1. Introduction
Bridges that are constructed over waterways, with piers located within the flow path, are more vulnerable to failure due to pier scour. Bridge pier scour is the erosion of bed material from the region surrounding the pier’s base [
1]. During high-flow events, large quantities of sediment can be removed from around and beneath the foundations of piers jeopardizing the structural integrity of the piers and consequently the bridge [
2]. There are approximately 500,000 bridges in the USA National Bridge Inventory (NBI) that are built over waterways, amounting to nearly 83% of all the bridges in the NBI [
2,
3]. In 2011, 23,034 of such bridges in the NBI were classified as scour-critical, which, according to the USA Federal Highway Administration (FHWA), means that they are predicted to fail if a given flood event were to occur [
2]. Therefore, bridge pier scour is an ongoing problem that affects many bridges in the USA, and elsewhere, ultimately endangering the safety of the public.
Bridge pier scour is caused by a three-dimensional flow separation induced by the pier. As the flow separates around the pier, three main scour-causing mechanisms arise: the horseshoe vortex upstream of the pier, flow acceleration along the sides of the pier, and wake vortices downstream of the pier (depending on the Reynolds number). When present, these components coincide and form the scour holes observed around bridge piers [
1,
4,
5,
6,
7]. Detailed studies were performed by Dey et al. and Tafarojnoruz and Lauria, which examined the flow behaviour around piers by means of experimental and numerical investigations, respectively [
8,
9]. These studies identify the contribution of each of the aforementioned scour-causing mechanisms as the scour hole around the pier develops.
Bridge pier scour can occur under two conditions: clear water or live bed. Unlike clear-water scour, live-bed scour has the contribution of sediment from upstream, which replenishes, to some degree, the scour hole around the pier [
2,
6]. For this paper, only clear-water conditions will be studied.
Many parameters influence bridge pier scour. Over the years, extensive research has taken place to better understand the influence of each parameter; however, there are still some parameters for which the knowledgebase is limited. One such parameter is the presence of an ice cover around the pier. Ice on the water’s surface, which acts as an upper boundary, has been found to change the flow behaviour beneath. Specifically, the velocity (
u) profile under an ice cover transitions from a logarithmic shape with the maximum
u near the surface to that which resembles a pipe flow with the maximum
u occurring at approximately mid-depth [
10]. Due to the location of the maximum
u shifting downwards within the flow depth, an increased
u gradient occurs near the bed, which can induce greater bed shear stresses. As a result, more scour can occur [
11,
12,
13].
Ice covers can possess a wide variety of shapes and configurations influenced by various factors such as weather and location. As a result, the roughness of the underside of an ice cover can vary drastically within a given reach and season, making the ice cover roughness highly irregular [
14,
15,
16]. If the ice is thermally grown only, the roughness will frequently be minimal, and the ice will have a smooth surface contacting the passing flow [
12]. On the other hand, mechanical thickening processes due to ice shoving can generate a rough ice cover [
17]. For example, most often, later in the season, during the breakup period, large pieces of ice can break apart into smaller ice fragments and, similar to debris, accumulate in random patterns generating a rough surface, sometimes referred to as an ice jam [
11]. Due to the unsafe conditions during field measurements of an ice jam roughness, limited data are available [
16]. However, Beltaos used a remote technique consisting of a floating sensor that was deployed beneath the ice jam [
14]. The floating sensor was carried with the flow while recording the elevation of the underside of the ice jam for multiple kilometers. The mean elevation and the fluctuations from the mean were translated into the average thickness and the hydraulic roughness of the ice cover, respectively. For the 20 data sets collected at various locations, the absolute roughness values (taken as the 84th percentile of the absolute deviation from the mean) varied from 0.24 to 1.15 m, averaging 0.79 m, while omitting one data point with an extremely large value of 2.34 m [
14]. Zare et al. utilized a bottom-mounted Acoustic Doppler Current Profiler (ADCP) over a four-month period that included spring breakup to measure
u profiles continuously at a fixed location in Nelson River, Canada [
10]. Hourly averaged
u profiles were fit to the log law to estimate the roughness of the bottom side of the ice surface, with equivalent Manning’s roughness values ranging from near zero (smooth) to as high as 0.08 (very rough). Therefore, an ice cover can possess an underside that is smooth or extremely rough.
The roughness of the underside of an ice cover, in conjunction with the bed roughness, has an influential role in shaping the corresponding
u profile. If the ice and the bed’s roughnesses possess similar magnitudes, the location of the maximum
u will occur at approximately mid-depth. However, if the surface has a greater roughness than the bed, the location of the maximum
u could occur below mid-depth. This can cause a higher
u gradient next to the bed, resulting in increased shear stress and potentially greater scour [
11,
12,
17].
Solid ice covers that span the entire width of the river, which is the focus of this study, can be either unattached or attached to the banks, such that the ice is floating or fixed, respectively. When the ice is floating, it can raise or lower depending on the stage of the river. However, when the ice is attached to the banks, it is unable to adjust to varying water levels causing pressurized conditions to occur beneath the ice. Furthermore, ice covers, especially ice jams, frequently grow in size in the vertical direction due to the mechanical thickening process [
18]. When an ice cover attached to the banks grows vertically, the cross-sectional area of the river channel beneath the ice decreases, and a rise in the upstream water level can occur. When this happens, the passing flow is confined to the smaller cross-sectional area causing there to be an increase in velocity, resulting in even greater scour [
12,
19,
20].
The Hydraulic Engineering Circular No. 18 (HEC-18) has identified that an ice formation or jam is a factor that affects the local scour depth around bridge piers, as a more severe scour condition can occur with a smaller flow rate (
Q). HEC-18 further states that there are many examples of foundation scour from accelerated flow beneath an ice covering, but limited field measurements of scour induced by ice jams exist. When designing a bridge, HEC-18 requires that ice effects be considered when calculating the maximum scour depth, but the HEC-18 pier scour equation does not consider ice covers. Instead, HEC-18 suggests obtaining scour data from nearby bridges to estimate the expected scour [
2]. This is not always an accurate approach. According to Wuebben, the resulting bathymetry after an ice covering has subsided may not represent the deepest scour, as a portion of the scour hole could have been refilled during the ice breakup period [
16].
Various researchers have identified ice covers as a parameter that requires further research, e.g., Ettema et al. ranked ice covers at a medium-level priority in terms of bridge pier scour research needs [
21]. Wu et al. performed a bridge pier scour study in the presence of a floating ice cover, which examined the effect of different pier diameters and water depths on scour depth and scour width [
13]. The limitations of this study were that pressurized ice covers and different ice cover roughnesses were not considered. Ackermann et al. studied bridge pier scour with a floating ice cover possessing either a smooth or a rough surface [
22]. Various velocities under both live-bed and clear-water conditions were tested. The limitations of this research were that pressurized conditions were not tested and only the maximum scour depth was recorded. Lastly, Hains and Zabilansky performed a thorough study of bridge pier scour under a floating and pressurized ice cover [
23]. The limitations of this study were that live-bed conditions were reached in a number of clear-water tests making the final scour data for these points unusable, and
Q was changed for each test rather than being kept constant.
It is understood that ice covers influence the hydrodynamics of the flow passing beneath, which can in turn increase bed erosion. However, the full extent of the impacts an ice cover has on bridge pier scour are not known. Therefore, the intent of this study is to expand the existing knowledgebase pertaining to bridge pier scour in the presence of an ice cover. In addition to maintaining a constant flow rate and remaining in the clear-water regime amongst all tests, the objectives are to:
Examine the differences in scour for floating versus fixed ice covers;
Investigate the effect of different levels of ice cover submergence (flow pressurization) on scour;
Evaluate the influence of both smooth and rough ice covers on scour.
4. Discussion
In this study, a number of conditions pertaining to the presence of surface ice around bridge piers, in regard to bridge pier scour, were analysed. One objective, which proved to be challenging to correctly replicate experimentally, was a floating ice cover. In the field, a floating ice cover can both freely adjust its elevation to accommodate the changing water levels, and it can also protrude into the passing flow to some degree, depending on the characteristics of the ice. Therefore, it is possible that a floating ice cover can generate accelerated flow beneath, which could result in similar scour behaviours to that of the mildly submerged fixed ice covers. To overcome this issue, a range of si levels were tested from just touching the surface to protruding into the flow 30% of y. Since the flow depth throughout each test remained constant and the ice cover was weighted to achieve the desired si, whether the ice cover was fixed in place or not would have no impact on the scour results. This implies that the submerged ice cover cases could represent a floating ice cover that protrudes into the flow.
Upon analysing the velocity and scour results with and without an ice cover, it is evident that the presence of any ice cover increases scour. Specifically, as an ice cover grows and becomes more submerged, the quantity of pier scour becomes greater. These scour results were confirmed by the general trend observed from the
u data, which indicated an increase in the near bed
u gradient as the ice covers became more submerged. Both a smooth and rough ice cover were tested, and this was to reach both ends of the possible roughness spectrum. It was discovered that the rough ice cover induced more scour than the smooth ice cover under all conditions. In addition, the quantity of scour also increased at a greater rate for the rough ice cover than the smooth ice cover, as
yi decreased. For these reasons, the rough ice cover should be used from a design perspective in order to achieve a conservative design. Since many pier scour equations do not take into consideration the effects of ice covers, the results presented in
Figure 10a can be used in addition to such equations to help provide further insight.
Wu et al. presented equations that relate nondimensional scour to pier Froude number, under open-channel conditions, as shown in Equation (5), and smooth ice cover conditions, as shown in Equation (6) [
13]:
Equation (6) from Wu et al. and Equation (3) presented in this paper both pertain to a smooth ice cover and possess a number of similarities (
Figure 10b) [
13]. Specifically, the constants are almost identical, and the coefficients are relatively close given the differences in the experimental conditions. This lends credence to both Equations (3) and (6), and indicates that the experimental results presented in this paper are reasonable. The experimental conditions of Wu et al. varied from this study in that
y = 0.108 m, 0.150 m, 0.210 m,
d50 = 0.00051 m, and
uavg = 0.24 m/s [
13]. It can be noted that of the different experimental conditions,
d50 and
y are not considered in Equations (3) and (6), only
uavg.
When comparing the equations pertaining to a smooth ice cover (Equations (3) and (6)) to Equation (4) presented in this paper, which pertains to a rough ice cover, there are notable differences. The constant and the coefficient for Equation (4) are significantly greater than that of Equations (3) and (6). Wu et al. stated that the presence of a smooth ice cover, when compared to no ice cover, has only a limited influence on the pier Froude number [
13]. Meanwhile, the results presented in this study (Equations (3) and (4)) show that the presence of a rough ice cover, in comparison to a smooth ice cover, has a substantial influence on the pier Froude number. Therefore, given the findings of this paper and that of Wu et al., the pier Froude number only becomes influenced with rough ice covers [
13].
The work presented in this paper contains a number of limitations which warrants further research. First, bridge pier scour is a complex process which is influenced by numerous parameters, and due to the scope of the project, not all parameters could be examined to their full extent. Specifically, only one Q, y, D, and d50 were tested, and all the tests lied within the clear-water regime. The second limitation to this paper is that only one size of ice covering was used. The ice covering extended in the upstream and downstream direction a far distance from the pier, as to replicate an infinitely long ice cover, but perhaps shorter ice covers, such as ones localized around the pier, could have a different effect on the quantity of pier scour. The third limitation is that the ice cover itself was artificial as the smooth surface consisted of treated plywood and the rough surface consisted of PVC panels. While using artificial materials improves constructability, and the chosen materials were intended to mimic natural river ice, it is possible that natural river ice would induce different flow characteristics. The last limitation is that only one u profile was measured for each test condition and it was collected in the center of the flume upstream of the pier. Additional u profiles would be beneficial as they would create a more detailed flow field; however, this is difficult to achieve with an ADV in the presence of a solid ice cover.
A point worth noting is that when the ice cover was submerged, the flow depth beneath the ice cover consequently reduced. In the absence of an ice cover, the authors acknowledge that a change in flow depth could influence the scour depth. Therefore, the scour depth experienced under a submerged ice cover could be a combination of both the flow pressurization and the reduction in flow depth. Further research is required to distinguish the contribution of both factors.