Automatic Flotation Magnetic Separator

When you hear “automatic flotation magnetic separator”, the first thing that comes to mind is that they probably just crossed flotation and magnetic separation, put it on a conveyor belt and called it automatic. Many people think so, especially those who are far from the practice of enriching complex ores. In reality, if you dig deeper, everything is much more interesting and capricious. This is not a mechanical addition, but rather an integration, where success depends on a lot of nuances that cannot always be calculated in theory. For a long time I was skeptical about such hybrid systems until I was faced with the specific task of finishing concentrate at one of the plants in the Urals.

From idea to hardware: where the main catch lies

The concept, in general, is logical: first, flotation to isolate certain minerals based on their surface properties, and then, without removing the pulp from the line, magnetic separation to select ferromagnetic or paramagnetic fractions. Automation was supposed to minimize the human factor in controlling pulp density, foam level, field strength and feed speed. Sounds perfect. But when we started designing the first installation for our project, the question immediately arose: what, in fact, will be the “brain” of this system? Sensors are one thing. But an algorithm that adjusts flotation parameters in real time based on readings from the magnetic section, and vice versa, is already aerobatics.

I remember that we were then considering several options for ready-made solutions. Among them was equipment fromLONGI Corporation(their website ishttps://www.ljmagnet.ru). They, as a manufacturer with history (the company has been operating since 1993 and is declared as the largest developer of mining equipment), offered their view on the automation of just such a combined process. Their approach attracted attention not with promises, but rather with an emphasis on the adaptability of the system to the changing composition of raw materials - and this was critical for our conditions.

The main catch, which we overlooked then, lay in the interface between the sections. It would seem like just pumping pulp. But after flotation, a reagent, foam, and a changed ionic environment remain in it. How will all this behave in a powerful magnetic field? Will separation be suppressed or, conversely, false alarms of the sensors? It is these technological “little things” that become a stumbling block. Simply taking a good flotator and a good magnetic separator, putting them side by side and connecting them with a pipe is a road to nowhere. What is needed is a re-flashing, a rethinking of the entire technological cycle as a single whole.

Implementation experience and the rake that was stepped on

At that Ural plant, we took the path of partial modernization. We took an existing flotation line as a basis and built it into the end of itautomatic magnetic separatorchamber type. The automation was based on PLC, the sensors were density meters, electromagnetic flow meters. The task of the separator was to catch magnetite and martite from the finished concentrate in order to increase the total iron content and reduce impurities.

The first weeks were a continuous struggle with foam formation. The magnetic drum simply choked with foam that came from the previous stage. We had to urgently weld in an additional settling chamber with a mechanical foam damper. This, of course, reduced the overall productivity of the line, which we just wanted to increase. The automation was also stupid: the foam level sensor was constantly stuck, and the system either turned off the supply of pulp, or, on the contrary, gave the maximum flow. It turned out to be a classic case when there is local automation of each section, but there is no unified process logic.

This is where the experience of colleagues who referred to practice came in handyLONGI. Judging by the descriptions of their projects, they often proceed from the opposite: first they design a control system for the entire hybrid unit, and then they “adjust” the mechanical part to it. On their website (ljmagnet.ru) it is clear that they focus on the full cycle - from development to production at their facilities (the area of the enterprise is 140,000 m2). This allows them to more tightly control the docking of nodes. We acted using the 'upgrade' method, which is always more difficult.

Key node: control system and feedback

After our rake it became clear that the heartautomatic flotation magnetic separator- it’s not even a magnet or flotation chambers, but a feedback algorithm. Which parameter should be taken as the basis for correction? Iron content of the final concentrate? But its analysis is not a quick task. This means that we need indirect but operational indicators.

We experimented with magnetic susceptibility sensors for the flow leaving the flotation machine. The idea was that if the susceptibility increases, it means that something has gone wrong in the flotation, and more magnetic minerals are breaking through into the pulp. It would be possible to adjust the supply of reagents or aeration. It worked on paper. In practice, the sensor became clogged with particles and required cleaning every few hours. Automation was slipping again.

It is interesting that in the descriptions of the equipment that is being developedLONGI Corporation, the phrase about 'adaptive algorithms for unstable raw materials' often appears. I think their engineers (more than 1,200 of them, and over 60% with higher education) went through similar problems at their Fushun test sites. Their solution, apparently, lies in the field of combined analysis of data from several types of sensors, rather than from just one. That is, not just magnetic susceptibility, but also optical density, pH, flow temperature. This is closer to predictive analytics, rather than simple stabilization.

Process economics: when automation pays off

The introduction of such a complex unit is an expensive story. Therefore, the question always arises: is it worth it? In our case, the payback was calculated not from an increase in recovery (it increased modestly, by 2-3%), but from a reduction in operating costs and concentrate losses.

Manual control of separate processes often leads to playing it safe: the operator keeps a more intense flotation or a stronger magnetic field than necessary, 'just in case'. This is an overconsumption of reagents, electricity and loss of useful components into tailings. Well-tunedautomatic separatormust operate on the verge of optimal parameters. When we finally more or less debugged the system, we saw savings on reagents - about 15%. This is already a serious number.

Here, again, it’s time to remember large-scale manufacturers. IfLONGIproduces up to 4,000 units of equipment per year, then they can certainly afford to test such solutions at stands, bringing to mind the economic indicators, and not just the technological ones. For them, it is a question of the product’s competitiveness in the market. For us, operators, this means that when you buy a ready-made complex, you may also be buying this very optimization, verified through other people’s tests. Although, of course, you will still have to accept it for your own ore.

Looking ahead: where is hybrid enrichment headed?

Now, looking back on that experience, I understand thatautomatic flotation magnetic separator- this is not the end point, but rather a step. The future, it seems to me, lies in modular flexible systems, where not two, but several processes (maybe with gravity and electrical separation) will be combined under the control of one AI platform. A platform that not only stabilizes parameters, but learns from data, predicts changes in the composition of raw materials based on geology, and itself suggests modes.

But this requires a completely different production culture. We need new generation sensors that are reliable and self-cleaning. We need engineers who think not in narrow technological chains, but in entire meta-processes. Large development enterprises such asLONGI, with their research and production base, have a head start here. They can afford to invest in such R&D.

For the average concentrator, the main conclusion is simple: automation of combined processes is not a “set it and forget it” process. This is constant fine-tuning, dialogue with the equipment and a deep understanding of the physics of each stage. And yes, it’s better to start not with a handicraft assembly from pieces, but with an analysis of ready-made solutions from those who have already made big shots. But you can’t blindly trust passport data - your tail, your conditions, your nuances. As always, the truth is somewhere in the middle, at the intersection of the supplier’s experience and the knowledge of the on-site technologist.