Overtopping-Induced Embankment Breaching Experiments: State-of-the-Art Review on Measurement and Instrumentation

Abstract

The breaching of embankments have devastating consequences for the economic, human, cultural, and environmental assets. One of the most widely used approaches for understanding the characteristics of embankment breaching is through laboratory and field-scale experiments. Despite the advancements in instrumentation and measurement techniques of embankment breaching experiments, there is a lack of comprehensive documentation. In this review, the advancements and state-of-the-art instrumentation and measurement techniques employed in overtopping-induced embankment breaching of laboratory and field-scale experiments are discussed. The key parameters commonly measured in experimental modeling are breach morphological changes, reservoir and flow depth, velocity, breach outflow, and pore water pressure. Instrumentation for breach morphological change detection, including mechanical, photography, photogrammetry, electronic sensors, and laser technologies, are reviewed. The various flow velocity measuring techniques, such as Particle Tracking Velocimetry (PTV), Particle Imaging Velocimetry (PIV), acoustic, and radar-based techniques, are discussed. Instrumentation for water level, flow rate, and pore pressure measurements are also briefly documented. The challenges and constraints encountered during embankment breaching experiments are discussed. The review further suggests future perspectives in improving the accuracy of breach detection, velocity, and pore pressure measurement techniques. Additionally, improving scale effects by incorporating geotechnical factors is also recommended.

1. Introduction

Embankment dam breaching can be triggered by different causes. Most breaching incidents occur due to erosion by flowing water either through or over the dam or foundation material, followed by other causes such as mass sliding and slumping resulting from pore pressure buildup in the embankment. Among these causes, overtopping contributes to most of the occurrences worldwide [1,2,3,4]. In recent years, the effect of climate change has accelerated the occurrence and amount of rainfall and thereby increased inflows to dam reservoirs [5]. When such inflows exceed the storage capacity, overtopping occurs and leads to the failure of embankment dams. Such failures can result in a massive release of water stored in the reservoir as breach outflow.

The breach outflow from embankment dam failures can have devastating consequences for the economic, human, cultural, and environmental assets of the society living downstream of the structure. A recent example is Libya’s catastrophic flooding on 11 September 2023, which was caused by the extremely high floods from the storm Daniel which resulted in the breaching of two cascaded dams located upstream of Derna city. The outflow from the failure of these dams led to more than 5800 deaths, 8000 missing, and around 45,000 inhabitants displaced [6,7]. It also swept a large chunk of Derna and other nearby cities’ infrastructure largely damaging the health, water, education, and other facilities in these areas. The economic losses were estimated to be close to USD 1.8 billion [7]. Other catastrophic dam failures in recent years include the failure of an earth dam in 2009 in Indonesia that caused the death of more than 100 people and the destruction of many residential buildings [8]. Similarly, the failure of the Patel dam in Kenya (2018), which led to nearly 50 causalities [9], is another example. A visual overview of past notable embankment breaching incidences is shown in Figure 1. This highlights the need for improved research methodologies and techniques to investigate the occurrence and behavior of breaching of embankment dams.

Understanding the breaching processes of embankment dams is believed to have a significant role in minimizing the consequences of dam failures. However, it is difficult to monitor the failure of real dams mainly due to the unpredictability of floods and the inaccessibility of dams during such flooding events. Thus, it is common to use analytical, numerical, or experimental models to drive inferential results of the breaching behavior of embankment dams. Experimental hydraulic models are primarily designed and built to represent real-world hydraulic structures to make effective use of technical and financial requirements solving problems pertinent to hydraulic engineering [14]. Over the past four decades, many experimental studies have been conducted to understand the behavior of embankment breaching due to overtopping at laboratory [15,16,17,18] and field scales [19,20,21].

Research in the field of experimental modeling of embankment dam breaching has advanced significantly in the past few decades, as have reviews on the subjects. Among the most recent reviews, a notable study by Amaral et al. [22] reviewed methods and instruments with a wide range of measurement parameters (morphology detection and image reconstruction, velocity, water level, and flow measurements). However, the scope of this study was limited to studies conducted at the Portuguese National Laboratory for Civil Engineering (LNEC). Zhong et al. [1] reviewed the failure modes of the embankment and landslide dams by compiling historical, experimental, and field investigations. Dezert et al. [23] compiled nine laboratory studies conducted at the hydraulic laboratory of the Norwegian University of Science and Technology (NTNU) targeted at investigating the effect of downstream slope riprap placement and its resistance to varying discharge on rockfill dams. Aureli et al. [24] documented numerous historical and real laboratory test cases of dam breaks useful for validating numerical models. Despite the progress in improving the knowledge in this regard, there is a lack of comprehensive documentation, in scope and depth, of the advances made in instruments and measurement techniques of dam breach experiments conducted over the past two decades.

In this paper, most of the studies reviewed focus on advancements in instrumentation and measurement techniques for laboratory and field experiments on earth and rockfill embankment breaching due to overtopping. Despite the differences between embankment dams and levees/dike embankments, breaching experiments conducted on both structures have similarities in the use of instruments and measurement parameters. Similarly, though landslide dams have different modes of formation compared to embankment dams, laboratory experiments performed on both structures regarding instrumentation and measurement have many common features. Hence, breakthrough experimental studies on levees, dikes, and landslide dams exhibiting structural and hydraulic (i.e., non-wave overtopping) similarity with embankment dams are included in this review, depending on their relevance and significance. The main objectives of this paper are (1) to review the advances in state-of-the-art instrumentation and measurement techniques used in dam breaching experiments worldwide over the past two decades and (2) to shed light on some of the major challenges faced in conducting experimental research on embankment dam breaching due overtopping, a loading condition caused by extremely high inflows.

2. Review Methodology

A literature search was conducted on embankment dam breaching experiments from electronic academic repositories such as Scopus and Google Scholar. The major sources used in this review are journal papers, conference proceedings, book chapters, theses, and instrument specifications. The literature search was conducted in two rounds.

The first round of the literature review started by setting up a search query followed by repeated refining. All references (title, abstract, keywords) that fulfill the primary search criteria were downloaded from Scopus for further screening. The articles were first screened by title. Title screening was based on their relevance to the review scope. Publications on less related topics, such as numerical modeling, wave-overtopping, seepage, and seismic-induced breaching experiments, were eliminated from the list. Next, abstract screening was performed; the main criteria were to check whether the experiments were used and whether they fall within the scope of this paper. Lastly, full-text screening was conducted, and publications that appeared to have less contribution to the advancement in instrumentation and measurement were dropped. In this round, a draft structure and outline of the paper was set. Moreover, the major contents for each section were extracted from each publication.

After making refinements, the search query used to locate relevant publications was (TITLE-ABS-KEY (‘embankments’ OR ‘embankment’ OR ‘dam’ OR ’rockfill’ OR ‘levee’ OR ‘dike’ OR ‘riprap’) AND TITLE-ABS-KEY (‘breach’ OR ‘breached’ OR ‘breaching’ OR ‘overtopping’ OR ‘overtopped’) AND TITLE-ABS-KEY (‘laboratory’ OR ‘flume’ OR ‘experiment’ OR ‘experimental’)), where TITLE, ABS, and KEY are the title, abstract, and keywords of the publication, respectively. In addition, the temporal scope of the search was limited to works published from 2000 to 2024.

The second round of the literature review was focused on reinforcing the prepared outline and contents from round one. Here, a search for publications relevant to each section was conducted on Google Scholar. Moreover, online platforms such as connectedpapers.com and www.researchrabbit.ai/ were used to trace related and derivative publications on a given topic. These emerging tools provide an interesting opportunity to exhaustively search for literature.

3. Embankment Breaching Parameters

The key parameters to measure during an embankment breaching experiment can be classified into geometric and hydraulic. These parameters are crucial to understand the breaching characteristics of embankment dams. The geometric parameters are related to the breach morphological changes such as breach width, depth, and how they evolve with time. The hydraulic parameters are water levels, inflow, breach outflow, flow velocity, and pore water pressure (Figure 2). The advances in instrumentation and measurement of these parameters are discussed in the sections below.

4. Advances in Breach Morphology Detection Techniques

Detecting the morphology of an evolving breach, either in laboratory experiments or field tests, has undergone continuous improvements over the past few decades. So far, many studies have been able to depict the breach geometry with sufficient accuracy both spatially and temporally. Some studies utilize simple 2D profiles in cross-sectional or longitudinal sections while others develop a 3D representation of the breach. The techniques employed in breach detection can be classified depending on the materials and methodology used. In this review, the techniques for breach detection are categorized as mechanical, laser-based, embedded sensors, photography, and photogrammetry methods. The classification is based on the dominant components used in the techniques, but some of the methods in use are rather a combination of one or more of these described techniques. For example, some studies employ mechanical methods supported by photography to monitor breach evolution. Breach morphology detection by embankment material, measurement dimension, and accuracy for selected instrumentation and measurement techniques is summarized in Table 1. In addition, summaries of the advantages, disadvantages, and limitations of the instruments and techniques used for breach morphology detection are presented in Table 2.

4.1. Mechanical Techniques

In many studies involving embankment breaching experiments, especially for large-scale and field-level tests, mechanical techniques are the most preferred method due to their simplicity and convenience. Mechanical tools are commonly assisted by image or video recordings from cameras of high resolution to trace the dynamics of breach with time as it is almost impossible to do it manually. For laboratory-scale experiments, using a simple ruler stamped on the walls of a flume is very common to measure the reduction in height of the crest level [25]. For 2D mapping or to develop full 3D breach morphology, multiple gauged rods/staff running from bridges above the embankment arranged longitudinally and/or cross-sectionally are used [26,27,28,29,30,31]. The gauged rods can be of moving type or fixed from the top. In a field-scale test, Hanson et al. [19] used point gauge instruments mounted on a rolling carriage to determine the breach cross-section, the shape of the breach bottom, and the elevation of water above the breach. Examples of typical mechanical techniques used for breach morphology detection are shown in Figure 3a for 2D and Figure 3b for 3D mapping.

Despite their widespread advantages, the use of mechanical techniques involves intrusive instruments, which can affect the water flow and erosion dynamics. For 2D and 3D breach detection, the accuracy of measurements is dependent on the number and diameter of gauges used. The greater the number of gauges, the higher the accuracy of breach geometry. But this comes at the expense of increased interference with the hydro-morpho dynamics of the flow. Thus, the number and orientation of gauges represent a compromise between measurement quality and the level of interference with the flow during the experiment.

Researchers have made efforts to minimize interference. For example, Niu et al. [15] used thin nylon strings, marked at regular intervals, distributed throughout the body of a landslide dam. The 3D of the evolving breach was then reconstructed by locating the position of the breach at each nylon line from images. Replacing the larger diameter/thickness gauging rods with the thin nylon gauges is believed to increase the accuracy of mapping the breach geometry by minimizing the contact area of gauges with the incoming flow. However, despite the efforts made to reduce the effect, due to its intrusive nature, interference of the gauging instrument lines with the flow is inevitable, and the uncertainty levels should be addressed carefully.

4.2. Light-Based Techniques

4.2.1. Laser-Based

With advancements in laser technology for geographic studies, laser-based techniques are now used to capture the dynamics of dam breach morphology. Various types of lasers have been documented for embankment breach experiments. These lasers can be categorized into sweeping-laser point [16], fixed laser sheet [17,32,33], sweeping laser sheet [18], and Light Detection and Ranging (LiDAR) scanners.

The sweeping laser-point technique (Figure 4a) first introduced by Spinewine et al. [16] used a mirror to project a laser point across a section, creating a laser sheet. The laser’s position imprinted on the evolving breach, along with the checkerboard placed at the center of the embankment, is captured by high-speed cameras. The images, in combination with the known position of points in the gridded perforations of the checkerboard, are used to reconstruct the 3D breach morphology. A maximum error of 30 mm in the vertical direction of the breach topography was reported. The sweeping laser completes one move in 5 s, which was assumed to be quasi-instantaneous. Given the rapid dynamics in the flow and morphology, capturing real-time changes is challenging because of the delay in completing one laser sweep.

An improved version of the sweeping laser-point technique was the fixed laser sheet projected onto the top of the embankment (Figure 4b). Combined with 2D and 3D cameras, this technique, applied by Amaral, Viseu and Ferreira [17], helped overcome some of the challenges faced, such as the delay in sweeping, when detecting the morphology of evolving breaches using the sweeping laser-point method.

The use of a sweeping laser plane, also known as Laser Profilometry, marked an improvement over the fixed laser sheet technique (Figure 5). It was first used by Rifai et al. [18] and later adopted for other similar studies [34,35,36,37] for breaching experiments in fluvial dikes. This technique can be used for continuous detection of 3D breach evolution and processes for laboratory model fluvial dikes with an approximate accuracy of 5 mm. It involves three steps, i.e., camera calibration, image processing, and geometric transformation [18]. The system was further improved to capture underwater breach progression besides its reduced cost [37]. When using LP, laser profiles on the top of the water surface tend to be affected by refraction. The reduction in accuracy should be corrected using appropriate techniques [38].

For large and prototype scales, LiDAR-based Terrestrial Laser Scanning (TLS) systems have been used to develop a 3D digital terrain model breach. Leica Station^® laser scanners have been used to monitor the breach progression of a large embankment [20] and a dike [39] breaching at 1 h intervals. A typical post-processing procedure for LiDAR data is summarized in Figure 6a. Figure 6b illustrates an example 3D terrain model scanned using LiDAR, along with the corresponding generated longitudinal profile of a dike breaching experiment conducted by Herrero-Huerta et al. [39].

Recently, laboratory-scale LiDAR-based technologies have emerged as a promising alternative for capturing embankment breaching. Ellithy et al. [41] employed a laser-based instrument, ‘Shallow Water LiDAR (SWL)’, to capture the 3D changes in the shape of an eroding bottom and water profile along a line above it. The advantages and disadvantages of these techniques are summarized in Table 2.

4.2.2. Fringe Projection Methods

For small-scale tests, the fringe projection method was used to capture the three-dimensional dynamic morphological processes during the breach test of non-cohesive embankments [42,43]. The fringe projection system is comprises a slide projector to project fringes of various colors and a camera system to capture continuous images (7.5 Hz) of the breach along with the fringes. As the breach progresses, the distortion in the fringes in the breached area becomes bigger and bigger. These distortions (displacement from the original location) in the fringe shapes are the basis for reconstructing the 3D representation of the breach. An accuracy of ±0.45 mm in breach width and ±0.140 in breach slope angle of the embankment is reported in this study. Furthermore, these methods are sensitive to errors due to refraction and turbidity.

4.2.3. Infrared-Light Methods

Another state-of-the-art method for breach morphology detection is through scanners, which mainly use infrared lights. These 3D depth sensing devices, such as the Kinect sensor, use visible (RGB) and infrared (IR) lights to detect the depth of objects and thereby capture the change in the shape of the breaches [17,44,45,46,47]. Experimental research using such techniques is still in the early stages of development.

4.3. Embedded Electronic Sensors

Researchers have utilized devices buried in the body of the embankment to sense movement in the embankment during breaching and record the time when the movement occurs [48,49]. Most sensors measure the variation in electrical potential as a function of change in breach morphology. The most common electronic devices used for such purposes are potentiometers and microswitch sensors [50], thermistors [51] (Figure 7), and acceleration probes [21,52]. Recent studies also report the use of radio-frequency identification (RFID) sensors to detect head-cut migration in earthen embankments [53].

4.4. Photography

The most basic and frequently used approach for drawing 2D breach profiles in embankment breaching experiments is to capture images from a set of cameras positioned overhead and sideways. These photographs are then processed to extract cross-sectional or longitudinal morphological changes. The use of Virtual Reality (VR) cameras and drones for large-scale modeling of embankments is also an emerging technique [54].

Detection of the boundary between the breach bottom and the water–sediment mixture can be performed manually or through computer applications. Given that breaching experiments often generate numerous images, manual detection is not practical. Therefore, computer software is typically preferred to save time and effort [55]. Zomorodian et al. [56] employed the Canny edge detection algorithm, which involved the conversion of video recordings into 5 s interval images, followed by image editing and filtering, to track the initiation and development of breaches on embankment dams.

Image distortion is a common issue in photography. Many efforts have been made to minimize its effect. Most studies follow orthorectification and calibration procedures of captured images using computer applications. In addition to reducing distortion, the utilization of grids, whether they are manually drawn or projected onto the downstream slope, proves beneficial in defining breach profiles. Many studies have employed grids drawn directly onto the downstream slope [48,49,57,58,59]. Using grids helps minimize distortion errors commonly encountered in photography, though the accuracy is significantly affected by grid spacing.

Table 1. Summary of selected instrumentation and measurement techniques for breach morphology detection by embankment material, measurement dimension, and accuracy.

Category	Instrument/Technique	Embankment Material	Measurement Dimensions	Experiment Scale	Accuracy
Mechanical methods	Gauged rods [19]	Cohesive earth-fill	2D	Field-scale	*
	Nylon strings [15]	Landslide	3D	Small-scale	*
Laser-based	Sweeping laser [16]	Sand dikes	3D	Large-scale	≈30 mm
	Fixed laser sheet [17]	Homogeneous earth-fill	2D and 3D	Large-scale	*
	Laser Profilometry Technique (LPT) [18]	Homogeneous fluvial dikes	3D	Small-scale	<5 mm
	Laser scanners (Leica C10) [20]	Grass-covered earth-fill	3D	Field-scale	4 mm **
Light-based	Fringe projection methods [43]	Non-cohesive embankment	3D	Small-scale	<0.5 mm
	Infrared-light (Microsoft Kinect sensor) [44]	Dike embankment	3D	Small-scale	0.3 mm to 0.4 mm
Embedded electronic sensors	Microswitch sensors [50]	Non-cohesive embankment	3D	Small-scale	*
	Thermistors [51]	Sand dikes	3D	Small-scale	10–25 mm
	Acceleration probes [21]	Non-homogeneous soil levees	3D	Field-scale	*
Photography [49]	HD camera	Rockfill and moraine dams	2D	Large-scale	*
Photogrammetry [60]	More than 3 HD cameras	Rockfill dam	3D	Small-scale	<10 mm

Notes: * Accuracy not specified in source, ** from product specifications.

Table 1. Summary of selected instrumentation and measurement techniques for breach morphology detection by embankment material, measurement dimension, and accuracy.

Category	Instrument/Technique	Embankment Material	Measurement Dimensions	Experiment Scale	Accuracy
Mechanical methods	Gauged rods [19]	Cohesive earth-fill	2D	Field-scale	*
	Nylon strings [15]	Landslide	3D	Small-scale	*
Laser-based	Sweeping laser [16]	Sand dikes	3D	Large-scale	≈30 mm
	Fixed laser sheet [17]	Homogeneous earth-fill	2D and 3D	Large-scale	*
	Laser Profilometry Technique (LPT) [18]	Homogeneous fluvial dikes	3D	Small-scale	<5 mm
	Laser scanners (Leica C10) [20]	Grass-covered earth-fill	3D	Field-scale	4 mm **
Light-based	Fringe projection methods [43]	Non-cohesive embankment	3D	Small-scale	<0.5 mm
	Infrared-light (Microsoft Kinect sensor) [44]	Dike embankment	3D	Small-scale	0.3 mm to 0.4 mm
Embedded electronic sensors	Microswitch sensors [50]	Non-cohesive embankment	3D	Small-scale	*
	Thermistors [51]	Sand dikes	3D	Small-scale	10–25 mm
	Acceleration probes [21]	Non-homogeneous soil levees	3D	Field-scale	*
Photography [49]	HD camera	Rockfill and moraine dams	2D	Large-scale	*
Photogrammetry [60]	More than 3 HD cameras	Rockfill dam	3D	Small-scale	<10 mm

Notes: * Accuracy not specified in source, ** from product specifications.

4.5. Photogrammetry

One of the recent techniques for detecting breach progression in embankment dam models involves the reconstruction of 3D maps from multiple 2D images captured by high-resolution cameras positioned at different angles (Figure 8). The number and quality of cameras used for breach morphology detection through photogrammetry have been increasing over the years, thereby improving the spatial and temporal resolution of the 3D representations.

The use of computer applications for photogrammetric methods of breach morphology detection has been increasing in the last few years. To date, various free and commercially available image processing software packages have been developed to generate 3D images from multiple 2D images. Some of these include Agisoft Metashape Professional Edition versions 1.7 and 1.8 [61], Agisoft Photoscan Professional v. 1.2.4 [62], Pix4D [63], 3-DM Analyst Research Suite 2.3 [64], Micmac [65], and Bundler [66]. Most of these softwares have been applied in geoscience applications and has proven to be a valuable tool for tasks such as terrain modeling, geological mapping, environmental monitoring, and natural hazard assessments.

The application of Unmanned Aerial Vehicles (UAVs) for large-scale photogrammetry of breaching experiments in levees [63] and embankment dams [67] has also appeared to be useful in recent years.

Overcoming Challenges in Photogrammetric Techniques

The first major challenge in reconstructing breach morphology using photogrammetric techniques is refraction. Frank and Hager [68] proposed a method to reduce the refraction errors by using a side camera capturing the water level above the breach in the traverse direction and rectifying the images obtained from three cameras. However, this method assumes that the magnitude of refraction errors in the vertical axis does not change with time during the breaching process. In reality, the amount of refraction changes with the fluctuation in the water surface. Another study by Walder et al. [64] introduced an underwater camera system to capture breach morphology from the upstream side, potentially reducing refraction errors inherent in photogrammetric methods, though the design complexity and need for waterproof cameras remain a challenge. Additionally, turbidity is a significant source of error, especially during the mass destabilization of embankments [61]. Image distortion is also a problem in photogrammetry. The distortion in individual images may even be pronounced after applying photogrammetric techniques. Therefore, the calibration of imaging devices is a very important step in photogrammetrc techniques.

Table 2. Summary of advantages, disadvantages, and limitations in instruments/techniques used for breach morphology detection.

Techniques/Instruments	Advantages	Disadvantages and Limitations
Mechanical methods [26,27,28,29]	Simple and convenient Suitable for large-scale field experiments Cost-effective	Intrusive Limited spatial resolution Less suitable for dynamic breach detection
Sweeping laser [16]	Non-intrusive Suitable for dynamic breach detection	Delayed capture of real-time morphological changes Must be used with photography Not ideal for 3D use Prone to refraction errors
Laser Profilometry Technique (LPT) [18,34,35,36,37]	Continuous 3D monitoring of breach dynamics Non-intrusive	Sensitive to surface reflectivity and refraction Environmentally sensitive
LiDAR scanners  [20,39]	Suitable for 3D monitoring Effective for large-scale and prototype experiments Fast data acquisition High accuracy and resolution	Expensive Requires extensive data processing Performance may vary in field experiments
Fringe projection methods [42,43]	High spatial resolution Continuous monitoring capability	Sensitive to water Affected by lighting conditions
Infrared-light (Microsoft Kinect) [17,44,45,46,47]	Continuous 3D breach dynamics monitoring Cost-effective compared to LiDAR Not affected by lighting	Not suitable for field-scale experiments Sensitive to water reflectivity
Embedded electronic sensors [21,50,51]	Ideal for dynamic monitoring Versatile	Limited use due to interference with the flow Low spatial resolution
Photography [48,49,55]	Simple Low-cost	Image distortion errors Not suitable for 3D monitoring
Photogrammetry [61,62,63]	High spatial resolution Continuous 3D monitoring More affordable than LiDAR Non-intrusive	Sensitive to refraction Affected by turbidity during breaching Prone to image distortion

Table 2. Summary of advantages, disadvantages, and limitations in instruments/techniques used for breach morphology detection.

Techniques/Instruments	Advantages	Disadvantages and Limitations
Mechanical methods [26,27,28,29]	Simple and convenient Suitable for large-scale field experiments Cost-effective	Intrusive Limited spatial resolution Less suitable for dynamic breach detection
Sweeping laser [16]	Non-intrusive Suitable for dynamic breach detection	Delayed capture of real-time morphological changes Must be used with photography Not ideal for 3D use Prone to refraction errors
Laser Profilometry Technique (LPT) [18,34,35,36,37]	Continuous 3D monitoring of breach dynamics Non-intrusive	Sensitive to surface reflectivity and refraction Environmentally sensitive
LiDAR scanners  [20,39]	Suitable for 3D monitoring Effective for large-scale and prototype experiments Fast data acquisition High accuracy and resolution	Expensive Requires extensive data processing Performance may vary in field experiments
Fringe projection methods [42,43]	High spatial resolution Continuous monitoring capability	Sensitive to water Affected by lighting conditions
Infrared-light (Microsoft Kinect) [17,44,45,46,47]	Continuous 3D breach dynamics monitoring Cost-effective compared to LiDAR Not affected by lighting	Not suitable for field-scale experiments Sensitive to water reflectivity
Embedded electronic sensors [21,50,51]	Ideal for dynamic monitoring Versatile	Limited use due to interference with the flow Low spatial resolution
Photography [48,49,55]	Simple Low-cost	Image distortion errors Not suitable for 3D monitoring
Photogrammetry [61,62,63]	High spatial resolution Continuous 3D monitoring More affordable than LiDAR Non-intrusive	Sensitive to refraction Affected by turbidity during breaching Prone to image distortion

Figure 8. Typical arrangement of cameras for 3D reconstruction of breach morphology from photogrammetry (inspired by [68]).

Figure 8. Typical arrangement of cameras for 3D reconstruction of breach morphology from photogrammetry (inspired by [68]).

5. Advances in Water Level Measurements

Accurate measurement of water flow and water depths in embankment breaching experiments is essential for understanding hydraulic processes and ensuring accurate results. Researchers have utilized various sensor technologies and methodologies to capture dynamic water depths. For instance, some use resistive gauges [17,69,70,71], capacitance sensors [50,58,72,73], ultrasonic sensors [17,26,30,43,61,71,74], and floating devices [43]. Ultrasonic sensors are the most commonly used instruments for water level measurements in embankment breach modeling experiments.

Ultrasonic sensors operate by emitting ultrasonic waves and measuring the time it takes for the waves to return after reflecting off the water surface. Resistive gauges function by detecting changes in electrical resistance, while capacitance sensors measure variations in capacitance between the probe and water, which correlates with changes in depth. In comparison, ultrasonic water level sensors generally have higher accuracies than resistive and capacitance-based sensors. In addition, resistive gauges are prone to measurement errors related to water conductivity. Capacitance-based gauges are less sensitive to water conductivity. On the other hand, the accuracy of ultrasonic sensors is highly affected by turbulence and the presence of debris.

To improve the accuracy of ultrasonic sensors, some studies have introduced wave suppressors and honeycomb wire meshes to improve flow uniformity, thereby reducing measurement errors [75,76]. Moreover, calibration plays a crucial role in ensuring the accuracy of water level measurement sensors, compensating for environmental factors and sensor-specific deviations. For instance, Amaral et al. [17] calibrated both resistive and ultrasonic sensors ahead of every dam breach test. Table 3 provides an overview of sensor types, sampling frequencies, accuracies, and measurement ranges.

Table 3. Summary of sensors used for measuring the depth of water in experimental dam modeling.

Study	Embankment/Dam Category	Type	Sampling Frequency (Hz)	Accuracy (mm)	Measuring Range (mm)
Amaral et al. [17]	Homogenous earth dams	Resistive probes	*	±0.3	100–1200
		Ultrasonic	200	±0.3 **	100–1000
Elalfy et al. [30]	Levee	Ultrasonic	290	±0.3	60–400
Feliciano Cestero et al. [77]	Levee	Ultrasonic	20	±0.3	100–1000
Gregoretti et al. [76]	Homogenous landslide dam	Ultrasonic	*	±1	*
Hiller et al. [78]	Downstream slope ripraps	Diver (pressure sensors)	10	±5	*
Hiller et al. [78]	Downstream slope ripraps	Ultrasonic	10	1 % of FS	350–3400 **
Liu et al. [79]	Dike	Ultrasonic	4	±0.2	*
Monteiro-Alves et al. [80]	Rockfill dams	Ultrasonic	*	±0.1	70-2000
Orendorff et al. [58]	Non-cohesive homogenous embankment	Capacitance probe	30	*	*
Dewals et al. [34]	Fluvial dikes	Ultrasonic
		mic+35/IU/TC	400	1% of FS **	350 **
		mic+130/IU/TC	200	1% of FS **	1300 **
Walsh et al. [81]	Non-cohesive embankment dams	Capacitance probe	100	*	*
Walder et al. [64], Bereta et al. [82]	Non-cohesive embankment dam	Submersible transducers	*	±0.10% FS **	10–210 **

Notes: * Accuracy not specified in source, ** from the product specification. FS = Full scale.

Table 3. Summary of sensors used for measuring the depth of water in experimental dam modeling.

Study	Embankment/Dam Category	Type	Sampling Frequency (Hz)	Accuracy (mm)	Measuring Range (mm)
Amaral et al. [17]	Homogenous earth dams	Resistive probes	*	±0.3	100–1200
		Ultrasonic	200	±0.3 **	100–1000
Elalfy et al. [30]	Levee	Ultrasonic	290	±0.3	60–400
Feliciano Cestero et al. [77]	Levee	Ultrasonic	20	±0.3	100–1000
Gregoretti et al. [76]	Homogenous landslide dam	Ultrasonic	*	±1	*
Hiller et al. [78]	Downstream slope ripraps	Diver (pressure sensors)	10	±5	*
Hiller et al. [78]	Downstream slope ripraps	Ultrasonic	10	1 % of FS	350–3400 **
Liu et al. [79]	Dike	Ultrasonic	4	±0.2	*
Monteiro-Alves et al. [80]	Rockfill dams	Ultrasonic	*	±0.1	70-2000
Orendorff et al. [58]	Non-cohesive homogenous embankment	Capacitance probe	30	*	*
Dewals et al. [34]	Fluvial dikes	Ultrasonic
		mic+35/IU/TC	400	1% of FS **	350 **
		mic+130/IU/TC	200	1% of FS **	1300 **
Walsh et al. [81]	Non-cohesive embankment dams	Capacitance probe	100	*	*
Walder et al. [64], Bereta et al. [82]	Non-cohesive embankment dam	Submersible transducers	*	±0.10% FS **	10–210 **

Notes: * Accuracy not specified in source, ** from the product specification. FS = Full scale.

6. Advances in Velocity Measurements

Direct measurement of flow velocity and capturing the flow hydrograph in embankment breaching experiments remains an important yet challenging task. The primary difficulty arises from the presence of supercritical flow downstream of the breach. Various studies have employed different methodologies to measure flow velocity in embankment dam failure experiments, including imaging-based techniques, acoustic, radar, etc. Table 4 summarizes the advantages, disadvantages, and limitations of these velocity measurement techniques and instruments.

6.1. Imaging-Based Methods

6.1.1. Particle Tracking Velocimetry (PTV)

Particle Tracking Velocimetry (PTV) involves tracking the movement of individual particles suspended in a fluid to characterize the velocity field. This method requires the seeding of tracers, a light source (usually a laser), high-quality cameras to capture the illuminated tracers, a calibration mechanism, and specialized software for data analysis. PTV has become a widely adopted technique for measuring velocity in various fluid mechanics applications, particularly at the microscale level. Recently, it has also been applied in embankment breach modeling experiments for studying flow dynamics in these settings [57,58,72,83].

The size, weight, and color contrast of seeding particles can influence the accuracy of velocity measurements using PTV. Older studies used various types of floating particles as tracers [84]. A more recent study by Rahman et al. [85] used green polyethylene fluorescent microspheres (1 g/cc density, 1 mm diameter) as seeding particles. These particles were designed to have a density similar to that of water, minimizing the effects of buoyancy force. This approach aimed to address the challenges posed by buoyancy, which often causes tracers to float on the water’s surface. However, when particles are suspended on the surface, it becomes difficult to capture velocity variations across the depth of flow. To enhance the visibility of the particles, Rahman et al. [85] employed fluorescent microspheres in combination with an LED backlight system, improving the contrast between the particles and the water.

In addition, Aleixo et al. [86] emphasized that both the amount and rate of seeding are crucial for accurate PTV results and computational efficiency. Higher quantities of uniformly seeded tracers typically yield better outcomes. However, an excessive number of particles can lead to two main drawbacks: increased computational time and the potential for particle aggregation, which introduces uncertainty into velocity measurements.

6.1.2. Particle Imaging Velocimetry (PIV)

Similar to PTV, PIV also involves acquiring sequential images using high-resolution cameras and detecting tracer particles in each frame to calculate the velocity field in two or three dimensions. The key difference between the two methods is that PTV tracks individual particles across consecutive frames to determine their trajectory and velocity, while PIV calculates the spatially averaged velocity field of all particles. Previous applications of PIV include levee breach [30,52], landslide dams [67], and earthen dam [64,87] model breaching experiments.

In comparison, both PIV and PTV have their strengths and limitations. With a higher concentration of tracers, PIV typically provides reliable velocity field measurements. However, when fewer tracer particles are used, PTV offers better spatial resolution [86]. Additionally, since PIV focuses more on detecting the overall pattern of particle motion rather than tracking individual particles, it is more suited to natural conditions where tracers may not be seeded. Therefore, PIV is often used in outdoor conditions, while PTV is more suitable for controlled laboratory settings [88].

Some studies have combined both methods, using each based on the specific requirements of the experiment. Amaral et al. [17] employed both PIV and PTV techniques to determine the surface velocity at and around the breach site. Large-scale Particle Imaging Velocimetry (LS-PIV) was used to measure the surface velocity of flow in the area around the breach location, without accounting for vertical variations in the velocity field. PTV was then used to capture the instantaneous velocity field at specific locations around the breach site, particularly to detect events such as the velocity surge during the mass collapse of breach slopes.

Large-scale Particle Imaging Velocimetry (LS-PIV) is a more expansive version of PIV, primarily used for large-scale and field experiments. Both PIV and LS-PIV operate on the same fundamental principle of using tracer particles to measure fluid flow, but they differ in scale and application. For example, Liu et al. [87] applied large-scale PIV to compute the surface velocity and breach hydrograph of an earthen embankment. Other studies have also adopted this method with reliable results in embankment breaching experiments [22,71]. In large-scale tests, inaccuracies can arise due to the lack of geometric rectification. This distortion occurs from the angle between the camera’s optical axis and the surface points on the flowing water. Moreover, LS-PIV techniques estimate surface velocities, which can lead to uncertainties in accurately representing the vertical velocity profile [89].

6.1.3. Stereoscopic PIV

Stereoscopic velocity measurement involves utilizing two or more cameras to capture images of a scene from different perspectives (Figure 9). By analyzing the disparities in these images, one can calculate the velocity of objects in three dimensions. A few studies have implemented the method of detecting the 3D velocity field in embankment and dam breaching experiments. A study by Eaket et al. [90] employed three cameras positioned at different locations in such a way that a given point is seen at least by two cameras and two images are taken at the same type by at least two cameras. The method employed tracking lightweight, small-diameter plastic particles. Elkholy et al. [91] used a similar approach with two cameras.

The use of such methods has the advantage of measuring velocity in three dimensions, allowing a more comprehensive understanding of flow dynamics compared to traditional two-dimensional techniques. However, their use in dam breaching experiments has been limited mainly due to their complex setup, larger data post-processing requirements, limited viewing lengths, increased cost, sensitivity to calibration errors, etc.

6.2. Acoustic Methods

Acoustic methods utilize the Doppler shift principle with sound waves to measure the speed and direction of flow at a certain depth below the water surface. These methods are in wide use for velocity measurements of various fluid flow conditions. The most commonly used type of acoustic device used in measuring velocity is the Acoustic Doppler Velocimetry (ADV) instrument [73,82] (Figure 10a).

The use of Acoustic Doppler Velocimeters (ADVs) in breaching experiments to measure flow velocities near the breach site is limited for several reasons. First, the highly variable velocity above embankment beds makes flow measurement challenging [57]. Second, most acoustic devices are fixed in place to measure flow velocities from a specific location. In breaching experiments, however, the water level progressively decreases, causing the instrument to be left above the water surface. A more detailed overview of the advantages, disadvantages, and limitations of these devices is provided in Table 4.

A few studies have also utilized ultrasonic instruments such as Ultrasonic Velocity Profilers (UVPs) for dam breach velocity measurements [91,94,95]. The UVP operates on the principle of detecting ultrasonic pulses reflected from particles in the fluid (see Figure 10b). When the UVP emits a signal that encounters a waterborne particle, a portion of the energy scatters upon the particle, generating echoes. The major requirement of this method is that it needs the presence of adequate suspended sediments to reflect the ultrasonic signals.

6.3. Radar-Based Techniques

One of the relatively latest non-contact methods of measuring velocity profiles in open channel systems is the Surface Velocity Radar (SVR) method (Figure 11). SVR uses radar technology by directing electromagnetic microwaves to the surface of the water and receiving the reflected waves. The change in frequency of the reflected waves (Doppler shift) is used to compute the velocity of flow [96]. However, their application in dam breach modeling experiments is still in the early stages of development. Chen et al. [97] employed hand-held SVR to measure the velocity of debris flow in a field-scale experiment of landslide dam failure.

Due to their accuracy and reliability, SVRs have great potential for use in future laboratory and prototype-scale experiments involving high flow rate measurements. Similar to LS-PIV, SVRs offer advantages in large-scale outdoor applications, as they can be used at any time of the day, unlike LS-PIV, which has limited use at night. However, the main disadvantage of SVR methods is their relatively low accuracy in very low-velocity applications [98].

6.4. Other Methods

There are also other less frequently used methods for measuring velocity in embankment breaching experiments. Some of them are electromagnetic velocity meters, stereoscopic methods, etc. Electromagnetic velocity sensors are devices used to measure the velocity of a fluid by utilizing the principles of electromagnetic induction. Chinnarasri et al. [28] used an electromagnetic velocity (EMV) meter to measure the velocity at the breach location of an embankment dam. Similarly, Zhu et al. [99] and Zhao et al. [59] employed three EMV sensors positioned along the flume length to measure the velocity of flow during the breaching of an embankment. These methods involve fixing the instrument at a specific depth above the breach bottom to obtain readings. Nevertheless, since the readings are collected from separate points, they have limitations in precisely capturing the highly dynamic velocities during breach events. In addition, these instruments might alter the flow dynamics, which is a characteristic of most intrusive methods.

Table 4. Summary of advantages, disadvantages and limitations, and accuracy of velocity measurement techniques.

Techniques	Advantages	Disadvantages and Limitations	Accuracy
PTV	High spatial resolution Highly suitable for laboratory conditions Non-intrusive	Higher computational demands Potential errors in tracking particles Less suitable for large-scale experiments Less accuracy when a higher concentration of tracers is applied Limitation when measuring near flume walls	Depends on the tracer seeding rate, camera calibration, flow conditions, scale of the experimental model, etc.
PIV	High spatial resolution Suitability for large-scale outdoor measurements using Large Scale PIV (LSPIV) Possibility for mapping velocity field in 3D (Stereoscopic PIV) Non-intrusive	Higher computational demands Complex setup and calibration Dependency on seeding particles Higher costs	Depends on the tracer seeding rate, interrogation window size, timeframe, flow conditions, scale of the experimental model, etc.
ADV	3D Velocity field Continuous velocity measurement Suitable for large and field-scale applications Better performance in turbid conditions as in embankment breaching	Intrusive Difficulty measuring the highly varying velocity during breaching Less suitable for movable bed conditions as in embankment breaching Interference with flume walls	Depends on flow conditions, probe location and alignment, acoustic signal quality, etc.
Ultrasonic Velocity Profiles (UVP)	Less flow disturbance Real-time velocity measurement	Requires suspended sediments for better results	Depends on the concentration of suspended sediments, distance from the probe, and flow conditions.

Table 4. Summary of advantages, disadvantages and limitations, and accuracy of velocity measurement techniques.

Techniques	Advantages	Disadvantages and Limitations	Accuracy
PTV	High spatial resolution Highly suitable for laboratory conditions Non-intrusive	Higher computational demands Potential errors in tracking particles Less suitable for large-scale experiments Less accuracy when a higher concentration of tracers is applied Limitation when measuring near flume walls	Depends on the tracer seeding rate, camera calibration, flow conditions, scale of the experimental model, etc.
PIV	High spatial resolution Suitability for large-scale outdoor measurements using Large Scale PIV (LSPIV) Possibility for mapping velocity field in 3D (Stereoscopic PIV) Non-intrusive	Higher computational demands Complex setup and calibration Dependency on seeding particles Higher costs	Depends on the tracer seeding rate, interrogation window size, timeframe, flow conditions, scale of the experimental model, etc.
ADV	3D Velocity field Continuous velocity measurement Suitable for large and field-scale applications Better performance in turbid conditions as in embankment breaching	Intrusive Difficulty measuring the highly varying velocity during breaching Less suitable for movable bed conditions as in embankment breaching Interference with flume walls	Depends on flow conditions, probe location and alignment, acoustic signal quality, etc.
Ultrasonic Velocity Profiles (UVP)	Less flow disturbance Real-time velocity measurement	Requires suspended sediments for better results	Depends on the concentration of suspended sediments, distance from the probe, and flow conditions.

7. Advances in Flowrate Measurement

Flow measurements in dam breach modeling are crucial for understanding the dynamics of flow and breach morphology at various locations, such as upstream (inflow hydrograph), downstream (outflow hydrograph), and at the breach site. These measurements are important for predicting flows resulting from dam failures or malfunctions in complex systems or cascading dams.

7.1. Inflow Measurement

Understanding inflow is crucial for assessing breach dynamics, predicting breach evolution, and evaluating outflow characteristics. Some studies have used mechanical instruments such as weirs. Older studies used devices such as the Annubar flowmeters, which measure pressure differentials between upstream and downstream sections of a channel with an accuracy range of ±3% [100,101]. Electromagnetic flowmeters, known for their non-intrusiveness and higher accuracy, are commonly used in laboratory flumes for inflow measurement [59,61,102]

7.2. Outflow Measurement

7.2.1. Weirs

Traditionally, outflow in dam breach experiments is measured using weirs of various geometries and sizes, including V-notch [50,103], sharp-crested rectangular [26,27,31], and trapezoidal weirs [42]. While weirs are simple to construct and provide reliable flow measurements, their ability to capture transient and rapidly changing water levels and velocities is limited in dam breach scenarios [57,64]. Additionally, their accuracy is influenced by upstream flow conditions, such as water level and flow velocity. Changes to bed morphology due to dam material deposition after breaching further complicate flow measurement accuracy.

7.2.2. Volume Balance Equation and Rating Curves

The volume balance method involves the conservation of mass principles to assess the flow dynamics during a breach. As such, it accounts for the inflow, outflow, and changes in storage within the dam to estimate breach outflow.

A rating curve is a graphical representation or mathematical relationship between the stage (water level) of a river, stream, or reservoir and the discharge (flow rate) at a specific location. These curves are developed based on field or laboratory measurements of both discharge and water levels. Hence, measuring the water level of the reservoir enables us to extrapolate the water remaining in the reservoir and estimate the amount of discharge using the volume balance method.

Volume balance methods are believed to give an accurate estimate of the volume of water released if given with accurate measurements of water level and good stage–reservoir relationships. While the volume balance method can estimate the overall volume and flow, it may not effectively capture the instant changes in flow rate exactly in the location where the breach develops [17,71].

7.2.3. Site-Specific Outflow Measurements

The aforementioned methods do not fully capture the highly variable flow dynamics in embankment breaching experiments [17]. Many prior studies estimate breach outflow by assuming a trapezoidal or semi-circular cross-section at the breach location. However, breach flow is influenced by complex and rapid morpho-dynamic changes, including surface erosion, undercut formation, and mass failure of the embankment. Furthermore, some studies have made measurements at locations distant from the breach. As noted by Amaral et al. [17], such discharge estimates are based on ‘non-local’ measurements, which may not accurately represent the dynamic processes occurring at the breach site.

To address this challenge, several researchers have developed enhanced methodologies for measuring transient breach outflow at or near the breach area. These advancements are partly attributed to improvements in flow velocity measurement techniques. For example, Orendorff et al. [57] attempted to determine the velocity profile using Particle Tracking Velocimetry (PTV), with the flow cross-section approximated as a rectangular breach. A subsequent study by Bento Ana et al. [71] introduced a ’direct’ method for measuring breach discharge by simultaneously assessing flow velocity using Large-Scale Particle Image Velocimetry (LS-PIV) and the cross-sectional area of flow. In this approach, the cross-sectional area was estimated using a laser sheet directed at the breach site, combined with high-speed, high-resolution image acquisition and analysis using filtering algorithms. Furthermore, Amaral et al. [17] proposed a method for computing breach outflow discharge based on on-site measurements of surface velocity and area. Local velocity estimates were obtained using non-intrusive PIV and PTV techniques at the breach site.

Estimates of discharge based on locally measured data provide more reliable and detailed insights into the breach process, capturing the hydro-morphological dynamics during overtopping. However, several challenges remain, including the difficulty in obtaining accurate flow velocity measurements, as the flow at such locations is often supercritical, highly turbid, and turbulent. Additionally, the method’s applicability is limited in small-scale experiments due to the complex instrumentation requirements [71].

8. Advances in Pore Pressure Measurement

The buildup of pore pressure within the body of the embankment has a very destabilizing role. The increase in pore pressure causes increased uplift pressure in the embankment. Pore pressure can have values ranging from negative (tensile stress) to positive (compressive stress) values depending on the level of seepage into the dam/foundation. Pore pressures exacerbate the failure of embankment dams, though the failure mode was due to overtopping.

Measurement of pore water pressures can be achieved by piezometers, micro-tensiometers, or a combination of both. Many studies have used different types, numbers, and arrangements of piezometers [80,104,105]. Walder et al. [64] utilized a combination of piezometers and micro-switch tensiometers on the dam body to establish the pressure-head profile, noting the challenge of frequent calibration due to drift problems while Kiplesund et al. [60], Kiplesund and Sigtryggsdottir [106] applied ten piezo-resistive pore pressure sensors in various setups connected to a small-diameter pipe system. Piezometer-based pore pressure measurements can have an error range as low as ±0.00001 kPa [107,108], although there is potential interference in the breach process due to their intrusive nature. The limitations in using piezometers are errors due to (1) delay, dependent on the length and other flow parameters in the piezometer pipe system, especially at the start and end of the measurements, and (2) entrance of air and debris into the system that might hamper the flow of water thereby increasing measurement uncertainty.

Tensiometers are also used in breaching experiments of various scopes and purposes, such as for homogeneous non-cohesive embankments [42,43,64], seismically triggered breaches [105], and other applications. Several studies have employed multiple micro-tensiometers attached to probes and transducer chips to measure pore water pressure dynamics on the downstream slopes of non-cohesive homogeneous embankments [72,85,109] (Figure 12). Tensiometers can achieve an accuracy of 90–95% in measuring water content at soil–water interfaces [42].

Recently, Höttges et al. [111] introduced a novel approach for measuring pore pressure using Distributed Fiber-Optic (DFO) pressure sensing at a laboratory scale, which is also applicable to prototype scales. This method detects changes in temperature or strain through Rayleigh scattering of light, enabling precise pressure measurements. Qiu et al. [29] employed pore pressure sensors embedded within the embankment body (Figure 13). While these methods allow for instant pore pressure recordings, their intrusive nature may negatively impact the dam material’s composition and compaction as well as interfere with the flow.

9. Challenges and Constraints in Experimental Research

Ensuring the accuracy of parameters used to describe embankment breaching characteristics in experimental studies is crucial. Despite significant progress in experiment-based research on embankment dam breaching, several challenges remain that require careful consideration and experimentation. These challenges include uncertainties related to scale effects, reproducibility and repeatability, assumptions regarding flow symmetry, seepage, and more [112,113]. Additionally, challenges associated with material grading are discussed in the following sub-sections.

9.1. Scale Effects

Scaling experimental models involving free surface flows, such as embankment dam breaching, requires fulfilling the similarity in geometric (both soil grain size and embankment dimensions), kinematic, and hydrodynamic characteristics [103]. The most commonly used similitude criterion to maintain the dynamic behavior of fluids between a model and a prototype is Froude’s scaling. For flow in embankment dam breaching experiments, where inertial and gravitational forces are dominant, Froude’s similarity is more applicable [61,74,80,114]. Uncertainties arising from size reduction that cannot be accounted for by a given similitude criterion are known as scale effects. Understanding and accounting for scale effects is therefore crucial in projecting experimental findings to real breaching scenarios, ensuring that the conclusions gained from smaller-scale experiments accurately reflect the behavior of full-scale embankments.

A robust scaling optimizes the space, time, cost, and technical requirements and reduces scale effects [115]. Maintaining the similarity of prototype and model embankment dams has been a challenging task for many researchers in the field. Many studies have reported a good validity of Froud’s similarity for embankment breaching modeling. For example, Hiller et al. [74] compared the effect of field scale to laboratory models for similarities in dam stability, flow regime, depth of overflow, and packing densities. They reported a good similarity in Froude’s number related to stability, flow regime, and packing factors between in situ and laboratory model experimental tests for dumped types of ripraps. On the other hand, for placed ripraps, in situ and model tests indicated good similarities for flow characteristics and overflow depths. Another study by Monteiro-Alves et al. [80] used Froude’s similarity rule and determined the scale effect of experimental rockfill dams subjected to overflow conditions. They found an error of 2.5% in unit flow discharges among experimental models. Moreover, Toledo et al. [116] compared the unit outflow and breaching head from one prototype and smaller model rockfill dams with three different downstream slopes (1.5, 2.5, and 3.5) and reported good agreement for two out of the three slopes. Froude’s similarity, due to its broader validity for free surface conditions, is largely accepted by many as a standard scaling method.

However, some researchers have raised concerns over the ‘default’ use of Froud’s number for all free-surface flow conditions. Al-Riffai [110] argued that embankment dam breaching, unlike many other free-surface flows, involves unsteady flow and movable-bed conditions, which cannot be adequately described by a single scaling criterion. Moreover, Froude’s similarity, particularly for homogeneous embankment dams, fails to sufficiently account for factors such as soil erodibility and cohesion, which are crucial to understanding the soil’s resistance to erosion [114]. A study by Abdellatif and El-Ghorab [26] compared laboratory experiments at different scales to examine the effects of scaling on breach processes. They found good similarity in breaching rates (depth and width) between large- and small-scale experiments; however, the smaller models showed shorter breach initiation periods than the larger ones. This suggests that forces causing drag stress and turbulence, which lead to sediment transport, are dependent on fluid viscosity, highlighting the necessity of applying Reynolds similitude in such cases [117].

To minimize discrepancies (scale effects) in simulating sediment transport, researchers have developed methods to improve the applicability of Froude’s similitude for dam breaching cases. For instance, Amaral et al. [114] introduced a correction factor to Froude’s similarity, allowing for larger grain sizes than those predicted by plain geometric similarities, thereby accounting for changes in compaction. Furthermore, Sadeghi et al. [103] and Schmocker et al. [117] employed a method of scaling particle size based on Zarn’s approach. This method involves a three-step process: (1) geometric scaling using the particle gradation curve of the prototype, (2) increasing particle size according to the Reynolds number to account for critical shear stress, and (3) filtering out particle sizes smaller than 0.22 mm to avoid excessive cohesion.

9.2. Laboratory Effects

Laboratory or model effects arise due to the boundary constraints, simplifications, and assumptions made on physical models. Like scale effects, these effects are undesired differences between prototype and laboratory constraints that do not exist in natural settings. In embankment breaching experiments, the presence of a flume wall and bed, the introduction of an initial breach, and inflow conditions can lead to changes in breaching patterns.

Laboratory flume walls are typically constructed from smooth materials such as glass, metal, or acrylic, which, compared to natural settings, exhibit relatively low roughness. Some researchers have neglected the influence of flume walls on breach characteristics for simplicity [43,50], while others have highlighted their effect on experimental models. Orendorff [73] reported that the presence of flume walls increased tailwater depth by constraining the shape of the breach, thus affecting the breaching characteristics.

On the other hand, it is very common to introduce initial breaches at the crest (pilot channel). Initial breaches come with varying locations, geometry, and sizes within the crest. Various researchers have selected different locations of the pilot channel, i.e., the middle [118], adjacent to the flume wall [50,73] where the sidewalls act as an axis of symmetry of the breach. The latter is made with the assumption that one half of the initial breach is carved on one side of the experimental dam so that the other half is omitted. Making the initial breaches adjacent to the flume walls might have the advantage of simplicity in observing and monitoring the vertical breaching process. However, such an optimistic assumption might lead to some unforeseen consequences. The constriction made by the glass wall might lead to higher acceleration and more erosion compared to an initial breach located at the center of the dam as well as potential changes in breach morphology, impacting the overall accuracy and reliability of the experimental outcomes. Hence, the effect of introducing initial breaches with different geometry, location, and size needs to be considered with due care.

Inflow condition is another important boundary condition in embankment breach characteristics of embankments that can lead to laboratory effects. Evangelista [55] reported that breach characteristics due to overtopping caused by a sudden change in inflow to the reservoir are significantly different from a gradual increase. In general, the effect of varying inflow conditions might impact the outflow hydrograph, the time to failure, and the shape of the final breach.

9.3. Repeatability of Experiments

Repeatability signifies the extent to which experiments or measurements can be precisely reproduced, ensuring the reliability and credibility of the obtained data. Walder et al. [64] suggested that experiments can be successfully repeated if construction is made carefully including exerting a quasi-identical compaction effort. Hence, conducting repeatability tests is crucial to accounting for the errors and assessing their impact in ensuring the reliability of experimental measurements executed in other test runs. Nevertheless, conducting repeatability tests remains an important but often overlooked aspect of experiment-based embankment breaching studies.

Some studies have relied on simple visual or graphical inspection of test runs to assert the repeatability of their experimental results [55,80,86]. Other studies have used more advanced techniques to validate repeatability through qualitative approaches. For instance, Coleman, Andrews, and Webby [50] investigated experiment repeatability by comparing outflow hydrographs from multiple iterations of sand embankment breaching experiments. Tabrizi et al. [119] reported good repeatability across three supposedly identical experiments investigating the effect of compaction on overtopping-induced embankment breaching, although without statistical support. Toledo et al. [116] conducted extensive repeatability tests on rockfill dam failure mechanisms, concluding that substantial variations occurred between quasi-identical tests. While many of these studies have limitations in quantifying repeatability, they represent valuable efforts in assessing experimental consistency.

A more robust approach to repeatability analysis, involving quantitative statistical results, could enhance the reliability of repeatability tests for embankment breach parameters. Monteiro-Alves et al. [80] conducted a set of tests to perform repeatability analysis on pore pressures developed at the downstream shoulder of rockfill dams, reporting a maximum difference of 0.021 m between tests. They also compared variations in discharge across test groups using descriptive statistical parameters, such as mean, standard deviation, and coefficient of variation. The results showed standard deviations ranging from 0.005 m to 0.037 m, with coefficients of variation ranging from 1.4% to 10.9%. Sharif Yusuf et al. [120] conducted repeatability tests on four embankment dams by comparing the non-dimensional depth versus time curves of quasi-identical experiments. Dezert et al. [23] evaluated the repeatability of large-scale rockfill dam breaching tests, reporting larger differences in standard deviations among tests.

Understandably, repeating experiments is primarily constrained by the scale, time, and budgetary requirements of a given embankment model. Therefore, it is relatively easier to perform repeatability analysis in small-scale experimental models. In contrast, replication of larger and field-scale models becomes more difficult due to their higher financial requirements. Nevertheless, it should be underlined that more reliable results are obtained when experimental findings are supported by sufficient replications and statistical evidence. The importance of repeatability checks is particularly critical in non-homogeneous embankments, where variability in material properties can significantly influence the results, compared to homogeneous embankments.

9.4. Seepage Losses During Breaching

Conducting purely overtopping-induced experimental tests has been challenging due to seepage through the experimental embankment. Most studies overlook the seepage flow from upstream to downstream; however, seepage, especially in non-cohesive embankments or those without a core, can cause a surge in pore pressure that may destabilize the downstream slope [121]. Dupont et al. [70] attempted to prevent seepage by covering the upstream face with a non-permeable material. However, this technique alters the embankment’s overall behavior. Other studies have introduced drainage filters at the downstream toe [50] or small-diameter perforations at the upstream section [113] to reduce pore pressure buildup. Despite these efforts, the effect of seepage on overtopping-induced embankment breaching experiments remains a challenge.

9.5. Effect of Embankment Material Gradation

Many experimental studies to date represent the embankment material of the prototype solely by the median diameter ($d_{50}$). However, the full range of material sizes must be considered. For example, Morris, Hassan, and Vaskinn [49] demonstrated that embankment materials with the same $d_{50}$ but differing grading ranges resulted in different breach formation characteristics. Therefore, it is essential to consider various material grading parameters (e.g., $d_{50}$, $d_{10}$, and $d_{60}$) as well as the overall range of diameters when selecting the material sizes for the model embankment.

10. Conclusions

Understanding the breaching characteristics of embankment dams using experimental models employing state-of-the-art instruments and techniques is believed to improve the research in the field. The following concluding remarks were drawn from the comprehensive review.

• Mechanical methods remain simple but lack precision and continuous monitoring capability. Laser-based techniques achieve high accuracy, particularly in small-scale or controlled environments, despite their higher costs. Light-based methods offer better spatial resolution but are sensitive to environmental factors like water and turbidity. Embedded sensors provide real-time data but introduce intrusiveness and accuracy variability. Photogrammetry stands out as versatile, balancing accuracy with ease of implementation but refraction, turbid water, and image distortion can limit its use.
• Resistive gauges, capacitance sensors, ultrasonic sensors, and floating devices have been used in water level measurements in embankment breaching experiments. Among these, ultrasonic sensors are the most widely used instruments for measuring water levels in embankment breach modeling experiments.
• For velocity measurements, Particle Tracking Velocimetry (PTV) provides high spatial resolution and performs well in controlled laboratory settings but loses accuracy with increased tracer density and near flume walls. Particle Image Velocimetry (PIV) supports large-scale outdoor experiments, though its complex setup, high computational demand, and reliance on tracers limit its practicality in dynamic, sediment-laden environments. Acoustic Doppler Velocimetry (ADV) captures 3D velocity fields effectively, especially in turbid, large-scale conditions, but its intrusive nature and sensitivity to rapidly changing breach fronts and movable beds reduce accuracy. Ultrasonic Velocity Profiling (UVP) offers real-time, minimally intrusive measurements but relies on suspended sediments for accuracy, making it less effective in clearer flows.
• Weirs, while simple and reliable, struggle to capture transient flow changes during embankment breach experiments. Volume balance methods offer accurate overall estimates but fail to track instantaneous flow variations at the breach location. On the other hand, site-specific outflow measurements provide detailed data, though challenges in accurately measuring flow velocity and their complexity in small-scale setups limit their applicability. Each method has its strengths and limitations, requiring careful consideration of the experiment’s specific needs and conditions.
• Pore pressure plays an important role in destabilizing embankment dams during overtopping incidents. Accurate measurement remains challenging due to sensor limitations like response delays and interference with breach processes. Piezometers offer high precision but are prone to errors from system design and debris intrusion, while tensiometers provide reliable water content readings yet require careful setup and calibration.
• Experimental research on embankment dam breaching, while essential for understanding breach mechanisms, remains constrained by factors such as scale effects, laboratory limitations, repeatability challenges, seepage, and material gradation complexities.

In general, this analysis underscores that no single technique universally excels across all embankment breaching experiment settings. The choice of instrument and technique depends on various factors, including but not limited to the desired accuracy and precision, size and scale of the physical model, suitability for transient and rapid flow variability, temporal resolution, and installation complexity.

11. Future Perspectives

From this state-of-the-art review on advancements in instrumentation and measurement techniques of embankment breaching experiments, the following future research directions were discovered.

• Most recent studies employed non-intrusive methods such as photogrammetry and laser technologies. However, results from these photogrammetric methods are subject to uncertainties due to reflection from the water surface and turbidity. On the other hand, laser technologies are less sensitive to reflection and turbidity [18]. Hence, integrating the two methods might improve the accuracy of breach morphology detection.
• For laboratory-scale experiments, the use of small-scale LiDAR technology capable of simultaneously taking measurements of the whole embankment surface could advance the research.
• Image-based velocity measurement techniques (such as PIV and PTV) have been widely used in embankment breaching experiments. Such methods are known to have errors associated with image distortion and particle tracking. However, these errors are not well reported in many of the studies on embankment breaching. Hence, future studies should consider improving the accuracy of such techniques by quantifying such errors.
• Measurement of pore pressure using piezometers has been part of many embankment breaching studies. However, depending on the sensor type, there may be delays in response to rapid changes in pore pressure, making it difficult to capture real-time data during sudden breaching events. Hence, conducting experiments with sensors capable of reducing the delay can produce better results.
• A few studies have attempted to address the uncertainties related to repeatability, seepage losses, and material grading assumptions of embankment breaching experiments. More research is needed in the future to address these issues.
• Most of the experiments on embankment breaching due to overtopping have been conducted using Froude’s scaling methods, which primarily account for hydraulic similarity. Future research should consider geotechnical factors, such as soil properties and material behavior, as they significantly influence embankment stability and breach dynamics. Incorporating these factors would improve the accuracy of scale modeling and better reflect real-world conditions.