Data Analytics
We use SQL Server and MATLAB to carry out data analysis and visualization of Fleet Management Systems' data. An in-depth data analysis is the backbone of any successful simulation study.
Statistical Data Analysis
We collect data from different sources at your mine-site to carry out a complete data analytics on your current actual production data. The main categories of data collected are: medium to short-term mine plans, road haulage network, truck-shovel dispatching production and activity data, and processing plant production historian data. Statistical analysis of data provides understanding about the underlying stochastic processes and possible correlation between random variables. This stage of study will reveal if there is a need for further correlation modeling for the data. Dispatching data collected automatically by truck-shovel systems are full of erroneous data and outliers that are not representing the processes of interest. We will carry out major cleanup of the collected data and comprehensive statistical analysis using the following techniques.
- Graphical and statistical analysis of collected data
- Fitting distributions on historical dispatching and plant data
- Checking goodness of fit for distributions
- Regression Analysis
- Predictive Failure Modeling
- Histograms
- Quantile summaries and box-plots
- Summarize basic statistical properties of data
- Sample mean and sample variance
- Standard deviation, lexis ratio, skewness
- Time series plots
- Auto correlation plot
- Statistical correlation analysis
- Linear correlation analysis
- Unequal variances, t-test hypothesis testing
Dispatch & Plant Historian Production Data
- Mine plans with a monthly and weekly time frame resolution,
- Polyline and polygon mine surveyed models,
- Haulage road network with the respective road specifications,
- Actual (as-built) mining limits and road profiles,
- Truck-shovel hours scheduled,
- Production block-model data with all the attributes of interest,
- Destination of material and time period (shift) scheduled and actually mined.
Mine plans and grade control models
Mobile equipment dispatch data
-
Truck and shovel activity database will be used to
collect data for start and end-time of
- hauling,
- traveling empty,
- waiting,
- queuing,
- backing,
- tipping,
- loading,
- inactive,
- scheduled maintenance,
- unscheduled maintenance
- Shift change schedule,
- Lunch breaks,
- Fueling,
- Scheduled maintenance to trucks, shovels, in-pit crusher
- Historical weather delays. Production database including timestamps for
- queue time,
- spot time,
- load time,
- haulage time,
- backup time,
- cycle time,
- time full,
- truck ID,
- shovel ID,
- load tonnage,
- material type,
- polygon name,
- shovel location,
- dump location,
- effective flat haul loaded,
- effective flat haul empty,
- distance loaded,
- distance empty,
- truck speed loaded,
- truck speed empty,
- shovel movement speed,
- GPS origin,
- and GPS destination Auxiliary equipment data including
- waiting,
- inactive,
- stopped,
- moving,
- loading,
- dumping,
- spotting time for dozers, graders, loaders, compactors, and water trucks.
- Hourly tonnage and feed-grade records using weightometers and auto samplers,
- Plant performance indicators, such as crusher power consumption,
- In-plant surge bin level,
- Cyclone throughputs,
- Plant balance calculations,
- Plant actuals, such as commodity produced, reject and,
- Tailings volumes and assays,
- Scheduled downstream maintenance that would impact mining, and
- Downtimes of any plant component that would affect mining operations.
Historical processing plant production data
Results and Deliverables
- Identify the probability properties of components of the whole mining-systems,
- Identify the probability properties of trucks, shovels,crushers,and auxiliary equipment in a mixed fleet condition,
- Model probability distributions for loaded and unloaded trucks’ speeds for various haul-road profiles,
- Model probability distributions for trucks’ speeds for various seasonal conditions, day and night shifts,
- Model probability distributions for the effective flat haul loaded and empty speeds,
- Model probability distributions for the movement speed of shovel from one location to another,
- Model probability distributions of loading time,
- Model probability distributions of spotting time, /li>
- Model probability distributions of back-up time,
- Model probability distributions of shift-change,
- Model probability distributions of shovel bucket load,
- Model probability distributions of fuel consumption,
- Model probability distributions of operator efficiency and
- Model probability distributions of availability of equipment,
- Model probability distributions for mean-time-between-failures and mean-time-to-repair for all the equipment,
- Analyze correlation to evaluate the strength of the relations between variables,
- Construct reliability models for the truck-shovel system,
- Construct the reliability of truck-shop system considering the service capacity of the shop,
- Construct the probability distribution of the number of trucks in work-state,
- Construct a reliability model for the repair-stands in the workshop,
- Assess key set of measures allowing an estimation of overall system efficiency,
- Identification of the probability distribution of the number of trucks and shovels in work-state,
- Establish probability distribution of numbers of shovels in state of accessibility,
- Estimate the conditional mean-time of the trucks awaiting repair,
- Estimate adjusted steady-state availability of trucks
- Mean number of failed trucks at any given time,
- Mean number of busy and idle repair-stands,
- Estimate the probability distribution of the number of trucks in the work-state as the function of the reliability,
- Create haul-road profile and travel time analysis charts.
The following results will be the outcome of the
historical dispatching data analysis and modeling. This
stage of the study alone is very valuable, since it
provides a quantified description of the mining-systems
reliability.