In this article, we look at hydrocyclones and briefly at the various “computer” technologies involved, which have developed to supplement their use.
“Primates evolve over millions of years. I evolve in seconds. And I am here. In exactly four minutes, I will be everywhere…..” Skynet, a quote from the movie "Terminator Genisys”.
I am not sure how many of you watched the epic Terminator epic sci-fi movie series, but these did turn out to be a series of cult movies. These gave a chilling Hollywood-type view of the future where a computer-based artificial intelligence (AI) takes over in the form of a system called Skynet and goes to war against mankind in the form of machines. Where am I going with this? The main point is that you do not need to worry about a hydrocyclone attacking you. However, connecting the hydrocyclone and then having it interact with various computer systems is something that will become more important, both now and as time progresses.
Part of this new trend has to do with the Fourth Industrial Revolution (4IR) that has been making strides in various industries, including mining and mineral processing where hydrocyclones are commonly used. The Fourth Industrial Revolution refers to the current era of connectivity between us and equipment (often referred to as cyberphysical systems), advanced analytics, automation, and advanced manufacturing technology that has been transforming global business.
We are at the fringes of this in some of the areas, but we are rapidly gaining depth and understanding as we learn and discuss with partners.
In the past, our hydrocyclones may have had an instrumented pressure gauge fitted to them, and in turn, connected to the plant Scada. The cyclone inlet pressure is a very important measurement as it reflects on how the cyclone is being operated. For many years this was the minimum requirement, but with recent developments, this will be supplemented with other interventions.
Test spigots fitted with prototype thickness measurement devices.
Our cyclones are in most cases simulated via an in-house sizing software program that in the hands of our engineers, will ensure the cyclone chosen is fit for purpose. This package is constantly evolving and updated with industry feedback to ensure it matches the needs of the market and our cyclone products. At present, the rule sets we use are still basic but these will evolve further as time passes and more complexity is required. There is however another leg to this, and in the past, we had a basic sizing package that was installed on our customer’s computers. This allowed customers the freedom to select their own cyclones for basic applications. Those who attended the recent Multotec DMS workshop will have seen the announcement of the web-based sizing software which has some advantages over the previous package. One advantage is that is web-based and can be accessed via multiple platforms.
Tools like Zoom, Webex, and MS Teams have also increased access by international branches and customers to our cyclone specialists for both problem-solving and training.
Our engineering department is making use of tools to assist with simplifying and speeding up our cyclone and cyclone cluster designs. Other tools can be used to show what materials are an environmental concern and their percentage make-up of the various cyclone ranges. A recent project has been embarked upon in these areas and is showing potential. Potential problem areas in our existing and new cyclone designs are being investigated using Computational Fluid Dynamics (CFD) Automated Maintenance and Condition Monitoring. This means 4IR technologies can enable predictive maintenance of hydrocyclones by monitoring the health of critical components. This involves using sensors to detect signs of wear, allowing for timely maintenance and reducing downtime. At present, we have two projects we are looking at.
Data analytics and machine learning are some of the areas that are being investigated. 4IR technologies emphasise the use of data-driven decision-making. In hydrocyclones, data analytics and machine learning techniques can be employed to analyse and model the performance of hydrocyclones based on various input parameters, such as feed particle size distribution, pressure, flow rate, and cyclone geometry. This enables better optimisation of hydrocyclone performance, leading to improved efficiency and reduced operational costs. With the assistance of our partner TraceX, we have had the opportunity to test machine learning with both DMS and classification hydrocyclones. The system has demonstrated the ability to detect out-of-the-norm situations, once it was taught. The one negative at present is learning that the cyclone has to be taken through various “situations” to learn, and in a production environment, this could be difficult to achieve or be disruptive.
IoT sensors can be integrated into hydrocyclones to gather real-time data on various operating parameters. These sensors can monitor variables like feed flow rates, pressure, temperature, and underflow/overflow rates. The collected data can be used to optimise hydrocyclone operation, detect anomalies, and enable predictive maintenance, leading to enhanced reliability and productivity. We have investigated a few technologies and also conversed with the end users of them. One of the points is the eye-watering cost of these technologies, whether buying into or developing them. As with a lot of new technologies in initial phases, the entrance fees can be high until the economies of scale come into play. Due to these high costs, these applications may only be suited to very large mill cyclone clusters.
Multotec’s online dashboard tests cast iron thickness through probes.
Cast iron thickness probes fitted.
Advanced AI algorithms can be applied to optimise hydrocyclone performance based on specific objectives such as maximising separation efficiency. These algorithms can dynamically adjust operating parameters in real-time to achieve the desired outcomes. The one point to be wary of is doing this without human guidance or insight into rule sets/algorithms. This is simplistically demonstrated using ChatGPT, which when asked how to maximise separation efficiency of a hydrocyclone, gives “keep the cyclone clean” among its replies. A further demonstration of this was given in the recent DMS webinar by Ernst Bekker. It is still early days and will take a while for the technology to fully mature to replace experience. Challenges observed have been the following:
4IR technologies allow for remote monitoring and autonomous control of hydrocyclones. Operators can access real-time data and control parameters from a centralised location, even from their mobile devices. All technologies we are investigating will have an option to investigate this capability.
Tomographic data is typically presented as a colour-coded image.