4 half-day sessions – June 22-25, 2026 – Level: Intermediate

Objectives

This course introduces you to Multiple-Point Statistics (MPS), a powerful simulation technique for modeling complex spatial variability using training images. Developed in collaboration with the University of Neuchâtel, the course combines theoretical foundations with hands-on practice using Isatis.neo and its integrated DeeSse engine. You’ll learn to select suitable training images, prepare your data, and generate realistic subsurface models, whether categorical or continuous. Ideal for applications in mining, hydrogeology, remote sensing, and reservoir modeling, MPS equips you to assess uncertainty and model features driven by geological morphology, such as channelized permeability or ore grades in vein deposits.

Course content

DAY 1

• General introduction
  – Introduction to the geostatistical approach
  – The concept behind conditioning data and training images
  – General principles and introduction to the Direct Sampling algorithm

• Hands-on exercises
  – Isatis.neo fundamentals
  – A first simple application of DeeSse for a stationary categorical and continuous case

• From stationary to non-stationary simulations
  – Understanding DeeSse parameters
  – Why training images are needed: how to obtain them and what properties they should have
  – Handling non-stationarity in the simulation grid
  – Multivariate simulations

• Hands-on exercises
  – A simple practical case study: the Areuse delta
  – How to generate a training image and an orientation trend to control simulations
  – Joint simulation of two variables

DAY 2

• Applying MPS to real data
  – How to handle non-stationarity using analog data
  – Discussion of examples involving secondary attributes: climate data, a bauxite mine in Australia, bedrock topography, and geophysics
  – Time-series simulation using the Direct Sampling technique

• Hands-on exercises
  – A practical 2D case study using secondary variables: the Herten aquifer (fluvioglacial deposit)
  – Filling gaps in satellite images using multivariate and multi-temporal techniques

• Modeling with elementary training images
  – Elementary training images and invariances
  – Application example for a mining site in South Africa
  – Multi-scale simulations based on Gaussian pyramids

• Hands-on exercises
  – Simple examples using elementary training images and invariances
  – Exploring pyramids
  – A first example with a 2D fluvioglacial facies model (Herten aquifer)

DAY 3

• Hands-on exercises: modeling a fluvioglacial deposit
  – Building elementary training images
  – Introduction to Python programming for task automation
  – Building the stratigraphic model
  – Modeling the fluvioglacial aquifer from borehole data

• An overview of advanced methods
  – Cross-validation
  – Multi-scale simulations on unstructured grids
  – Inequalities and block conditioning
  – Connectivity conditioning

Outlines

• Balanced learning approach: The course combines theory with practical applications, ensuring concepts are understood and applied effectively.

• Hands-on software training: Engage in computer-based exercises using Isatis.neo software, reinforcing learning through real-world data scenarios.

• Personalized feedback: Receive individualized guidance and feedback from experienced trainers during online sessions to support your learning journey.

• Comprehensive resources: Access detailed course materials, including documentation, journal files, and datasets, to reinforce learning and facilitate application post-training.

Who should attend

This course is tailored for professionals and researchers involved in spatial modeling who want to enhance their ability to simulate complex geological structures and facies distributions using Multiple-Point Statistics (MPS). Ideal participants include:

• Researchers and academics
  Engaged in spatial data analysis, stochastic simulation, or geoscientific modeling who want to explore MPS in practical workflows.

• Geologists & geomodellers
  Working in mining, oil & gas, or hydrogeology who need to model intricate geological patterns – such as channels, fractures, or stratigraphy – that are difficult to capture with traditional variogram-based approaches.

• Reservoir engineers
  Focused on building realistic facies or property models that improve reservoir characterization and flow simulations.

• Environmental & hydrogeological scientists
  Needing to simulate spatial heterogeneities in aquifer systems with geological realism.

• Geostatisticians and data scientists
  Looking to deepen their expertise in MPS and apply advanced simulation techniques using training images and high-resolution geological analogs.

• Consultants and technical advisors
  Supporting clients with subsurface modeling projects who want to stay at the forefront of geostatistical innovation.

Prerequisites

None.
A theoretical understanding of geostatistical approaches is an advantage.

Explore Geovariances’ training offerings

Download our catalog to view all the courses we offer

Geovariances – Datamine France provides training for mining professionals seeking to strengthen their geostatistics expertise. Our courses blend theory with hands-on practice.

Flexible delivery formats, including online, hybrid, and face-to-face, are complemented by on-demand training, available for both in-company and public sessions, and tailored to your specific needs.

Develop skills in key areas such as:
– Local and recoverable resource estimation
– Uncertainty and risk analysis
– Resource classification
– Geological domain modeling
– Drill Hole Spacing Analysis
– Machine Learning

On-demand hands-on sessions on Isatis.neo and Isatis.py allow you to directly apply geostatistical methods in industry-standard software.

I download the full training catalog

The post Multiple-Point Statistics simulations (MPS) with Isatis.neo appeared first on Datamine.

4 half-day sessions – December 7-10, 2026 – Level: Intermediate

Objectives

This course introduces you to Multiple-Point Statistics (MPS), a powerful simulation technique for modeling complex spatial variability using training images. Developed in collaboration with the University of Neuchâtel, the course combines theoretical foundations with hands-on practice using Isatis.neo and its integrated DeeSse engine. You’ll learn to select suitable training images, prepare your data, and generate realistic subsurface models, whether categorical or continuous. Ideal for applications in mining, hydrogeology, remote sensing, and reservoir modeling, MPS equips you to assess uncertainty and model features driven by geological morphology, such as channelized permeability or ore grades in vein deposits.

Course content

DAY 1

• General introduction
  – Introduction to the geostatistical approach
  – The concept behind conditioning data and training images
  – General principles and introduction to the Direct Sampling algorithm

• Hands-on exercises
  – Isatis.neo fundamentals
  – A first simple application of DeeSse for a stationary categorical and continuous case

• From stationary to non-stationary simulations
  – Understanding DeeSse parameters
  – Why training images are needed: how to obtain them and what properties they should have
  – Handling non-stationarity in the simulation grid
  – Multivariate simulations

• Hands-on exercises
  – A simple practical case study: the Areuse delta
  – How to generate a training image and an orientation trend to control simulations
  – Joint simulation of two variables

DAY 2

• Applying MPS to real data
  – How to handle non-stationarity using analog data
  – Discussion of examples involving secondary attributes: climate data, a bauxite mine in Australia, bedrock topography, and geophysics
  – Time-series simulation using the Direct Sampling technique

• Hands-on exercises
  – A practical 2D case study using secondary variables: the Herten aquifer (fluvioglacial deposit)
  – Filling gaps in satellite images using multivariate and multi-temporal techniques

• Modeling with elementary training images
  – Elementary training images and invariances
  – Application example for a mining site in South Africa
  – Multi-scale simulations based on Gaussian pyramids

• Hands-on exercises
  – Simple examples using elementary training images and invariances
  – Exploring pyramids
  – A first example with a 2D fluvioglacial facies model (Herten aquifer)

DAY 3

• Hands-on exercises: modeling a fluvioglacial deposit
  – Building elementary training images
  – Introduction to Python programming for task automation
  – Building the stratigraphic model
  – Modeling the fluvioglacial aquifer from borehole data

• An overview of advanced methods
  – Cross-validation
  – Multi-scale simulations on unstructured grids
  – Inequalities and block conditioning
  – Connectivity conditioning

Outlines

• Balanced learning approach: The course combines theory with practical applications, ensuring concepts are understood and applied effectively.

• Hands-on software training: Engage in computer-based exercises using Isatis.neo software, reinforcing learning through real-world data scenarios.

• Personalized feedback: Receive individualized guidance and feedback from experienced trainers during online sessions to support your learning journey.

• Comprehensive resources: Access detailed course materials, including documentation, journal files, and datasets, to reinforce learning and facilitate application post-training.

Who should attend

This course is tailored for professionals and researchers involved in spatial modeling who want to enhance their ability to simulate complex geological structures and facies distributions using Multiple-Point Statistics (MPS). Ideal participants include:

• Researchers and academics
  Engaged in spatial data analysis, stochastic simulation, or geoscientific modeling who want to explore MPS in practical workflows.

• Geologists & geomodellers
  Working in mining, oil & gas, or hydrogeology who need to model intricate geological patterns – such as channels, fractures, or stratigraphy – that are difficult to capture with traditional variogram-based approaches.

• Reservoir engineers
  Focused on building realistic facies or property models that improve reservoir characterization and flow simulations.

• Environmental & hydrogeological scientists
  Needing to simulate spatial heterogeneities in aquifer systems with geological realism.

• Geostatisticians and data scientists
  Looking to deepen their expertise in MPS and apply advanced simulation techniques using training images and high-resolution geological analogs.

• Consultants and technical advisors
  Supporting clients with subsurface modeling projects who want to stay at the forefront of geostatistical innovation.

Prerequisites

None.
A theoretical understanding of geostatistical approaches is an advantage.

Explore Geovariances’ training offerings

Download our catalog to view all the courses we offer

Geovariances – Datamine France provides training for mining professionals seeking to strengthen their geostatistics expertise. Our courses blend theory with hands-on practice.

Flexible delivery formats, including online, hybrid, and face-to-face, are complemented by on-demand training, available for both in-company and public sessions, and tailored to your specific needs.

Develop skills in key areas such as:
– Local and recoverable resource estimation
– Uncertainty and risk analysis
– Resource classification
– Geological domain modeling
– Drill Hole Spacing Analysis
– Machine Learning

On-demand hands-on sessions on Isatis.neo and Isatis.py allow you to directly apply geostatistical methods in industry-standard software.

I download the full training catalog

The post Multiple-Point Statistics simulations (MPS) with Isatis.neo appeared first on Datamine.

Headquartered in Brisbane, Australia, Commit Works’ suite of products enables mine sites to digitally manage shift planning, resource allocation, task coordination and production tracking within a single unified platform.

Commit Works’ flagship product CiteOps is an enterprise-grade platform built to connect short- to medium-term planning with frontline execution. The software delivers real-time visibility, enhances cross-team coordination, and enables continuous feedback loops that improve productivity and operational performance. By empowering teams with real-time, data-driven insights, the platform helps reduce downtime, optimise resource allocation, and improve overall efficiency.

The strategic investment strengthens Datamine’s ability to support digital transformation across the mining value chain. By integrating Commit Works’ capabilities, Datamine expands its portfolio to better connect planning and production, delivering a more seamless and comprehensive solution for mine operations.

“This strategic investment opens the door to significant growth by leveraging the global reach and deep expertise of the broader Datamine organisation and is a powerful validation of the exceptional software we have built, marking the beginning of an exciting new phase of global growth,” said James Aleman, Commit Works CEO. “Our product is highly complementary to the existing Datamine solutions our customers already use, making it a valuable addition to their ecosystem”, he explains.

For Datamine customers, the partnership bridges a gap between planning and production, enabling greater clarity, discipline, and real-time visibility in operational planning and execution.

“By supporting frontline workers and supervisors with a robust, reliable shift planning capability, we are helping our customers drive continued improvements in operational performance, reduce risk, and consistently deliver better outcomes. Improved compliance to plan, real-time visibility, and workforce management are critical to modern mining operations, and this is exactly what this solution delivers” said John Bailey, CEO of Datamine.

About Commit Works

Commit Works is a leader in operational planning and frontline execution software for the mining industry. Its platform digitises shift-level planning, coordination, and performance management to help sites improve productivity, strengthen plan compliance, align teams, and reduce day-to-day variability. Used by mining operations globally, Commit Works enables disciplined routines, real-time visibility, and sustained performance improvement.

The post Datamine Announces Strategic Investment in Commit Works to Strengthen Operational Planning and Execution Capabilities appeared first on Datamine.

The same can be true of a single planning horizon. For example, let’s look at strategic planning. Even though strategic planning may be a silo in the overall mine planning process, it can also have its own silos. To identify these silos, let’s think about the many value levers in a strategic plan.

1. Mine schedule: when a block is mined.
2. Destination schedule: where a block goes and how it is processed, if ore.
3. Haulage plan: how trucks are used to achieve the mine plan.
4. Waste management plan: where specifically to dump waste material given options of multiple ex-pit and in-pit dumps.
5. Capital expenditure plan: how to most effectively invest capital in equipment and infrastructure.

Many mining companies treat some of these value levers as silos from a mine planning point of view. When one silo’s planning is complete, that part of the plan is used as content for the next silo. This approach is time-consuming, inefficient, and can lose a tonne of business value. It doesn’t have to be that way!

In an earlier article, we showed how you can use all strategic planning value levers simultaneously instead of sequentially across silos. We used Minemax Scheduler for that exercise.

How does your company consider strategic mine planning value levers? Are there still silos or are you on the journey to an integrated approach to strategic mine planning? If you’d like find out how to start that journey, reach out to us.

The post Breaking Down the Silos in Mine Planning appeared first on Datamine.

For most mine planners developing an extraction schedule, planning for tailings just isn’t a real priority. Let the mill deal with that, they say, I’ve got bigger fish to fry. However in strategic mine planning, when we’re looking at large horizons of time and the varying inter-dependencies of the components that make up the operation, the topic must be addressed.

Have you ever been asked to work out a sequenced delivery of pit material to help construct new tailings dams? What if that material has to be of a specific composition, and it’s not readily available in the mining faces? What if multiple cells of this facility will be constructed over time, requiring you to plan for a long term requirement? And finally, what if the mill can’t come on line until this tailings facility is ready –  meaning that any delay in the delivery of material will directly affect revenue generation and the bottom line for the business?

Here we have a complex situation that demands smart, integrated planning of multiple downstream components over the life of the project. It’s pretty tough to accomplish it in a spreadsheet or a basic heuristic scheduling package. However, this is the kind of thing Minemax Scheduler was made to help you do. Read on to see how easy it is.

Defining the Problem

Consider a typical situation of an open pit with multiple phases feeding a single mill that produces gold and tailings.

In this situation, we need to solve for cut-off grade, therefore a block from the mine can go anywhere that gives the greatest NPV, subject to the conditions and constraints in each period. For every block that’s processed, there’s a calculated mass of tailings produced with a given volume, which must be stored in a facility with a fixed capacity.  That tailings facility requires a designed embankment structure (dam) that must be constructed of a certain kind of waste material that is only available in certain parts of the pit.

Since we don’t know when anything will occur until we solve the problem, our model must incorporate the requirement to send suitable material to build this structure before a single block can be processed in the mill.

Model Definition

To incorporate destination requirements as well as timing decisions, we model mill (Mill_Tonnes) and waste destinations (Waste_Tonnes) as alternative options using the decision tree diagram in Minemax Scheduler. This means that during the optimization, the economic value and impact of multiple parallel constraints determine whether material is sent to the mill for processing, or is considered raw waste.

If material is sent to the mill for processing, processed waste (Tailings) is produced and placed in the tailings basin (vol-Tailings) and is costed accordingly.

Raw waste (Waste_Tonnes) that is not going in the mill is modelled using an alternative decision. It can be either sent to the waste dump (vol-Waste Dump), or the tailings dam (vol-Tailings Dam). There is an obvious haulage difference between the two options, and so we model this using the truck hours (TH) processes Waste_TH_Tailings and Waste_TH_Dump.

The basin location (vol_Tailings) is unavailable for storing tailings until the tailings dam is finished. Minemax Scheduler handles this through location-to-location precedences between the dam and the basin, such that the dam is completed prior to the basin becoming available.

This arrangement allows for all possible destinations of waste material to be considered when optimizing the total mine sequence problem, and all the associated costs of doing so, along with known design/storage capacity limitations that restrict the solution.

Optimization of the Model

In seeking maximum NPV, the model will assess all possible destinations of material with consideration of this important tailings design requirement. Since the mill is the sole revenue-producing agent in this model, and given the power of discounting, the model will seek to construct that embankment structure as soon as possible, and this will influence the overall sequence of the pit such that the critical material types are made available that enable this amongst the other competing factors.

You can check to ensure your model is working through evaluation of multiple dashboards. The 3D visualization can show you the progress of your waste placement in each of the three locations. You can see the effect of the delay at the mill through a movement chart indicating mill tonnages or tailings tonnages.

You can interrogate the build-up of volume to reach design capacity in each of the waste locations, including the tailings dam, using the remaining reserves table and dump tables. If you run out of capacity in the tailings basin, the mill will stop producing; likewise, if you do not have enough material to construct a tailings dam structure, the mill will never operate.

Summary

This simple example illustrates just how important it is to consider all the inter-dependent components of your operation – including tailings – throughout the scheduling process in order to achieve the best possible value.

With Minemax Scheduler, you can integrate and optimize every aspect of your strategic plan. Contact us for more information.

The post Modeling Tailings within Your Mine Schedule Optimization appeared first on Datamine.

We all know that when a leach pad is used, the metal production is delayed – sometimes for several years. If we want to account for that in our schedule and our financials, we need to model it in the software before we optimize. Tweaking the schedule after the fact is inefficient and prone to human error. Furthermore, by manually adjusting a schedule after the optimization is complete, you may find that your result is no longer compatible with the dynamics of your model or the operational realities you are trying to achieve.

In this article, we ‘ll show you a new, easy way to model leach lag in Minemax Scheduler strategic mine planning software. This new functionality lets you include time-delayed leach metal production within your schedule optimization so you can find a practical and achievable schedule with no need for further manual schedule adjustments.

Interested? It won’t take long – modeling time-delayed leach recoveries in Minemax Scheduler is straightforward and quick.

Introducing the example

We’ll use a simple example of a gold mine operation with one mill and one leach pad. The ore delivered to the mill produces metal in the same year of mining whereas the metal production is delayed following a 5-year leach cycle: Year 1: 30% recovery, Year 2: 30% recovery, Year 3: 20% recovery, Year 4: 15% recovery and Year 5: 5% recovery.

Defining the model

In the screenshot below, you can see this example modeled in Minemax Scheduler.

There are processes created for Mill, Leach and Waste tonnes which are defined as alternative decisions. This means Minemax Scheduler decides based on the economic value of each block whether material is sent to mill, leach or to a waste dump. When material is delivered to mill and leach then mill metal and leach metal are produced and the total metal is the sum of these two products.

Now let’s have a look at how to include time-delayed recoveries in Minemax Scheduler. The process is actually very simple.

Including time-delayed recoveries

On the define screen, if you click on the ‘Leach_Metal’ process in the list of processes on the left, you’ll notice a check box called ‘delayed recovery’. This little box is all you need to enable time-delayed recoveries for the ‘Leach_Metal’ process.

Once the ‘delayed recovery’ check box is checked, you can select the ‘Leach_Metal’ process on the ‘delayed recoveries’ screen under financials in the scenario menu. The screenshot below shows a complete entry of leach recoveries for our example. It only took a few seconds to set this up.

Minemax Scheduler uses these leach recoveries combined with the financials defined on the ‘cost & revenue’ screen to optimize the schedule.

Optimizing the schedule with leach lag

While the schedule optimization is running, all aspects of your strategic mine schedule are simultaneously optimized while meeting all your precedences and constraints. Minemax Scheduler integrates trucking, the waste dump sequence, stockpiles, cut-off and cut-over grades, material destinations, capex and now time-delayed leach recoveries.

After the optimization is finished, we can compare leach ore delivered to the leach pad (the chart on the left) with metal recovered from the leach pad (the chart on the right) stacked by pit.

Let’s take pit 5 (coloured in green) as an example. Here, ore is mined and delivered to the leach pad in 2022, but metal is recovered from the leach pad over 5 years following the defined leach cycle. You can examine the remaining pits and see the same pattern.

The result is a practical and achievable schedule including time-delayed leach recoveries. There’s no need for any time consuming post-processing.  In the end, you have more time to evaluate additional scenarios, and greater confidence that you are giving accurate data to your management.

Summary

We’ve shown you how to model delayed leach recoveries in Minemax Scheduler so they are processed as a part of the schedule optimization, giving you a practical and achievable schedule. It’s easy, there’s no messy and troublesome post-processing, and you have the guarantee that the schedule achieves the best NPV.

Want to learn more? Contact us for an online or in-person demonstration.

The post Including time-delayed leach metal production in your strategic mine plan appeared first on Datamine.

What I’ve also discovered is that there is a lot of noise that hinders the mine planner from knowing exactly how their primary tools perform the mine scheduling task. This noise enters the picture in all kinds of ways, and over time I believe it has clouded our perspective on how we are dealing with the problem at hand. Let’s look at this from the ground up – there are really two basic methods of creating a deterministic mine schedule: manual or automated.

Manual Scheduling

Manual scheduling is just that; while usually augmented with a mine design software interface, all decisions are made by the hand of the mine planner. As a result, the success of the resultant schedule is almost always reliant on the capability/expertise/experience of the individual.

Automated Scheduling

Automated scheduling is a way of expressing the mine scheduling problem in software to generate a solution for the engineer. Because software has evolved greatly over time, there are assorted methods for integrating important mining parameters into software and producing the desired solution.

This is where the noise comes in.

To try and distill this down further, let’s expand on the two broad types of automated scheduling methods: Heuristics and Optimization.

Heuristic Scheduling

Many of the automated scheduling software packages in the market today use some form of heuristic approach or even a combination of different heuristic methods. You can be assured you are working with a heuristic if you notice words such as genetic or fitness or target-based or meta-whatever in the description. Unfortunately, you will also notice in these same contexts the misuse of the word optimization.

Heuristics are sometimes favored due to their speed in processing big problems. They also give users the ability to target something, such as material tonnage or grade, or possibly some measure of value. All of these things are important to a mine planner.

Heuristic methods, however, cannot mathematically guarantee that the solution will meet a single or a combination of constraints (whether that’s maximum milling tonnage, minimum milling grade, maximum truck hours, or others). That’s a bit of a tough spot for mine planners, as they are often dealing with hard constraints.

Further, because of the way the underlying algorithms are seeking paths to solutions, they also sometimes have issues with repeatability. Repeatability is critically important if your work is to support a business plan which may later be audited by a third party. By using a heuristic method, you are not only challenged to arrive at a solution that meets your conditions, but you are also faced with the real possibility of getting a different answer when you click the “solve” button again.

So what else is there?

Optimized Scheduling

Enter ‘real’ optimization. This is an automated means of generating the mathematically optimum solution after considering all other feasible alternatives. Feasibility in mine scheduling equates to enforcing spatial precedences and honoring all production or grade constraints. This method is far more computationally intensive as the software’s underlying algorithms have to search through the entire feasible solution space to isolate the ‘best’ solution, whatever the definition of ‘best’ may be. For many strategic mine scheduling problems, ‘best’ usually means maximizing NPV.

Some mine planners may have discounted schedule optimization because the processing of large problems takes much more time than a heuristic approach. I fully appreciate that time is a limited commodity. The application of optimization technology has evolved greatly over the past few years, and the way a software application translates the complexities of the mine scheduling problem into those solving engines has improved significantly in overall speed/performance.

Minemax Scheduler is one of the mine scheduling solutions on the market that uses NPV schedule optimization to generate practical and optimal solutions for mine planners of nearly all mineral commodities. Through modern optimization technology, Scheduler delivers a solution that is mathematically guaranteed to meet the constraints imposed.

Also, because of the nature of the process, it is repeatable. To find out more on how we harness the power of optimization to solve the myriad of complications in the mine scheduling space, view our integrated approach example.

At the end of the day, the best tool for a mine planner depends ultimately on what they are dealing with. I hope this short dialog has helped clarify some of the realities in this space.

Still confused by the noise? Check out this really good write-up on the different optimization approaches for mine planners.

The post Clearing the noise around automated mine scheduling tools appeared first on Datamine.

In this article we take a look at another aspect of the problem which deals with moving waste to its final destination in the waste dump. In particular, we will explain how developing your mining and waste dump schedules in isolation may cause problems, and will discuss how integrated waste dump and pit schedule optimization, recently implemented in Minemax Scheduler, improves the accuracy and practicality of your strategic mine plans.

Waste Trucking Is Important

Trucking requirements vary according to the location of the mining block in the pit, so modelling mining capacity using truck hours based on cycle times and truck capacity is important. It enables us to set limits on available truck hours rather than on total movements in the schedule optimization, which helps to develop a more practical schedule.

Waste dump scheduling also has a significant influence on trucking requirements. So it is equally important to model waste movements using truck hours based on cycle times to the specific waste dump location in order to develop accurate schedules. This is even more critical for larger operations with bigger waste dumps and/or higher strip ratios, since higher waste volumes and/or longer waste hauls mean a higher proportion of the total trucking capacity needs to be allocated to waste movement.

Many open pit mines are effectively large-scale, bulk earthmoving operations where the valuable mineral that goes on to further processing can be only a relatively small proportion of the total material movement.

The Traditional Approach

Traditionally, mine planners focus their efforts on developing mine schedules with only rough consideration and approximations used for waste dumping. This can mean that whilst attempts are made to ensure that schedules are optimized for ore and waste extraction in order to find the best value, the optimization is restricted to a degree because it only includes estimated waste movements.

The detailed waste dumping schedule is typically then developed later on, as a separate process to the mine schedule optimization. Typically, two of the main approaches to estimate waste dumping are:

Considering Waste Movements Expressed in Tonnage

This basic approach uses a maximum material movement limit based on digging and hauling capability in tonnes per period and does not consider depth dependent waste trucking requirements from the pit out and to the waste dump at all. This approach will lead to the highest degree of inaccuracy.

Considering Waste Trucking to the Waste Dump Centroid

A more advanced approach is to consider waste trucking requirements on a per block basis from the pit to some average location in the waste dump. This is quite often the centroid of the waste dump. This means that we calculate waste truck hours based on cycle times from the correct location in the pit to the dump centroid. This method is more accurate than the basic approach but still has a considerable degree of inaccuracy.

Problems with the Traditional Approach

Unfortunately, many people are not aware of the risks associated with the traditional approach using estimated waste dumping. The problem is that if we carry out the detailed waste dump scheduling as a separate process after the extraction schedule is optimized, we might find out that the estimated truck fleet from the extraction schedule is very different to the truck numbers required to move the waste to the actual location on the waste dump.

If we overestimate the trucks in our schedule, this means that we will end up with unused trucks which would result in a loss of value. If we underestimate the trucks in our schedule, we won’t have enough trucks to move the waste, and the schedule will become impractical.

In both situations, we have to return to the mine scheduling process and re-adjust the original schedule to satisfy the newly identified trucking requirements. If we do this, there is a high chance that this new schedule will no longer be optimal and we are very likely to lose Net Present Value. So what can we do to avoid losing business value?

Integrated Waste and Pit Schedule Optimization

The Integrated Approach rectifies the imperfections of the traditional approach by integrating detailed waste and mine scheduling within one single optimization process. The Integrated Waste and Pit Schedule Optimization concept was first introduced in Minemax Scheduler 6.1. This concept is based on modelling a detailed waste dump design, divided into lifts and then further subdivided into waste dump blocks (or “cells”).

It is very easy to import a waste dump model into Minemax Scheduler using standard design package formats, and once imported, you can view the waste dump design in 3-D, together with the pit/pushback visualizations.

You can also set up precedences for the waste dump cells to ensure the correct horizontal and vertical construction sequence of the dump. By doing this, the waste dump cells with their associated storage capacities and physical location attributes can be included in the optimization algorithm, in a similar fashion to the pit blocks.

You can then let Minemax Scheduler find an optimal solution that jointly considers the block’s source locations as well as its range of possible final destinations. Using this integrated approach will lead to a truly optimal solution that maximizes the net present value of the project. The integrated approach is particularly beneficial for modelling accurate trucking requirements because we calculate truck hours based on correct cycle times to the final waste dump location rather than “rough estimates” to an arbitrary point.

This allows us to use the correct total truck hours as a maximum capacity constraint for the optimization process. To improve the accuracy even more, we can also use trucking costs as an input to the financial model used by Scheduler to make its optimization decisions. Not convinced yet? Let’s take a closer look at some examples.

Traditional vs Integrated Approach

To compare these two approaches, we modelled an example of a mine operation with one pit, one waste dump and one processing plant to optimize a 12-year schedule in Minemax Scheduler.

The Traditional Approach

First, we modelled our example using estimated waste trucking to demonstrate the traditional approach. For this, we used estimated cycle times from the block pit location out to a centroid of the waste dump, determined the total truck hours and used them to constrain the capacity of the schedule. Then we ran an optimization on the model to find the best schedule that maximizes NPV, and viewed reports with trucking requirements per time period.

The movement chart for the traditional approach shows an optimal schedule utilizing a maximum of 50 trucks to move the ore and waste material.

The Integrated Approach

Next, we modified the basic model using the integrated approach. We imported a detailed waste dump model split into separate dump lifts and smaller “cells” on each dump lift, set appropriate precedences between the cells and calculated waste truck hours based on cycle times to the individual cells in the waste dump. Then we optimized the model and viewed the reports with the movement chart showing the trucking requirements per year.

The movement chart for the integrated approach shows a need for up to two additional trucks between years 2010 and 2014 when compared to the traditional approach.

This means that if we used the traditional approach for this scenario, we would have underestimated the number of trucks in our schedule, leaving us one or two trucks short of being able to move all the material according to the planned schedule. So the mine plan that was produced using the traditional approach is in effect, not practically achievable.

Total NPV Value Comparison

When we compare both schedules, the traditional approach estimates a higher NPV than the more accurate NPV calculated using the integrated approach. However, we have already seen that the schedule estimated using the traditional approach isn’t practical.

In fact, if we choose the traditional approach, we may end up with significantly less value than originally expected.

Summary

We have showed you via our basic waste dumping exercise how easy it is to produce more accurate and practical mine schedules when you use Minemax Scheduler to develop a model with detailed waste placement considering associated trucking requirements. With this integrated process, you will need to do less manual tweaking of schedules and ultimately you will have better utilization of your resources and greater confidence that your schedule is both optimal and practically achievable.

Modelling Complex Waste Dumps

Using Minemax Scheduler with integrated waste and pit schedule optimization, you can also handle more complex waste dumping scenarios in your mine planning projects and operations. For example, you can model multiple waste dump scenarios, and in-pit waste dumping (backfilling). You can also combine detailed waste dumping with alternative decisions where the classification of ore and waste material (or cut-off grade) is not predetermined but is rather an outcome of the schedule optimization.

Would you like to learn more about modelling complex waste dumps in Minemax Scheduler? Minemax provides a hands-on training course in advanced waste dump modelling techniques in Minemax Scheduler. Contact us for more details.

The post How integrated waste scheduling improves the accuracy of your mine plans appeared first on Datamine.

So how do you reduce the number of blocks in your model, and how do you do it quickly, easily, and safely?

Enter the Reblocking

When modelling a mine, the standard representation of the available material is a block model: a regularised 3D grid where each grid cell (or block) has a number of recorded properties. These properties are the quantities and qualities of the material within that block. As you know, for a large ore body and the surrounding waste, this can mean a very large number of blocks.

One way to reduce the number of blocks in an existing block model is reblocking. This is a technique that aggregates some blocks together into larger blocks. The new blocks are the weighted sum of the original blocks that they represent. Two common forms of reblocking are geometric reblocking, and attribute-based reblocking. Each has advantages and disadvantages, so let’s look at these reblocking options in a little more detail.

Geometric Reblocking

Given that material within a block model is already aggregated into abstract blocks, the simplest way to reduce the number of blocks is simply to make those blocks bigger. By merging neighbouring blocks together, we can do this without going back to the original statistical model. Material within these blocks is aggregated together by adding quantities, and blending attributes (weighted by quantity).

We must take care when merging blocks to consider material from each pushback separately, so as to avoid situations where a block could belong to multiple pushbacks.

Advantages

• The original geometric structure of the model is retained
• Block merging can be controlled in each dimension (easting, northing, bench) to get the exact block size and count required
• Estimating the resulting number of blocks is easy

Disadvantages

• Any fine details (such as thin veins or ore) are lost, as high-grade blocks are smoothed out and aggregated into neighbouring lower grade material
• Reducing the number of blocks too far will result in very large, very average-grade (and potentially unprofitable) blocks
• Blocks at pushback boundaries may overlap each other, and so are not strictly geometrically correct

Attribute-Based Reblocking

Where geometric reblocking merges geospatially neighbouring blocks, attribute-based reblocking merges attribute-neighbouring blocks. In the simplest case, we choose a single attribute (typically the most important attribute, such as the primary grade) and subdivide the range of this attribute into a number of sub-ranges or “bins”. Reblocking examines each block within the same pushback and bench for the value of the chosen attribute, and chooses the appropriate bin for each. Once reblocking has examined the complete set of blocks, it aggregates all blocks that are put into a bin.

There is typically no reason to have multiple blocks if all of the material is sent to the same destination, and the qualities of the material have no bearing on constraints or financial results. Therefore, we can use a single bin for attribute values that have little distinction (such as waste material). For material with attribute values that impact decisions more closely, such as those close to cut-off grades or for critical contaminants, we can use smaller bins. If the model has more than one important attribute – such as in a multimetallic deposit – reblocking can be performed for multiple attributes. In this case, we specify ranges for each of the important attributes. We then combine these ranges and create a multidimensional partitioning of bins. As before, reblocking places each block in a single bin, but now that bin is determined by the values of each of the important attributes.

Advantages

• The block count can be reduced significantly, while controlling which materials are aggregated

Disadvantages

• All geometry within the bench is lost (so scheduling should no longer be performed at the block level)
• Bad choices for binning ranges can lead to grade smoothing, or aggregation of material that should remain separable
• It is sometimes difficult to estimate the resulting number of blocks, as the number of utilised bins on each bench is not known in advance

Everything Simplified

Reblocking is an easy way to simplify your model. Geometric reblocking is useful if you need to maintain the geometric structure of the model, while attribute-based reblocking is useful if you need to maintain value specificity. We’ll be talking more about the impact that reblocking can have on schedule optimization in a future article. We’d be happy to talk to you about how you can use reblocking to get more value out of your strategic mine plan – just ask!

The post Buried in billions of blocks? Reblock! appeared first on Datamine.

Too much simplicity on the other hand can lead to false conclusions and incorrect results. Finding the right balance is essential. One particular challenge within the strategic modelling space is modelling nonlinearities with respect to material type and/or grade as they appear in the many aspects of processing. These include processing throughput, processing recoveries, reagent consumption, power consumption or beneficiation of grades.

In this article, we will show you how to deal with nonlinearities caused by processing material blends and we will model this behaviour using the Minemax Scheduler strategic mine planning tool.

Why Nonlinearities Complicate Things

To start with, let’s consider mill recovery for a blend of two different material types. In our example, we will be producing a 2:3 blend of materials A and B. For this situation, we will be processing 20,000 tonnes of material A at a grade of 2g/t and 30,000 tonnes of material B at the same grade of 2g/t. We know that the mill recovery for material A is 80% while the mill recovery for material B is 90%.

If nonlinearities are not accurately modelled then the strategic mine schedule may not be optimal or even valid.

If we assume a linear relationship of recovery for these material types, then this will give the following results:

Contained Metal A = (20,000*2) = 40,000 g
Contained Metal B = (30,000 *2) = 60,000 g
Total Contained Metal A+B = 100,000 g

Recovered Metal A = (20,000*2*0.8) = 32,000 g
Recovered Metal B = (30,000 *2*0.9) = 54,000 g
Total Recovered Metal A+B = 86,000 g
Total Recovery A+B = 86,000/100,000 = 86%

However, if the actual recovery of a 2:3 blend is 83%, we will get a different amount of recovered metal:
Total Recovered Metal A+B = 50,000 *2*0.83 = 83,000 g

This simple example illustrates clearly that if we use a simplified linear representation of recoveries, we will incorrectly estimate recovered metal by 3,000 g. If nonlinearities are not accurately modelled for strategic schedule optimization, then the schedule may not be optimal or, even more importantly, may not be valid. The degree to which the result is sub-optimal or invalid is related to the degree of nonlinearity.

Simple Linear Modelling

Let’s model the above example in Minemax Scheduler using a linear modelling approach. To do this, we model mill and waste dump destinations as alternative processes. This means that during the optimization we calculate a profitability for each block, which informs whether material should be processed in a mill, or sent to a waste dump in order to maximize the total net present value. The decision tree for this situation is shown in the screenshot below:

For material going to the mill, we calculate the recovered metal of each block which is represented by the ‘MillMetal’ process in the process decision tree. In this approach, the mill recovery is a function of the blend of material types and grade, similar to our simple linear example above.

Simple linear modelling of nonlinearities will most definitely produce incorrect results.

For this modelling approach, the recovered metal is only correct if all material going to the mill in a time period is the same material type and grade range for all blocks. Unfortunately this is hardly ever the case and so this approach will most definitely produce incorrect results. So what can we do about this?

Linear Approximation of Nonlinear Modelling

One approach to dealing with nonlinearities is to use a linear approximation by creating a piecewise linear function. This means we define a number of discrete blends of materials A and B for which we know the correct mill recovery. For our example of a tonnage based blend of two material types ‘material A’ and ‘material B’, let’s assume that we know the correct mill recovery for 7 discrete blends with blend ratios of 20:80, 30:70, 40:60, 50:50, 60:40, 70:30, and 80:20.

To model processes for these blends in Minemax Scheduler, we base them on a percentage of material A tonnes contained in the blend. We can model it this way because for 2 material types, material B represents the remaining percentage to balance out the total of 100%. For instance, for blend 1 (blend ratio 20:80) we will define a process called ‘Blend 1: A:20%’ which means it contains 20% of material A tonnes and 80% of material B tonnes.

In this approach, we can’t define a recovery curve, so the definition is piecewise. Each recovery range is therefore discrete, and covers the specified blend +/- 5%. For example, the recovery for ‘Blend 1: A:20%’ also covers blends containing 15% – 25% of material A. By doing this, we ensure that all possible blends are covered by one (and only one) recovery definition.

The recovered metal for each blend is based on the recovery for that particular blend applied to the grade of each block in the blend, and contributes to the total ‘MillMetal’ process.

Piecewise linear approximation of nonlinearities increases the validity of our schedule.

There is still one missing piece in the puzzle as we need to ensure that the mill receives a given blend using the blend ratio. In Minemax Scheduler, we use blending constraints to do this. To constrain the model, we need to define an attribute for each blend that represents the percentage of material A in the blend and then constrain it with the appropriate range. That allows for the specified blend +/- 5% margin. In the example below, we can see a model setup for constraints of blends 1 and 2: Blend 1: 15 ≤ material A% < 25 Blend 2: 25 ≤ material A% < 35

During the optimization, Minemax Scheduler ensures that the resulting schedule satisfies these constraints and gives us the guarantee that we have found a truly optimal schedule.

What Should You Know About Nonlinear Modelling?

There are a few things you should know about models produced using a linear approximation:

1. Remember that linear approximation is only an approximation, not an exact model, and the key is to define blend ratio processes appropriate for the degree of nonlinearity in the model.
2. Be aware that the number of blend ratio products can become very large as more material types and possibly grade ranges are added.
3. In real models, you should apply linear approximation to each property of milling you wish to model, not only to recovery. For example, this would be processing throughput, reagent consumption, power consumption or beneficiation of grades.
4. Reblocking can introduce inappropriate linear combinations of blocks even before optimization occurs. For this reason, at the time of import, consider reblocking strategies that bin blocks consistently with the blend ratios being modelled.

Summary

There is no doubt that dealing with nonlinearities in processing models is a complicated task. If we simplify the model to use a linear representation of the processing model, then we are most likely to lose some accuracy in our model which can lead to sub-optimal schedule results. On the other hand, if we consider an appropriate level of piecewise linear approximation for nonlinearities in our model, there is a good chance that we will increase the validity and net present value of our schedule and find a truly optimal solution. Do you struggle with nonlinearities in your strategic model? Contact us and we will be happy to discuss your specific situation in more detail.

The post Modelling Mill Nonlinearities in Strategic Schedule Optimization appeared first on Datamine.