**Home | Plantwide Solutions | SAG/Ball | Flotation | Ore Sizing and Crushing| Thickener | Froth Flotation and Feed belt Camera Systems | Success Stories **

# Asarco Ray, Plant-wide Control

Asarco Ray is one of our newer plant-wide efficiency projects. The grinding side of the expert control project will be completed shortly, and we are only awaiting the addition of a Mill Scanner to the SAG mill before it is considered 100% complete. Even in its current state, the grinding circuit has seen a significant improvement from our SAG and ball mill expert control strategies implemented on the KSX expert system. Our 30 years of experience combined with KSX’s power optimization tools can improve any grinding (SAG and ball mill) circuit without needing any additional instrumentation. The neural networks, genetic algorithms, fuzzy logic and industry-best rule templates result in cutting edge adaptive expert control for SAG mills. Through a series of randomized on/off tests, the initial results were extremely promising and KSX expert control demonstrated tighter control as measured by the standard deviation with higher average throughput in the SAG mills. Currently, the SAG mill expert control system is operational roughly 90% of the time with excellent results. Below is a very basic overview of the strategies and typical results we have seen.

**Objectives**

- Protect the SAG mill
- Maintain smooth and stable operation of the SAG mill (SAG mill optimization)
- Maximize SAG mill tonnage

The KSX expert system will make changes to the manipulated variables to try and maintain the control variables within their limits, while simultaneously trying to maintain the highest possible feed rate.

**Manipulated Variables (Directly Changeable)**

- Feed Rate
- SAG Mill Speed
- Feed Water

**Control Variables (Not Directly Changeable)**

- Bearing Pressure
- SAG / Ball Mill Power
- SAG Mill Sound
- SAG Mill Density
- Recycle Tonnage

**Rules**

The expert system rules calculate a new set point for each manipulated variable every minute. Data is read from the DCS every 15 seconds, and calculations and filters are performed every 30 seconds.

**Feed Rules**

Each decision cycle the expert system calculates a change in the feed set point. This change is applied (added to or subtracted from the current set point).

A negative set point change will be applied if **any** of the following conditions are true:

- Bearing pressure is
*high* - SAG mill power is
*high* - Recycle is
*high*

A positive set point change will be applied if **all** of the following conditions are true:

- Bearing pressure is
*ok* - Sag mill power is
*ok* - Recycle is
*ok*

**Speed Rules**

Currently because of hard ore conditions the SAG mill speed is run at 100% in fixed speed mode only. In the future, if soft or fine ore conditions are restored, we will commission rules to control the SAG mill speed. Each decision cycle the expert system will calculate a change in the SAG mill speed set point. This change is applied (added to or subtracted from the current set point).

A negative set point change will be applied if **any** of the following conditions are true:

- SAG mill feed is at its maximum, mill density is
*high*and bearing pressure remains*low* - SAG mill feed is at its maximum and mill sound is high

A positive set point change will be applied if **all** of the following conditions are true:

- SAG mill feed is not at its maximum

**SAG Mill Water Rules**

Each decision cycle the expert system calculates a change in the SAG mill water set point. This change is applied (added to or subtracted from the current set point).

A negative set point change will be applied if the SAG mill density is low. A negative set point change will also be applied if the mill density is within its range and **any** of the following conditions are true:

- SAG mill power is
*high* - SAG mill sound is
*high* - Recycle tonnage is
*high* - SAG mill feed is at its maximum, and bearing pressure remains
*low*

A positive set point change will be applied if the mill density is high. A positive set point change will be also applied if the mill density is within its range and **all** of the following conditions are true:

- SAG mill power is
*ok* - SAG mill sound is
*ok* - Recycle tonnage is
*ok* - Bearing pressure is
*high*

**Fuzzy Logic**In the above rules, references are made to values like *ok*, *high*, *very* *high* etc. These values are called fuzzy variables, and are defined within the expert system by shapes called fuzzy sets. The fuzzy sets allow the expert system to gauge how *high *something is. The calculated response (set point change) varies in magnitude based on the truth value (how *high*) of the control variable. So the *high-*er the control variable is, the larger the corresponding change. The fuzzy value for each control variable is calculated based on the distance from the actual value to the limit, and the rate of change, or slope of the trend for that variable. That is to say, if a control variable is still below its limit, but is moving rapidly towards it, it may be determined to be *high*. Likewise, a control variable may be above its limit, but moving rapidly down, thus may be determined to be *ok*.The expert system calculates the fuzzy values for each of the control variables, then calculates the set point change for each of the manipulated variables that will best meet the objectives of protecting the mill, controlling smoothly and maximizing tonnage.

Because there is a rock sizing camera system being used at Ray, we use feed size to determine bearing pressure ranges since we are relying on bearing pressure as a proxy to estimate volumetric mill loading. This means that different adaptive bearing pressure ranges will be targeted based on measured feed size. The mill will be watched closely, looking at the magnitudes, and the rates of change of the control variables (1st and 2nd derivative) to make determinations about the current feed rate, volumetric loading, and grinding efficiency. For example, as ore size increases, the strategy will seek to run lower bearing pressure to maximize the lifting of the large ore in the mill, so that the falling rock has the greatest kinetic energy as it cascades. This approach will break large ore much more efficiently than attrition grinding. With finer ore, the strategy will seek out a higher optimal volumetric loading target to utilize the benefits of attrition grinding to maximize throughput.

**See Products Used In this Project**

Antofagasta – Los Pelamres”> |