Journal of Hydraulic Research

ISSN: 0022-1686 (Print) 1814-2079 (Online) Journal homepage: http://www.tandfonline.com/loi/tjhr20

Supercritical flow in junction manholes under

invert- and obvert-aligned set-ups

Gaetano Crispino, Michael Pfister & Corrado Gisonni

To cite this article: Gaetano Crispino, Michael Pfister & Corrado Gisonni (2018): Supercritical flow
in junction manholes under invert- and obvert-aligned set-ups, Journal of Hydraulic Research

To link to this article: https://doi.org/10.1080/00221686.2018.1494056

Published online: 22 Oct 2018.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tjhr20
 Journal of Hydraulic Research, 2018
https://doi.org/10.1080/00221686.2018.1494056
© 2018 International Association for Hydro-Environment Engineering and Research

Research paper

Supercritical flow in junction manholes under invert- and obvert-aligned set-ups

GAETANO CRISPINO (IAHR Member), PhD, Department of Engineering, Università della Campania “Luigi Vanvitelli”,
Aversa, Italy
Email: [email protected] (author for correspondence)

MICHAEL PFISTER (IAHR Member), Professor, Civil Engineering Department, Haute Ecole d’Ingénierie et d’Architecture,
Fribourg (HEIA-FR, HES-SO), Switzerland
Email: michael.pfi[email protected]

CORRADO GISONNI (IAHR Member), Full Professor, Department of Engineering, Università della Campania “Luigi
Vanvitelli”, Aversa, Italy
Email: [email protected]

ABSTRACT
Junction manholes are a part of urban drainage infrastructures. They merge inlet branches into an outlet branch. Former studies focused on junction
flow by considering a specific manhole layout: identical diameters were assigned to the branches and the invert was flush. Nevertheless, engineers
are often involved in designing junctions under generalized geometries, with different branch diameters and, sometimes, bottom offsets at manhole
inlets. For these junction arrangements, the empirical relations documented in literature are not applicable. An experimental campaign was thus
conducted to determine the main flow features of junction manholes under generalized set-ups. The results include a diagram of state to be used if
junctions operate under mixed flow conditions. Information regarding the main wave patterns, with detailed discussion of wave features and heights,
is provided to support the hydraulic design of junction manholes under the supercritical flow regime.

Keywords: Experimental facility; hydraulic model; laboratory study; shock wave; sewer hydraulics; sewer manhole; supercritical
flow

1 Introduction Reynolds stress turbulent models is complicated for practical

applications. Empirical relations derived from physical model
Junction manholes are components of urban drainage systems. investigations are more frequently adopted to design or verify
They combine individual sewer branches into a single sewer col- junction manholes.
lector, converting linear systems to ramified sewers. Together As for many others sewer structures, the hydraulic behaviour
with inspection manholes, junction manholes are presumably of junctions strictly depends on the energetic content of the
the most common manhole in sewer nets. A reliable prediction approach flows. If the approach flows are subcritical, then the
of the main flow features in junctions is fundamental and, at the junction flow field is prone to be described with theoretical
same time, challenging. In fact, the combination of two or more approaches (Gurram, Karki, & Hager, 1997; Hager, 1987; Pfis-
free-surface approach flows results in a complex3D flow struc- ter & Gisonni, 2014). Nevertheless, uniform flows entering
ture, with local phenomena as separation and recirculation flow sewer manholes are often supercritical under steep sewer slopes
zones, shock waves and, partially, hydraulic jumps. Some stud- exceeding around 1%. For these conditions, any change of sewer
ies in the literature describe 2D and 3D numerical representa- geometry (contraction, expansion, junction or bottom slope vari-
tions of the junction flow field (Geberemariam, 2017; Krüger & ation) or roughness may initiate shock waves. A shock wave
Rutschmann, 2000; Motlagh, Nazemi, Sadraddini, Abbaspour, is a standing wave involving a relatively large surface pertur-
& Motlagh, 2013; Ramamurthy, Han, & Biron, 2013). How- bation (Gisonni & Hager, 2012; Hager, 2010). In the worst
ever, the application of computational fluid dynamics using case, the breakdown of the supercritical flow occurs with an

Received 7 October 2017; accepted 25 June 2018/Currently open for discussion.

ISSN 0022-1686 print/ISSN 1814-2079 online

http://www.tandfonline.com
1
2 G. Crispino et al. Journal of Hydraulic Research (2018)

abrupt transition from free-surface to pressurized air-water flow. Gisonni (2014), the junction was adapted allowing for estab-
The prediction of the shock wave height is thus essential for lishing identical branch obvert elevations. The junction model
supercritical flows. is sketched in Fig. 1b. It consisted of two upstream branches
An early experimental analysis of wave features was pre- including a straight (subscript o) and a lateral (subscript L), all
sented by Schwalt and Hager (1995) and Schwalt (1996). Three having circular cross-section. The upstream branches merged
standing waves across 30° and 60° junctions with rectangu- under a junction angle of δ = 45°. A downstream branch (sub-
lar branches were described. Del Giudice and Hager (2001) script u), with a circular cross-section of diameter Du = 0.240
and Gisonni and Hager (2002) studied in detail the flow pat- m, exited from the junction manhole. The upstream branch
tern in 45° and 90° standard junction manholes approached diameters were varied as 0.123, 0.190 and 0.240 m. The junction
with circular conduits with equal diameters, respectively. Zhao, manhole presented a U-shaped cross-section and its benches
Zhu, and Rajaratnam (2004) physically reproduced the prob- were 1.50 Du high. The branches and most of the perimeter walls
lematic 90° junction manhole in Calgary (Canada). Saldarriaga, of the model were made of transparent PVC. Bottom offsets
Bermúdez, and Rubio (2012), Gisonni and Pfister (2015) and d = Du − Di (with i = o, L) were inserted at the manhole inlets
Saldarriaga, Rincon, Moscote, and Trujillo (2017) provided if one or both upstream branch diameters were smaller than the
shock wave occurrence conditions and preliminary design cri- downstream one. Given the branch diameter values, d was equal
teria for symmetric and asymmetric junction manholes with to 0.21 Du and 0.49 Du . It can be thus considered as a small drop
approach supercritical flows, sometimes equipped with bottom (De Martino, Gisonni, & Hager, 2002), because d < 0.60 Du .
drops at manhole inlets. Pfister and Gisonni (2014) focused on The branches and the junction structure were horizontal and
local head losses of sub- and supercritical flows of 45° and 90° friction losses were neglected due to the short length. A jet-box
junction manholes with circular branches and different upstream was placed upstream of each upstream branch, at a distance of
branch diameters. about 1.50 m from the manhole inlets. It consists of a sandwich-
The above-mentioned investigations considered junction type structure allowing to fix a certain partial filling ratio
manholes for flush inverts (i.e. invert-aligned junction) with y = h/D, with h as flow depth, independent on discharge Q.
equal upstream and downstream branch diameters. To the The jet-box assured a smooth transition from pressurized to free-
knowledge of the authors, only Saldarriaga et al. (2012) intro- surface flows, without shock waves and velocity concentrations
duced small bottom drops at manhole inlets. However, the downstream of the jet boxes.
upstream branch diameters were equal to the downstream diam- Discharges were measured by inductive flowmeters with a
eter, so that the obverts of the upstream and downstream full-scale (FS) accuracy of ± 0.5%. Flow depths were mea-
branches were not aligned. Oppositely, junction manholes are sured with point gauges of accuracy around ± 1 mm due to
often aligned at the upstream and downstream branch obverts flow turbulence and the presence of air–water flow. The shock
(i.e. obvert-aligned junction), resulting in small invert offsets. wave height measurement was less accurate because of the tur-
This set-up is preferable because it assures a hydraulic dis- bulent fluctuations of the wave maximum, especially for large
connection between the upstream and downstream branches. A Froude numbers F. Flow depths in the upstream and down-
backwater effect propagating into the upstream branches is less stream branches were measured across the openings along the
probable. branch obverts.
This paper presents an experimental approach for junctions.
A physical model study on an obvert-aligned junction man-
2.2 Test programme
hole was performed. The obverts of a straight, a lateral and a
downstream branch were aligned, and different sets of branch The test programme served to identify the main hydraulic fea-
diameters were utilized. Detailed information on the main flow tures of supercritical junction flows. More than 150 tests on the
pattern and shock wave features are provided. The experi- above-described physical model were performed by varying the
mental results are also compared to observations in similar governing parameters. Junction set-ups from A to H were con-
junction models under invert-aligned set-ups, including constant sidered as specified in Table 1. The discharge Q follows from a
or different upstream branch diameters. Empirical equations are combination of F, y and D (Gisonni & Hager, 2012) as:
provided to estimate the relevant supercritical flow features of
junction manholes approached by circular branches.
Q = F (g y 4 D5 )0.5 (1)

2 Experimental set-up
where g is gravity acceleration. Equation (1) is valid for circular
pipes within the usual range 0.30 < y < 0.95.
2.1 Physical model
Tests characterized by junction failures (namely choking
The physical model (Fig. 1a) was operated at the Laboratory of of one or two branches, breakdown with hydraulic jump,
Hydraulic Structures (LCH) of École Polytechnique Fédérale de manhole up-surging) are excluded herein. In the following,
Lausanne (EPFL). Starting from the facility used by Pfister and the present data on obvert-aligned junctions are compared to
Journal of Hydraulic Research (2018) Supercritical flow in junction manholes 3

Figure 1 (a) Physical model of junction manhole at LCH-EPFL:  1 straight branch (subscript o),  2 lateral branch (subscript L), 
3 U-shaped
junction manhole, 4 downstream branch (subscript u),  5 collecting sump (here for: Do = 0.123 m, DL = 0.240 m and Du = 0.240 m); (b) plan
(above), longitudinal (below on the left) and transverse (below on the right) section view of 45° junction (grey = U-shaped portion; white = circular
branch portion; ____ = example of free-surface profiles)

Table 1 Test programme for Du = 0.240 m, with minimum (min) and maximum (max) values of experimental parameters

Set-up Number of tests Do (m) DL (m) yo (–) yL (–) Fo (–) FL (–)

A 30 0.123 0.123 0.31–0.65 0.31–0.54 1.44–6.89 1.44–6.68

B 34 0.190 0.123 0.34–0.63 0.30–0.54 1.23–5.22 1.65–7.45
C 48 0.240 0.123 0.17–0.56 0.36–0.53 1.65–4.08 1.61–5.09
D 19 0.123 0.190 0.35–0.56 0.29–0.55 1.64–5.82 1.92–5.01
E 22 0.190 0.190 0.30–0.54 0.27–0.47 1.90–5.55 2.28–6.88
F 18 0.240 0.190 0.26–0.33 0.25–0.48 3.34–4.94 1.56–7.91
G 14 0.123 0.240 0.35–0.58 0.28–0.30 1.44–5.88 3.17–5.28
H 25 0.190 0.240 0.31–0.36 0.28–0.32 4.50–5.67 3.14–3.60
min 14 0.123 0.123 0.17 1.23 0.25 1.44
max 48 0.240 0.240 0.65 6.89 0.54 7.45

unpublished data derived from physical (Gökok, 2013; Nieder- set-up. Niedermann (2013) and Gökok (2013) carried out tests
mann, 2013) and numerical (Köestler, 2014) tests on junctions. with variable approach flow features for δ = 45° and 90°,
All these investigations considered an invert-aligned junction respectively. The tested branch diameters were identical to the
4 G. Crispino et al. Journal of Hydraulic Research (2018)

Table 2 Flow types and limit parameters of Gökok (2013), Niedermann (2013) and Köestler (2014)

Author Flow type δ (°) parameter Do (m) DL (m) yo (–) yL (–) Fo (–) FL (–)

Gökok (2013) (I)-(II)-(III)-(IV) 90 min 0.123 0.123 0.15 0.19 0.10 0.10
max 0.240 0.240 1.00 1.00 15.71 6.55
Niedermann (2013) (I)-(II)-(III)-(IV) 45 min 0.123 0.123 0.17 0.21 0.13 0.19
max 0.240 0.240 0.94 0.94 9.93 10.33
Köestler (2014) (I) 30–120 min 0.123 0.123 0.21 0.19 2.61 1.63
max 0.240 0.240 0.57 0.58 5.86 7.26

diameters used in the present model. The junction invert was 3 Diagram of state
flush and consequently offsets at the obvert of the manhole inlets
were present. Köestler (2014) simulated spatial flow fields in 3.1 Standard junctions with identical branch diameters
a numerical model with FLOW-3D (Flow Science, 2012). The
For sewers involving air–water flows, various mechanisms may
model reproduced the geometry of the aforementioned invert-
imply the transition between flow regimes (Chanson, 2009;
aligned junction set-up, with different Do , DL and δ. The latter
Falvey, 1980; Granata, de Marinis, & Gargano, 2015). A dia-
ranged between 30° and 120°. The tests carried out by Gökok
gram of state is thus useful to describe the individual hydraulic
(2013), Niedermann (2013) and Köestler (2014) include the
behaviour. For junctions, the diagram of state shows the flow
following junction flow types (Hager, 2010):
type establishing under given hydraulic and geometric bound-
ary conditions. Del Giudice and Hager (2001) and Gisonni and
• Type (I): supercritical flows in both upstream branches; Hager (2002) introduced a diagram of state for standard 45° and
• Type (II): supercritical flow in the straight branch and sub- 90° junction manholes. They selected yo Fo and yL FL cosδ as
critical flow in the lateral branch; governing parameters for defining the flow type. In fact, y F can
• Type (III): subcritical flow in the straight branch and super- be formulated non-dimensionally equal to the relative dynamic
critical flow in the lateral branch; force component:
• Type (IV): subcritical flows in both upstream branches.
Pd = (Q V)/(g D3 ) (2)

In addition, tests considering exclusively a lateral or a by assuming a ± 10% approximation of A = D2 y 1.4 for circular
straight approach flow were conducted by Gökok (2013) and branches. Seven flow regions were defined:
Niedermann (2013). Although not establishing merging flow
processes, these scenarios may provoke flow conditions with (1) subcritical flows entering junction manhole;
relevant shock wave heights. Limits and flow types of the exper- (2) supercritical straight branch and subcritical lateral branch;
iments conducted by Gökok (2013), Niedermann (2013) and (3) transitional to supercritical flow in both upstream branches;
Köestler (2014) are listed in Table 2. (4) both upstream branches under supercritical flow conditions;
(5) choking either of lateral or straight branches;
(6) subcritical straight branch and supercritical lateral branch;
2.3 Scale effects
and
Scale effects as a consequence of the considered reference (7) choking of downstream branch.
forces can affect the model data. For the Froude similar-
ity, as usually applied in open-channel hydraulics, Gökok
3.2 Invert-aligned junctions
(2013) stated that scale effects due to surface tension in
the herein exploited junction model are reduced if provid- The experimental data of Niedermann (2013), Gökok (2013)
ing a flow depth h ≥ 0.04 m. This limitation is also com- and Köestler (2014) on invert-aligned junctions with different
patible to Del Giudice and Hager (2001). Accordingly, only D, y, F and δ are included in the diagram of state shown in
tests with a sufficient flow depth were retained herein. These Fig. 2. The data derived from Del Giudice and Hager (2001)
limitations agree with the necessity to guarantee sufficient and Gisonni and Hager (2002) are also plotted. Here, yo Fo
Reynolds numbers R = 4 Rh V/υ, which varied between 7.5 × and yL FL cosδ are multiplied by the diameter ratio β i = Di /Du
104 ≤ Ro ≤ 3.0 × 105 and 9.0 × 104 ≤ RL ≤ 3.4 × 105 , with (i = o, L) to account for the variability of Do and DL .
Rh as hydraulic radius, V = Q/A as average flow velocity, A Figure 2 indicates that the flow regions being initially defined
as the cross-sectional area and υ as water viscosity (Pfister & for a standard junction manhole need to be adapted, when con-
Chanson, 2014). sidering the generalized geometry with different Do and DL .
Journal of Hydraulic Research (2018) Supercritical flow in junction manholes 5

Figure 2 Diagram of state (Del Giudice & Hager, 2001) for invert-aligned junctions under flow type: (a) (I), (b) (II), (c) (III), (d) (IV) (: Del
Giudice & Hager, 2001; ◦: Gisonni & Hager, 2002; : Niedermann, 2013; •: Gökok, 2013; ♦: Köestler, 2014)

In particular, Fig. 2a shows that not all points are included 1.20
in the flow region (4) (grey-shaded region), as expected for (7) set-up A
both supercritical upstream branches. Some points fall into the set-up B
flow regions (1), (3) and (7) despite stable supercritical flows (5)
set-up C
persisted both in the junction manhole, and a choking of the 0.80 (6) (4)
downstream branch was never observed during these test-runs. set-up D
(-)

Figure 2b regroups the experimental tests under flow type (II). set-up E
L Lcos

The points should be all entirely restrained within the flow (3) set-up F
region (2) (grey-shaded area). Instead, some points are included 0.40
in the flow region (3), but no transitional flows were observed set-up G
FL

for these tests. Finally, Fig. 2c and 2d refers to the flow types (1) (2) set-up H
(III) and (IV). As shown, a satisfying conformity subsists, given
that the corresponding flow regions (5) and (1) (grey-shaded 0.00
areas) enclose the data in accordance with their definition. Few 0.00 1.00 2.00 3.00
points are close to the boundary lines. Fo o o (-)
Figure 3 Diagram of state (Del Giudice & Hager, 2001) for
obvert-aligned junctions under junction Type (I) (for flow region
3.3 Obvert-aligned junctions numbers see main text and for set-ups Table 1)

The flow patterns observed in the obvert-aligned junction model

are plotted in the diagram of state of Fig. 3. As visible, not all and Köestler (2014) some points are included in the upper por-
the points are located in the flow region (4) (grey-shaded area), tion of flow region (1) and in the flow region (5). Few points
although only supercritical approach flows were tested. In par- with β L = 0.79 (set-ups E and F) are located in the flow region
ticular, for β L = 0.51 (set-ups A, B and C) most of the points (7), but in these cases the free-surface flow regime was still
are located in the flow region (3). Similarly to Gökok (2013) preserved without choking in the downstream branch.
6 G. Crispino et al. Journal of Hydraulic Research (2018)

Figure 4 Diagram of state for generalized junction manhole set-ups. Data points derived from the present physical investigation and from Del
Giudice and Hager (2001), Gisonni and Hager (2002), Gökok (2013), Niedermann (2013) and Köestler (2014). The dotted area indicates a region
not covered by experimental points. For numbers see main text

3.4 Adapted diagram of state for generalized junction set-ups yL FL βL cos δ = −0.70 (yo Fo βo ) + 0.42 (3)

An adapted diagram of state (Fig. 4) is presented based on the

Equation (3) is valid for 0.00 < yo Fo β o < 0.60.
results derived by physical and numerical investigations on var-
ious arrangements of junction models. This diagram of state is
• Flow region (2) is again related to the flow type (II). It is
thus applicable for determining the flow types to be expected in
delimited on the left by Eq. (3), if yL FL β L cos δ < 0.147,
junction manholes with:
and by the vertical line with equation:

• different junction manhole angles δ, ranging between 30° and

yo Fo βo = 0.39 (4)
120°;
• rounded or sharp-crested outer wall deflection at the con-
nection between the lateral branch and the junction manhole if yL FL β L cos δ > 0.147. The region limited by 0.30 <
bench; yL FL β L cos δ < 0.50 (light-grey shaded area in Fig. 4)
• constant downstream branch diameter Du ; includes both flow types (I) and (II). It can be thus considered
• different upstream branch diameters Do and DL ; as a transition region. The horizontal boundary line separat-
• bottom or top offsets at the manhole inlets for Di < Du (i.e. ing flow regions (2) and (4) is defined to include about 90%
i = o, L); of the overall experimental data-points under flow type (II).
• upstream and downstream branches and junction manhole Its equation is:
almost horizontal. yL FL βL cos δ = 0.40 (5)

As shown in Fig. 4, flow regions (1), (2), (4), (6) and (7) are • Flow region (4) comprises the flow type (I) points. Its domain
preserved, even if modified when compared with the original is slightly larger than the flow region previously defined,
diagram. The boundary lines separating the flow regions are because it encloses the former flow region (5), which is not
defined based on the distribution of the experimental data of the considered by corresponding physical observations in invert-
identical flow type. In particular: and obvert-aligned junctions with different Do and DL . Flow
region (4) is under-delimited by the horizontal boundary line
• Flow region (1) still includes data under flow type (IV). Nev- with Eq. (5) (about 10% of the flow type (I) points are under
ertheless, the region is made smaller by subtracting an upper this limit), left-delimited by the vertical boundary line with
portion from the original area, given that both subcritical flow Eq. (4) and up-delimited by a horizontal line with equation:
points are not included there (Fig. 2d). The equation of the
upper boundary line of the flow region (1) is: yL FL βL cos δ = 1.25 (6)
Journal of Hydraulic Research (2018) Supercritical flow in junction manholes 7

Flow region (6) comprehends flow type (III) points. This flow
region is restrained by the boundary lines with Eq. (4), on the
right, and Eq. (6), above. The same is under-delimited by the
boundary line with Eq. (3), for yo Fo β o < 0.087, and by the
boundary line of equation (yo Fo β o > 0.087):

yL FL βL cos δ = +0.26 (yo Fo βo ) + 0.16 (7)

• Flow region (7) refers to choking of the downstream branch.

Figure 5 Shock waves patterns in a supercritical junction flow under
The inferior boundary line of Eq. (6) was marginally raised
a certain junction angle δ (the black arrows indicate flow directions)
to enclose the experimental points related to flow conditions
with severe heights of shock waves but exempt from the
occurrence of choking flow in the downstream branch. 4.2 Shock wave A

Invert-aligned junctions
The variation of Do and DL does not affect the wave A. Nie-
4 Shock waves dermann (2013) and Gökok (2013) observed wave A mainly for
the straight branch being active, as represented in Fig. 6a. For
4.1 Standard junctions with identical branch diameter two supercritical approach flows, the formation of the wave A
The following shock waves (Fig. 5) are observed in a junction is rare, instead, occurring only for Do > DL .
manhole approached by one or two supercritical flows under a For a discharge only in the straight branch, Del Giudice and
certain δ (Del Giudice & Hager, 2001; Gisonni & Hager, 2002; Hager (2001) found that the maximum relative wave height
Schwalt & Hager, 1995): YA = hA /ho − 1 increases linearly with Fo as:

YA = 2/3 (Fo − 1) (8)

• Wave A, with a maximum height hA , occurs as a result of flow
expansion for supercritical straight flow only or if the straight up to YA ≤ 3.00. Vice versa, fixing a value of YA = 3.00 guaran-
flow is dominant compared to the lateral flow. tees safe operation against bench manhole overtopping. Figure
• Wave B is due to the impact of the lateral flow onto the 7a shows that the wave A maxima, as observed by Nieder-
opposite bench wall. It is a “wall” wave, with a maximum mann (2013) and Gökok (2013) for only active straight branch,
height hB being typically observed for lateral flow only or in slightly still increase with Fo . Experimental data derived from
junctions approached by supercritical flows in both branches. Del Giudice and Hager (2001) and Gisonni and Hager (2002)
• Wave C originates from the deviation of the lateral flow are also included in Fig. 7a. As shown, δ and yo do not affect
caused by the straight flow, similarly to an abrupt wall deflec- YA . Equation (8) is reliable up to Fo = 4.00. For Fo > 4.00, the
tion. This wave usually starts at the junction point J (Fig. 1) scatter between observed and predicted YA is significant, prob-
and its maximum height is hC . ably because the wave becomes thin and turbulent making its
• Wave D arises from the reflection of wave B. It can be located measurement less accurate. It is noteworthy that the assumption
at the end of the manhole or in the downstream branch, YA = 3.50 was recorded as an upper limit for the wave A height
depending on the position and magnitude of wave B. Wave for Fo > 6.00.
D is smaller than wave B. Its height affects the downstream If two supercritical flows enter the junction, then YA ranges
branch discharge capacity. between 0.50 and 2.00, resulting in a maximum wave A height
• Wave E forms along the lateral branch due to its curvature. It significantly smaller than other shock waves.
occurs for junctions with δ > 45°. It is similar to a bend wave
but of reduced height. Obvert-aligned junctions
• A swell occurs due to the transition from U-shaped to cir- In the obvert-aligned junction model, if dL > do and FL > 3.00,
cular profile at the manhole outlet. A pronounced swell may the lateral flow passes upon the straight flow and collides on
provoke a backwater flow in the manhole. The swell height the opposite bench wall. The straight flow then loses the side
directly affects the discharge capacity of the junction. constraint by the lateral flow and impinges onto the inner wall
deflection (Fig. 6b). Under these conditions, the relative wave A
height is around 30% approximated (Fig. 7b) by:
Given the content of the present investigation, the shock waves
A, B and C are described in the following. YA = 0.80 (Fo − 1) (9)
8 G. Crispino et al. Journal of Hydraulic Research (2018)

Figure 6 Upstream view of the manhole showing shock wave A occurring (a) for only straight flow in 45° junction manhole (Do = 0.240
m, yo = 0.52, Fo = 3.42, DL = 0.240 m) , (b) for flow type (I) in the obvert-aligned 45° junction (Do = 0.190 m, yo = 0.37, Fo = 4.62,
DL = 0.123 m, yL = 0.37, FL = 6.14) with formation of two independent shock waves (shock wave A highlighted by the grey arrow; the white
arrows denote flow directions)

Figure 7 Relative height YA of wave A as a function of Fo for (a) only discharge in the straight branch in invert-aligned junctions; (b) flow type (I)
in obvert-aligned junctions

Following Eq. (9), YA in obvert-aligned junction manholes under dictates the bench wall height. According to Del Giudice and
flow type (I) is larger than the corresponding height given by Eq. Hager (2001) and Gisonni and Hager (2002), for 45° junctions
(8) for invert-aligned junctions operating under a discharge only under flow types (I), (II) and (III), the relative wave B height
in the straight branch. YB = hB /hL – 1 is a function of FL as:
The hydraulic prevalence of the straight flow against the lat-
eral flow is typically defined with the discharge ratio QL /Qo . YB = CB (FL − 1) (10a)
Saldarriaga et al. (2012) observed wave A under flow type (I)
for QL /Qo ≤ 0.10. According to the present observations, this
criterion does not apply for obvert-aligned junctions with a bot-
YB = 0.80 FL (10b)
tom offset at the manhole lateral inlet, because wave A occurred
for both QL /Qo < 0.10 and QL /Qo > 0.10. with hB as the wave B height, hL as the lateral branch flow depth
and CB as a coefficient. Equation (10a) is applicable for flow
4.3 Shock wave B types (I), with CB = 1.14, and (II), with CB = 1.00. Equation
(10b) is valid for flow type (III), instead. Equation (10a) can be
Invert-aligned junctions also used for 90° junction manholes under flow type (I) (Gisonni
Schwalt and Hager (1995) stated that wave B is the highest wave & Hager, 2002) using again CB = 1.14. This evidence proves
for flow type (I). Therefore, the estimation of wave B height that δ does not affect the height of wave B.
Journal of Hydraulic Research (2018) Supercritical flow in junction manholes 9

Figure 8 Relative wave B height YB as function of FL in invert-aligned 45° and 90° junctions for (a) flow type (I); (b) flow type (III)

Figure 9 (a) Relative wave B height YB as function of FL in invert-aligned junction manholes with different junction angles δ under flow type (I);
(b) relative wave B height YB as function of FL for 45° obvert-aligned junction model

Niedermann (2013) and Gökok (2013) considered the wave flow type (III), the relative maximum wave B heights are well
B formation under flow types (I) and (III). For flow type (III), approximated by Eq. (10b).
the straight flow disturbs the impact of the supercritical lat- Figure 9a includes the values derived numerically by
eral flow onto the opposite bench wall and reduces the wave B Köestler (2014). As shown, YB augments by increasing FL and
height. For 90° junctions, the wave B is comparable with wave is independent of δ. Differently from the physical model obser-
E (Fig. 5). For particular tests with two supercritical flows, the vations, Eq. (10a) underestimates YB , as given by the numerical
wave B maximum was reaching the top of the bench. Conse- model, by 25%. Contrarily, if FL < 3.00, then the average devi-
quently, the remaining freeboard was small. This circumstance ation is smaller, being about 4%. Equation (10a) gives wave B
was frequently detected for FL > 3.50 and for small DL , if 90° heights slightly smaller than the numerical model predictions.
junctions are considered (Gökok, 2013). For 45° junction man- Compared to the physical model data, the numerical model is
holes (Niedermann, 2013) the wave B maximum was slightly thus limited to provide the maximum shock wave height cor-
smaller than in 90° junctions. rectly if the flow regime is highly turbulent, as expected for
YB is plotted against FL in Fig. 8, for flow type (I) (Fig. 8a) large FL .
and flow type (III) (Fig. 8b) as for 45° and 90° invert-aligned
junctions. The wave heights tend to augment by increasing FL ,
and the variability of DL does not influence the magnitude of Obvert-aligned junctions
YB . For flow type (I), Eq. (10a) predicts YB accurately only for According to the experiments, the presence of the inlet bottom
FL < 4.00. This is presumably due to the loss of accuracy in offsets in the obvert-aligned junction does not alter signifi-
the wave height measurement for lateral flows with large F. For cantly the wave B features. As for wave A, if β L = 0.51 and
10 G. Crispino et al. Journal of Hydraulic Research (2018)

wave B is not disturbed by the straight approach flow, giv-

ing a maximum height exceeding the invert-aligned junctions.
Therefore, the relative wave B heights are underestimated.

Junctions under invert- and obvert-aligned set-ups

As soon as a bottom offset is present at one or two manhole
inlets, the balance of the pressure forces acting on the control
volume should account for the static force Ps generated by the

Figure 10 (a) Permanent water cushion below the flow issued by the
straight branch (yo = 0.58, Fo = 1.44, yL = 0.29, FL = 3.24; white
arrows denote flow direction); (b) longitudinal view of the junction
inlet, below

Figure 11 Shock wave B height hB /Du as function of SL for invert-

(I) or obvert- (O) aligned junctions with different δ, D, y and F

FL > 3.00, then the flow issued by the lateral branch over-
steps the straight flow, directly impinging onto the opposite
bench wall (Fig. 7b). Under these conditions, the interaction
between the approach flows is reduced, as compared to the Figure 12 Shock wave C: (a) upstream and (b) downstream view
invert-aligned junctions. Shock wave B develops without any of the shock front originating near to the junction point and end-
ing at the collision against the manhole bench wall (shock wave B)
disturbance. Moreover, if the shock wave B occurs near the end (Do = 0.240 m, yo = 0.27, Fo = 6.28, DL = 0.240 m, yL = 0.30,
of the junction, then the flow collides with the manhole end wall FL = 3.36); top view of the shock front for: (c) Do = 0.240 m,
producing a massive spray. This undesirable flow condition was yo = 0.18, Fo = 7.47, DL = 0.123 m, yL = 0.50, FL = 2.37 and (d)
systematically observed for FL > 5.00, independent of Fo . Do = 0.240 m, yo = 0.35, Fo = 3.61, DL = 0.123 m, yL = 0.50,
As previously noticed for 45° and 90° invert-aligned junc- FL = 3.99; (e) top view of the 45° obvert-aligned junction manhole
in operation, with the absence of the shock wave C (Do = 0.123 m,
tions, YB increases significantly with FL (Fig. 9b). The average
yo = 0.28, Fo = 7.90, DL = 0.123 m yL = 0.41, FL = 4.93). The
accuracy of Eq. (10a) is ± 25%, with extreme deviations of white dashed line highlights the trace of the shock front, grey circle
about 95%. The average accuracy decreases significantly by denotes the impact onto the manhole bench wall; white arrows indicate
considering the largest dL (β L = 0.51). In this set-up, the shock flow direction
Journal of Hydraulic Research (2018) Supercritical flow in junction manholes 11

permanent water cushion below the flow issued by the upstream of the lateral approach flow features. For flow type (I) the
branches (Fig. 10). The water cushion height hwc depends on following relation was proposed:
the approach flow conditions, mainly on Fi and hi (i = o, L).
The dynamic force Pd is added to Ps acting on the inner drop YC = FL − 1 (14)
face (Fig. 11) resulting in the specific force S. Ps is computed as
(Gisonni & Hager, 2012): Based on the experimental results of Del Giudice and Hager
(2001), wave C is smaller than wave B and therefore irrelevant
PS = 0.50 (hwc /Du )2.50 (11) for the design of 45° junction manholes. For δ = 90°, wave C is
mixed with the terminal portion of the wave E along the lateral
The height hwc of the water cushion at the lateral inlet ranges branch. The definition of wave C thus loses importance when
between 0.60 dL and 0.80 dL , according to the present obser- designing 90° junctions.
vations. Assuming that a dL -high water cushion is present, the The experimental investigations conducted on 45° invert-
specific force SL of the lateral approach flow is: aligned junctions confirm the description of wave C found in
literature. Niedermann (2013) observed the shock wave C under
SL = [0.50 (dL /Du )2.50 + yL FL βL ] cos δ (12) flow type (I). The shock front of shock angle θ , originated by
the collision of supercritical approach flows, rotates the straight
flow and impacts the opposite manhole bench wall (Fig 12a and
The wave B height can be correlated to SL , as defined according
12b). In accordance with the physical model tests, shape and
to Eq. (12). Figure 11 shows the non-dimensional wave B height
shock angle of wave C depend strictly on the approach flow
hB /Du as a function of SL for both invert- and obvert-aligned
conditions F and y. If the straight flow prevails over the lat-
junction manholes under junction flow types (I) and (III). The
eral flow, then the shock front is extended, with θ < δ. The
equation:
collision point against the bench wall is thus located more down-
hB /Du = 1.50 SL (13) stream (Fig. 12c). Oppositely, for junctions mainly governed by
the lateral flow, the shock wave is more compact and charac-
describes the experimental points with maximum deviations of terized by a rounded crest front and a rearward collision point
± 35%. If dL = 0, then the wave B height is only a function of (Fig. 12d).
Pd . Different from Eqs (10a) and (10b), Eq. (13) is an explicit
function of d, valid for both flow types (I) and (III). Moreover, Obvert-aligned junctions
it provides more accurate prediction of the maximum wave B In the obvert-aligned junction, the shock wave C was rarely
height. observed. For β o = 0.51 the formation of this wave is inhibited.
As shown in Fig. 12e, the approach flow issued by the straight
4.4 Shock wave C branch passes upon the lateral flow without any flow impact and
deflection. This condition is clearly observed for Fo > 3.00 and
Invert-aligned junctions β L = 0.51 (largest bottom drop at the lateral inlet) whereas the
Del Giudice and Hager (2001) proposed to evaluate the relative limit value increases up to 4.00, approximately, if β L was equal
maximum wave C height YC = hC /hL – 1 only as a function to 0.80 and 1.00. In all other set-ups, the impact between the two

Figure 13 (a) YC , (b) hB /hC as function of FL in the 45° junction manhole

12 G. Crispino et al. Journal of Hydraulic Research (2018)

supercritical approach flows results in a continuous front shock. Acknowledgements

However, the height of the shock wave C is smaller than the
maximum height reached by the shock wave B, especially for The authors thank Mr Cédric Bron, lead technician at LCH,
β L = 0.51. for his support during the model tests. The Laboratory of
Figure 13a refers to wave C data for flow type (I). In pres- Hydraulics, Hydrology and Glaciology (VAW) of ETH Zurich
ence of a lateral bottom offset, the relative wave C becomes provided model elements.
YC = hC /(hL + dL ) – 1. As visible, Eq. (14) estimates YC with
an acceptable accuracy for FL < 5.0. Conversely, the differ- Funding
ence between measured and predicted data tends to augment
by increasing FL , because of the significant turbulence and The project was financially supported by the City of Zurich
air entrainment in the model flow. The scatter is also rele- (ERZ Entsorgung + Recycling Zurich).
vant for β L = 0.51. In this case, Eq. (14) overestimates the
wave C height because of the weak flow interaction between
the two supercritical flows. As underlined by Del Giudice Notation
and Hager (2001), wave B and C heights are both correlated
to lateral approach flow features. They are thus comparable. A = cross-sectional area (m2 )
Figure 13b confirms that wave C is smaller than wave B, C = coefficient (–)
given that the ratio hB /hC is about 1.25% and is larger for d = bottom offset (m)
β L = 0.51. D = diameter (m)
F = Froude number (–)
g = gravity acceleration (m s−2 )
5 Conclusions h = flow depth (m)
P = specific force component (–)
The present research considers a junction manhole with different Q = discharge (m3 s−1 )
upstream branch diameters D, filling ratios y, Froude numbers R = Reynolds number ( − )
F and junction angles δ under both obvert- and invert-aligned Ra = axial curvature radius (m)
set-ups. The discussion mainly focuses on junction manholes Rh = hydraulic radius (m)
approached by one or two supercritical flows. The results of S = specific force (–)
this study provide a description of the general junction flow V = average flow velocity (m s−1 )
pattern. x = streamwise coordinate (m)
Based on physical model tests, an adapted diagram of state y = partial filling ratio (–)
for generalized junction manhole set-ups is presented. Five Y = relative wave height (–)
flow regions are defined, depending on the dynamic force com- β = diameter ratio (–)
ponents of the approach flows. Differently from the original δ = junction angle (°)
diagram, the new flow regions are analytically defined. This θ = shock angle (°)
advance will allow the flow type established in the junction υ = water viscosity (m2 s−1 )
manhole to be detailed more easily and precisely.
The hydraulic features of shock waves typically occurring
Subscripts
in supercritical flows in junctions are described. The junction
angle is proved not to be a governing parameter in the evalu-
A = wave A
ation of wave maximum heights. For invert-aligned junctions
B = wave B
the variation of upstream branch diameters is not sufficient to
C = wave C
vary the occurrence conditions of shock waves. As to obvert-
d = dynamic
aligned junctions, the shock wave patterns are slightly changed.
L = lateral branch
In particular, wave A occurs also for junction regimes with-
o = straight branch
out dominant straight flows. Oppositely, shock wave C does not
s = static
occur when large bottom offsets are foreseen at the straight inlet
u = downstream branch
and its Froude numbers are large. In accordance with previous
wc = water cushion
investigations, shock wave B is the largest wave. It dictates the
definition of the manhole freeboard independently of the even-
tual presence of bottom or top offsets at the manhole inlets. This ORCID
research proposes a relation to estimate the corresponding wave
B maximum only as a function of the lateral dynamic force Gaetano Crispino http://orcid.org/0000-0002-3889-1115
component. Corrado Gisonni http://orcid.org/0000-0002-9220-2149
Journal of Hydraulic Research (2018) Supercritical flow in junction manholes 13

References Köestler, V. J. (2014). Numerical simulation of supercritical

flow in junction manholes (Unpublished master’s thesis).
Chanson, H. (2009). Turbulent air-water flows in hydraulic Laboratory of Hydraulic Constructions, Ecole Polytechnique
structures: Dynamic similarity and scale effects. Environmen- Fédérale de Lausanne, Lausanne.
tal Fluid Mechanics, 9(2), 125–142. Krüger, S., & Rutschmann, P. (2000, August). Numerical sen-
De Martino, F., Gisonni, C., & Hager, W. H. (2002). Drop in sitivity analysis of supercritical confluences to inaccuracy
combined sewer manhole for supercritical flow. Journal of of upstream boundary conditions. In Proceedings of the 4th
Irrigation and Drainage Engineering, 128(6), 397–400. international conference on hydroinformatics (p. 253). Iowa
Del Giudice, G., & Hager, W. H. (2001). Supercritical flow in City: Iowa Institute of Hydraulic Research.
45° junction manhole. Journal of Irrigation and Drainage Motlagh, Y. Y., Nazemi, A. H., Sadraddini, A. A., Abbaspour,
Engineering, 127(2), 100–108. A., & Motlagh, S. Y. (2013). Numerical investigation of the
Falvey, H. T. (1980). Air-water flow in hydraulic structures. effects of combining sewer junction characteristics on the
Denver, CO: US Bureau of Reclamation. hydraulic parameters of flow in fully surcharged condition.
Flow Science. (2012). FLOW-3D (10.1.0). Santa Fe, NM: Flow Water and Environment Journal, 27, 301–316.
Science. Niedermann, E. (2013). Physical model investigation of super-
Geberemariam, T. K. (2017). Numerical analysis of stormwa- critical flow in 45° junction manholes (Unpublished mas-
ter flow conditions and separation zone at open-channel ter’s thesis). Laboratory of Hydraulic Constructions, Ecole
junctions. Journal of Irrigation and Drainage Engineering, Polytechnique Fédérale de Lausanne, Lausanne.
143(1), 05016009, 1–9. Pfister, M., & Chanson, H. (2014). Two-phase air-water flows:
Gisonni, C., & Hager, W. H. (2002). Supercritical flow in the Scale effects in physical modeling. Journal of Hydrodynam-
90° junction manhole. Urban Water, 4(4), 363–372. ics, Ser. B, 26(2), 291–298.
Gisonni, C., & Hager, W. H. (2012). Idraulica dei sistemi fog- Pfister, M., & Gisonni, C. (2014). Head losses in junction man-
nari: Dalla teoria alla pratica [Wastewater hydraulics: From holes for free surface flows in circular conduits. Journal of
theory to practice]. Milan: Springer-Verlag. Hydraulic Engineering, 140(9), 06014015, 1–6.
Gisonni, C., & Pfister, M. (2015). Discussion of ‘Hydraulic Ramamurthy, A. S., Han, S. S., & Biron, P. M. (2013). Three-
behaviour of junction manholes under supercritical flow con- dimensional simulation parameters for 90° open channel bend
ditions’ by Saldarriaga, J., Bermúdez, N. and Rubio, D. P.. flows. Journal of Computing in Civil Engineering, 27(3),
Journal of Hydraulic Research, 53(2), 286–289. 282–291.
Gökok, T. (2013). Physical model investigation of supercriti- Saldarriaga, J., Bermúdez, N., & Rubio D. P. (2012). Hydraulic
cal flow in junction manholes. (Unpublished master’s thesis). behaviour of junction manholes under supercritical flow con-
Laboratory of Hydraulic Constructions, Ecole Polytechnique ditions. Journal of Hydraulic Research, 50(6), 631–636.
Fédérale de Lausanne, Lausanne. Saldarriaga, J., Rincon, G., Moscote, G., & Trujillo, M. (2017).
Granata, F., de Marinis, G., & Gargano, R. (2015). Air-water Symmetric junction manholes under supercritical flow condi-
flows in circular drop manholes. Urban Water Journal, 12(6), tions. Journal of Hydraulic Research, 55(1), 135–142.
477–487. Schwalt, M. (1996). Vereinigung schiessender abflüsse [Com-
Gurram, S. K., Karki, K. S., & Hager, W. H. (1997). Subcriti- bining outflows]. Wasser-Abwasser, 137(6), 326–330.
cal junction flow. Journal of Hydraulic Engineering, 123(5), Schwalt, M., & Hager, W. H. (1995). Experiments to
447–455. supercritical junction flow. Experiments in Fluids, 18(6),
Hager, W. H. (1987). Discussion of ‘separation zone at open- 429–437.
channel junctions’ by James L. Best and Ian Reid (November, Zhao, C.-H., Zhu, D. Z., & Rajaratnam, N. (2004). Supercriti-
1984). Journal of Hydraulic Engineering, 113(4), 539–543. cal sewer flows at a combining junction: A model study of
Hager, W. H. (2010). Wastewater hydraulics: Theory and prac- the Edworthy trunk junction, Calgary, Alberta. Journal of
tice (2nd ed.). Berlin: Springer-Verlag. Environmental Engineering and Science, 3(5), 343–353.