Nothing Special   »   [go: up one dir, main page]
Next Article in Journal
Petrography and Geochemistry of the Upper Cretaceous Volcaniclastic Deposits of the Haţeg Basin (Southern Carpathians): Inferences on Petrogenesis and Magma Origin
Previous Article in Journal
Ore Genesis of the Langcun Porphyry W-Mo Deposit, South China: Constraints from Fluid Inclusion and H-O-S Isotopics
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Minerals
Volume 15
Issue 2
10.3390/min15020110
minerals-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Roping Prediction Versus Detection: Could Prediction Be Possible?

by
Lin Yang
1,2,
Lei Chen
1,2,*,
Difan Tang
1,
Massimiliano Zanin
2,3,
Chris Aldrich
2,4 and
Richmond Asamoah
3
1
School of Electrical & Mechanical Engineering, The University of Adelaide, Adelaide, SA 5005, Australia
2
ARC Training Centre for Integrated Operations for Complex Resources, Adelaide, SA 5005, Australia
3
Future Industries Institute, University of South Australia, Mawson Lakes, SA 5095, Australia
4
Western Australian School of Mines, Curtin University, Perth, WA 6845, Australia
*
Author to whom correspondence should be addressed.
Minerals 2025, 15(2), 110; https://doi.org/10.3390/min15020110
Submission received: 2 January 2025 / Revised: 17 January 2025 / Accepted: 18 January 2025 / Published: 23 January 2025
Browse Figures
Figure 1
<p>The structure of a hydrocyclone.</p> "> Figure 2
<p>The schematic of the roping detection equipment innovated by Hulbert [<a href="#B10-minerals-15-00110" class="html-bibr">10</a>].</p> "> Figure 3
<p>The schematic of the roping detection equipment innovated by Rakesh et al. [<a href="#B23-minerals-15-00110" class="html-bibr">23</a>].</p> "> Figure 4
<p>Schematic of experimental setup (redrawn from Nayak et al. [<a href="#B37-minerals-15-00110" class="html-bibr">37</a>]).</p> "> Figure 5
<p>Schematic of a 125 mm hydrocyclone test rig (redrawn from Hou et al. [<a href="#B55-minerals-15-00110" class="html-bibr">55</a>]).</p> "> Figure 6
<p>Principle of the laser–optical measuring device (redrawn from Neesse et al. [<a href="#B9-minerals-15-00110" class="html-bibr">9</a>]).</p> "> Figure 7
<p>The schematic of the underflow width measurement (modified from Janse van Vuuren et al. [<a href="#B83-minerals-15-00110" class="html-bibr">83</a>]).</p> "> Figure 8
<p>Air core in hydrocyclone (redrawn from Li et al. [<a href="#B88-minerals-15-00110" class="html-bibr">88</a>]). The blue line represents the path of water flow.</p> ">
Versions Notes

Abstract

:
Roping is a hydrocyclone failure mode that reduces separation efficiency, negatively impacting both the comminution circuit and downstream flotation processes. Therefore, detection of roping as early as possible is crucial in maintaining the normal performance of physical separation and linked processes. Most importantly, instead of detecting roping after it happens, could roping be predicted even before it arises? This review examines various detection methods, including mechanical, tomography, vibration, acoustic, and image processing, highlighting their cost and ability to monitor parameters like air core size, spray angle, and solid concentration. While most current methods detect roping only after it happens, predictive approaches could save time and costs. A promising solution combines pressure and vibration sensing with advanced signal processing, showing early potential to transform roping prediction and improve operational efficiency. This review highlights research gaps across various methods, underscores the importance of developing predictive capabilities for hydrocyclone operations, and outlines the essential conditions and future priorities for achieving roping prediction.

1. Introduction

The hydrocyclone is an important piece of classification equipment implemented between the comminution and flotation stages due to its simple structure and large capacity in the mineral process [1]. It is one of the most important pieces of equipment in the mineral industry and has been proven to be very effective in the fine separation of sizes [2].
Figure 1 shows a typical hydrocyclone, which comprises an inlet section for the entry of slurry, inducing a swirling motion. As the slurry moves into the conical section, the diameter narrows, accelerating the mixture and generating centrifugal forces. Heavier particles are pushed toward the cone wall, forming a dense underflow stream, whereas lighter particles migrate toward the center, forming the overflow stream. The vortex finder, a cylindrical section at the top of the cone, segregates the clarified liquid from coarser particles. The underflow section collects the separated particles for further processing, and the overflow section allows the clarified liquid to exit [1,3].
Roping in hydrocyclones is a complex phenomenon influenced by both design and operating variables, making it a common issue in daily operations. When roping occurs, the underflow becomes restricted, leading to a significant reduction in separation efficiency. Particles larger than the intended product size are misdirected to the overflow, contaminating it and compromising downstream processes. This effect is particularly problematic in systems where multiple hydrocyclones share a common overflow tank, as roping in one unit can contaminate the entire tank and disrupt subsequent processes like flotation. Both design and operating variables significantly contribute to the onset of roping in hydrocyclones. Key design factors include a low underflow-to-overflow diameter ratio [4], which disrupts the internal flow dynamics, making the separation process less efficient. Operating variables such as increased feed pressure [5], elevated feed solid concentration [6], and finer feed particle size [7] exacerbate roping by altering the balance of forces within the hydrocyclone. Addressing roping requires not only optimized operating conditions but also design adjustments tailored to specific material and flow properties to mitigate its impact on process efficiency.
Although roping has been studied by many researchers, only limited work has been done to solve the issue, which remains a challenge [5]. This is because there are too many factors that affect the characteristics of the slurry, such as feed pressure, solid concentration, and feed particle size [5,8,9], and they keep changing during the process.
Nowadays, most mines rely on visual inspection and lab analysis to determine the performance of the hydrocyclones, which is empirical, and there is always a delay in realizing the problem. Hence, it is important to find a more reliable way to detect and even predict the occurrence of roping.
This review will encompass the presentation and evaluation of current hydrocyclone monitoring technologies. Simultaneously, we will delve into the occurrence of roping, examining previous scholarly research on this phenomenon. Most crucially, our focus will be on discussing and analyzing the potential for and research directions in predicting roping.

2. Roping Detection Techniques

Throughout the past two decades, numerous methods have been introduced for the real-time assessment of hydrocyclone behavior and attributes. Although some of these techniques have made it to the commercial market, the integration of online hydrocyclone monitoring into process control systems remains relatively limited. As a result, online monitoring technology cannot be classified as a fully matured technology at this stage.

2.1. Mechanical Method

Hulbert [10] introduced a mechanical device installed at the apex of the hydrocyclone to monitor the shape of the underflow discharge. As shown in Figure 2, the abrasion-resistant component of this device traces the spray shape of the underflow discharge, allowing the angle detector to measure the spray angle. This pioneering method saw successful application in 1986 within the mill circuit at the Vaal Reefs Gold Mine. However, its broader industrial adoption was hindered by high maintenance demands and limited durability.

2.2. Tomography

Tomography refers to a class of imaging and diagnostic techniques used to visualize the internal structure of objects or substances without physically dissecting them. These methods are widely used in fields such as medical imaging, geophysics, and materials science to gain insight into the three-dimensional distribution of properties or materials within a target object [11]. Tomography techniques use data from multiple angles or directions to reconstruct cross-sectional images or volumetric representations, providing valuable information for analysis, diagnosis, and decision making. There are several tomography techniques, each tailored to specific applications, and they play a pivotal role in nondestructive testing, medical diagnostics, and various scientific investigations.
Among all tomography techniques, electrical resistance tomography [12,13,14,15], electrical impedance tomography [16,17], and ultrasound tomography [15,18,19] have been studied by many scholars for hydrocyclone working condition detection and performance monitoring.

2.2.1. Electrical Resistance Tomography (ERT)

Electrical resistance tomography (ERT) is a technique for creating images by reconstructing the electrical conductivity distribution within a system by measuring the resistive component of electrical impedance [20]. Through meticulous design, such instrumentation has the potential for online application to monitor and regulate the functioning of established process equipment, resulting in notable benefits such as substantial energy conservation [21]. Due to its advantages, some scholars have applied ERT to hydrocyclone performance monitoring and working condition detection [12,13,14,15].
Williams et al. [13] investigated the size and behavior of the air core within a 44 mm hydrocyclone. They applied an ERT-based online technique with 15 electrodes arranged in a single plane and a sampling frequency of approximately 22 Hz. Using quantitative and qualitative algorithms, the measured data were reconstructed, and the air core size was calculated. Their findings demonstrated that ERT could effectively monitor the size and position of the air core in the hydrocyclone in real time. However, their study had several limitations: 1. the air core size could be overestimated when it was located in the central area; 2. the experiment was conducted using water only, limiting its applicability to real-world conditions; 3. the calculated air core size was not verified in the study.
In a subsequent study, Williams et al. [14] pioneered the development of an online monitoring system based on ERT for industrial hydrocyclones employed in the clay refining process. The system was tested on a 50 mm diameter hydrocyclone, and the feed slurry has up to 15 wt% with a cut size of 7 μm. The experimental results showed the system’s ability for discharge fault detection and air core size measurement. Although clay slurry was tested in this study, several constraints remain: 1. the solid concentration of the slurry used was lower than that typically found in the real-world mining industry; 2. only one solid sample (non-conductive) was tested in this study.
Similarly, Dyakowski et al. [22] made significant progress using ERT, successfully detecting various underflow discharge faults such as roping, spigot detachment, and choking. They also measured air core sizes under different hydrocyclone operating conditions and directly calculated solid concentration profiles through parametric reconstruction of conductivity data. However, the results indicated that the clarity of the ERT images needs improvement and the measurement can be adversely affected by high-conductivity materials.
Rakesh et al. [23] proposed an ERT system for hydrocyclone air core measurement. Moreover, they managed to predict roping using computational fluid dynamics (CFD). In their study, a high-speed dual-plane ERT system (ITS z8000) was installed on the conical section of a 3-inch hydrocyclone, as shown in Figure 3. The system has a data acquisition speed of 1000 dual frames per second. A series of tests was conducted using water with different vortex finder diameters (18 and 25 mm), spigot diameters (10, 12.5, 15, and 20 mm), and pressures (5, 10, 15, 20, 25, and 30 psi). The experimental results for the air core diameter were compared with CFD and high-speed camera data, with average differences of 5.68% and 2.52%, respectively. This study demonstrated that ERT can measure the air core size with reasonable accuracy. However, when the system was tested with a 10 wt% solid slurry, the ERT image quality was noticeably affected. Therefore, improved anti-interference designs and algorithms are needed to mitigate the effects of high-concentration slurry and noise, enabling better imaging quality.
Vakamalla et al. [24] used ERT and a high-speed camera to experimentally validate the air core information predicted by their CFD model while considering the inclination effect on hydrocyclone performance.
To evaluate the performance of different ERT reconstruction algorithms, Diddi et al. [25] conducted a series of studies on hydrocyclone air core measurement. They began by assessing two non-iterative methods (monotonicity and factorization algorithms) using both phantom and experimental data. Subsequently, they compared the performance of these methods with Gauss–Newton and Total Variation algorithms. Additionally, they developed a new threshold method to obtain crisp radius values. In summary, different algorithms related to hydrocyclone air core measurement using the ERT technique were compared and analyzed in this study, which deepened readers’ understanding of ERT reconstruction algorithms. However, there are still some limitations: 1. The study focused only on specific algorithms. 2. The hydrocyclone data used in this study consisted of water or low-concentration slurry as the feed, which may not represent the complexity of real-world hydrocyclone operations in industries such as mining. In addition, the study pointed out that the main challenges of image reconstruction algorithms include the need for enhanced algorithms that can provide a sharp distinction between the air and water phases, as well as faster simulation times.
The studies mentioned above show the potential application of ERT monitoring techniques in evaluating the performance and operational status of hydrocyclones. However, further research is needed to enhance the accuracy of estimating the air core’s characteristics, especially in the central region of the hydrocyclone [13]. Additionally, a trade-off exists between the accuracy of the reconstructed image and the time required for reconstruction, making the technique either time-consuming or less precise [23].
Several commercial products utilizing the tomography method have now been developed. CycloneSense™ [26] is a prominent example, designed to measure the shape, size, and position of the hydrocyclone air core. Mounted on the hydrocyclone underflow, it uses an array of 12 electrodes to continuously monitor the air core inside the hydrocyclone. This reflects the maturity of tomography technology to some extent. However, the specific performance of this product still needs to be evaluated under varying operational conditions and with different types of materials.
The studies mentioned above demonstrate the potential of ERT monitoring techniques for assessing hydrocyclone performance and operational status. However, further research is needed to achieve the following objectives:
  • Improve the accuracy of air core characterization, particularly in the central region of the hydrocyclone [13].
  • Develop solutions to balance the trade-off between reconstruction accuracy and processing time [23].
  • Minimize sensitivity to conductive media variations [23,25].

2.2.2. Electrical Impedance Tomography (EIT)

Electrical impedance tomography (EIT) employs electrical currents and surface voltage measurements to evaluate the conductivity distribution within a body [27]. It measures both resistance and reactance. This concept originated in the early twentieth century, initially introduced as a method for exploring underground mineral deposits. The potential application of EIT techniques to monitor the performance and detect the working condition of hydrocyclones has been widely studied [16,17].
Williams et al. [16] and Gutierrez et al. [17] implemented an EIT system comprising 16 electrodes placed near the inlet of a 44 mm hydrocyclone to enable real-time monitoring and control of underflow discharge. Their experiments explored a range of calcium silicate slurry concentrations (0 to 35 wt%) and feed flow rates (0.4 to 0.6 dm3/s). The reconstructed EIT images provided detailed insights into the internal dynamics of the hydrocyclone, revealing a clear correlation between the air core size and the discharge angle. This relationship underscores the potential of EIT for optimizing hydrocyclone performance by enabling precise control of operational parameters, particularly under varying slurry conditions. However, further research is needed to investigate the sensitivity of EIT to both conductive and non-conductive slurries.
Antonio et al. [28] evaluated three methods (artificial neural networks, Gaussian profiling, and radial basis functions) for improving EIT image analysis. The experiment setup was the same as that used by Gutierrez et al. [17]. The study concluded that each method has its limitations: the neural network method requires large data for training, Gaussian profiling is limited to offline use, and the radial basis function method is time-consuming for image reconstruction and analysis.
In summary, the EIT technique has potential for monitoring hydrocyclone performance and condition. However, challenges include difficulties in image reconstruction due to variations in conductivity, the complexity of electrode array installation, and potential intrusiveness. Additionally, achieving high spatial resolution often requires more computational time.

2.2.3. Ultrasound Tomography (UT)

Ultrasound tomography (UT) is a crucial technique capable of reconstructing the distribution of a gas–liquid two-phase flow [29,30]. It uses ultrasonic waves, typically in the frequency range of 2 MHz to 15 MHz, to generate images of a target object. Some efforts have been made to monitor the performance of hydrocyclones employing ultrasonic sensors to observe the behavior of the air core [15,18].
Schlaberg et al. [18] applied a UT technique by installing 16 ultrasound transducers around a plane near the top of the conical section of a 50 mm diameter hydrocyclone. The system successfully visualized the air core’s size, shape, and position. The study highlighted ultrasound tomography’s potential for real-time monitoring and control. However, industrial implementation faces challenges due to hardware complexity and computational demands.
The UT technique offers a way to monitor air core behavior in conditions where optical methods may fail. However, there are some limitations to this technique:
  • A high sampling rate and a large number of sensors are required for improving accuracy.
  • Data loss can occur when tracking moving objects [29], limiting the industrial feasibility of using UT for monitoring hydrocyclone performance and condition.

2.3. Vibrational Method

Vibrational methods for fault detection and diagnosis operate on the principle that mechanical faults in equipment induce characteristic changes in vibration patterns. Advances in sensors and data processing techniques have enabled this method to continuously collect vibration data from machinery in real time using accelerometers, piezoelectric sensors, or other vibration-sensitive devices. By processing vibration signals using various methods—such as fast Fourier transform (FFT), root mean square (RMS), wavelet transforms, and time-domain analysis—faults can be identified, localized, and, in some cases, predicted before they lead to system failure. The typical sampling frequency ranges for vibrational methods are usually between 1 kHz and 20 kHz. This approach is particularly useful for rotating machinery, such as motors, pumps, turbines, and hydrocyclones.
In recent years, vibration-based monitoring techniques have been studied and applied in different industries for fault detection and diagnosis [31,32,33,34,35]. These methods have been particularly effective in identifying specific operational issues in hydrocyclones, including roping and choking.
Dubey et al. [36] used a piezoelectric accelerometer at the hydrocyclone apex to collect hydrocyclone vibration signals. By changing the feed solid concentration, they were able to collect signals under different operating conditions (from spray to roping) with a sampling frequency of 20 kHz. Applying the FFT, which can be performed in real time, revealed that acceleration amplitudes in the 60 to 120 Hz range increased significantly when operating near the roping condition. In contrast, these amplitudes remained relatively low within the same frequency range when the hydrocyclone was in a spray state. The authors concluded that vibration-based monitoring holds promise as an effective tool for assessing operating conditions in hydrocyclones. This study used single-density particles (silica sand), so the effect of varying particle density was not tested. It is unclear if the spray and roping discharge criteria apply to multi-density particles.
Nayak et al. [37] used a tri-axial accelerometer near the underflow of the hydrocyclone to detect early-stage choking. The schematic of their experimental setup is shown in Figure 4. In their experiment, the solid concentration was controlled by adding the iron ore fines with the water from 9.09 to 31.03 wt%. The working status of the hydrocyclone changed as the concentration of the slurry changed. The vibration signals collected by the accelerometer were first decomposed using the empirical mode decomposition method [38]. Then, the FFT and the power spectral density of the high-frequency decomposed parts were calculated using Matlab. The results showed that choking can be monitored by observing the frequency peak within a certain frequency band in FFT spectra. This study provides some inspiration for the application of vibration methods to monitor the working conditions of a hydrocyclone. However, this result is only applicable in the laboratory when the feed is known and stable. Additionally, the hydrocyclone feed concentration of iron ore is typically maintained between 30 % and 40 % by weight, whereas the concentration tested in this study was relatively lower.
Wang et al. [39] explored blockage detection in a laboratory-scale hydrocyclone through the application of a vibrational technique that relies on wavelet denoising [40,41] and the discrete-time Fourier transform method [42,43,44]. Four acceleration sensors were used with the sampling frequency of 25 kHz. Their study demonstrated that the acceleration amplitude exhibited sinusoidal fluctuations over time at blockage degrees of 0%, 50%, and 100%, while the hydrocyclone’s vibration frequency increased as the throughput rose. This study concluded that the integration of both the time domain and the frequency domain proved to be highly effective in identifying the operational status and promptly identifying blockage faults.
The studies mentioned above demonstrate that the vibration-based monitoring technique serves as a valuable tool for predicting choking and roping conditions in hydrocyclones. However, there are some areas that require further attention: 1. Conducting a comprehensive investigation into the vibration patterns of hydrocyclones. 2. Developing a robust calibration-based algorithm that can be effectively applied to industrial hydrocyclones.

2.4. Acoustic Emission (AE) Method

The acoustic emission (AE) technique has emerged as a valuable asset in the mining domain, offering a non-invasive and real-time method for detecting and assessing structural integrity, thereby enhancing safety and operational efficiency. The ability of AE to detect and analyze stress-induced acoustic signals resulting from rock fracture [45,46,47], equipment wear [48,49,50,51], and structural instability [52,53] is invaluable in maintaining a safe and productive mining environment. By providing early warnings of potential failures and structural anomalies, AE aids in minimizing downtime, optimizing maintenance strategies, and ensuring the safety of personnel working in hazardous mining environments. At the same time, AE technology is also being studied as a tool to monitor the operational status of hydrocyclones. The typical sampling frequency for AE systems spans from 200 kHz to 10 MHz, significantly higher than that of vibrational methods. However, these higher frequencies impose greater demands on sensors and data acquisition systems, requiring more advanced and robust technology to maintain accuracy and reliability. The sampling frequency generally depends on the mechanical property of hydrocyclones when structural-borne acoustic signals are measured. The higher sampling frequency can help distinguish hydrocyclone-related signals from background noise and capture finer details of the acoustic emissions.
Hou et al. [54] developed a PC-based AE signal processing system to monitor a 5-inch hydrocyclone performance. The silica sample (HPF-5) they used in the test, from Hepworth Minerals and Chemicals Ltd. (England, UK), had a cut size of 13.1 μm. The AE sensor was attached to the middle of the conical section of the hydrocyclone using epoxy adhesives. Different solid concentrations (from 10 to 40 wt%) and feed pressures (from 8 to 18 psi) were tested in their experiment, and the AE signals under different working conditions with a sampling frequency of 2000 Hz were recorded. The findings indicated that the spectral features of the signal exhibited sensitivity to changes in operating parameters such as feed pressure, mass flow rate, and solid concentration. Another case study on the non-intrusive passive AE sensor to monitor the hydrocyclone’s performance in real time was also conducted by Hou et al. [55]. A single piezoelectric sensor was used in this study. They used the same sampling frequency (2000 Hz) and test conditions (10 to 40 wt% solid concentration and 8 to 18 psi feed pressure) as in the study in Hou et al. [54]. The test sample was silica (HPF2) from Hepworth Minerals and Chemicals Ltd., which had a cut size of 13.1 μm. The schematic of the test rig is shown in Figure 5. The quantitative relationships between various operating parameters (such as feed pressure, mass flow rate, and solid concentration) of the hydrocyclone and the statistical and spectral attributes of the acoustic signals were established. Additionally, this study indicated that the AE method can be used for hydrocyclone fault detection, such as roping and blockage. The results proved the method’s effectiveness for hydrocyclone online control and monitoring.
Neesse et al. [9] employed an acoustic-based sensing technique to monitor the underflow discharge patterns of the hydrocyclone, specifically focusing on spray and roping discharge. The results indicated that the spray discharge pattern exhibited higher acceleration amplitudes in frequency ranges from 100 to 600 Hz, while the rope discharge showed damped acceleration amplitudes within the same frequency range. Additionally, the authors noted that there were higher acceleration amplitudes in the relatively low frequency ranges from 1 to 100 Hz. This study showed the correlation between the AE signal and the working condition of hydrocyclone. However, the reason behind the AE signals in the frequency domain needs more research (Hao et al. [56]). While the acoustic signal-based sensing technique provides data regarding the operational status of the hydrocyclone, specifically in terms of roping and spray discharge, it still has constraints as it is significantly affected by the materials and geometry of hydrocyclones. Moreover, it is also necessary to prevent the mixing of noise signals from surrounding areas. Therefore, a robust algorithm should be developed to reduce the interference of noise signals from the working environment. Some common noise reduction algorithms for the AE method and their limitations are listed in Table 1.

2.5. Image Processing Method

The utilization of image processing techniques in the mining industry represents a groundbreaking leap in how to analyze and manage various mining operations. Image processing involves the manipulation and interpretation of digital images to extract valuable information, and it has found widespread application in the mining sector. By employing imaging technology, the mining industry can gain insights into ore quality [64,65], ore size [66,67], and ore composition [68,69,70,71]. Image processing also facilitates automated monitoring and control of processes, such as ore classification [72,73] and equipment health monitoring [74,75,76,77]. From geological exploration to real-time monitoring of mining sites, the integration of image processing techniques offers a comprehensive approach to enhancing productivity, resource management, and overall performance in the mining industry.
Image processing technology has also been studied and applied for monitoring the operation of hydrocyclones [9,36,78,79,80,81,82,83]. Petersen et al. [78] investigated the hydrocyclone underflow discharge pattern using video graphic analysis. In this investigation, a vision system for online hydrocyclone underflow angles measurement was introduced, with initial tests conducted in both laboratory and industrial settings. The authors employed a simple matrix subtraction method, which included three parameters: the darker range, the brighter range, and the intensity difference. This method was used to extract the edge of the hydrocyclone underflow shape, as the hydrocyclone underflow spray angle is an observable parameter that allows for the immediate identification of spray or roping operations.
Neesse et al. [9] demonstrated a laser–optical measuring system. As depicted in Figure 6, a laser beam was projected onto the underflow stream, with a charge-coupled device (CCD) camera capturing the reflection of the beam. Since the discharge angle changes as the working status changes, the reflect length was different under different working conditions. From the pixel-scattering in both X and Y directions, spray and roping appear in different regions in the figure and are easily distinguishable.
Janse van Vuuren et al. [83] provided an approach to monitor the operational conditions of hydrocyclones using video recordings of the underflow discharge. In this study, a CCD camera was used to record the hydrocyclone underflow footage. Then, the underflow shape was analyzed by measuring its width along a horizontal line in the image, utilizing various noise reduction techniques and detecting flow boundaries through motion analysis. The metal ore samples they used included gold, ilmenite, and platinum group metals from the Merensky reef in South Africa. Figure 7 shows the schematic of their underflow width measurement. The results indicated that the underflow width decreases as the operation status transits from spray to roping. This proved that the image processing technique can be used as a method for roping detection. However, there were some problems when extracting the underflow discharge width from the videos, for example, when the contrast differences and foreground noise were extremely low. Therefore, a well-lit environment and effective and reliable algorithms are needed.
Recent breakthroughs in computer vision tasks have been attributed to the remarkable progress of deep convolutional neural networks, which makes producing more meaningful state detection and managing significant noise possible. In earlier image-based methods [9,78,83], it was necessary to explicitly estimate the hydrocyclone’s spray angle from the image, typically involving edge reconstruction or an approximation based on the spray width at a certain distance from the spigot. However, with convolutional neural networks, the raw image of the operational hydrocyclone can be directly processed. Giglia and Aldrich [84] validated the potential and advantages of using convolutional neural networks for monitoring hydrocyclone working status. In their study, two feature extractors (VGG-16 and VGG-19) from Simonyan and Zisserman [85] were used. The developed model successfully generated meaningful predictions regarding the operational state from video footage of unseen experimental runs and effectively managed some level of image noise. In addition, the model required no recalibration to the site for new installations, and the ability to accommodate scenarios with moving cameras. However, this method still needs high-quality datasets to train the model, and it is difficult to obtain diverse and well-labeled datasets in real-world situations. Additionally, capturing high-quality images requires strict environmental conditions, such as control over dust, vibration, and lighting.
The studies mentioned demonstrate the utility of image processing-based monitoring techniques for assessing the working condition and performance of hydrocyclones. However, some research work is still needed to optimize the industrial application of this technology. For example, image extraction algorithms need to adapt to different environments and types of mineral slurry.
In summary, various technologies are available for roping detection, but their practical implementation faces significant challenges. Table 2 shows the principles and limitations of different methods. As can be seen from the table, methods like ERT and EIT lack versatility for different mineral types. AE and image processing techniques require controlled environments, while mechanical methods incur high maintenance costs. Additionally, vibrational methods present calibration difficulties, limiting their effectiveness in industrial applications.

3. Roping Prediction

To predict roping, various experimental studies have been conducted utilizing distinct methodologies, including mechanical energy balance, investigations of water-only hydrocyclones, and measurements of the air core using different equipment in both laboratory and on-site settings, to understand roping. Scholars attempt to associate roping with other easily observable variables and establish predictive models.
Concha et al. [4] presented an equation based on a phenomenological model derived by their predecessors. The formula for calculating the air core diameter ( d a ) in a hydrocyclone was established using geometric and operating parameters, including the surface tension of the liquid–air interface, liquid viscosity, the external pressure drop, and the diameters of the overflow and underflow ( d o and d u ), as shown in Figure 8. They applied this equation to estimate the air core diameter in a 6-inch hydrocyclone, varying the underflow-to-overflow volume flow ratio from 2 to 14. Using data from Plitt et al. [86] and Bustamante [87], they plotted the relationship between the air core-to-apex ratio and the apex-to-vortex diameter ratio, highlighting different operational states. The results indicated that roping in the hydrocyclone could be predicted.
However, this study has limitations. The proposed air core model is only applicable to hydrocyclones operating with water. Additionally, the model requires real-time measurement of a difficult-to-obtain parameter: the surface tension at the liquid–air interface. As a result, the model may not accurately predict the air core diameter when dealing with varying ore–water slurries in practical classification processes.
Mazumdar et al. [82] correlated the hydrocyclone cone ratio with the spray angle. They tested various combinations of spigot diameters (3.2 to 6.4 mm) and vortex finder diameters (8 to 11 mm) across feed inlet pressures ranging from 10 to 50 psi. Their results showed that, with a constant cone ratio, the spray angle increased as the Reynolds number rose. They concluded that the spray angle could be reliably correlated with the inlet Reynolds number. However, since their study was conducted using water only, its applicability to real-world scenarios is limited.
In recent years, CFD was used to study roping in a hydrocyclone. Davailles et al. [89] made a comprehensive study on hydrocyclone roping. Both an experimental study and CFD simulations were conducted. In this study, they mainly focus on feed solid concentrations ranging from 10% to 50%, especially concentrations exceeding 30%. To validate the CFD simulations, they used a cone ratio ( d o / d u ) of 0.55, which matched well with experimental data obtained from a 100 mm diameter hydrocyclone under pressures varying from 60 to 250 kPa. This result is consistent with the result of Heiskanen [8], who proved that roping can be detected by monitoring the overflow mass flow rate and solid concentration.
Pérez et al. [5] verified some findings from previous studies and identified inlet pressure and particle diameter as key factors influencing the transition from spray to roping using CFD analysis. Their results indicated that an increase in particle diameter leads to a decrease in air core diameter and discharge angle during the transition to roping. This occurs due to a reduced spinning velocity or a higher solid fraction concentration at the apex as inlet pressure increases.
Through the above research, we know that scholars have conducted experiments, simulations, and other studies on the variables that may cause roping, attempting to identify the more critical factors that cause roping. Their aim has been to identify the most crucial factors leading to roping and to reveal and predict this phenomenon. Although these studies have not provided a comprehensive explanation for roping due to its complexity, they provide good guidance on how to monitor roping, as mentioned in previous section. From these studies, it can be concluded that there are two main limitations that limit the prediction of roping:
  • Sensor development: continuous, consistent, and accurate monitoring of the essential operation parameters of hydrocyclones requires further advancement of sensing technologies.
  • Mechanistic understanding: a limited understanding of the underlying mechanisms of roping prevents accurate identification of key parameters necessary for developing predictive models.

4. Possibility Analysis of Roping Prediction

As shown above, different technologies have been studied for hydrocyclone roping detection. Although they may not be widely used in practical applications, they have been technically proven to detect the occurrence of roping. However, researchers are not merely content with detecting roping; they have consistently aspired to develop methods for predicting its occurrence. Detection is when roping occurs, while prediction is one step ahead of roping occurring. If the roping can be prevented before its separation efficiency is affected, the efficiency of the entire subsequent flotation process could be ensured. Therefore, researchers have conducted extensive studies to understand the underlying causes and characteristics of roping and to explore the potential for its prediction. Many factors have been found to have a correlation with roping, such as feed pressure, the feed and underflow mass flow rate, the air core, the solid concentration, the hydrocyclone’s geometry, and the feed particle size distribution [5,9]. While these studies have made decent progress in understanding roping, the complex nature of roping and the interactions of various parameters make prediction extremely difficult.
So, is it possible to predict roping? Or, what further work and research are needed to achieve the prediction of roping? We believe that predicting the occurrence of roping is possible. For instance, Tyeb et al. [90] demonstrated promising results by leveraging vibration analysis integrated with advanced machine learning techniques to predict roping in hydrocyclones. Their study utilized vibrometry to capture high-resolution vibration signals under various operational conditions, which were subsequently analyzed using one-dimensional convolutional neural networks (1D-CNNs) to identify patterns indicative of roping. The binary and ternary models they developed using 5 s vibration signals achieved 97.31% and 94.94% prediction accuracies, respectively. However, the feed information (particle size distribution, solid concentration, and feed pressure) is required for training the models.
Therefore, achieving consistent and reliable roping prediction requires the development of a series of advanced sensors. As previously mentioned, the occurrence of roping is influenced by numerous factors, with the most significant challenge stemming from the constantly fluctuating feed in practical applications. In such dynamic conditions, even if predictive models based on specific variables are developed, the lack of effective methods to accurately and continuously monitor these variables in real time during standard hydrocyclone operations remains a critical limitation. For example, some studies [17,18] mentioned in the previous section proved that there was a correlation between the air core and roping. Concha et al. [4] have established the predictive model for predicting the air core based on certain pre-known geometric and operating parameters (such as the liquid–air interface, the viscosity of the liquid, the external pressure drop, etc.). However, due to the limitations of various detection methods, these parameters cannot be detected in real time during the daily hydrocyclone operation. This results in some predictive models not being able to be verified and applied in practical work. Therefore, only when the feed or the other critical parameters can be measured accurately and in a timely fashion will a predictive model for an online system be developed.

5. Conclusions

A comprehensive review of the literature on hydrocyclone online condition monitoring with a focus on roping leads to the following conclusions:
  • All the methods for roping detection have certain limitations, making it difficult for them to be widely applied in practice: Mechanical methods face challenges related to high maintenance and durability concerns, while ERT and EIT methods are sensitive to conductive slurry and struggle with balancing reconstruction accuracy and processing time. Ultrasound tomography requires a high sampling rate and numerous sensors, making implementation complex. Vibrational methods lack robust algorithms and a comprehensive understanding of underlying patterns. Acoustic emission methods are prone to contamination by environmental noise, and image processing methods depend heavily on precise knowledge of ore characteristics, which may not always adapt to changing mineral attributes, leading to accuracy issues.
  • Research on the phenomenon of roping in hydrocyclones remains sparse, with current studies offering limited coverage of the diverse and interconnected factors influencing its occurrence. Despite significant efforts, no model developed so far has proven capable of handling the complexities and variabilities of real-world operational scenarios. The dynamic nature of feed conditions, slurry properties, and hydrocyclone geometry pose significant challenges for practical implementation.
  • While the prediction of roping is theoretically achievable, it is contingent on the availability of rich, accurate, and real-time data from the process. This level of detailed monitoring remains beyond the capabilities of existing technologies, leaving prediction efforts constrained by incomplete or insufficiently reliable information. Developing effective predictive models also requires a deeper understanding of the mechanisms underlying roping, coupled with robust algorithms capable of processing the intricate interactions of operational variables.
  • A pivotal advancement needed in this domain is the development of a suite of high-precision, durable, and reliable sensors tailored to monitor critical parameters in dynamic and industrial environments. Such sensors must be capable of operating under harsh conditions, such as high-pressure flows and varying solid concentrations, while maintaining the sensitivity and accuracy required for predictive analytics. By enabling the continuous collection of comprehensive data, these sensors would provide the foundation for building predictive models that can forecast roping events with a high degree of reliability and timeliness, thus enabling proactive operational adjustments and minimizing process disruptions. Only with such technological advancements can the vision of accurate and actionable roping prediction be realized, transforming process efficiency and reliability in industrial applications.

Author Contributions

Conceptualization, L.Y., L.C. and D.T.; methodology, L.Y., L.C. and D.T.; writing—original draft preparation, L.Y.; writing—review and editing, L.Y., L.C., D.T., M.Z., C.A. and R.A.; visualization, L.Y.; supervision, L.C., D.T., M.Z., C.A. and R.A.; project administration, L.C. All authors have read and agreed to the published version of the manuscript.

Funding

The research of Lin Yang was supported by the Australian Research Council Integrated Operations for Complex Resources Industrial Transformation Training Centre (project number IC190100017) and funded by universities, industry, and the Australian Government.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
AEAcoustic emission
CCDCharge-coupled device
CFDComputational fluid dynamics
ECTElectrical Capacitance Tomography
ERTElectrical resistance tomography
EITElectrical impedance tomography
FFTFast Fourier transform
RMSRoot mean square
UTUltrasound tomography

References

  1. Revuelta, M.B. Mineral Resources: From Exploration to Sustainability Assessment; Springer: Berlin/Heidelberg, Germany, 2017. [Google Scholar]
  2. Wills, B.A.; Napier-Munn, T. 9-Classification. In Wills’ Mineral Processing Technology; Elsevier: Amsterdam, The Netherlands, 2005; pp. 203–224. [Google Scholar] [CrossRef]
  3. Gorain, B.K. Physical Processing: Innovations in Mineral Processing. In Innovative Process Development in Metallurgical Industry; Springer: Berlin/Heidelberg, Germany, 2016; pp. 9–65. [Google Scholar] [CrossRef]
  4. Concha, F.; Barrientos, A.; Montero, J.; Sampaio, R. Air core and roping in hydrocyclones. Int. J. Miner. Process. 1996, 44–45, 743–749. [Google Scholar] [CrossRef]
  5. Pérez, D.; Cornejo, P.; Rodríguez, C.; Concha, F. Transition from spray to roping in hydrocyclones. Miner. Eng. 2018, 123, 71–84. [Google Scholar] [CrossRef]
  6. Laguitton, D. The SPOC Manual—Simulated Processing of Ore and Coal. CANMET, EMR, Canada, 1985. [Google Scholar]
  7. Daza, J.; Cornejo, P.; Rodríguez, C.; Betancourt, F.; Concha, F. Influence of the feed particle size distribution on roping in hydrocyclones. Miner. Eng. 2020, 157, 106583. [Google Scholar] [CrossRef]
  8. Heiskanen, K. Experimental hydrocyclone roping models. Chem. Eng. J. 2000, 80, 289–293. [Google Scholar] [CrossRef]
  9. Neesse, T.; Schneider, M.; Golyk, V.; Tiefel, H. Measuring the operating state of the hydrocyclone. Miner. Eng. 2004, 17, 697–703, Hydrocyclones ’03. [Google Scholar] [CrossRef]
  10. Hulbert, D. Measurement Method and Apparatus for Hydrocyclones. 1991. Available online: https://data.epo.org/publication-server/rest/v1.0/publication-dates/19930224/patents/EP0522215NWA3/document.pdf (accessed on 17 January 2025).
  11. Weissleder, R.; Wittenberg, J.; Harisinghani, M.G.; Chen, J.W. (Eds.) Imaging Physics. In Primer of Diagnostic Imaging, 5th ed.; Mosby: Philadelphia, PA, USA, 2011; Chapter 14; pp. 690–746. [Google Scholar] [CrossRef]
  12. Williams, R.A.; Dickin, F.; Dyakowski, T.; Ilyas, O.; Abdullah, Z.; Beck, M. Looking into mineral process plant? Miner. Eng. 1992, 5, 867–881. [Google Scholar] [CrossRef]
  13. Williams, R.A.; Ilyas, O.; Dyakowski, T.; Dickin, F.; Gutierrez, J.; Wang, M.; Beck, M.; Shah, C.; Rushton, A. Air core imaging in cyclonic separators: Implications for separator design and modelling. Chem. Eng. J. Biochem. Eng. J. 1995, 56, 135–141. [Google Scholar] [CrossRef]
  14. Williams, R.A.; Jia, X.; West, R.M.; Wang, M.; Cullivan, J.; Bond, J.; Faulks, I.; Dyakowski, T.; Wang, S.; Climpson, N.; et al. Industrial monitoring of hydrocyclone operation using electrical resistance tomography. Miner. Eng. 1999, 12, 1245–1252. [Google Scholar] [CrossRef]
  15. West, R.M.; Jia, X.; Williams, R.A. Parametric modelling in industrial process tomography. Chem. Eng. J. 2000, 77, 31–36. [Google Scholar] [CrossRef]
  16. Williams, R.A.; Dickin, F.; Gutierrez, J.; Dyakowski, T.; Beck, M. Using electrical impedance tomography for controlling hydrocyclone underflow discharge. Control Eng. Pract. 1997, 5, 253–256. [Google Scholar] [CrossRef]
  17. Gutierrez, J.; Dyakowski, T.; Beck, M.; Williams, R.A. Using electrical impedance tomography for controlling hydrocyclone underflow discharge. Powder Technol. 2000, 108, 180–184. [Google Scholar] [CrossRef]
  18. Schlaberg, H.I.; Podd, F.J.; Hoyle, B.S. Ultrasound process tomography system for hydrocyclones. Ultrasonics 2000, 38, 813–816. [Google Scholar] [CrossRef]
  19. Olson, T.J.; Waterman, R.J. Hydrocyclone Roping Detector and Method. U.S. Patent 6,983,850, 10 January 2006. [Google Scholar]
  20. Daily, W.; Ramirez, A.; Binley, A.; LeBrecque, D. Electrical resistance tomography. Lead. Edge 2004, 23, 409–512. [Google Scholar] [CrossRef]
  21. Dickin, F.; Wang, M. Electrical resistance tomography for process applications. Meas. Sci. Technol. 1996, 7, 247. [Google Scholar] [CrossRef]
  22. Dyakowski, T.; Jeanmeure, L.F.; Jaworski, A.J. Applications of electrical tomography for gas–solids and liquid–solids flows—A review. Powder Technol. 2000, 112, 174–192. [Google Scholar] [CrossRef]
  23. Rakesh, A.; Kumar Reddy, V.T.S.R.; Narasimha, M. Air-Core Size Measurement of Operating Hydrocyclone by Electrical Resistance Tomography. Chem. Eng. Technol. 2014, 37, 795–805. [Google Scholar] [CrossRef]
  24. Vakamalla, T.R.; Kumbhar, K.S.; Gujjula, R.; Mangadoddy, N. Computational and experimental study of the effect of inclination on hydrocyclone performance. Sep. Purif. Technol. 2014, 138, 104–117. [Google Scholar] [CrossRef]
  25. Diddi, S.; Jampana, P.V.; Mangadoddy, N. Evaluation of two noniterative electrical resistance tomography (ERT) reconstruction algorithms for air-Core measurements in Hydrocyclone. Ind. Eng. Chem. Res. 2022, 61, 18017–18029. [Google Scholar] [CrossRef]
  26. Metso. CycloneSense™. 2024. Available online: https://www.metso.com/portfolio/cyclonesense/ (accessed on 9 November 2024).
  27. Lobo, B.; Hermosa, C.; Abella, A.; Gordo, F. Electrical impedance tomography. Ann. Transl. Med. 2018, 6, 26. [Google Scholar] [CrossRef]
  28. Antonio, J.; Gnecchi, G.; Pineda, E. Evaluation of Methods for On-line Image Analysis of Electrical Impedance Tomography Images. In Proceedings of the Electronics, Robotics and Automotive Mechanics Conference (CERMA 2007), Cuernavaca, Mexico, 25–28 September 2007; pp. 283–288. [Google Scholar] [CrossRef]
  29. Xu, L.; Han, Y.; Xu, L.A.; Yang, J. Application of ultrasonic tomography to monitoring gas/liquid flow. Chem. Eng. Sci. 1997, 52, 2171–2183. [Google Scholar] [CrossRef]
  30. Tan, C.; Dong, F. Sensor Instrumentation for Flow Measurement. In Encyclopedia of Sensors and Biosensors, 1st ed.; Narayan, R., Ed.; Elsevier: Oxford, UK, 2023; pp. 536–554. [Google Scholar] [CrossRef]
  31. Tsantiotis, D.C.; Costopoulos, T.N.; Zotos, I.S. Maintenance of Mechanical Drivers and Fault Diagnosis of Machine Elements. In Proceedings of the 2nd International Conference “From Scientific Computing to Computational Engineering”, 2nd IC-SCCE, Athens, Greece, 5–8 July 2006; pp. 5–8. [Google Scholar]
  32. Kumar, B.K.; Diwakar, G.; Satynarayana, M. Determination of unbalance in rotating machine using vibration signature analysis. Int. J. Mod. Eng. Res. 2012, 2, 33415–33421. [Google Scholar]
  33. Ojha, S.; Sarangi, D.; Pal, B.; Biswal, B. Performance monitoring of vibration in belt conveyor system. J. Eng. Res. Appl. 2014, 4, 22–31. [Google Scholar]
  34. Bansal, P.; Vedaraj, I.S.R.; Student, P.G. Monitoring and Analysis of Vibration Signal in Machine Tool Structures. Int. J. Eng. Dev. Res. 2014, 2, 2310–2317. [Google Scholar]
  35. Algule, S.R.; Hujare, D.P. Experimental study of unbalance in shaft rotor system using vibration signature analysis. Int. J. 2015, 3, 124–130. [Google Scholar]
  36. Dubey, R.; Singh, G.; Majumder, A. Roping: Is it an optimum dewatering performance condition in a hydrocyclone? Powder Technol. 2017, 321, 218–231. [Google Scholar] [CrossRef]
  37. Nayak, D.K.; Das, D.P.; Behera, S.K.; Das, S.P.; Sadangi, J.K. Prevention of hydrocyclone choking through detection of sub-Hz frequency shift of vibration signal. J. Vib. Eng. Technol. 2020, 8, 517–528. [Google Scholar] [CrossRef]
  38. Huang, N.E.; Shen, Z.; Long, S.R.; Wu, M.C.; Shih, H.H.; Zheng, Q.; Yen, N.C.; Tung, C.C.; Liu, H.H. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proc. R. Soc. Lond. Ser. A Math. Phys. Eng. Sci. 1998, 454, 903–995. [Google Scholar] [CrossRef]
  39. Wang, G.; Liu, Q.; Wang, C.; Dong, L.; Dai, D.; Shen, L. Study of Blockage Diagnosis for Hydrocyclone Using Vibration-Based Technique Based on Wavelet Denoising and Discrete-Time Fourier Transform Method. Processes 2020, 8, 440. [Google Scholar] [CrossRef]
  40. Mei, S.; Liu, X.; Mei, S. Cell-filtering-based multi-scale Shannon–Cosine wavelet denoising method for locust slice images. Int. J. Wavelets Multiresolution Inf. Process. 2019, 17, 1950035. [Google Scholar] [CrossRef]
  41. Ong, P.M.; Galvez, M.C.; Vallar, E.; Shiina, T. Wavelet Denoising Applied to Light Emitting Diode Lidar Signal. Adv. Sci. Lett. 2017, 23, 1374–1378. [Google Scholar] [CrossRef]
  42. Smith, B.C. Fundamentals of Fourier Transform Infrared Spectroscopy; CRC Press: Boca Raton, FL, USA, 2011. [Google Scholar]
  43. Candan, C.; Kutay, M.A.; Ozaktas, H.M. The discrete fractional Fourier transform. IEEE Trans. Signal Process. 2000, 48, 1329–1337. [Google Scholar] [CrossRef]
  44. Osgood, B. The Fourier transform and its applications. Lect. Notes EE 2009, 261, 20. [Google Scholar]
  45. Lockner, D. The role of acoustic emission in the study of rock fracture. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 1993, 30, 883–899. [Google Scholar] [CrossRef]
  46. Du, K.; Li, X.; Tao, M.; Wang, S. Experimental study on acoustic emission (AE) characteristics and crack classification during rock fracture in several basic lab tests. Int. J. Rock Mech. Min. Sci. 2020, 133, 104411. [Google Scholar] [CrossRef]
  47. Rodríguez, P.; Celestino, T.B. Application of acoustic emission monitoring and signal analysis to the qualitative and quantitative characterization of the fracturing process in rocks. Eng. Fract. Mech. 2019, 210, 54–69. [Google Scholar] [CrossRef]
  48. König, F.; Marheineke, J.; Jacobs, G.; Sous, C.; Zuo, M.J.; Tian, Z. Data-driven wear monitoring for sliding bearings using acoustic emission signals and long short-term memory neural networks. Wear 2021, 476, 203616. [Google Scholar] [CrossRef]
  49. König, F.; Sous, C.; Chaib, A.O.; Jacobs, G. Machine learning based anomaly detection and classification of acoustic emission events for wear monitoring in sliding bearing systems. Tribol. Int. 2021, 155, 106811. [Google Scholar] [CrossRef]
  50. Shen, C.H. Acoustic emission based grinding wheel wear monitoring: Signal processing and feature extraction. Appl. Acoust. 2022, 196, 108863. [Google Scholar] [CrossRef]
  51. Twardowski, P.; Tabaszewski, M.; Wiciak-Pikuła, M.; Felusiak-Czyryca, A. Identification of tool wear using acoustic emission signal and machine learning methods. Precis. Eng. 2021, 72, 738–744. [Google Scholar] [CrossRef]
  52. Ferrando Chacon, J.L. Fault Detection in Rotating Machinery Using Acoustic Emission. Ph.D. Thesis, Brunel University London, Uxbridge, UK, 2015. [Google Scholar]
  53. Zhang, Y.; Lu, W.; Chu, F. Planet gear fault localization for wind turbine gearbox using acoustic emission signals. Renew. Energy 2017, 109, 449–460. [Google Scholar] [CrossRef]
  54. Hou, R.; Hunt, A.; Williams, R. Acoustic monitoring of hydrocyclone performance. Miner. Eng. 1998, 11, 1047–1059. [Google Scholar] [CrossRef]
  55. Hou, R.; Hunt, A.; Williams, R. Acoustic monitoring of hydrocyclones. Powder Technol. 2002, 124, 176–187. [Google Scholar] [CrossRef]
  56. Hao, Q.; Shen, Y.; Wang, Y.; Zhang, X. Detection method based on Kalman filter for high speed rail defect AE signal on wheel-rail rolling rig. Nondestruct. Test. Eval. 2018, 33, 75–91. [Google Scholar] [CrossRef]
  57. Kim, S.J.; Kim, S.; Lee, S.; Youn, B.D.; Kim, T. Effective band-selection algorithm for rolling element bearing diagnosis using AE sensor data under noisy conditions. Struct. Multidiscip. Optim. 2022, 65, 275. [Google Scholar] [CrossRef]
  58. Zhou, C.; Zhang, Y. Particle filter based noise removal method for acoustic emission signals. Mech. Syst. Signal Process. 2012, 28, 63–77. [Google Scholar] [CrossRef]
  59. Zhao, J.; Ma, Y. Denoising Method of AE Signal by Using Wavelet Analysis. In Proceedings of the 15th World Conference on Non-Destructive Testing, Rome, Italy, 15–21 October 2000; pp. 15–21. [Google Scholar]
  60. Grosse, C.U.; Finck, F.; Kurz, J.H.; Reinhardt, H.W. Improvements of AE technique using wavelet algorithms, coherence functions and automatic data analysis. Constr. Build. Mater. 2004, 18, 203–213. [Google Scholar] [CrossRef]
  61. Prajna, K.; Mukhopadhyay, C. Fractional Fourier transform based adaptive filtering techniques for acoustic emission signal enhancement. J. Nondestruct. Eval. 2020, 39, 14. [Google Scholar] [CrossRef]
  62. Farooq, Y.; Savaş, S. Noise removal from the image using convolutional neural networks-based denoising auto encoder. J. Emerg. Comput. Technol. 2024, 3, 21–28. [Google Scholar] [CrossRef]
  63. Almazrouei, E.; Gianini, G.; Almoosa, N.; Damiani, E. A deep learning approach to radio signal denoising. In Proceedings of the 2019 IEEE Wireless Communications and Networking Conference Workshop (WCNCW), Marrakech, Morocco, 15–18 April 2019; pp. 1–8. [Google Scholar]
  64. Chatterjee, S.; Bhattacherjee, A.; Samanta, B.; Pal, S.K. Image-based quality monitoring system of limestone ore grades. Comput. Ind. 2010, 61, 391–408. [Google Scholar] [CrossRef]
  65. Perez, C.A.; Estévez, P.A.; Vera, P.A.; Castillo, L.E.; Aravena, C.M.; Schulz, D.A.; Medina, L.E. Ore grade estimation by feature selection and voting using boundary detection in digital image analysis. Int. J. Miner. Process. 2011, 101, 28–36. [Google Scholar] [CrossRef]
  66. Mukherjee, D.P.; Potapovich, Y.; Levner, I.; Zhang, H. Ore image segmentation by learning image and shape features. Pattern Recognit. Lett. 2009, 30, 615–622. [Google Scholar] [CrossRef]
  67. Ma, X.; Zhang, P.; Man, X.; Ou, L. A new belt ore image segmentation method based on the convolutional neural network and the image-processing technology. Minerals 2020, 10, 1115. [Google Scholar] [CrossRef]
  68. Oestreich, J.; Tolley, W.; Rice, D. The development of a color sensor system to measure mineral compositions. Miner. Eng. 1995, 8, 31–39. [Google Scholar] [CrossRef]
  69. Casali, A.; Gonzalez, G.; Vallebuona, G.; Perez, C.; Vargas, R. Grindability soft-sensors based on lithological composition and on-line measurements. Miner. Eng. 2001, 14, 689–700. [Google Scholar] [CrossRef]
  70. Tessier, J.; Duchesne, C.; Bartolacci, G. A machine vision approach to on-line estimation of run-of-mine ore composition on conveyor belts. Miner. Eng. 2007, 20, 1129–1144. [Google Scholar] [CrossRef]
  71. Khorram, F.; Memarian, H.; Tokhmechi, B.; Soltanian-zadeh, H. Limestone chemical components estimation using image processing and pattern recognition techniques. J. Min. Environ. 2011, 2, 126–135. [Google Scholar]
  72. Lepistö, L.; Kunttu, I.; Visa, A. Rock image classification using color features in Gabor space. J. Electron. Imaging 2005, 14, 040503. [Google Scholar] [CrossRef]
  73. Murtagh, F.; Starck, J.L. Wavelet and curvelet moments for image classification: Application to aggregate mixture grading. Pattern Recognit. Lett. 2008, 29, 1557–1564. [Google Scholar] [CrossRef]
  74. Bharati, M.H.; MacGregor, J.F. Multivariate image analysis for real-time process monitoring and control. Ind. Eng. Chem. Res. 1998, 37, 4715–4724. [Google Scholar] [CrossRef]
  75. Duchesne, C. Multivariate image analysis in mineral processing. In Advanced Control and Supervision of Mineral Processing Plants; Springer: Berlin/Heidelberg, Germany, 2010; pp. 85–142. [Google Scholar]
  76. Massaro, A.; Galiano, A. Image Processing and Post-Data Mining Processing for Security in Industrial Applications: Security in Industry. In Handbook of Research on Intelligent Data Processing and Information Security Systems; IGI Global: Hershey, PA, USA, 2020; pp. 117–146. [Google Scholar]
  77. Gordan, M.; Sabbagh-Yazdi, S.R.; Ismail, Z.; Ghaedi, K.; Carroll, P.; McCrum, D.; Samali, B. State-of-the-art review on advancements of data mining in structural health monitoring. Measurement 2022, 193, 110939. [Google Scholar] [CrossRef]
  78. Petersen, K.; Aldrich, C.; Van Deventer, J.; McInnes, C.; Stange, W. Hydrocyclone underflow monitoring using image processing methods. Miner. Eng. 1996, 9, 301–315. [Google Scholar] [CrossRef]
  79. Aldrich, C.; Uahengo, F.; Kistner, M. Estimation of particle size in hydrocyclone underflow streams by use of Multivariate Image Analysis. Miner. Eng. 2015, 70, 14–19. [Google Scholar] [CrossRef]
  80. Dubey, R.; Climent, E.; Banerjee, C.; Majumder, A.K. Performance monitoring of a hydrocyclone based on underflow discharge angle. Int. J. Miner. Process. 2016, 154, 41–52. [Google Scholar] [CrossRef]
  81. Krishna, V.; Sripriya, R.; Kumar, V.; Chakraborty, S.; Meikap, B. Identification and prediction of air core diameter in a hydrocyclone by a novel online sensor based on digital signal processing technique. Chem. Eng. Process. Process Intensif. 2010, 49, 165–176. [Google Scholar] [CrossRef]
  82. Mazumdar, A.; Dubey, R.; Banerjee, C.; Sengupta, K.; Majumder, A. A study on the characteristics of spray angle formation in a 2 inch hydrocyclone using water only. Int. J. Miner. Process. 2014, 126, 141–145. [Google Scholar] [CrossRef]
  83. Janse van Vuuren, M.; Aldrich, C.; Auret, L. Detecting changes in the operational states of hydrocyclones. Miner. Eng. 2011, 24, 1532–1544. [Google Scholar] [CrossRef]
  84. Giglia, K.; Aldrich, C. Operational state detection in hydrocyclones with convolutional neural networks and transfer learning. Miner. Eng. 2020, 149, 106211. [Google Scholar] [CrossRef]
  85. Simonyan, K.; Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv 2014, arXiv:1409.1556. [Google Scholar]
  86. Plitt, L.; Flintoff, B.; Stuffco, T. Roping in hydrocyclones. In Proceedings of the 3rd International Conference on Hydrocyclones, Oxford, UK, 30 September–2 October 1987; Elsevier: Amsterdam, The Netherlands, 1987; pp. 21–34. [Google Scholar]
  87. Bustamante, M. Effect of the Hydrocyclone Geometry on Normal Operation Conditions. Ph.D. Thesis, University of Concepción, Concepción, Chile, 1991. (In Spanish). [Google Scholar]
  88. Li, Y.; Liu, C.; Zhang, T.; Li, D.; Zheng, L. Experimental and numerical study of a hydrocyclone with the modification of geometrical structure. Can. J. Chem. Eng. 2018, 96, 2638–2649. [Google Scholar] [CrossRef]
  89. Davailles, A.; Climent, E.; Bourgeois, F.; Majumder, A.K. Analysis of swirling flow in hydrocyclones operating under dense regime. Miner. Eng. 2012, 31, 32–41. [Google Scholar] [CrossRef]
  90. Tyeb, M.; Mishra, S.; Singh, A.; Majumder, A. Prediction of operating state of hydrocyclones using vibrometry and 1D convolutional neural networks. Adv. Powder Technol. 2024, 35, 104337. [Google Scholar] [CrossRef]
Minerals 15 00110 g001
Figure 1. The structure of a hydrocyclone.
Figure 1. The structure of a hydrocyclone.
Minerals 15 00110 g001
Minerals 15 00110 g002
Figure 2. The schematic of the roping detection equipment innovated by Hulbert [10].
Figure 2. The schematic of the roping detection equipment innovated by Hulbert [10].
Minerals 15 00110 g002
Minerals 15 00110 g003
Figure 3. The schematic of the roping detection equipment innovated by Rakesh et al. [23].
Figure 3. The schematic of the roping detection equipment innovated by Rakesh et al. [23].
Minerals 15 00110 g003
Minerals 15 00110 g004
Figure 4. Schematic of experimental setup (redrawn from Nayak et al. [37]).
Figure 4. Schematic of experimental setup (redrawn from Nayak et al. [37]).
Minerals 15 00110 g004
Minerals 15 00110 g005
Figure 5. Schematic of a 125 mm hydrocyclone test rig (redrawn from Hou et al. [55]).
Figure 5. Schematic of a 125 mm hydrocyclone test rig (redrawn from Hou et al. [55]).
Minerals 15 00110 g005
Minerals 15 00110 g006
Figure 6. Principle of the laser–optical measuring device (redrawn from Neesse et al. [9]).
Figure 6. Principle of the laser–optical measuring device (redrawn from Neesse et al. [9]).
Minerals 15 00110 g006
Minerals 15 00110 g007
Figure 7. The schematic of the underflow width measurement (modified from Janse van Vuuren et al. [83]).
Figure 7. The schematic of the underflow width measurement (modified from Janse van Vuuren et al. [83]).
Minerals 15 00110 g007
Minerals 15 00110 g008
Figure 8. Air core in hydrocyclone (redrawn from Li et al. [88]). The blue line represents the path of water flow.
Figure 8. Air core in hydrocyclone (redrawn from Li et al. [88]). The blue line represents the path of water flow.
Minerals 15 00110 g008
Table 1. Limitations of noise reduction algorithms for AE method.
Table 1. Limitations of noise reduction algorithms for AE method.
AlgorithmLimitationImpactReference
Bandpass FilteringRequires precise knowledge of signal frequency range.Risk of losing critical signal components or retaining noise if range is incorrectly set.Kim et al. [57], Zhou and Zhang [58]
Wavelet DenoisingChoice of wavelet function and decomposition level affects performance.Can result in loss of important features or incomplete noise suppression.Zhao and Ma [59], Grosse et al. [60]
Adaptive FilteringRequires a reference noise signal, which may not always be available.Ineffective in unpredictable or broadband noise environments.Prajna and Mukhopadhyay [61]
Deep Learning-Based DenoisingRequires large labeled datasets for training; computationally intensive.High initial costs and unsuitable for real-time industrial applications without optimization.Farooq and Savaş [62], Almazrouei et al. [63]
Kalman FilteringAssumes linear and Gaussian noise, requiring accurate model parameters.Suboptimal in complex or non-linear noise environments.Hao et al. [56]
Table 2. Principles and limitations of different methods.
Table 2. Principles and limitations of different methods.
MethodPrincipleLimitation
Mechanical method (Section 2.1)Detecting through designed mechanical structure• High maintenance cost
ERT
(Section 2.2.1)		Evaluating the conductivity distribution by measuring the resistance• Inacurate in the hydrocyclone central region
• Trade-off between reconstruction accuracy and processing time
• Sensitive to conductive media
TomographyEIT
(Section 2.2.2)		Evaluating the conductivity distribution by measuring resistance and reactance• Sensitive to conductive media
• Complexity of electrode array installation
• Computational time required for high resolution
UT
(Section 2.2.3)		Utilizing ultrasonic waves to generate images of the target object• A high sampling rate and a large number of sensors are required
• Data loss can occur when tracking moving objects
Vibrational method
(Section 2.3)		Monitoring changes in vibration patterns induced by mechanical faults in equipment• Need more understanding of vibration patterns
• A robust calibration-based algorithm is required
Acoustic emission (AE) method
(Section 2.4)		Monitoring the changes in the emission signals that correspond to variations in the internal structure• Can be affected by the materials and geometry of hydrocyclone
• Sensitive to background noise
Image processing method
(Section 2.5)		Extracting valuable information from a digital image• Requires strict environmental conditions
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Yang, L.; Chen, L.; Tang, D.; Zanin, M.; Aldrich, C.; Asamoah, R. Roping Prediction Versus Detection: Could Prediction Be Possible? Minerals 2025, 15, 110. https://doi.org/10.3390/min15020110

AMA Style

Yang L, Chen L, Tang D, Zanin M, Aldrich C, Asamoah R. Roping Prediction Versus Detection: Could Prediction Be Possible? Minerals. 2025; 15(2):110. https://doi.org/10.3390/min15020110

Chicago/Turabian Style

Yang, Lin, Lei Chen, Difan Tang, Massimiliano Zanin, Chris Aldrich, and Richmond Asamoah. 2025. "Roping Prediction Versus Detection: Could Prediction Be Possible?" Minerals 15, no. 2: 110. https://doi.org/10.3390/min15020110

APA Style

Yang, L., Chen, L., Tang, D., Zanin, M., Aldrich, C., & Asamoah, R. (2025). Roping Prediction Versus Detection: Could Prediction Be Possible? Minerals, 15(2), 110. https://doi.org/10.3390/min15020110

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop