Underwater Acoustic Remote SensingFor Predictive Maintenance Modeling of Dams

Kenneth J. LaBryChief Scientist – Acoustics

Underwater Acoustics International

I. Introduction

We are now living in a time where robotics and remote sensing, are utilized to unburden human activity from risk exposure and repetitive effort as well as to provide cataloged and characterized information from remote sensing observation, which is used to make better decisions in all aspects of society from medicine to transportation and civil infrastructure. This paper proposes to implement a combined effort of robotics, remote sensing and digital data management for Predictive Maintenance modeling of dams.

There are 84,000 Dams in the United States, 1,756 of these are also hydroelectric power generating facilities. These Dams are an average of 52 years old. Most of these dams have unknown disposition of underwater components and un-mapped stages of degradation. The examination of any Dam is a singular exercise. Some Dams may be of similar design but no two Dams are quite alike with each Dam requiring a tailored examination plan. This paper presents a system and methodology for the application of high definition Underwater Acoustic Remote Sensing in the inspection of submerged dam structure components and the interfaces of those structures with the surrounding water bottom and shows how the information acquired can be presented in a platform that is of value to the owner-operator.

The paper demonstrates the pitfalls and problems associated with acoustics and sonar utilization in substructure inspection, the challenges inherent in shallow environments, confined spaces and difficult access, the environmental difficulties, cost effectiveness, and benefits, as well as result capabilities. It outlines the basic acoustic principles involved in the inspection of substructures, the development of remote sensing equipment capable of generating the necessary resolution and definition for shallow environments, as well as the development of the techniques and methodologies necessary for proper execution of substructure inspections and comprehensive shallow water-bottom surface mapping. The discussion also encompasses remarks regarding the economic and safety advantages of remote sensing in substructure inspections.

Case study examples are shown and briefly discussed. The examples depict integrated modeling capabilities for illustrating the inspection results, the utilization of reiterative modeling to track conditional changes and provide a knowledge-based predictive basis for maintenance and rehabilitation scheduling.

II. Application

The inspection of underwater structural components of dams, locks, floodgates and other water control structures is crucial if infrastructure components are to maintain adequate

performance throughout their life cycle and attain maximum longevity. With the recognized value of these assets and their irreplaceability, it has become a priority to develop predictive asset maintenance programs through monitoring and mapping structure conditions and performing comparisons of subsequent condition mapping to develop predictive degradation patterns. The technology to perform inspections in the underwater environment, and gather this knowledge, has been limited to visual, video and acoustic profiling. The common, traditional methods of divers and robotic vehicles that perform largely visual and tactile surveys of the structures, and acoustic echosounder profiling methods for structure examination, have severe limitations due to the subjective nature of the data acquisition and the need for data adjustment. The visual inspection methodology, whether by diver or utilizing an underwater video camera, deployed, either by a diver or on a Remotely Operated Vehicle (ROV), is problematic under inclement environmental conditions such as high current flow, extensive structural surfaces, and low to no visibility. Due to the significant impact of environmental conditions on results and findings, these surveys tend to be extremely subjective and provide only generalized comparison information from successive inspections and investigations. These typically cursory investigations are very limited in perspective because of positional accuracy correlation and visibility limitations. Due to these challenges, they provide a poor representation for baseline control of infrastructure condition. The limited information provided by these methods, provides insufficient data for adequate change tracking. The problems encountered by underwater video, even in the best water clarity are due to the limited field of view during any video frame and lack of perspective because of this limitation. If only a small component of the structure is within the field of view then difficulties arise in accurately modeling the structure surface in order to enable successive investigations that can be used to quantitatively determine degradation rates. The same applies to tactile diving inspections, which carry the added risk of safety of dive personnel exposed to a water environment that can have differential pressure and subsequent flow.

Acoustic profiling studies have the problem of providing a sparse data set of specific point reflections from which images are rendered. There are many physical reasons for this, some of which are due to the principles of wave mechanics involved with acoustic pulses, which restrict the visualization and resolution capabilities of this methodology. Because of these physical limitations the image rendering from this method, will always lack fine detail on close inspection. While the result is a point cloud, which looks representable when viewing a large section of the data set, compressed into a small area the size of a typical computer screen, on close inspection, it is evident that there are gaps of up to several feet in the data. The application of multi-beam acoustic profiling systems also has similar inherent deficiencies and additional problems arising from acoustic propagation in confined spaces. The issue for Multibeam acoustic systems, is that in confined spaces with multiple acoustic reflective planes in close proximity there is the potential for significant ghosting from signal multipath that can generate false anomalies and mask existing anomalies without any trail indicating that there may be a problem with the data. Both of these profiling methods have serious issues with profile point location in penstocks or tunnels, due to the reliance on an inertial sensor that is operating in a closed tube, with no known reference points for updates to the sensor to compensate for precession drift. This means that the data set is usually “normalized”. This is when the points are shifted so that they form the pattern of a nice tube. This is valid if the data set is adjusted using a common reference position, otherwise the only justification is that it looks good and correlates to what is expected. There is no scientific validation for this type of manipulation of data. This can very well lead to features missed by this method. There have

been efforts to implement multi-beam imaging sonars, with limited success. Profiling sonars perform best when they are used in defining patterns that can show sediment build-up and large debris accumulation, or major structural failures.

The development of a different method based on a modified steered beam sonar sensor, that provides image visualization down pipe and along a wall surface plane, has added a new dimension and capability to underwater acoustic remote sensing. These systems have the ability to produce very high definition sonar imagery as well as very precise localized, acoustic profile measurement in difficult environments. They yield excellent performance in confined environments with multiple acoustically reflective planes in close proximity to the transducer emitter/receiver, and are a superior alternative to other current methods of underwater structure inspections. This technique utilizes wave propagation parallel to the surface plane of interest and extracts data from a different aspect of the reflected acoustic wave. The technique provides an investigation with full comprehensive coverage of the surfaces of interest and provides a control baseline for future inspections.

III. Acoustic Results for multi-axis steered beam sonars

The results are a spatially referenced, high-resolution image rendering of submerged surfaces with a broad overall perspective. This is utilized in comparative analysis software to track changes. The results are then used in a predictive maintenance model. The results also produce a full coverage visualization of the structure with no gaps, and provide data over-sampling and overlap for validation. This technique also provides a very good visualization of the corners of intersecting planes, which is found to be problematic for other techniques previously discussed.

IV. Delivery of Information

The best data is of little value if it is not utilized. One of the most important discoveries in the dissemination of the results from these underwater investigations is that, while the quality of the data acquisition and interpretation are important, the digestibility of the data presented to the client or project owner is crucial to providing a useful tool for operational efficiency. If the data acquired is to be of value, results must be delivered in an easily accessible, easily digestible format that provides actionable information about the condition of the structure and water bottom interface. As a solution to providing a 21st Century delivery platform for the results, the implementation of a secure on-line mapping service that incorporates all of the results in a geographic format with an aerial reference background is suggested. This provides for a readily available, easily digestible format with actionable information that can be viewed and queried without special software or intimate knowledge of Geographic Information Systems or CAD software operations.

V. Case Examples

The following diagrams depict examples of results obtained with proper Underwater Acoustic Imaging Technology and the actionable information, change-tracking potential provided by those results.

Figure 1 shows the results that comprised a comprehensive 3-D model of a Dam system, incorporating data from Underwater Acoustic Imaging, Acoustic water bottom profiling, and mobile LIDAR into a 3-D model representation that defined spalling and degradation on the exposed sections of the concrete spillway apron and the slab and buttress dam. Downstream water channelization failure and the erosion patterns resulting from this issue, that were causing undermining of the adjacent earthen embankment, are shown as well as upstream sediment deposition and accretion at the heel of the dam.

Figure 1. Dam System Evaluation with Erosion Pattern Mapping

Figure 2 shows an example of a partially flooded tubular water intake inspection, where the conduit is corrugated metal pipe. This provides information that is spatially referenced prodicing positional measurement and dimensional characterization of observed features that are used for comparative analysis with successive investigations.

Figure 2. Flooded Penstock Internal Investigations (Corrugated Metal Conduit)

Figure 3 and Figure 4 show different visualization perspective examples of a flooded concrete lined metal penstock. This inspection documented various concrete coating deformities such as cracks, spalling and de-lamination. This provides information that is spatially referenced prodicing positional measurement and dimensional characterization of observed features that are used for comparative analysis with successive investigations. The advantages of the rectangular representation is that it allows for an even distribution of the sonar data over the observed surface and also profides the ability to overlay the imagery on geographic information such as aerial imagery for landmark correlation.

Figure 3. Flooded Penstock Internal Investigations (Concrete lined Penstock) Cylindrical visualization.

Figure 4. Flooded Penstock Internal Investigations (Concrete lined Penstock) Rectangular visualization

Center – bottom of Penstock

945’

1020’

870’

180° 0°360°

Figure 5 shows an example of progressive change tracking of erosion holes at the upstream base of a dam. The time span is a 60-day period with an increasing a pool elevation condition change. This example demonstrates the effectiveness of successive monitoring investigations to evaluate the rapidity with which a conditional change is occurring and the value that knowledge provides in making scheduling decisions for repair and remediation.

Figure 5. Erosion hole progression shown at the heel of a concrete dam spillway structure.

Change tracking of observed anomaly over a 60-day interval with a conditional change of 2’ greater reservoir level

Figure 6 shows a very effective method of presenting the results as actionable information that can be easily digested and readily accessed. This is by the presentation of the information in a web based mapping service. This provides a geographically referenced aerial imagery background so that information is easily related to landmark references by engineering and operations staff. The example shown is that of Fenstermaker’s Map Analyst service, which provides a platform for readily accessed actionable information with change tracking and imbedded retrieval of all information related to observations. This scheme to observe and measure current conditions and then monitor them over time to track changes and evaluate the rate of change, can provide valuable decision making assistance for when or if remedial action will be necessary. This can lead to a much more effective use of maintenance budgets and large capital expenditure refurbishments.

Figure 6. 2-dimensional data re-mapping to overlay aerial imagery with tunnel crown and invert sections. Displayed extract shows a cavity in the concrete tunnel across the invert area. The corresponding cylindrical 3-D image depiction is inset.