Numerical simulation study of the sonic detection and imaging methods for fish top in downhole environments Authors: Zhongzhi Hu, Dingrui Yang, Wenxing Wu, Xiang Gao, Lei Wang Published: 2025 ABSTRACT AbstractTo address the difficulty and low efficiency of fishing and docking fractured drill tools in deep and ultra-deep wells, a line-array scanning method using acoustic transducers was employed to generate images of the fish top. Based on borehole acoustic theory, a casing-fish annulus space model was established. The acoustic wave propagation characteristics and one-dimensional linear array receiving properties were studied, with reflection echoes serving as the evaluation index. Key parameters such as sound source frequency, detection distance, fish wall thickness, and array spacing were analyzed in detail. Effective strategies to enhance sound wave transmission distance and image quality were discussed. Acoustic imaging results for a Ø244.5 mm casing and Ø127.0 mm drill pipe indicated that the horizontal section image error relative to the true center of the fish was 11.7%, while the error between the image center and the true center of the fish top was 2%. For inclined sections, The image of the fish revealed the inclined orientation. The findings indicated that acoustic imaging of the fish top was achieved using fewer acoustic transducers, providing a novel solution for enhancing fishing efficiency and serving as a valuable reference for future research on downhole tools and systems.Keywords: Fractured drilling tools, Fishing operations, Ultrasound imaging, Fish top imaging, Numerical simulationSubject terms: Mechanical engineering, Petrol Introduction Drilling tool fractures in deep and ultra-deep wells are a significant form of failure. When a drilling tool fractures, salvage operations must be implemented swiftly to avoid the collapse of open-hole sections, tool burial, or even well abandonment1. Conventional fishing technologies struggle to obtain the top position of the dropped drill tool, and the success of the docking between the fishing tool and the dropped drill tool depends on changes in drill weight and pump pressure signal. This process is both difficult and inefficient. Additionally, as well depth increases, the overhang and pump pressure signals weaken during the fishing process, thereby escalating the difficulty of retrieving the drilling tool2. Therefore, real-time monitoring of the relative position between the fishing tool and the fish top is essential to enhance the success rate and efficiency of fishing operations. After the drill pipe breaks, it is common to infer the cause of the fall based on the shape of the drill pipe at the drop junction. If downhole conditions are determined, appropriate fishing tools can be selected3. However, incorrect tool selection may exacerbate the issue, leading to further crack expansion, thereby complicating the accident. In cases where predicting the fall of the fish through the drilling interface is infeasible, traditional fall detection technology involves determining the size, shape, and depth of the fallen fish in the well. This is achieved by pressing a lead mold against the top of the fallen fish or the blocked position4. Lead molds have chemical stability in complex downhole fluids, but the marks they leave are often insufficiently specific for accurate assessment of the fish top. As a result, the state of the fish top is often assessed based on experience. In addition to lead printing, active magnetic detection, electromagnetic wave detection, downhole TV, and acoustic detection technologies are also used for downhole fish detection. Bin Zhong Zhou et al.5 employed traditional radar to create images of falling objects in the well by analyzing reflected and scattered electromagnetic signals. He used guided waves to detect targets along the well axis, thereby achieving excellent results. However, the main radiation direction of the system remains radial, with only a small part of the energy propagating along the shaft. In comparison with downhole TV imaging, electromagnetic imaging technology is more limited in the application of falling object detection. It is primarily used to identify object types, detect underground environments, and prevent collisions. Downhole TV imaging6 is a logging technology that directly observes downhole conditions in open holes and cased wells. This includes visible downhole TV, short-wave infrared downhole TV, and Visuray downhole TV. However, it is constrained by the well-fluid environment, which can include mud, sediment, and oil–water mixtures, thereby requiring washing for clear water visibility. Short-wave infrared TV can penetrate crude oil but not water7. While X-ray scattering imaging is unaffected by well fluid8, its resolution and visual quality are significantly inferior to visible downhole TV. Miniaturization X-ray detection equipment remains a challenge. Recently, ultrasonic imaging for fish location detection has gained attention. In industrial and medical ultrasound applications, phased array ultrasound has achieved notable advancements in both imaging algorithms and hardware9,10. In medical ultrasound testing, rectangular array elements (acoustic transducers) are commonly used for three-dimensional (3D) imaging. Matrix array enables arbitrary focus in 3D space and direct image voxels11. In industry applications, line arrays, due to their simple structure, strong directivity, and wide frequency range, are more commonly used for 3D imaging through mechanical scanning12. Annular arrays, spherical surfaces, and concave surfaces also have successful cases13,14, but their limitations in production cost, circuit complexity, and processing technology hinder their widespread use. The 3D imaging capabilities of linear arrays are sufficient to fuse discrete ultrasonic data into 3D comprehensive images. Given the complex downhole environmental conditions during drilling, replacing drilling fluid in the wellbore with a transparent medium is impractical. Acoustic detection technology, leveraging the characteristics of acoustic imaging, holds significant potential for guiding the alignment of downhole tools with fish tops in the retrieval of fractured drilling tools. This study investigated the feasibility and influencing factors of acoustic imaging in a drilling wellbore, especially under the condition of a limited array arrangement of acoustic transducers, using numerical simulation methods. Initially, factors affecting the detection of echoes from falling fish were analyzed, including the influences of distance, falling fish wall thickness, transducer excitation frequency, and array spacing on echo amplitude and signal extraction. Furthermore, the optimal detection range, frequency limits, and ideal array spacing were determined. Given these data, a numerical model was established, and a 3D imaging method utilizing a linear array was established. Finally, the imaging results were compared with the model to evaluate accuracy, thereby confirming the feasibility and reliability of the method. Thus, considering the stringent size limitations of downhole tools and instruments, this study introduced an innovative approach for imaging the fish on top of a fractured drill pipe using a one-dimensional line array. It also defined the relationship between the number and density of transducers and the imaging quality, thereby offering valuable insights for the subsequent research on related methods and technologies. Principles Full matrix imaging is an ultrasonic array post-processing technology for image reconstruction based on data captured from the full matrix. It enables focusing at any point in the detection area, thereby offering excellent resolution across the entire image. This overcomes the limitation of conventional acoustic detection, which only achieves high resolution within the focused region15. The method consists of two primary tasks: reflected wave collection and data processing. The assembled offset reflection points delineate the reflection interface, which accurately defines its position and size. Given a limited amount of scanned data, the method enhances image contrast and applies interpolation to improve resolution, thereby achieving a resolution approximately equal to the wavelength of the reflected wave. Full matrix imaging The full matrix imaging algorithm divides the two-dimensional (2D) imaging plane into grids. Upon utilizing the full matrix synthesis algorithm, the beam was focused on any grid, and the amplitudes of each array combination were superimposed under this condition to obtain the image amplitude for that grid16. For an array with N2 elements (acoustic transducers, hereafter referred to as transducers), the grid image amplitudes were derived by superimposing the amplitude of group A-scan time-domain signals. Furthermore, by traversing all the grids in the imaging region, a 2D image was drawn according to the image amplitude at each grid. Full matrix imaging is thus achieved focusing on any grid within the imaging target region during the post-processing. The matrix data Aij collected is expressed as: The geometric relationship for imaging divides the linear array center x into an M×M grid on the 2D image plane. When the beam is focused on the grid with coordinates (x, y), the image amplitude for the grid is given by the integral: Where: k(x, z) is the focus delay index value, which is used to obtain the corresponding amplitude from the A-scan time-domain signal fij(k(x, z)) : Where xi and zi are the transducers center coordinates, fs is the sampling frequency, and c is the medium sound velocity. In cased wells, the refraction of sound waves into the formation is neglected, assuming the propagation medium is only the fluid in the well. In the propagation process, sound waves experienced losses due to medium absorption and scattering. These effects were considered comprehensively, with no compensation applied for small attenuation. Based on sound field distribution theory, the relationship between amplitude attenuation and propagation distance is as follows: Where A0 represents the amplitude of the transmitted signal, and Ax represents the amplitude at a propagation distance (dx). dtx and drx are the lengths of the transmission and receiving paths. r1 is a function of the attenuation of the amplitude and the square root of the propagation distance. r2 is a function of the attenuation along the propagation path, with α as the attenuation coefficient related to the medium properties and path length. Thus, the corrected time-domain signal gij(k) is expressed as: Where i and j represent the indices of the transmitting and receiving arrays. The corrected signal incorporates a distance compensation factor R2(α, dtx, drx). To further minimize amplitude gaps, a logarithmic transformation is applied: Where I(x,z) is the pixel matrix containing sound pressure information. A linear array was employed to achieve faster 3D ultrasound imaging, where the speed of 2D imaging plays a crucial role. Sampling points were arranged in a linear array at a fixed interval, and the linear array probe was systematically scanned to obtain 2D images. These discrete 2D data points were mapped to cylindrical coordinates, thereby reconstructing a 3D ultrasonic image of the fish top. Figure 1 shows the flow chart of the imaging algorithm after data collection. The process involves 2D full matrix imaging of each virtual slice, which was followed by image gain processing to enhance the imaging quality. A 3D reconstruction was achieved by encoding information according to each map position. The Hipper transform enhanced the accuracy of data acquisition, and the imaging dynamic range was set to 60 dB. Threshold filtering was applied to the resulting image to eliminate background noise. The pixels of the 2D image were mapped to 3D space according to the actual spatial relationships, and the formed 3D pixels were interpolated. Subsequently, using a value of -10 dB as the isosurface, the 3D envelope line was drawn, and the image was further interpolated and smoothed. Flow chart of full matrix imaging algorithm for reconstructing the fish top. Research methodology To explore and establish an acoustic imaging method for the fish top in confined underground spaces, the simulation software COMSOL Multiphysics was used for model establishment and numerical analysis. The “Acoustic-piezoelectric interaction, transient” physical field interface was applied. A pulse function, defined as a voltage pulse, was substituted into the time-domain wave equation as the sound source. The wave propagation direction was then solved using this equation. The active sound field adhered to the time-domain wave equations: Where ρ is the density of the media, c is the sound velocity of the media, pt is the background sound pressure, and pi is the total sound pressure. Numerical simulation scheme The actual propagation distance of the acoustic signal was affected by the environment and its properties. When sound waves propagate in the wellbore environment, many factors affect the reception of reflected waves, such as the sound source frequency, propagation distance, fish wall thickness, and the receiving position, properties of the medium, as well as other physical parameters. In this study, the effects of source frequency, fish wall thickness, detection distance, and array distance on the reception of reflected wave signal were primarily considered. The array size was 3 mm, and the number of arrays was selected as 16. The detection distance was initially set to 1 m, the filling medium in the wellbore was water, and the sound speed was 1500 m/s. Given variations in water temperature, pressure, and other factors17, the sound propagation speed changed, thereby affecting the acquisition of imaging echoes. Furthermore, considering various factors such as computing time, resources, and grid independence, the spatial dimension was selected according to the center of gravity for the research problem. The spatial dimension is selected as a 2D two-dimensional axisymmetric model when studying detection distance, sound source frequency, and fish wall thickness; and as a 3D model when studying array spacing. Geometrical model An annular space composed of Ø244.5 mm casing (inner diameter 222.6 mm) and Ø127.0 mm drill pipe (outer diameter) was selected to assess the acoustic wave propagation rule under the influence of multiple factors. The single-factor analysis method was employed to analyze the effects of sound source frequency, array spacing, fish wall thickness, and detection distance on sound wave transmission. The reference values of each parameter are presented in Table 1. Table 1Benchmark parameters of numerical simulation.ParametersUnitsValueSound sourceFrequency (f0)kHz100Drill pipeThickness (t)mm9.19Detection range (L)mm100Array elementNumber--1Element spacing (d)mm1Array width (w)mm64Element width (d0)mm3WaterVelocity(v1)m/s1495SteelLongitudinal wavevelocity(v2)m/s5900PZT-5 HDensitieskg/m³7500 Benchmark parameters of numerical simulation. Longitudinal wave velocity(v2) Figure 2 shows the casing-falling fish annulus space model, Fig. 2-a is the side view of the casing-fish annulus space model, and Fig. 2-b is the top view drawn with an enlarged scale. The detection distance (L) is defined as the surface distance between the bottom of the transducer and the fish top. Array width (d) is defined as the distance between the first and the last transducer center; the array distance (w) is defined as the distance between two adjacent transducer centers. Perfect-matched layers (PML) were set at both ends of the model to simulate the actual borehole length. Casing-fish annulus space model (a) three-dimensional space model (b) Linear acoustic transducer arrangement. Boundary conditions The casing and drill pipes were equivalent to hollow cylinders, with the end face of the broken drill pipe serving as the starting end face and the z-axis as the center axis of the wellbore. A transducer was arranged at the L position of the drill pipe end face. The perfectly matched layer (PML) was the 20 mm end of the geometric model, which extended along the Z-axis. The simulated process of the model is as follows: the pulse sound excited by the transducer scattered at the bottom of the well and the casing wall, which was received by the transducer. In the constructed model, the transducer material was PZT-5 H, the filling material in the well was water, and the acoustic velocity in the medium was 1500 m/s. Subsequently, considering the large impedance difference between the water and the well wall, the boundary of the hard sound field was assumed around the water area. The fish at the bottom of the well was set as linear elastic material with isotropy. The PML simulated an infinitely long bottom-hole environment, thereby absorbing excess top and bottom model reflections. Furthermore, according to the simulation process, all solid surfaces except the transducer surface were subjected to fixed constraints. The falling fish and the transducer were solid mechanical physical fields, while the electrostatic field involved only the transducer through circuit excitation. The structural mechanics and pressure acoustics were coupled at the contact surface between fish and water, and the piezoelectric effect was confined inside the transducer, thereby forming a physical field of acoustic-solid-electric coupling. The excitation of the transducer was a pulse signal, and the excitation duration was 2 e–6 s. The excitation function was set as the function shown in Eq. (9), and the function image is shown in Fig. 3. Element excitation signal. Where T0 is the period of the excitation signal of the sound source, and the transmitting time of the pulse signal was controlled. Thus, f0 is the sound source frequency. Meshing and independence verification For different parts of the mesh refinement, in the waters near the boundary of the perfect matching layer, the structured mesh was used to extensively eliminate the influence of non-uniform reflection waves. The accuracy of the model calculation was affected by the setting of physical boundary properties and grid division. The transducer uses a fine tetrahedral mesh, the boundary conditions of the model were also set, and the structured mesh was generated through mapping. Furthermore, considering the large size gap between the grid transducer and the water area, it was preferred to divide the grid according to the transducer size. In the transient numerical solution, it was necessary to determine the non-correlation between the number of grids, time step, and calculated results. Set the time step by referring to the relationship between the time step and the minimum network Formats (9) and (10). For the solution of acoustic wave in a well, the most preferred was the reflected wave. Thus, the vertical wall reflected a wave with a similar depth as the transducer, which was selected to verify the grid independence. The monitoring time is the time required for the sound wave to reflect the measurement point at the vertical plane (t = 147 µs). The mesh sizes are presented in Table 2. Table 2Mesh size.Grid refinementObjectGrid size (mm)MaximumMinimumUltrafineTransducer0.50.1Water0.750.21Fish0.50.1FineTransducer1.50.5Water2.250.65Fish1.50.5RegularTransducer2.50.65Water2.750.75Fish1.50.5Transducer3.50.75CoarseWater3.650.85Fish30.75Transducer4.50.85Extra coarseWater4.950.95Fish4.50.85 Mesh size. Where represents the minimum model mesh size, is the minimum wavelength size, is the maximum frequency, and is the time step of the solver. Figure 4 shows the reflected waves under different grids, which indicated that 1.5 mm and 2.5 mm grids effectively captured reflected waves. Reasonable grid density saved computing resources. Under the same simulated working conditions, it was desirable to use low-quality mesh while maintaining accuracy. Therefore, the mesh size of the 2D axisymmetric model selected in this study was ~ 24,385, the mesh size of the 3D model was ~ 637,382, and the time step was 1 µs. Grid independence verification. Fundamental assumption To reduce the complexity of the difficulty, the model was simplified. The basic assumptions are as follows: The probed object was an isotropic reflector array, the secondary reflection was ignored, and its reflectivity was a scalar independent of the angle of the incident ultrasonic beam. The probed object was an isotropic reflector array, the secondary reflection was ignored, and its reflectivity was a scalar independent of the angle of the incident ultrasonic beam. Each array exhibited a high transmission efficiency (receiving sensitivity), and there was a good consistency between each array. Each array exhibited a high transmission efficiency (receiving sensitivity), and there was a good consistency between each array. It was assumed that the filling medium in the well was water. Thus, in the short-distance detection range, the medium was of constant temperature, incompressible state, and low viscosity. Thus, no attenuation due to scattering and absorption, and the attenuation coefficient α was zero. It was assumed that the filling medium in the well was water. Thus, in the short-distance detection range, the medium was of constant temperature, incompressible state, and low viscosity. Thus, no attenuation due to scattering and absorption, and the attenuation coefficient α was zero. Interpretation of result The influence of distance on echo reception Based on the theory of acoustic wave propagation, defining the superior detection distance was conducive to compensating echoes, and distinguishing the reflected waves of the casing wall from the multiple mode waves. Therefore, to analyze the correlation between the amplitude of the echo wave and the propagation distance, and to clarify the nature of acoustic wave axial propagation in the borehole, the reference value modeling was used. Thus, L increased from 100 mm to 500 mm for comparative analysis in the free field and casing respectively. Figures 5(a–b) show the sound pressure distribution of reflected sound waves at different distances under two boundary conditions, respectively. It was observed that the borehole reflected amplitude negatively correlated with distance as in the free field. This was attributable to the propagation path of the reflected wave received under borehole constraint, which was not a straight round trip from the transducer to the falling fish (also known as waveguide). In the process of sound transmission and reception, the reflected wave reflected several times at the wall of the well, which affected the extraction of the reflected wave from the falling fish. Reflected echoes at different distances between the free field and borehole (a) in the borehole (b) in the free field. The free field of fish reflection echo at 200–400 mm corresponded to that in the borehole, which served as an advantageous range for detection. The reflected echo was easily separated from multiple reflections on the borehole wall, and the reflected echo did not need additional filtering. For a single transducer with a coaxial line arrangement within 100 mm, the influence of fish shape on sound was investigated, and the sound reflected multiple times under borehole constraints, thereby forming a detection blind spot. Thus, sound waves in borehole confinement exhibited an enhanced effect, decaying more slowly than free fields, thereby enabling longer-range detection. Furthermore, it is necessary to analyze the waveform characteristics of different mode waves in long-distance detection. In the actual detection, different detection methods or fishing tools were used to eliminate the influence of detection blind spots. Effect of wall thickness on echo reception In the detection of falling objects, the accurate identification of the size of falling objects is an important factor in determining the salvage strategy. To observe the size effect of the falling object on the reflected wave, the distance between the falling object and the probe was kept at 200 mm, the probe excited at 100 kHz, and the wall thickness of the cylindrical falling object changed, thereby observing the change of the reflected wave. Figure 6 shows the sound pressure echo distribution corresponding to the Ø127 mm drill pipe thickness (t) of 9.19 mm, 7.52 mm, 6.43 mm, and 5.59 mm. The results indicated that the reflected wave amplitude increased with thickness. Therefore, no significant change was observed in the reflected amplitude with increasing or decreasing wall thickness of 1 mm. Simultaneously, the increase in wall thickness led to a variety of mode waves, which necessitated filtering by setting the energy threshold. Results of different wall thicknesses. The influence of frequency on echo reception Acoustic frequency significantly impacts imaging resolution and beam attenuation rate, which largely determines the effectiveness and accuracy of acoustic detection of falling objects. Higher frequencies enhance resolution, thereby enabling more detailed detection. However, excessively high detection frequencies require suppression of side lobes, and the increased sound wave attenuation hampers detection performance. Therefore, it is necessary to select the appropriate acoustic wave transducer emission frequency to detect the shape of the fish top. Figure 7 shows the distribution of echo sound pressure in the time domain at different frequencies over a fixed detection distance, which is determined through numerical simulation. At a detection distance of 200 mm, the reflected wave gradually became more obvious with the increased probe excitation frequency. The reflected amplitudes at 300 kHz and 400 kHz excitation frequencies were similar. When the excitation frequency was 200 kHz, the reflected wave appeared clearer than at 100 kHz, thereby facilitating subsequent recognition and extraction of the reflected wave. Pulse echoes of different sound source frequencies. Figure 8 shows the response amplitude-frequency graph for different excitation frequencies obtained through Fourier transform. At a transmitting frequency of 100 kHz, the sound main lobe was relatively wide. As the frequency increased to 200 kHz, the main lobe narrows, thereby resulting in improved directivity. Higher frequencies not only enhanced directivity but also increased side lobe intensity and decreased lateral resolution. In the Ø244.5 mm casing (inner diameter 222.6 mm), the sound source exhibited consistent spectral characteristics between 200 kHz and 400 kHz. At 200 kHz, side lobe intensity was relatively low, which contributed to improved resolution. Pulse echo spectra of different frequencies. The distance between array elements and echo reception High frequencies not only narrow the sound beam but also lead to larger attenuation amplitude during sound wave propagation. Therefore, the imaging system did not increase the ultrasonic frequency without limitation to pursue the resolution, thereby focusing on sound beam accuracy. Optimizing the related parameters of the phased bursts is also an important way to improve the horizontal resolution of the detection system. Furthermore, the more the number of elements, the clearer the imaging effect. However, considering the limited size of the pipeline and circuit complexity reduction, 16 arrays were selected with only the array space varying from 1 mm to 11 mm. Figures 9(a–f) show the unfiltered imaging sound pressure cloud image obtained through numerical simulation. The red area of sound pressure represents the area where sound waves were reflected by objects, and the blue area represents the area where silent waves were reflected. The yellow part was the artifact formed using the imaging algorithm. The black wire frame was the projection of the drill pipe under the virtual section. Cloud image of sound pressure with array spacing d at (a) 1 mm, (b) 3 mm, (c) 5 mm, (d) 7 mm, (e) 9 mm, and (f) 11 mm. The results indicated that the wider the spacing, the higher the lateral resolution of the array, and the closer the mapping was to the drill pipe section. At an array spacing of 1–5 mm ( the array width gradually approached the drill pipe diameter), the array lateral resolution changed significantly. At an array spacing of 7–11 mm, the array lateral resolution changed slightly, and the imaging was almost consistent with the section plane. It is concluded that to improve the imaging resolution, the array width was increased in a specific range. When the array diameter remained unchanged, the array spacing of 5 mm (the array width of 115 mm) was the ideal state. Imaging result Given the above conclusions, the full matrix imaging technology was introduced to falling fish detection, utilizing a linear array arranged along the X-axis for scanning. The acoustic transducer conducted a full matrix simulation with a step length of 12 mm, thereby extracting temporal data from the reflected waves of falling fish. A 3D envelope was generated, using an isosurface value of -10. In this process, the average data processing time of the algorithm remained within 0.08 s. Figure 10 represents a schematic diagram of the detection process, while Fig. 11-a–f depicts results at various virtual slice positions. In these figures, the horizontal axis (x) represents the diameter of the pipe at the section position, and the vertical axis (z) represents the depth of the fish. Areas with high sound pressure, shown in red, indicated the locations of the fish. These results indicated that full matrix imaging effectively produced planar images of the fish object. Additionally, by changing the position of the measurement and superimposing the images of the measurement lines, the shape of the fallen object in the well was detected. Schematic diagram of simulated probe slice. Schematic diagram of sections at positions y (a) 57 mm, (b) 47 mm, (c) 37 mm, (d) 27 mm, (e) 17 mm, and (f) 7 mm. Figure 12 compares imaging results with the actual drill pipe. The blue sections highlight the possible positions of the fish calculated according to the imaging data. The parameter y indicates the distance between the linear array and the center of the drill pipe. Sub Figure (a) shows imaging results for a scanning distance interval Δy of 10 mm; Figure (b) shows the imaging results for a scanning distance interval Δy of 5 mm. The results indicated that the smaller the scanning distance Δy improved the accuracy of the imaging results relative to the fractured drill string underground. However, a larger scanning interval yielded useful imaging data of the fish underground during actual detection. Images of the fish top view at different scanning distances: (a) 10 mm and (b) 5 mm. According to Eq. (12), the relative error between the virtual and real imaging areas was ~ 11.7%. The imaging center coordinates (-4.7, 1.62, 57.71),compared to the real coordinates of (0,0, 57). The relative errors along the x, y, and z axes were 2%,1.05%,and 1.02%, respectively. Where E represents the percentage difference between the virtual image and the real image, Sim represents the virtual imaging; and Sreal represents the real image. Figure 13 shows the imaging of the inclined fish top view. In this figure, gray sections represent the casing. After multiple reflections of the sound waves occurred at the wellhead, they were received by the acoustic sensors; thus, the imaging density was used to assess the cross-sectional imaging. The results indicate that full matrix imaging effectively produced a planar image of the collected fallen fish object. Additionally, by changing the measurement position and overlaying the images of the measurement lines, the shape of the fallen objects inside the well was detected. The image of the top of the tilted falling fish. For the research on falling fish in the non-horizontal section of the well (the angle between the falling fish section and the shaft was 30°). The same method was used, and the final imaging results are shown in Fig. 12. The results indicated that the large-angle inclined section did not appear fully. However, the tip of the fish top and part of the low side appeared. The 3D image tilt angle was 24.6°, the relative true inclination error was ~ 20% and the Z-axis imaging of the falling fish tip ranged from 122 mm to 142 mm. The Z-axis imaging of the low side ranged from 150 mm to 175 mm, the real section tip was from 116.42 mm to 121.05 mm, and the low side position was from 175.03 mm to 179.65 mm. This primary cause of the imaging error lay in the multiple internal reflections of sound waves within the fish, thereby resulting in artifacts. For the research on falling fish in the non-horizontal section of the well (the angle between the falling fish section and the shaft was 30°). The same method was used, and the final imaging results are shown in Fig. 12. The image overlaps almost exactly with the top of the fish. The results show that the relative error of image area is 13.2%. The imaging center coordinates are accurate relative to the real coordinates. Discussion In this study, when the sound source was close, the incidence angle of each transducer varied, and the emission angle was calculated geometrically. Conversely, when the sound source is far away, the large radius of the spherical wave results in nearly parallel wave propagation paths. In such cases, the beam angles were equal, and the relationship between the path length and array distance followed the cosine function. The dividing line between the near field and the far field was determined by the array (D). For a linear array (d). To distinguish the shape of the top of the fish, it is necessary to arrange the transducer array within the near field. The transducer transmitted sound waves at the highest possible frequency, and the array width was maximized. In free space, the theoretical boundary between near field and far field of an array width of 108 mm and center frequency of 100 kHz was ~ 1.5 m. However, in pipeline detection, to ensure effective detection, the actual detection distance was relatively shorter. Accurate performance of the algorithm relied on selecting array elements with high receiving sensitivity and exceptional consistency18. Enhancements to imaging were achieved by increasing the number of array elements, array width, and array spacing within a reasonable range. However, practical constraints, such as the complexity of data acquisition, signal transmission, and hole size, make extending the detection range challenging. Therefore, the optimization focus for transducer arrays was to maximize the data collected using a finite number of elements. In downhole imaging, the ideal echo waveform only reflected the peak corresponding to the top or part of a specific part of the fish surface, thereby enabling clear visualization of the surface contour. However, artifacts often appear in the image due to the reflection of the acoustic beam and the characteristics of the reflective interface. Some artifacts were removed through prior estimation of the falling fish. In theory, the first peak of reflection represented the signal reflected from the top interface of the fish, which corresponded to the initial surface first contour encountered by the sound beam. Therefore, removing data after the first peak effectively eliminated removable artifacts. However, issues arose when the amplitude defined as the peak value of the pseudo-signal exceeded that defined as the peak value of the real signal, and the amplitude of the wave collected at different angles was significantly different. Therefore, when using threshold filtering, the energy threshold was defined as 1.6 times the mean threshold of the entire scan line. The first point exceeding this threshold was regarded as the starting point of the first peak signal, and waveform data beyond this point was retained for further processing. Limitations and future work This study ignored changes in bottom-hole temperature and irregularities in borehole walls such as open holes. However, borehole temperature significantly impacted sound velocity, particularly with increasing detection depth. Follow-up experiments were conducted to calibrate mud temperature and test mud sound velocity. Further exploration is needed to investigate the effects of irregular borehole walls, including open holes. In practical applications, the soft and hard properties, as well as the formation porosity, can generate a variety of complex waves that interfere with the echo. Thus, the direct wave reflected by the tube and fish can not be separated by analyzing only the dominant detection distance. A more effective approach will be to study the behavior of multiple reflected waves to enhance the signal, thereby enhancing the detection distance. Additionally, it is necessary to analyze the attenuation factors in sound propagation to compensate for energy lost due to attenuation in the reflected waves. For multi-group or multi-line array imaging with fishing tools in downhole exploration, the line array must be arranged sparsely to maximize imaging resolution while ensuring adequate coverage. The probe should be developed to match existing fishing tools, and the anchor structure, such as the hook and slip, may require custom design. However, the use of a threshold of 1.6 times the mean across the scan line can effectively distinguish the first peak. In practice, empirically setting the threshold size and selecting waveform data may not be entirely accurate. Adaptive threshold algorithms based on statistical methods or physical principles may be necessary to improve this process. In actual downhole exploration, the transmission medium was water or downhole mud. The goal is to detect fish or objects that have fallen into the well, which may include rocks or irregular well walls. Some rocks may be removed through cleaning processes. For large objects, the condition of the fish top or objects should be determined according to the reflected acoustic information, with intervention measures taken afterward. Non-metallic rock chunks in open holes may be the result of wall collapse and may not have a salvage value. Conclusion This study proposed a method based on a one-dimensional linear array of acoustic transducers to image the top of a fractured drill pipe at the bottom of the well. Numerical simulations and full-matrix acoustic imaging methods were employed to achieve 3D imaging of the fracture pipe top section model. The simulation results indicated the feasibility of using acoustic imaging to visualize the top of the fractured drill pipe. The method shows promising potential for further research, particularly in enhancing imaging quality, underground guided fishing technology, and related tools development. Key insights from this research include: The energy of the sound wave reflected by the falling fish negatively correlated with the distance. The amplitude of the reflected waves was enhanced by the well wall, with larger falling objects generating stronger reflection waves. While the amplitude changes of the reflection waves from thicker drill pipes were less obvious, more waves of different modes were generated. Additionally, the frequency of the sound wave was proportional to the amplitude of the reflected wave. As the frequency increased, the primary lobe of the reflected wave became more concentrated with an increasing number of side lobes. Therefore, it is necessary to select the appropriate excitation frequency combined with specific conditions in the actual detection. The energy of the sound wave reflected by the falling fish negatively correlated with the distance. The amplitude of the reflected waves was enhanced by the well wall, with larger falling objects generating stronger reflection waves. While the amplitude changes of the reflection waves from thicker drill pipes were less obvious, more waves of different modes were generated. Additionally, the frequency of the sound wave was proportional to the amplitude of the reflected wave. As the frequency increased, the primary lobe of the reflected wave became more concentrated with an increasing number of side lobes. Therefore, it is necessary to select the appropriate excitation frequency combined with specific conditions in the actual detection. As the size of the drilling tool and hole changes, increasing the frequency and array spacing enhances data acquisition and the final appearance of the fish. However, the limited space at the bottom of the hole imposed higher demands on the arrangement of array transducers. Subsequent research should focus on optimizing the use of limited data to produce clear images of fish at the bottom of the well. As the size of the drilling tool and hole changes, increasing the frequency and array spacing enhances data acquisition and the final appearance of the fish. However, the limited space at the bottom of the hole imposed higher demands on the arrangement of array transducers. Subsequent research should focus on optimizing the use of limited data to produce clear images of fish at the bottom of the well. Acknowledgements This study was financially supported by the Science and technology project of Sichuan Province (2021YJ0354), and Sichuan University of Science & Engineering for its software and hardware support. These supports are gratefully acknowledged. Author contributions D.Y., Z.H., W.W., X.G. and L.W. made significant contributions to the preparation of the manuscript, D.Y. and Z.H. performed the data analyses and wrote the main body of the manuscript, W.W., X.G. and L.W. assisted with the analyses and provided constructive discussions. All authors reviewed the manuscript. Data availability The datasets used and analyzed during the current study available from the corresponding author on reasonable request. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets used and analyzed during the current study available from the corresponding author on reasonable request.