3.2.4. Identification of AE Signals from the Hsu-Nielsen Source

Simulation of AE signals by the Hsu-Nielsen source was carried out at various points of the lateral beam of the bridge structure inside one of the bridge sections. The distance from the sensor to the Hsu-Nielsen source was from 0.15 to 15 m. Totally about 100 pulses were excited (Figure 8).

**Figure 8.** Signals from Hsu-Nielsen source at 1 m distance from AE sensor.

For detection of signals from the Hsu-Nielsen source, a planar location was constructed (Figure 9). It is revealed that the signals from Hsu-Nielsen source are with certainty located in their actual places and form compact clusters, in contrast to noise. This confirms the possibility of identifying signals from defects based on location data.

**Figure 9.** Planar location. Hsu-Nielsen source signals are marked by red, noise is marked by orange. Color of points indicates total number of events in this coordinates. Yellow points: 3–10 events, orange points: 11–25 events, red points: 26–100 events. Blue circle – location cluster of signals from Hsu-Nielsen source.

Since the bridge is a structure of testing characterized by a high level of acoustic noise, a locational cluster corresponding to a defect can be masked by a large number of false locations of noise events. The complexity of the detection algorithm lies in the fact that the values of the AE parameters of the noise and useful signals vary in fairly wide and overlapping ranges. This is due to the variety of noise sources and different distances between the sensor and points where the defect or simulator is placed from 0 to 15 m.

The difference of parameters is best shown in the two-dimensional plots of characteristics, for example, on the scatter plot diagram "Counts/Duration versus Duration" (Figure 10a) and the scatter plot diagram "Energy versus Counts/Duration" (Figure 10b). As can be seen from the chart, the AE impulses emitted at the distance of 7 m from the sensor have the parameter distribution other than the noise distribution. When the distance increases over 7 m, the differences disappear, the AE impulse parameters become close to those of noise [13].

**Figure 10.** (**a**) Scatter plot diagram of AE signal energy and duration; (**b**) scatter plot diagram of amplitude and parameter "counts/duration." Impulses corresponding to noise are marked black, while impulses from the Hsu-Nielsen source emitted at distances of 0–2 m, 2–7 m and 7–15 m from AE sensors are marked green, yellow and red, respectively.

#### 3.2.5. SHM-System

SHM-system for this application was designed and installed on the bridge. AE sensors, vibration transducers, strain gauges, temperature sensors, modules for collecting, measuring and processing signals were installed on the span of the bridge structure (box beams). Low-frequency AE sensors, inclinometers, crack opening sensor were installed on the supports. In addition, SHM-system is equipped with video cameras to monitor the traffic situation, a weather station to monitor wind loads and to determine the presence of precipitation for AE signals filtering and a communication system.

#### *3.3. High-Temperature Process Pipeline*

On the active high-temperature pipeline without decommissioning, studies were produced to assess the feasibility of conducting AE monitoring. Application of AE monitoring of high-temperature or cyclic thermal loaded structures is an effective method for defects detection. Several interesting applications are described in the papers [15–18]. The tested object consisted of pipe sections with a diameter of 50 to 80 mm with a total length of 8 m. Individual pipe sections were connected with both flanged and welded joints. The working medium of the pipeline was molten salt, the temperature of which could reach 600 ◦C. To synchronize the AE data with the operating mode of the pipeline, one thermocouple was installed. 8 AE sensors were installed through waveguides. Simulations of AE signals were carried out once per day by emitting from a sensor to monitor the attenuation factor.

When the pipeline was cold, the propagation velocity and attenuation coefficient were measured, which were 4900 m/s and 1.7–2.7 dB/m, respectively. Then the pipeline was put into operation. AE data were taken in various operating modes of the pipeline for 63 days, including in the total amount for 15.5 days in a stationary high-temperature regime. The process of heating the pipeline was long and took up to several days. During the heating process, the melt began to flow into the pipeline. The most suitable was the use of the frequency range 100–500 kHz and the threshold in the region of 40–55 dBAE. It was found that the activity and amplitudes of noise increased significantly with the pipeline temperature rising at a rate exceeding 1 ◦C/min, especially when clamping fixtures for waveguides were used (Figure 11a). Amplitudes of noise reached values of up to 100 dBAE. It is assumed that the occurrence of noise is associated with the thermal expansion of the pipeline, which induced friction on the thermocouple and other items on the surface of the pipe. During the work in the stationary mode, the noise level was reduced to an acceptable one and amounted to about 50 dBAE.

It is revealed that the attenuation coefficient increases with the accumulation of sediments on the inner surface of the pipeline and can reach 58 dB/m. With this level of attenuation, the maximum distance between the AE sensors could not be more than 0.7 m. The value of the propagation velocity for acoustic waves was unchanged.

In the course of the work, the melt leakage was observed as a result of the destruction of the flange joint gasket. In this case, the AE system recorded an increase in the duration of the AE signals a few hours before the leak invisible under the layer of thermal insulation began to be observed visually (Figure 11b).

As a result of the conducted studies, the possibility of AE monitoring of the most critical sections of the pipeline was confirmed: the detection and location of crack-like defects and detection of leaks on this pipeline by the AE method can be accomplished.

**Figure 11.** Amplitudes of noise depend on the pipeline temperature (**a**); the melt leakage (**b**). Colors of points indicates different channels.

## *3.4. SHM-System for Oil Refinery Equipment with Defects Presence*

At an oil refinery, non-permissible defects were found in the upper part of the gas adsorber of the circulating gas (commissioned in 1970). From 1999 to 2009, the lamination was known to exist in the main upper end plate (the area of the lamination ~0.4 m2, the depth of the lamination: 6.1–12.4 mm). On the same unit, the defect of the planar type (crack) in the weld of the shell to the end plate (length: 140 mm, depth: 12–14 mm) was also present (Figure 12). This gas adsorber has the following parameters: working pressure: 4.2 MPa, operating temperature +45 ◦C, wall thickness: 22.0–24.0 mm, the inner diameter: 1.156 m and the height: 16.45 m.

A preliminary decision was made to install a monitoring system for the upper part of the adsorber. The main task of the monitoring system was to ensure safe operation of the adsorber before its replacement within a year.

Prior to installation of the monitoring system during the overhaul in the spring of 2012, its AE testing was carried out (Figures 12 and 13). AE sources of class I and II were registered, the locations of which were correlated with the place of previously detected defects on the vessel head of the adsorber (Figure 12). This allowed us to assume the possibility of further expansion of the lamination zone during the life extension period of the adsorber.

Based on the results of the AE testing, the previously approved decision on the installation of the monitoring system for the vessel head of the gas adsorber was confirmed. The expanded SHM-system included 4 AE sensors and a weather station (Figure 14). The threshold data acquisition approach was used, and the threshold value was about 45 dBAE.

**Figure 12.** Previously found defects of gas absorber head plate: lamination in the vessel head (gray areas) and locations of the registered AE source (orange). Color of points indicates total number of events in this coordinates. Yellow points: 3–10 events, orange points: 11–25 events, red points: 26–100 events.

**Figure 13.** Gas adsorber schematic drawing with places of AE sensors mounting.

**Figure 14.** SHM-system for gas adsorber: (**a**) overall view of the head part; (**b**) AE sensor mounting place; (**c**) workplace of operator; (**d**) dialogue box of the software A-Line Mon. Color of points indicates total number of events in this coordinates. Green points: 1–2 events, yellow points: 3–10 events, orange points: 11–25 events, red points: 26–100
