Shukry (1950) whose paper investigated the
Minor Loss Coefficients 531

characteristics of spiral flow in bends in rectangular channels. Shukry found

that bend losses were a function of the following variables:
Re Reynold’s Number
θ Angle of curvature
R/b Ratio between curve radius, R, and channel width, b
Y/b Ratio between flow depth, y, and channel width, b
The routines in EXTRAN do not currently allow for a loss coefficient
that changes dynamically with velocity (represented by Re) or depth, so a
simplified method of estimating a loss coefficient from Shukry’s work
should begin with θ and R/b, both of which are functions of channel
geometry, then check to see if differences from velocity and depth are
significant.
Shukry’s recommendation was to fix Re, find R/b for the section, find K
using the curves he had developed, then make corrections for the curve angle
and depth. Since his findings showed that losses for high bend ratios are
negligible at any angle, this approach appears to be sound. Shukry presented
his results in a series of charts comparing two of the variables against the
loss coefficient in a family of curves, with measured data points shown.

Derivation of Bend Loss Coefficients from Shukry’s Study

The analysis which follows is based on the data points alone, read from the
charts. Points in bold shown in the tables that follow were measured by
Shukry, the others have been interpolated. Using data from the chart with
Re fixed at 31,500, the following measurements and interpolations were
made, shown in Tables 23.8 – 23.11. Review of Shukry’s results shows that
in typical open channel systems, bend losses are negligible when the curve
angle is less than 45 o and the R/b ratio is greater than 3.0.

Table 23.8: Shukry Data for y/b = 1.20

R/b
θ/180 0.50 1.00 2.00 3.00
1.00 .82 .25 .07 .01
.75 .75 .23 .06 .01
.50 .72 .20 .08 .01
.25 .07 .02 .01 .00
.125 .00 .00 .00 .00
532 Minor Loss Coefficients

Table 23.9 Shukry Data for y/b = 1.00.

R/b
θ/180 0.50 1.00 2.00 3.00
1.00 .95 .29 .08 .01
.75 .88 .27 .07 .01
.50 .90 .22 .04 .01
.25 .10 .03 .01 .00
.125 .00 .00 .00 .00

Table 23.10 Shukry Data for y/b = 0.80.

R/b
θ/180 0.50 1.00 2.00 3.00
1.00 1.11 .34 .09 .01
.75 1.05 .32 .09 .01
.50 .95 .29 .08 .01
.25 .13 .04 .01 .00
.125 .00 .00 .00 .00

Table 23.11: Shukry Data for y/b = 0.60

R/b
θ/180 0.50 1.00 2.00 3.00
1.00 1.28 .39 .11 .01
.75 1.21 .37 .10 .01
.50 1.10 .32 .09 .01
.25 .16 .05 .01 .00
.125 .00 .00 .00 .00

To use the experimental results most effectively in EXTRAN, it would

be ideal to develop an equation relating the loss coefficient to channel
geometry. There are no simple equations that reproduce all the values in the
table. Boundaries were placed on the applicability of the equations based on
typical situations for storm drains. Simplifying assumptions were also
made. First, because EXTRAN does not vary the head loss with depth, an
assumption was made that no change would be made based on the variable
y/b. For the equation to be developed, y/b was assumed to be 0.60, which
gives the highest losses. The assumptions include:
Minor Loss Coefficients 533

r/B > 1.0

Angle between 45 and 90 degrees
y/b = 0.60
Following the procedure used by USBR (1977), we first derive an
equation relating the loss coefficient to the r/B ratio for a bend angle of
90 degrees. There is no simple form which reproduces the values exactly.
A reasonable approximation can be made with the following equation, for
values of r/b > 1.0:
r
−1.2
K1 = e B
(23.21)

The result of this equation compared to Shukry’s data is shown in the

following table:

Table 23.12: Bend loss coefficients derived from Shukry's data.

R/b
y/b 1.00 2.00 3.00
1.2 .20 .08 .01
1.0 .22 .04 .01
0.8 .29 .08 .01
0.6 .32 .09 .01
Avg .26 .07 .01
Avg 0.6 and 0.8 .31 .09 .01
Eqn .30 .09 .03

The results of the equation are reasonably close to the average of the
losses at all depths, and quite close to the losses for the lower depth ratios
most likely to be found in drainage systems. They are too high for the r/b
ratio of 3.0, but drop off at higher ratios, just as Shukry’s data did.
The next step with Shukry’s data is to make an adjustment for the bend
angle. Again, this is done by making an approximation based on the
measured data. There are three data points in the range commonly found in
storm drains: 22.5, 45 and 90 degrees. There are three data points for
extreme bends where the flow doubles back: 90, 135, and 180 degrees.
When graphed, there is a distinct break point at 90 degrees, with a linear
relationship on either side. There is also a distinct breakpoint for angles less
than 45 degrees.
534 Minor Loss Coefficients

For angles between 22.5 and 45 degrees, the following equation is a

linear approximation:
K 2 = 0.005θ − 0.11 (23.22)

For angles between 45 and 90 degrees, the following equation is a linear

approximation:
K 2 = 0.019θ − 0.71 (23.23)

For angles larger than 90 degrees, this equation is a linear approximation:

K 2 = 0.003θ + 0.77 (23.24)

Figure 23.1 shows a comparison between Shukry’s data and these
equations for y/b = 0.6.
0.35
Eqn r/B = 1.0

0.30 Eqn r/B = 2.0

Eqn r/B = 3.0

Shukry r/B = 1.0

0.25
Shukry r/B = 2.0

Shukry r/B = 3.0

0.20
K
0.15

0.10

0.05

0.00
0 10 20 30 40 50 60 70 80 90 100
Angle

Figure 23.1 Comparison of derived equation with Shukry’s data.

23.5.3 Culvert Entrance and Exit

The accepted standard for losses at culverts is FHWA (1985): Hydraulic

design of highway culverts which is an update of US Bureau of Public Roads
or Federal Highway Administration documents for culvert design and
calculation of losses.
Minor Loss Coefficients 535

Table 23.13 Culvert loss coefficients (USBPR, 1961).

Type of Structure and Design of Entrance ke

Pipe, Concrete
Mitered to conform to fill slope 0.7
* End-Section conforming to fill slope 0.5
Projecting from fill, sq cut end 0.5
Headwall or headwall and wingwalls
Square-edge 0.5
Rounded (radius = 1/12D) 0.2
Socket end of pipe (groove-end) 0.2
Projecting from fill, socket end (groove-end) 0.2
Beveled edges, 33.7E or 45E bevels 0.2
Side- or slope-tapered inlet 0.2
Pipe, or Pipe-Arch, Corrugated Metal
Projecting from fill (no headwall) 0.9
Mitered to conform to fill slope, paved or unpaved slope 0.7
Headwall or headwall and wingwalls square-edge 0.5
* End-Section conforming to fill slope 0.5
Beveled edges, 33.7E or 45E bevels 0.2
Side- or slope-tapered inlet 0.2
Box, Reinforced Concrete
Wingwalls parallel (extension of sides)
Square-edged at crown 0.7
Wingwalls at 10E to 25E or 30E to 75E to barrel
Square-edged at crown 0.5
Headwall parallel to embankment (no wingwalls)
Square-edged on 3 edges 0.5
Rounded on 3 edges to radius of 1/12 barrel
dimension, or beveled edges on 3 sides 0.2
Wingwalls at 30E to 75E to barrel
Crown edge rounded to radius of 1/12 barrel
dimension, or beveled top edge 0.2
Side- or slope-tapered inlet 0.2
536 Minor Loss Coefficients

Virtually all of the handbooks and design criteria reviewed use these
coefficients for culvert losses. Table 23.13 shows values for various end
sections and pipe materials.
This publication makes the assumption that the exit loss coefficient is
1.0, meaning that all the velocity head is lost. This is appropriate for
discharges to ponds or lakes, but if the culvert discharges to a free flowing
stream or channel, the modeler should consider using the coefficients for
transition losses instead.
The minimum loss coefficient that should be used is 0.30, which is
Chow’s value for simplified straight-line transitions. It should be adjusted
for change in cross-sectional area as discussed earlier.

23.6 Coefficients for Pressure Flow in Junctions

Though there are many types of junctions with many combinations of pipes
entering and leaving, researchers have concentrated on a subset of those that
meet in a T, listed in order of complexity. Research which has addressed
these aspects of junction flow is summarized in Table 23.14.
• straight flow through a junction,
• flow turning 90° in a junction,
• lateral on a through main with different proportions of flow in
the lateral, and
• opposed laterals.
•

Table 23.14 Research results reported for junction losses (pressure flow)

Townsend
Johns ton
Marsalek
Sangs ter

Pedersen

Blaisdell

Serre
Hare

Straight through x x x x
90o turn x x x
Lateral on through main x x x x x x
Opposed laterals x x
Manhole benching x x x x
Inlet x
Minor Loss Coefficients 537

Most researchers have concentrated on methods of reducing losses

through benching or shaping the invert to guide the flow and developed
calculation procedures as a secondary goal to their studies. Several have
conducted a dimensional analysis to identify the independent variables
governing junction losses.
Between this analysis and the experimental results, a consensus can be
seen on which independent variables have the most effect on junction losses.
During experimental runs, many flow and geometric parameters were
varied; however, not all researchers varied all the parameters or the same
ones.
All found that there is no single loss coefficient that applies to every flow
entering and exiting a junction. Instead, junction geometry plays an
important role and K will vary depending on the size and configuration of
the pipes and junction. The most important variables are:
• ratio of junction width to the downstream pipe diameter (b/Dd),
• ratio of upstream and downstream pipe diameters (Du/Dd), and
• benching or invert shaping.

23.6.1 Straight Flow through a Junction

Four studies were found which developed experimental measurements of

head loss in straight-through manholes (two-pipe junctions with neither
lateral inflows nor bends). All three independent variables were studied;
however, none of the studies varied all three. Variations of Du/Dd and b/Dd
were studied by Sangster et al. (1958), while b/Dd and invert benching were
varied in research by Marsalek (1984), Johnston and Volker (1990) and
Pedersen and Mark (1990).
Sangster et al. presented results in two forms. The first was a set of
experimental results from 15 tests varying Du/Dd and b/Dd with one curve
drawn to fit the data, and second was as a set of curves to be used for design.
Table 23.15 shows these values. The design column (1) uses data for b/Dd =
1.0. For comparison, values for sudden expansion and contraction are
shown in column (3), taken fromKing and Brater’s tables for V = 15.0 fps.
This velocity was chosen to match the Froude numbers of Sangster’s study
at typical diameters for storm drain systems; 24” to 108”.
These results are shown two ways in the table. Columns (1) (2) and (3)
show the raw data. Columns (4) (5) and (6) show the coefficients
normalized as energy loss coefficients applied to the downstream velocity
head. The values become more similar as the difference between pipe sizes
538 Minor Loss Coefficients

is greater, indicating that manholes with a large disparity in pipe sizes act
somewhat like sudden expansions and contractions. One can also note that
Sangster’s design data are very close to the fitted experimental curve.

Table 23.15 Variation of upstream and downstream pipe diameter.

Du/Dd Dd/Du (1) (2) (3) (4) (5) (6) (7)

2.50 0.4 1.4 0.38 0.43 0.43
2.00 0.5 1.3 0.34 0.36 0.34
1.67 0.6 1.2 0.27 0.33 0.27
1.43 0.7 1.00 1.05 0.19 0.24 0.29 0.19
1.25 0.8 0.80 0.83 0.11 0.21 0.24 0.11
1.11 0.9 0.45 0.44 0.04 0.16 0.15 0.04
1.00 1.0 0.00 0.07 0.00 0.00 0.07 0.00 0.20
0.9 1.11 -0.45 -0.44 0.05 0.07 0.08 0.08 0.32
0.8 1.25 -1.10 -1.10 0.12 0.34 0.34 0.19 0.49
0.7 1.43 -2.10 -2.10 0.25 1.08 1.08 1.04 1.23
0.6 1.67 -3.60 -3.55 0.37 3.12 3.07 2.87
0.5 2.00 -6.00 0.51 9.00 8.16
(1) Kp, Sangster, et al., fitted curve to experimental data
(2) Kp, Sangster, et al., design charts
(3) Ke, King and Brater Tables 6-7 and 6-10 for v=15.0 fps’
(4) Ke for velocity head, Sangster, et al., fitted curve
(5) Ke for velocity head, Sangster, et al., design charts
(6) Ke, King and Brater, for velocity head in the downstream pipe
(7) Ke, for velocity head, Hare,

Ratio of upstream and downstream pipe diameters

For a single b/Dd, where b/Dd = 1, there are two sources of data for
determining the effect of Du/Dd on the loss coefficient: Sangster et al (1958)
and King and Brater (1963). The latter data are for sudden expansions and
contractions not associated with junctions, however. Hare (1983) reported
similar results for b/Dd = 2. Table 23.15 shows these results. The first
columns show the data reported in the research, and the last columns show
the data normalized as energy loss coefficients applied to the velocity head
in the downstream pipe. The data from Sangster et al (1958) track closely
with the equation developed by King. The Hare data are somewhat higher,
but include potential losses from the differing manhole geometry.
The similarity between the equation developed by King and Sangster's
experimental results is relatively close for contractions of less than 0.7 and
Minor Loss Coefficients 539

expansions of greater than 1.25. The use of Archer's equation in this range
seems appropriate for transitions in junctions.
In the narrower range typical of most storm drain junctions, the Archer
equation gives loss coefficient values lower than the experimental results.
The results of the comparison show that the values developed by King
for expansion and contraction in closed pipes (column 3) is a good
approximation for the losses attributable only to the change in pipe
diameters in flow through a junction.

Ratio of junction width to the downstream pipe diameter

Effects of junction width (b/Dd) for Du/Dd = 1.0 were studied by all the
researchers cited and all found the same basic linear relationship between
b/Dd and K with a shape factor, α, which is an empirical factor based on
experimental measurements.
Review of Sangster’s design curves show that junction width effects can
be expressed as additions to the effects of variations in pipe size
(Table 23.16), with α varying somewhat with the ratio of pipe sizes, again
trending lower as pipe size disparity grows. The other researchers found α
to be about 0.10 or 0.12 (See Table 23.17).

Table 23.16 Effects of junction width and pipe size on Ke (Sangster et al

1958).

b/Dd
Dd/Du 1.0 2.0 3.0 α
0.8 0.24 0.29 0.34 0.05
0.9 0.15 0.22 0.29 0.07
1.0 0.07 0.14 0.21 0.07
1.11 0.08 0.15 0.21 0.07
1.25 0.34 0.39 0.44 0.05
1.43 1.08 1.10 1.12 0.02

Table 23.17 Shape factor relating b/Dd to K.

Source α
Marsalek (1984) 0.12
Hare reported by Johnston and Volker (1990) 0.10
Johnston and Volker (1990) 0.10
Pedersen and Mark (1990) 0.12
540 Minor Loss Coefficients

The experimental results show that this factor should be added to the loss
coefficient for pipe size transitions as an additional parameter for the overall
junction loss coefficient.
b
K 2 = 0.10 (23.25)
Dd

Benching or invert shaping

Four of the researchers varied types of invert shaping to estimate the effects.
All agreed that benching/shaping the manhole invert was the most
significant factor affecting losses in any type of junction. The most
commonly studied methods of invert shaping were to add a bench to guide
the flow at half the depth and at the full depth of the entering/exiting pipes.
For two-pipe systems (straight-through flow and 90° turns within a manhole)
the ratio of junction width to downstream pipe diameter (b/Dd) was
important, as was the ratio of upstream and downstream pipe diameters
(Du/Dd). Insignificant or minor factors included manhole shape (square,
round, or rectangular), depth of submergence, Froude number, and Reynolds
number.
For three-pipe systems, the ratio of flows between lateral and upstream
mains (Qu/Qb) and the ratio of the lateral and downstream pipe diameters
(Db/Dd) were also significant. Total flow was varied by Sangster (1958) and
found to have no effect on losses.
The experimental results of benching clearly show a reduction in head
losses. Unfortunately, none of the studies varied all three of the most
important variables in a single experiment. The most comprehensive results
of benching were measured when upstream and downstream pipe sizes were
the same.

Table 23.18 Reduction factors for invert benching.

Source None Half Full

Pedersen and Mark (1990) 1.00 0.58 0.21
Johnston and Volker (1990) 1.00 0.60 0.56
Marsalek (1984) 1.00 0.70 0.52

The effects of benching were studied for Dd/Du = 1.0, where they showed
a clear reduction in the amount of head loss and the loss coefficient. Table
23.18 presents the results as a factor that can be applied to the loss
Minor Loss Coefficients 541

coefficient where Dd/Du = 1.0. Extrapolations to other pipe diameter ratios

may not be valid. It should be noted that these values have been developed
by the author, averaging results for different velocities or manhole shapes.
The resulting contribution to determining K for a junction is to multiply
the earlier value by a reduction factor, K3. A value of 0.60 is representative
of the findings of all three researchers for either the full or half-benched
condition, with the exception of Pedersen and Mark (1990) for full benching.

Computation of Junction Loss Coefficient

Summarizing the research and discussion, there are three factors that enter
into calculating a loss coefficient for a junction in a storm drain network
where the flow goes straight through, with no laterals or change in direction.
The calculation takes the form of
K j = (K 1 + K 2 )K 3 (23.26)
where K3, the loss due to expansion / contraction, is taken from King and
Brater's values for V = 15 fps (Table 23.15, column 6) and
b
K 2 = 0.10 (23.27)
Dd
represents the loss due to difference in pipe and manhole size, and
K 3 = 0.60 (23.28)
represents the reduction in loss due to benching the invert.

23.7 Coefficients for Free Surface Flow in

Junctions
Experimental results for free surface flow in junctions were difficult to find.
Most researchers have concentrated on estimating losses in sanitary sewer
systems where pressure flow is a more frequent occurrence and where
failures can have more serious health consequences.

23.7.1 Straight Flow through a Junction

Marsalek (1985) conducted experiments for both pressure flow and free
surface flow and reported differing values depending on manhole
542 Minor Loss Coefficients

construction and pipe diameter. Losses were calculated using the equation
based on the downstream velocity head. Loss coefficients were given for
circular manholes in a graph, which showed W/d vs. K related as follows.

Table 23.19 Coefficients for straight flow through a junction (free flow).

W/d No shaping Half benching Full benching

2.3 0.29 0.12
2.0 0.22 0.16
1.6 0.16
1.3 0.13
1.0 0.12

23.7.2 Flow Turning in a Junction

Marsalek (1985) gave values of loss coefficients for 90 degree bends in

manholes. Again, these are based on size of the manhole, diameter of pipe,
and amount of invert shaping or benching. The values for free-surface flow
are approximately 2/3 of the values he found for submerged flow:

Table 23.20 Coefficients for 90 degree bend through a junction (free

flow).

W/d No shaping Half benching Full benching

2.3 1.1 1.1 0.7

It is assumed that these coefficients combine all entrance, bend, and exit
losses into one number.

23.8 Conclusion and Recommendations

The purpose of this chapter was to review commonly used coefficients for
modeling minor losses in storm drains using EXTRAN, to find experimental
verification to determine if they are being used in the appropriate situations,
and if not, to develop better coefficients based on experimental data. The
literature review found much more information for pressure flow scenarios
than for free surface flow. There was also more information for transitions
in pipes than for transitions through junctions such as manholes.
 Minor Loss Coefficients 543

The first recommendation from the chapter is to urge researchers to

conduct new experimental studies with the goal of developing loss
coefficients that can be used for better estimates of junction losses in pipes
with free surface flow.
The review also recommends that modelers should use the coefficients
and methods of estimating them shown in Table 23.21, which have been
developed in this review from the best available experimental results
available.

Table 23.21 Recommended coefficients.

Type of Loss Source Table or Eqn

Pressure Flow
Entrance Merritt Table 23.3
Exit Assumed = 1.0
Expansion and Contraction King Table 23.4
Bends USBR Table 23.5
Free Surface Flow
Expansion and Contractions Chow Table 23.7
Bends Derived from Shukry Eqns 23.22-23.24
Culvert entrance FHWA Table 23.13
Pressure Flow, Junctions
Straight through King Table 23.15 Col 6
Expansion and Contraction King Table 23.15 Col 6
Bends
Free Surface Flow, Junctions
Straight through Marsalek Table 23.19
Bends Marsalek Table 23.20

References
American Iron and Steel Institute (1980). Modern sewer design. AISI, Washington DC.
American Society of Civil Engineers (1969). “Design and construction of sanitary and
storm sewers”. Manual of engineering practice No. 37 ASCE, New York, NY.
Camp, Thomas R. (1932) “Hydraulics of sewer transitions.” Journal of the Boston
Society of Civil Engineers, 19:325, Boston, MA.
City of Austin (1987). Austin drainage criteria manual, 2nd ed. Watershed Management
Division, Austin TX.
Chow, Ven Te (1959). Open channel hydraulics McGraw-Hill, New York, NY.
Creager, William P. and Joel D. Justin (1950). Hydroelectric handbook, 2nd ed. John
Wiley and Sons, New York, NY.
544 Minor Loss Coefficients

Daugherty, Robert L, Joseph B. Franzini, and E. John Finnemore (1985) Fluid

Mechanics with engineering applications, 8th ed. McGraw-Hill, New York, NY.
French, Richard H (1985). Open channel hydraulics McGraw-Hill, New York, NY.
Hare, C.M. (1983) “Magnitude of hydraulic losses at junctions in piped drainage
systems.” Civil Engineering Trans., Institution of Engineers, 25 (1):77
Henderson, F. M. (1966). Open channel flow. McGraw-Hill, New York, NY.
Herr, L.A., and Bossy, H.G. (1965) “Capacity charts for the hydraulic design of
culverts.” Bureau of Public Roads Hydraulic Engineering Circular No. 10, US
Government Printing Office, Washington, DC.
Herr, L.A., and Bossy, H.G. (1964) “Hydraulic charts for the selection of highway
culverts.” Bureau of Public Roads Hydraulic Engineering Circular No. 5,
Washington, DC.
“High-velocity flows in open channels: a symposium.” (1951) Transactions of the
American Society of Civil Engineers, 116:265, New York, NY.
Hinds, Julian (1928). “The hydraulic design of flume and siphon transitions.”
Transactions of the American Society for Civil Engineers, 92:1423, New York, NY.
“Hydraulic design of highway culverts.” (1985). Federal Highway Administration
Hydraulc Design Series No. 5 Washington, DC. FHWA-IP-85-15.
Johnston, Archie J., and Raymond E. Volker (1990). “Head losses at junction boxes” Jnl
Hydr Engr 116(3):326.
King and Brater (1963). Handbook of Hydraulics, 5th Ed. McGraw-Hill, New York, NY.
Linsley and Franzini (1979) Water Resources Engineering, 3rd Ed. McGraw-Hill, NY.
Marsalek, J. (1985). “Head losses at selected sewer manholes.” American Public Works
Association Special Report No. 52, Chicago, IL.
Maryland State Highway Administration (1981). Highway drainage manual. MSHA,
Baltimore, MD.
Merritt (1983). Standard handbook for civil engineers, 3rd Ed. McGraw-Hill, New York,
NY.
Mockmore, C.E. (1944) “Flow round bends in stable channels.” .” Transactions of the
American Society of Civil Engineers 109:593. New York, NY.
Sangster, W.M., Wood, H.W., Smerdon, E.T., and Bossy, H.G. (1958). “Pressure
changes at storm drainage junctions.” Bulletin No. 41, University of Missouri
Engineering experiment Stations, Columbia, MO.
Serre, Marc, A. Jacob Odgaard, and Rex A. Elder (1994). “Energy loss at combining pipe
junction” Jnl Hydr Engr 120(7):808
Shukry, Ahmed (1950) “Flow round bends in an open flume.” Transactions of the
American Society of Civil Engineers 115:75. New York, NY.
Task Force on Flow in Large Conduits (1965). “Factors influencing flow in large
conduits” Jnl of the Hydr Div, ASCE 91(HY6):123.
Taylor, Edward H. (1944). “Flow characteristics at rectangular open-channel junctions.”
Transactions of the American Society of Civil Engineers, 109:893, New York, NY.
Townsend, Ron D., and Jan R. Prins (1978). “Performance of model storm sewer
junctions.” Jnl of the Hydr Div, ASCE 104(HY1):99.
U.S. Bureau of Public Roads (1961). “Design charts for open channel flow.” Hydraulic
Design Series No. 3 Washington, DC.
Minor Loss Coefficients 545

U.S. Bureau of Reclamation (1965). Friction factors for large conduits flowing full
USBR Engineering Monograph 7, Washington, DC.
U.S. Bureau of Reclamation (1977) Design of small dams. Washington, DC
Urban Drainage and Flood Control District (1969) Urban storm drainage criteria manual
Denver Regional Council of Governments, Denver CO.
Urban Water Resources Research Council (1992) “Design and construction of urban
stormwater management systems.” Manual of engineering practice No. 77 ASCE,
New York, NY.
Virginia Department of Transportation (1988). Drainage manual VDOT, Richmond VA.
Cited in King and Brater
Anderson, A.G., and L.G. Straub (1948). Hydraulics of Conduit Bends St Anthony
Falls Hydraulic Lab Bulletin 1, Minneapolis, MN
Beij, K.H. (1938). “Pressure losses for fluid flow in 90 o pipe bends” National
Bureau of Standards Research Paper RP 1110 Washington DC.
Fuller, W.E. (1913) Loss of head in bends J New England Water Works Assoc.
Cited in Linsley and Frazini
Crane Corporation (1965). Flow of Fluids, Technical Paper 40
546 Minor Loss Coefficients