Introduction
To include blockage effects in hydraulic culvert design, Bradley et al. (
2005) recommend modifying existing hydraulic design methods, accounting for changes to the inlet geometry, surface roughness, and cross section shape and area. While roughness and cross section parameters for the culvert barrel can be modified using existing methods, empirical entrance loss coefficients and head-discharge relationships for partially blocked culvert inlets are limited. Inlet blockage conditions can be classified into bottom-up, top-down, and porous plug blockage conditions (
Weeks and Rigby 2016;
Ollett et al. 2017). Several studies have determined the entrance loss coefficients and head-discharge relationships for gated culverts, representative of top-down blockage of the inlet (
Horowitz and Matys 2015;
Zeng et al. 2020;
Cantero-Chinchilla et al. 2023). Kranc et al. (
1990) tested the effects of culvert inlet grate blockage, and Tullis et al. (
2008) tested the effects of pipe culvert embedment using physical model experiments. Empirical design values accounting for bottom-up blockage effects are presently not available, but the theoretically derived ELM and RAM have been proposed for estimating inlet blockage effects and imply that are more resilient to blockage effects (
Weeks et al. 2009;
Witheridge 2009).
Under outlet control conditions, the entrance loss coefficients for circular inlets are reported to be Reynolds invariant for
(
Idelchik 1986) and scale invariant for pipe diameters over 300 mm (
Tullis et al. 2008). Under submerged inlet control conditions, the use of vented inlets has been shown to reduce subatmospheric air pressure in the culvert barrel, thereby ensuring significant scale invariance for culvert diameters larger than 140 mm (
French 1957,
1961). Kolerski and Wielgat (
2014) used scale model experiments to show that the approach flow velocity is significantly unaffected by the local inlet flow field
upstream from the inlet for pipe culverts.
Experimental Setup
A physical culvert model was built in the hydraulic laboratory at the Norwegian University of Science and Technology (NTNU) (Fig.
2). Water was supplied through two 300 mm pipes, with built-in Siemens SITRANS FM MAG 5000 flow meters. The flow entered the head tank through a plastic mesh diffusion chamber, and the approach headwater elevation was measured relative to the inlet section invert using a millimeter scale located 1,125 mm upstream from the inlet face section where the approach flow was assumed unaffected by the local flow field near the inlet (
Kolerski and Wielgat 2014). The culvert barrel was constructed from a circular PVC pipe of internal diameter
and length
. For the outlet control (type 4 flow) conditions, a nominal zero slope was used for the culvert barrel, and the tailwater level was controlled using a weir placed at the downstream end of the tailwater tank to ensure a submerged outlet. For inlet control (type 1 and 5 flow), a nominal slope of
and a 400 mm vertical drop downstream from the outlet were used to ensure supercritical flow conditions in the culvert barrel and a free outlet. To avoid scale effects related to subatmospheric air pressure in the barrel and approach flow conditions, a 70 mm PVC pipe was installed into the inlet crown, 300 mm downstream of the headwall (
French 1961). Seven commonly used pipe culvert inlets were tested and installed with exposed inlet invert edges (Fig.
3 and Table
1). The culvert barrel was supported by beams and vertical columns in order to ensure a constant barrel slope, and four WIKA S-11 pressure sensors were mounted along the culvert barrel invert in order to determine the piezometric grade line through the culvert barrel (Fig.
2).
The different blockage elements used are shown in Fig.
4 and represent a span of idealized versions of bottom-up blockage conditions observed by the corresponding author during culvert inspections during and after floods. The blockage elements were made from plywood and PVC foam, with either untreated foam surfaces or rocks of
or 8 mm glued to the top and upstream sides. The rock sizes were chosen so that the blockage shape was not significantly altered, allowing for evaluation of both shape and roughness effects. No rocks were used within 18 mm upstream of the inlet face to ensure that the blockage ratio could be accurately determined. Photos of the model setup are shown in Figs.
5–7.
Experimental Procedure
For each experiment, a single blockage element was used, and measurements were taken at different headwater elevations on a rising stage. The headwater was allowed to stabilize before measurements were taken over 60 s, at 100 Hz. The total head (
) was calculated based on the measured piezometric head (
) and the average velocity (
). The headwater elevation (
) was determined using the inlet invert as a datum, assuming no friction losses between the measurement section and the inlet. Three different sets of experiments were conducted for both outlet control (type 4 flow) and inlet control (type 1 and 5 flow) conditions. First, the unblocked inlets were tested in order to compare the experimental results to previous studies (
French 1955,
1961;
Idelchik 1986) and to determine the unblocked discharge (
) used in the
ratio. The effect of the blockage ratio was then tested using 18 mm smooth blockage plates and varying blockage ratios (Fig.
8). Finally, the effects of blockage shape and roughness were tested using the blockage elements shown in Fig.
4 (Fig.
9). Blockage elements of
(
) and
(
) were used to evaluate shape and roughness effects over a span of blockage ratios. For the nonmitered inlets, blockage elements of width 750 mm (2
) were placed in front of the inlet section. For mitered inlets of nonconstant cross section areas (such as partial blockage), the inlet section cannot be uniquely defined (
French 1961). The blockage elements were therefore fitted inside the mitered part of the inlet and tested both at the upstream mitered edge (US) and the downstream crown edge (DS) of the inlet (Fig.
10). Due to the large number of blockage conditions tested, only smooth blockage surfaces were used for the mitered inlet. In all cases, the blockage ratio was calculated based on the blockage and culvert barrel area perpendicular to the culvert barrel axis.
Outlet control conditions were tested using type 4 flow. The piezometric grade line was evaluated to confirm that it was located above the barrel crown and that the inlet and outlet were submerged for the lowest discharge for each experiment, indicating pressure flow conditions. The entrance head loss was calculated as the difference between the total approach head in the head tank, and the average total head at the inlet section projected from the location of the four pressure sensors in the pipe, accounting for friction losses. The friction slope was determined using Eqs. (
2) and (
3) and a culvert barrel roughness height of
(
Rocha et al. 2017). The entrance loss coefficient was calculated for 5–7 values of
in the range
, and the average entrance loss coefficient was calculated as the simple average for each experiment. Eq. (
5) and regression analysis were used to determine the relationship between the blockage ratio and the entrance loss coefficient, and these relationships were then compared to ELM and the experimental results for embedded culverts given in Tullis et al. (
2008). The results are given as the average entrance loss coefficient for the unblocked and blockage ratio experiments. Due to the large number of experiments, only the average, maximum, and minimum entrance loss coefficients were evaluated for the blockage shape and roughness experiments. Finally, the maximum
values were compared to ELM and embedded culverts. The full set of entrance loss coefficients is included in Table
S1.
Inlet control conditions were tested using type 1 and 5 flow. The Froude number was evaluated to confirm supercritical flow in the culvert barrel using the piezometric head as the characteristic length and the average flow velocity as the characteristic velocity. Inlet submergence was determined based on the semidimensionless discharge for the unblocked part of the inlet (
), using
as the criterion for type 1 flow and
for type 5 flow (
Schall et al. 2012). For each experiment, eight measurements of
and
were taken in the range
. Different head-discharge equations were tested, and it was found that Eq. (
7) gave the lowest estimation errors. The head-discharge relationship was determined based on the full cross section of the barrel for direct comparison between the unblocked and blocked conditions, and
was calculated for corresponding inlets and values of
. The resulting
ratios were compared to RAM and the experimental results for embedded culverts from (
Tullis et al. 2008). The results from the unblocked inlet control experiments are given as comparisons of
to the results of French (
1961) and as
ratios for the blockage ratio experiments. Due to the large number of experiments, only the average, maximum, and minimum
ratios were evaluated for the blockage shape and roughness experiments. Finally, the minimum
values were compared to RAM and embedded culverts. The full set of head-discharge equations is included in Table
S2.
To adjust for small differences in blockage height (
) between experiments, the entrance loss coefficients and head-discharge relationships for the shape and roughness experiments were adjusted based on
, for direct comparison using
and 0.748. For both inlet and outlet control conditions, repeatability was tested for the
bevel inlet, which due to the high efficiency gave the largest relative errors based on differences between experiments. Due to the large number of experiments, repeatability was not tested further, but the uncertainty was estimated using Eq. (
9) and compared to the estimation errors for the complete set of entrance loss coefficients and head-discharge relationships. The accuracy of the measurement equipment and uncertainty of all model parameters are given in Table
12 in the Appendix. To evaluate the validity of the results, the Reynolds number for the culvert barrel was evaluated for the outlet control (type 4 flow) experiments. The barrel diameter was used as the characteristic length, and the dynamic viscosity was determined based on the measured water temperature (
White 2016).
Application of Results for Different USGS Flow Types
Given the variation in in-situ blockage conditions, it is recommended to use the maximum
values, and minimum
values as a conservative estimate of blockage effects. For outlet control conditions, the present study has covered outlet control with submerged inlet and outlet (type 4 flow), and the
values are valid for pressure flow conditions (type 4, 6, and 7 flow) (
Schall et al. 2012). For these flow conditions, bottom-up blockage conditions can be estimated similarly as for ELM, using values of
and Eq. (
5). As
has been shown to vary considerably for
, the
values are not valid for outlet control flow with an unsubmerged inlet (type 2 and 3 flow) (
Smith and Oak 1995;
Tullis et al. 2008), and further study of blockage effects for these flow types is warranted. For inlet control, the present study has covered both unsubmerged and submerged conditions (type 1 and 5 flow), and bottom-up blockage effects can be estimated similarly to RAM.
In this study, the blockage element was treated as part of the inlet, and it was found that the results could be accurately described using a constant value for the entrance loss coefficient and 5th-degree polynomial equations for the head-discharge relationship, similar to existing design methods for unblocked culverts. The present study has therefore not investigated the classification of novel flow types specific to culverts with bottom-up inlet blockages, but the large number of flow types identified for gated culverts in Cantero-Chinchilla et al. (
2023) indicates that there may also exist flow types beyond type 1–7 for this condition. In particular, unsubmerged outlet control (type 2 or 3 flow) for an unblocked culvert might change to inlet control (type 1 or 5) due to the presence of bottom-up blockage when the flow depth in the culvert barrel is lower than the blockage height. Further work on the classification of flow types for culverts with bottom-up blockage is therefore warranted.
The blocked entrance loss coefficients and discharge ratios were determined for comparison to ELM and RAM and affected the headwater elevation in different ways. Under outlet control conditions, the entrance head loss is one of several head losses that influence the headwater elevation. The effect of on the headwater elevation will therefore vary with the culvert design, but given that can approach values two to three orders of magnitude higher than for unblocked inlets, it can be expected that entrance losses will constitute a large fraction of the total head loss at higher blockage ratios. For inlet control (type 1 and 5 flow), the ratio modifies the head-discharge relationship directly, resulting in significantly higher headwater elevations for a given discharge.