Python: A key tool for data analysis in mineral processing

In mineral processing, visualizing plant data is a daily requirement: histograms, time trends, shift comparisons, correlations between variables, and “forensic metallurgy” when diagnosing unexpected behaviours. Using Excel for these tasks often involves repetitive formatting, manual filtering and error-prone formulas, all time-consuming activities that scale poorly as the dataset grows.

Python changes this dynamic. Using libraries such as pandas, matplotlib and seaborn, teams can integrate multiple data sources into a unified structure, automate the creation of standard graphics and refresh entire analyses with a single action (as shown in Figure 1). This approach not only guarantees consistency across all outputs but also eliminates repetitive manual work, enabling faster, more reliable insights for operational decision-making.

This approach has an immediate impact on both productivity and analytical quality. Instead of spending hours preparing charts, teams can focus on what truly matters: interpreting trends, testing hypotheses and identifying operational improvements. Beyond speed, Python unlocks capabilities that traditional tools like Excel cannot match. For example, it can generate dynamic time-series visualizations combined with histograms, apply advanced filtering across multiple variables in seconds and create density plots to reveal where data clusters occur. These are capabilities that are difficult to achieve with traditional tools such as Excel. These advanced techniques provide deeper insights into process behaviour and help uncover patterns that would otherwise remain hidden.

The following examples are based on real projects carried out by BBA for plants in Canada and Chile. All datasets have been modified to preserve confidentiality.

Case 1 – Automated generation of operational graphics

Operational graphics, such as pressure trends, particle size profiles, level signals and shift-to-shift comparisons, are essential for daily monitoring and diagnosis. In many plants, maintaining these visuals in Excel consumes a significant portion of metallurgists’ time.
To address this, BBA developed a series of Python scripts capable of automatically generating:

• histograms and distribution analyses.
• minute-by-minute and hourly trends.
• standard graphics for each shift.
• comparative charts for weekly or monthly performance.

By simply loading a new dataset, all graphics are updated instantly. This approach has significantly reduced the manual workload (saving several hours per week at some sites) while ensuring consistent, reproducible visuals for daily operational discussions. For example, Figure 2 illustrates a time series chart alongside a histogram, both generated in under a minute. Automating these tasks allows teams to reclaim valuable time, improve efficiency and focus on higher-priority activities.

This automation also accelerated a practice often referred to as “forensic metallurgy,” pinpointing when anomalies began, isolating abnormal periods and linking them to specific operational conditions. For instance, the following figure highlights key variables in a crushing operation at a mining site. These types of visualizations are essential for identifying correlations, and when multiple stages or sites need to be analyzed, Python becomes an invaluable tool. It enables teams to quickly display trends across different shifts or seasonal periods, tasks that would otherwise require hours of manual work using traditional tools such as Excel.

Case 2 – Cycle detection and stage-based analysis

Many metallurgical processes operate in cycles, such as filtration filters, carbon desorption columns, batch furnaces, etc. Analyzing each cycle manually, identifying start and end points, separating stages and comparing performance across shifts can be slow and subjective.

Using Python, BBA developed an automated cycle-detection algorithm (Figure 4) capable of:

• identifying individual cycles directly from signal behaviour.
• segmenting each cycle into its operational stages.
• generating reports for daily, weekly and monthly performance.

This eliminated interpretation inconsistencies and revealed deviations from design operating conditions that were not visible with traditional methods. The result was a clearer understanding of process stability and opportunities for optimization (Figure 5).

Case 3 – Density plots and advanced multivariable correlations

Understanding relationships between variables such as P80 and recovery can be challenging in Excel, especially when the dataset is large or when a third variable must be included as a filter.

Python enabled the development of density plots and multivariable correlation analyses to visualize:

• where the data clusters most intensely.
• how outliers influence process performance.
• how variables interact under different operating conditions.

For example, separating the P80-recovery relationship by feed grade revealed that optimal performance consistently occurred below a specific P80 threshold, regardless of the ore grade. This insight helped refine grinding targets and stabilize metallurgical outcomes.

While Python enables these powerful analyses, proper interpretation requires process expertise. At BBA, data analysts and metallurgists work together to ensure that conclusions are technically sound and aligned with plant reality.

Modern mining demands the ability to interpret large volumes of plant data quickly to make accurate, timely decisions. Python has proven to be a powerful tool for automating analysis, reducing manual effort and improving the consistency and depth of operational insights. When properly structured, data-driven workflows allow engineers to focus less on repetitive tasks and more on diagnosing process behaviour, identifying optimization opportunities and supporting reliable metallurgical decision-making.

The impact of these methodologies grows significantly when combined with process expertise, operational context and a rigorous analytical framework. At BBA, multidisciplinary collaboration is essential to this approach. Our process engineers, metallurgists and automation specialists work together to ensure that analytical tools integrate seamlessly with plant realities, control strategies and operational objectives. This synergy enables us to design solutions that not only analyze data but also strengthen decision-making and enhance day‑to‑day plant performance.

These integrated solutions increase both the efficiency and effectiveness of our clients’ operations, ensuring that high‑quality information is always available to guide operational improvements. Although this article focuses on data analytics rather than sustainability, it’s worth noting that improved operational stability and optimized process control can directly contribute to more energy‑efficient mineral processing.

Looking forward, the industry is moving toward real‑time data intelligence, advanced automation and predictive modelling, topics that will be explored in future publications. Python provides a robust foundation for that evolution, and BBA is ready to support operations at every step, from data strategy and automation to advanced analytics and metallurgical optimization.