2. Import Ore blocks only and an initial topography surface. Missing blocks above the topography were classified as air, and missing blocks below the topography identified as waste.

In Version 4 Ore and Waste blocks always have to be imported, i.e. only option 1 above is available. Although this means a larger input model than for V3 may sometimes be required it reduces the errors associated with importing topographies.

Q. When should I use the re-blocking option?

A. When your block model is relatively large (more than 5M cells), the cells are small (less than 10m in any direction) and you do not import a Mining or Property Surface.

The disadvantage is that importing surfaces with re-blocking may result in inaccurate surface representation in Studio NPVS. If you do need to re-block the model and import surfaces, import the surfaces independently of the block model.

Pit Optimization

Pushbacks

Q. What is a pushback?

A. A Pushback is, in the context of Studio NPVS, a collection of spatially connected blocks with more or less regular geometry that, after some minor modifications at the design stage, could become a practically minable pit. In most cases, no single pushback will ever represent an actual mine topography because two or more pushbacks will be mined at the same time. The actual mine topographies are generated by the Scheduler and can be saved as annual surfaces.

Scheduler

Q. How does Studio NPVS determine the 'best' schedule? What is the logic behind this?

A. Simply put, in your Scheduler Settings you define an Optimization Objective, which can be 'large NPV' or 'tracking' (staying as close as possible to an ideal value of a target). The Optimization Objective is a measure of "viability" of a schedule. For example, a schedule with larger NPV is better than one with smaller NPV. The best schedule is the one that yields the largest NPV (or best tracks the target) among all schedules defined in the Search Tree (in this case it is Search Tree II). The tracking objective is modeled by a classical "quadratic penalty function".

Q. The Scheduler sometimes fails to find a solution or finds a solution that does not strictly satisfy the constraints; why is that?

A1. The solution satisfying the constraints does not exist; this is certain in cases when the “no solution found” message is displayed.

A2. A solution may exist but the Scheduler has been unable to check all the alternatives due to the size of the problem; more likely for problems with more than 200 “activities” (scheduling units).

The Scheduler looks for solutions satisfying all constraints. When such a solution cannot be found a constraint that cannot be met is relaxed (the lower limit is decreased or the upper limit is increased depending on what is appropriate) and the search continues. After relaxing constraints about 10 times a failure to find solution is reported.

Q. What to do in cases where no solution is found?

A. Switch off some of the constraint or relax the limits (decrease the lower limit or increase the upper limit); try smaller activities and/or increase “search extent” parameter. If this does not work you may have to try different pushbacks. Once an imperfect solution is found, it can often be improved by other means.

Q. A solution is found but it violates one or more constraints at some periods; what can be done?

A1. If the problem is complex, it is possible that a solution exists but the Scheduler has failed to find it. Try the “re-scheduling” option; the search will be restarted from the current solution. This strategy is especially effective if the solution is satisfactory for some initial periods, for example, for the first 8 years, and deteriorates afterwards. In such cases, you should restart the search for a schedule from year 6 or 7.

A2. Try to vary the limits over time using tighter limits. For example, if the acceptable strip ratio is between 0 and 2.0 for the entire mine life but tends to be around 2 for the first 4 years and around 1.0 thereafter, it is more effective to send the limits to between 1.9 and 2.0 for the first 4 years and to between 0.9 and 1.0 thereafter than to set it to between 0 and 2.0 for the entire mine life.

Q. It is sometimes claimed that other schedulers find solutions in cases when Studio NPVS fails to do so. Is this possible?

A. If the claim is true it probably refers to a different kind of solutions than those offered by Studio NPVS.

Studio NPVS always produces practically mineable mining sequences based on the bench by bench mining of pushbacks. Many other commercial schedulers, as well as a straightforward spreadsheet approach work with the so-called “grade-tonnage curves”, that is material streams that come from multiple benches. The resulting schedules are often impractical because they do not follow the normal bench by bench mining practice. The difference between Studio NPVS and “other schedulers” may be that Studio NPVS tells you that there is no schedule satisfying the requirements while the “other schedulers” tell you that there are schedules but they cannot be implemented in practice.

Q. Why does NPVS Scheduler relax constraints at all, would it not be better just to say that a schedule does not exist?

A. No, because an imperfect schedule can be improved as indicated above while when there is no schedule there is little information on what to try next.

Q. When does Studio NPVS start to relax targets to find a viable solution?

A. When it cannot find solution in the "Search Phase" on Tree I or on Tree II even after relaxing the target limits several times (how many times exactly depends on various factors and is a parameter arrived at with experience).

Material Allocation Optimizer (MAO)

Q. What does MFO do?

A. The Scheduler generates a schedule assuming that all ore types are processed by their optimal processing methods as defined within the Economic Model. In the presence of strict processing capacities and ore quality requirements, the assumption of predefined ore destinations may have to be relaxed in order to meet an acceptable schedule. At the basic level MAO does just that, redirecting the ore, thereby reducing NPV, to sub-optimal destinations and allowing for stockpiling. In addition, MAO can be used to schedule the production of multiple blended products, for example iron ore blended to different specifications for different customers.

Q. Can MAO increase revenue or NPV?

A. No, the revenue and NPV produced by MAO will usually be lower than those reported by the Scheduler because some of the ore may go to non-optimal destinations to satisfy constraints and because of the cost of stockpiling.

Material Flow Optimizer (MFO)

Q. What does MFO do?

A. MFO checks if the NPV obtained by the Scheduler can be increased by increasing annual mining rates (mining faster) without changing processing capacities. If this is possible, the new optimal mining rates are calculated year by year and a new schedule is generated. The reason for the NPV increase is that extra high grade ore becomes available early on, so the revenues in initial years of the mine life, which have the greatest impact on NPV, are increased.

MFO does not change the order in which the blocks are to be mined as established by the Scheduler but does change some other schedule properties. These changes include:

● Higher ore cut-off grades. Since the ore output is increased while the processing capacities remain unchanged, the plant capacities are filled with the highest grade ore available and the remaining ore is sent to sub-optimal destinations, stockpiled or dumped.

● The ore destinations. For example, the ore that would normally be milled is sent to a leach pad and the ore that would normally be leached is stockpiled or dumped.

● Increased mining rates in the early years of the mine life.

● Reduced mine life.

● Reduced revenues over the entire mine life due to costs of using sub-optimal destinations, stockpile rehandling costs and possible dumping of ore that could be profitably processed.

● Timing of capital expenditures depending on mining rates.

● Haulage optimization carried on by the Scheduler may become invalid (the same is true for MAO).

Q. Why do other cut-off grade optimization packages sometimes generate higher NPV increases than MFO?

A. Most or all of the other commercially available software carry out optimization using “grade-tonnage curves” that represent ore tonnages divided into grade classes. This type of data ignores the location of ore and may lead to impractical schedules requiring, for example, mining several benches within the same pushback (phase) at the same time. In contrast, MFO always generates a practical mining schedule thus avoiding the overestimation of NPV by ignoring strategies that are impossible to implement in mining practice. The MFO NPV estimate might in some cases be lower than that produced by other systems, but unlike other estimates, it is always realistic.

Multi-Mine Scheduler (MMS)

Q. Are schedules dependent on the order in which mines have been imported into project?

A. No, they are not. Schedules are independent of the mine order.

Q. How many mines can MMS handle?

A. This depends on the size of the mine models, the total number of pushbacks in all of the mines and how strict the scheduling requirements are.

There are two types of limitations, one is common to all Studio NPVS projects and is concerned with computing resources, mostly RAM, and the other is the complexity of the scheduling problem as defined by the number of scheduling units called “activities”. The advantage of Studio NPVS is the scalability of the activities; they can be as small as a fraction of pushback-bench and as big as an entire pushback. Bigger activities imply smaller number of activities and reduce the complexity of the scheduling problem. Scheduling under strict requirements (many constraints with tight limits) calls for smaller activities increasing the overall complexity.

General Issues

Working with large block models

The size of the block models that can be handled by Studio NPVS is limited by the available computer resources, especially RAM.

The demand for resources grows with the number of blocks in the model, the number of fields in the model and the number of rock types. To keep the demand for resources under control when dealing with large block models one should take the following precautions:

● Import only the fields that are relevant for optimization omitting the fields relevant for reporting only. If necessary, you can later create a new project, import all the fields and create reports using the results (ultimate pit, pushbacks, annual pits etc.) obtained in the first project.

● Reduce the number of rock types to the minimum; for Studio NPVS only the differences between rock types pertaining to processing (different methods, recoveries or costs) matter.

● If the attributes (elements) other than products are used only for processing formulas and need not be reported, uncheck “Attributes by Rock Type” box in Economic Settings Options dialog.

● Use the re-blocking option on model import. For life of mine optimization it rarely makes sense to have blocks smaller than 10m by 10m by 10m, and for many mines much larger blocks are adequate.

MAO/MFO and the ultimate pit

MAO changes ore destinations, MFO changes destinations and cut-off grades so both optimizations change block values that have been used to generate the ultimate pit, which implies that the ultimate pit may no longer be optimal. To update the pit, pushbacks and the schedule you can do the following:

1. Save the MAO/MFO results as a Datamine file. The model in the file includes all the imported model fields plus revenue/cost fields, ore destination field and some others. Depending on NPVS version it may also contain “net value” field; if not, create it as NPVREVEN-NPVPCOST-NPVMCOST. Also depending on NPVS version, the model may contain “blocks” representing stockpiles; if so these blocks have to be deleted.

2. Import the model into a new project with “net value” field as the only product and in Economic Settings Options dialog choose “profit model” option.

3. Run the optimization programs.

Studio NPVS and 'Search Tree I'

When Studio NPVS reports it is 'building a search tree', it is searching for an optimal sequence of activities by checking targets against the accumulated values of attributes. Searching, in the context of Studio NPVS always refers to the search for one feasible solution.

Studio NPVS and 'Search Tree II'

The nodes in Tree I are connected by arcs representing activities; the duration of an activity varies from 0 to about 1/4 of the selected time period (year). Search Tree II is built on Search Tree I in such a way that the arcs between any two nodes represent exactly one year. This makes checking the targets per period possible. The search for schedules on Tree II starts with original (un-relaxed) target limits.

Searching, Expanding and Updating

In the context of Studio NPVS;

● "Searching" means looking for just one feasible solution; that is, for a sequence of activities satisfying target limits and sinking rates. In this phase target limits may be relaxed.

● "Expanding" means adding more solutions to the one already found; the targets limits are not relaxed anymore. This phase expands the scope for NPV maximization (or for optimizing another objective chosen in the Scheduler Settings).

● "Updating" means adding information to the search tree about connections between tree nodes; this farther expands the scope for objective optimization.

JMN 20065_01_EN