In a mineral processing plant, what often gives production managers the biggest headache is not a sudden equipment shutdown, but rather that the equipment is all running while production indices remain persistently unstable.

Today the concentrate grade is normal, but tomorrow it suddenly drops; yesterday the recovery rate was decent, but today the tailings grade has increased. Operators continuously adjust reagents, water addition, and feed rates. Sometimes the indices improve, only to fluctuate again after a while.

Where exactly is the problem?

If the causes are simply attributed to changes in ore properties, unstable grinding fineness, or unreasonable reagent addition, the analysis is not deep enough. What truly needs to be clarified is: which variable changed first, through what mechanism it affected subsequent processes, and ultimately how it was reflected in the concentrate grade and recovery rate?

Only by clarifying this causal chain can we move from passive adjustment to evidence-based control.

1. First, Distinguish: Grade Fluctuation, Recovery Fluctuation, and Throughput Fluctuation Are Not the Same Problem
In mineral processing production, a decrease in concentrate grade, a decrease in recovery rate, and a decrease in throughput are often collectively referred to as “poor indices,” but the causes of the three may be completely different.

1. Decrease in concentrate grade
This may be related to insufficient mineral liberation, gangue entrainment, decreased flotation selectivity, an inappropriate reagent regime, or changes in feed properties.
For example, some valuable minerals are not fully liberated and may enter the concentrate together with gangue; if froth entrainment increases, more gangue may also enter the concentrate, causing a decrease in grade.

2. Decrease in recovery rate
The recovery rate focuses on how much of the valuable metal contained in the raw ore ultimately enters the concentrate.
If valuable minerals are not fully liberated, flotation kinetics deteriorate, or operating conditions do not adapt to ore changes, more valuable metals may enter the tailings.

3. Decrease in throughput
A decrease in throughput may result from insufficient feed, increased ore hardness, limited mill capacity, equipment failure, or bottlenecks in downstream processes.
A decrease in throughput does not necessarily mean a decrease in grade or recovery; conversely, blindly increasing throughput may also push the grinding and flotation systems beyond their reasonable operating ranges.

Therefore, before analyzing index fluctuations, it is essential to first clarify exactly which index has changed and whether the other indices have changed simultaneously.

2. What Truly Affects Production Indices Is Often the Transfer Relationship Between Upstream and Downstream Processes
Taking a typical grinding–classification–flotation flowsheet as an example, production indices are not determined by a single piece of equipment alone, but by multiple stages working together.

The main influence pathways can be understood as:
Changes in raw ore properties → Changes in grinding load → Changes in classification state → Changes in overflow particle size → Changes in flotation conditions → Changes in concentrate grade and recovery rate.

For example, if the hardness of a certain batch of ore increases, the effective processing capacity of the mill may decrease at a constant feed rate, and the product particle size will gradually become coarser.
If the operators fail to detect the change in overflow particle size in time and continue to operate under the original flotation conditions, some valuable minerals may be difficult to recover effectively due to insufficient liberation.
At this point, the tailings grade may rise, and the recovery rate will decrease.
If the operators only see the decrease in recovery rate and directly increase the collector dosage, it may not solve the problem of insufficient liberation, and may even increase reagent costs or affect concentrate selectivity.

This is a typical case of upstream causes and downstream symptoms.

When flotation indices show problems, it does not necessarily mean the problem originated in the flotation stage.

3. Why Do Production Indices Still Fluctuate Even When Equipment Parameters Are Normal?
This is a problem that many mineral processing plants easily overlook.

The PLC or DCS screen showing that the feed rate, cyclone pressure, pump sump level, and flotation cell level are all within the set ranges does not mean the entire process is in a stable state.

The reason is that normal process parameters only indicate that certain variables have met the set requirements; they cannot prove that mineral liberation, classification performance, and flotation selectivity are all normal.

Taking the hydrocyclone as an example, pressure within the normal range only indicates that one important operating condition basically meets the requirements, and cannot directly prove that the overflow particle size is qualified.
Even at the same pressure, changes in feed density, ore particle size composition, cyclone wear conditions, and internal flow state may lead to different classification results.

Similarly, maintaining a stable flotation cell level does not mean that the concentrate grade and recovery rate will inevitably be stable. Factors such as pulp properties, mineral floatability, reagent action, and froth entrainment may still alter the separation results.

Therefore, judging whether the process is stable should at least be divided into three levels:

• Equipment level: Whether the equipment is operating normally, and whether current, pressure, flow, and interlock status are abnormal.
• Process level: Whether process conditions such as feed rate, density, particle size, pH, and liquid level are stable.
• Result level: Whether the concentrate grade, recovery rate, tailings grade, and throughput meet the targets.

Only by linking these three levels can we avoid the situation of “the screen is normal, the indices are abnormal, but we don’t know where the problem is.”

4. Why Do Indices Sometimes Become More Unstable the More Operators Adjust?
On site, once indices decline, operators tend to immediately adjust a certain parameter.

If the concentrate grade drops, they reduce the reagent; if the recovery rate drops, they increase the collector; if the froth deteriorates, they adjust the air input or liquid level.

These operations are not necessarily wrong in themselves; the problem is that without first determining the cause, one may mistake the result for the cause.

For example, a decrease in recovery rate may be caused by coarser grinding fineness, or by changes in ore properties, abnormal reagent addition, or inappropriate flotation conditions.
If the root cause is insufficient grinding fineness, simply increasing the collector dosage may not be effective.

In addition, the flotation process has a certain response time. After a reagent adjustment, the pulp needs to go through a certain transport and reaction process before it is ultimately reflected in the concentrate or tailings indices.
If operators make a second or third adjustment before the results have fully manifested, continuous over-adjustment may occur, causing parameters to fluctuate back and forth.

This is also why automatic control cannot be simply understood as “as soon as a parameter deviates, the valve acts immediately.”
A reasonable control strategy should consider measurement reliability, process lag, actuator response, and the interactions among variables, and should set appropriate adjustment speeds, control ranges, and necessary operating constraints.

5. How to Find the Causes of Production Index Fluctuations from Data?
Truly valuable analysis is not simply recording “today the grade dropped,” but placing production indices and process parameters on the same timeline.

Suppose a certain processing plant experiences a decrease in recovery rate over a period of time. The analysis can be conducted according to the following approach.

Step 1: Confirm whether the index change is real.
Check the sampling times, assay results, metering data, and statistical criteria for the raw ore, concentrate, and tailings to avoid drawing erroneous conclusions due to insufficient sampling representativeness or data errors.

Step 2: Review the process trends before the anomaly occurred.
Focus on data such as feed rate, mill load, water addition, pump sump level, cyclone pressure, and overflow particle size.
What should be noted is not only the values at the time of the anomaly, but also which variables first showed obvious changes before the anomaly occurred.

Step 3: Consider the time lag of the process.
There is usually a certain time interval from when changes occur in the grinding system, to when the relevant pulp enters the flotation, and then to when the differences are reflected in the concentrate and tailings sampling.
If material transport time, process response time, and sampling cycles are not taken into account, downstream index changes may be incorrectly attributed to operations occurring at the same moment.

Step 4: Judge the cause by comparing multiple indices.
If the overflow particle size becomes coarser while the tailings grade rises, priority should be given to checking mineral liberation and flotation adaptability, but this still needs to be confirmed in combination with ore properties and other data.
If the overflow particle size is basically stable but the actual reagent flow rate is abnormal, priority should be given to checking the reagent dosing equipment, reagent preparation, and related control loops.
These are only troubleshooting directions; conclusions cannot be drawn directly based on a single phenomenon alone.

Step 5: Verify the cause, rather than only looking at the results after adjustment.
After determining the possible cause, it should be verified in conjunction with process personnel analysis through controlled adjustments, comparative data, or necessary tests, and it should be observed whether the improvement is sustained.
If multiple parameters are changed at the same time, even if the indices improve, it is difficult to determine which measure actually played a role.

6. How Can Automation Systems Truly Help Stabilize Production Indices?
The value of an automation system is not just transmitting on-site parameters to the central control room, but connecting detection, control, trend analysis, and process evaluation.

This can be advanced from three levels.

First level: Stable control of key parameters.
Establish reliable measurement and control loops for variables suitable for closed-loop control, such as feed rate, pump sump level, water addition, flotation cell level, pH, and reagent flow rate.
First ensure that instruments are accurate, actuators are reliable, and control loops operate stably, and then gradually carry out multi-loop coordination.

Second level: Correlation analysis of process parameters.
Combine trend data from grinding, classification, and flotation to analyze the relationships among feed rate, mill load, overflow particle size, and flotation indices.
When abnormal combinations of parameters occur, the system can prompt operators to focus on checking the relevant stages, rather than only issuing an isolated high-pressure, low-level, or flow alarm.

Third level: Continuous evaluation of production indices.
Combine indices such as concentrate grade, recovery rate, tailings grade, throughput, and unit energy consumption with production data to evaluate whether the automatic control measures have truly improved the production results.

It should be emphasized that PLC and DCS themselves will not automatically identify all changes in ore properties, nor can they accurately calculate all mineral processing indices solely based on conventional equipment signals. When necessary, online particle size or grade analysis, sampling and assaying, and process models should be combined to gradually establish a more complete analysis and optimization system.

7. Rather Than Pursuing Complex Algorithms, Three Things Should Be Done First
For mineral processing plants preparing to implement automation upgrades, it is recommended to start with basic work.

First, establish a list of key parameters. Clarify which parameters directly affect process stability, which measurement points must be reliable, and which indices need to be confirmed through assaying.

Second, establish an abnormal event record. Record not only when the indices dropped, but also the preceding changes in feed, operational adjustments, instrument abnormalities, and equipment status, gradually accumulating data that can be used for analysis.

Third, clarify control objectives and evaluation methods. Automatic control should not only assess automatic operation time, but should also focus on whether parameter fluctuations, frequency of manual intervention, production indices, and unit energy consumption have been improved.

Only when the basic data are reliable, the control loops are stable, and the process relationships are clear, does it make sense to further carry out advanced control and intelligent optimization.