Webinar: Final Product Grouping in Multimetal & Rare Earth Pit Optimization

# Final Product Grouping in Multi-Metal & Rare Earth Pit Optimization

Hi, everyone! Thank you for joining us today. My name is Anya, and I’m part of the business development team at K-MINE. Before we begin, I’d like to invite you to follow us on LinkedIn and YouTube, so you can stay updated and see what’s coming next.

Today, we’re talking about pit optimization for multi-metal and rare earth deposits — and specifically, what happens when the final product is grouped into one number versus five — what that changes, and what it doesn’t.

Before we get into today’s topic, I want to take just a moment to tell you a little about K-MINE — especially if this is your first webinar with us. K-MINE has been working in the mining industry since 1994, starting in Ukraine and later expanding into Europe, the United States, and Canada. What makes our team unique is the combination of software expertise and real mining experience. Our specialists include software developers, geologists, and mining engineers. We work with companies at very different stages — from early exploration projects all the way to established mining operations. And regardless of the project size, our focus is simple: to provide practical, reliable, and cost-effective solutions that actually fit the client’s technical and operational needs.

So, how does K-MINE actually work in practice? We build software specifically for the mining industry — with tools that support a project from early stages of exploration all the way through to mine planning and day-to-day production. The platform is modular, which means teams can choose only the tools they actually need — whether they’re working with open-pit or underground operations. And as the project grows, K-MINE grows with it. You can start with just one module and gradually expand to a much broader software environment. Our tools can help teams manage and interpret geological data, design exploration drilling programs, assess and prioritize targets, and build both short-term and long-term mine plans. K-MINE can also connect with IoT and dispatch systems, giving teams better visibility into what is happening in real time — and helping improve coordination, decision-making, and overall operational efficiency.

What makes K-MINE different is that our software is built around real mining projects. A lot of what we develop comes directly from the challenges our consulting team sees in the field — and from the feedback we get from geologists, engineers, and our clients. So for us, software and consulting are closely connected. Our team includes qualified persons who prepare technical studies and reports under international standards, including NI 43-101, S-K 1300, and JORC. That work includes mineral resource estimates, PEAs, PFSs, feasibility studies, and definitive feasibility studies. We work across many hard-rock commodities, with especially strong experience in critical minerals. And that experience shapes how we build our software. We’re not developing features just because they look good on a screen. We’re thinking about real data, real project constraints, and the real decisions mining and exploration teams have to make every day.

So, here’s where we’re going today. First — rare earths: what they are, who uses them, and where the market stands. Second — the core question of today: one number… or five? Third — what that difference actually changes in your project. And finally — K-MINE itself, end to end — from the block model all the way to the pit shell.

Let me quickly explain the structure, so you know exactly what we’re covering today. There are really two parts. The first is the idea. We took the same deposit and ran the pit optimization twice. Once using a copper equivalent — so everything is reduced to one number. And once using five separate final products. Then we compared the results: the pit outline, the schedule, the revenue, and the risk. Some of the results were expected. Some were not. And I’ll show you both — including the part that did not go in our favor. The second part is the tool itself — how K-MINE handles this process end to end. And before we go any further, I want to give you the main conclusion now: using one number instead of five barely changes the pit — but it can completely change how you see the risk.

We’ll start with rare earths — and there’s a reason for that. Everything we’re going to discuss about multi-product deposits also applies to something more familiar, like a copper-zinc mine. But in those deposits, the effect can be subtle. With rare earths, it’s much easier to see. So we’ll start with the more extreme example — and then come back to the more familiar one. Rare earths are a group of seventeen metals with very similar chemical properties. And first, the name can be a little misleading. They’re not necessarily rare because there isn’t much of them. In fact, cerium is more abundant in the Earth’s crust than copper. The real challenge is separation. These elements behave so similarly that separating one from another can require dozens of processing stages. So historically, the problem wasn’t simply finding rare earths. The difficult part was separating them efficiently. And this is where the economics become especially interesting. Out of those seventeen elements, four tend to carry most of the value: neodymium, praseodymium, dysprosium, and terbium. They’re often referred to as the magnet four. The rest still has to be mined, hauled, and processed — but much of it contributes far less to the final revenue. And that is exactly why the way we value these deposits matters so much.

Permanent magnets are what make rare earths strategically important. They’re used in electric vehicle motors, wind turbine generators, drones, aircraft systems, and even the speakers in the headphones you may be wearing right now. By volume, magnets account for roughly half of rare earth demand. But by value, they represent almost all of it. And that’s where the imbalance becomes important. Cerium and lanthanum make up a large share of what actually comes out of the ground. But they’re relatively low-value products. They may sell for only one or two dollars per kilogram. Meanwhile, neodymium and praseodymium can sell for around one hundred dollars per kilogram. So keep this idea in mind — because it’s really the key point here. The market does not value these elements in the same proportions that geology gives them to us. The orebody has one natural composition. The market has a completely different value structure. And those two things do not necessarily line up.

Now, a quick look at the history — because it tells us something important about how this market behaves. And today, that behavior really matters. The story starts back in 1787 with a small Swedish village called Ytterby. Remarkably, one quarry there eventually gave its name to four different elements. Then, from the mid-1960s through the mid-1980s, the global market was dominated by one mine: Mountain Pass, in California. At the time, its importance was driven largely by europium — used to create the red color in television screens. But when television technology changed, the market changed with it. The orebody stayed exactly the same. Its economic value did not. Then, in 1983, the neodymium-iron-boron magnet was invented. And that single innovation became the foundation for much of the rare earth demand we see today. Then came 2010, when supply was disrupted for political reasons. The result was dramatic. Dysprosium went from about ninety dollars per kilogram to more than two thousand three hundred — and then later collapsed again. So the main point from this slide is this: in rare earths, it’s not only the prices that change. The relationship between those prices changes too. And those price ratios can shift dramatically within just a few years. Keep that in mind — because it becomes very important when we get to pit optimization.

Now let’s look at who actually controls this market — because the answer is a little more complicated than the headlines sometimes suggest. When it comes to mining, supply is actually more distributed than many people realize. Last year, China produced around 270 thousand tonnes of rare earth oxide. But the United States produced about 51 thousand, Australia — 29 thousand, and Myanmar — 22 thousand, out of roughly 390 thousand tonnes globally. So China clearly has a very strong position — but when we’re talking about the rock itself, it’s not a complete monopoly. The picture changes dramatically once we move further downstream. For separation, China controls roughly 91 percent. And for magnet manufacturing — around 94 percent. Just look at those three numbers for a moment. As we move from mining, to separation, to the final product, the supply chain becomes narrower and narrower. And the closer we get to the product that somebody actually buys, the fewer players there are. That’s one of the reasons rare earth prices can move so dramatically — even when the amount of ore being mined hasn’t changed very much. Just look at neodymium-praseodymium this year alone. Around 53 dollars per kilogram in January. 137 in April. And about 97 in August. That’s not a change over ten years. That happened in just seven months. And I want you to remember that — because in the second half, we’re going to bring this exact problem back to your deposit and see what it means for pit optimization.

Now, this is probably the most important point from the first part of the presentation — so I want to slow down here. At Mountain Pass, neodymium and praseodymium make up only about 16 percent of the basket by mass. But in terms of value? They represent around 93 percent. And the recovery numbers tell another important part of the story. Take cerium. It makes up roughly half of the material, but recovery is only around 11 and a half percent. And that’s not necessarily because it’s difficult to recover. It’s because economically, there’s simply not much reason to recover more of it. The market already has more cerium than it needs. And Mountain Pass is not an isolated example. At Ngualla in Tanzania, roughly 22 percent of the basket by mass accounts for more than 93 percent of the value. At Nolans in Australia, about 26 percent by mass generates more than 85 percent of the revenue. Different continents. Different projects. Different mineralogy. But essentially the same pattern. A relatively small part of the material creates most of the economic value. So here’s the key takeaway: if we look only at something like average grade, we’re looking primarily at mass. But mass and value are not the same thing. And this is where the conversation moves beyond rare earths. Because look at the example at the bottom of the slide. That’s a much more traditional polymetallic deposit — copper, zinc, lead, gold, and silver. Five different products. But economically, copper and gold alone represent roughly three quarters of the value. So this is not really a rare earth problem. It’s a multi-product problem. Rare earths simply make it much easier to see. And with a conventional polymetallic deposit, that imbalance can be much less obvious — which is exactly why we need to understand what happens when we reduce five different products to just one number.

So now, let’s open the model and see how this works in practice. What you’re looking at here is an actual block from the model we’ll use throughout the rest of the presentation. This block contains gold at 1.07 grams per tonne, silver at 3.56, copper at 1.14 percent, plus smaller amounts of zinc and lead. But then there’s one more number: copper equivalent — at 1.93 percent. And this is where things get interesting. Because the pit optimizer doesn’t really see gold, silver, copper, zinc, or lead. Whether we’re using Lerchs-Grossmann or Pseudoflow, the algorithm ultimately needs just one thing: how much is this block worth? Positive… or negative. One economic value. So before optimization even begins, those five metals have to be translated into one number. And one of the most common ways to do that is through a metal equivalent — in this case, copper equivalent. But here’s the important part. The moment we create that one number, we’ve already made a whole series of economic assumptions. We’ve decided which metals contribute value, how much of each metal we expect to recover, and what each one is worth relative to the others. All of those decisions are already built into that CuEq value before the optimizer ever sees the block. And that is exactly what we’re going to examine next: what happens when we take five different economic variables and compress them into one.

And here’s something that’s surprisingly easy to miss — even if you’ve worked with metal equivalents for years. There isn’t just one way to calculate that CuEq column. In fact, there are several approaches in use today — including in published technical reports. The first is the simplest. You take the price of gold, compare it to the price of copper, adjust for the units, and convert everything into a copper equivalent. That’s the version you often see in press releases. The second approach goes a step further. It includes recovery. Because in reality, we don’t recover every metal at the same rate. In our example, gold recovery is around 65 percent, copper around 80, and silver around 60. So even if two metals look equivalent based on price, they may not be equivalent once they go through the plant. The second formula tries to correct for that. And then there’s a third approach. Here, you only count the main value drivers — in our case, copper and gold. Zinc, lead, and silver are treated more like a bonus. That may sound overly conservative, but if by-product revenue is uncertain, it can actually be a defensible choice. And look at what happens. Same rock. Same prices. Same day. But three different CuEq values: 1.56, 1.37, and 1.05. That’s a very large spread for what is supposedly the same block. And this is why industry guidance has been warning about this for years. If you weight only by price, you are effectively assuming that all metals are equally recoverable, equally payable, and equally marketable. In practice, they almost never are. So the practical takeaway is very simple: when someone tells you this deposit is 1.4 percent copper equivalent — the first question should be: based on what?

Let’s make that a little more concrete — because a 48 percent difference can sound theoretical until it actually changes a decision. Imagine we have two blocks. The first is very simple — about 1.4 percent copper, with almost nothing else. The second is much more polymetallic — lower copper, but it also contains gold, silver, zinc, and some lead. Now let’s value both blocks using the price-only formula. The polymetallic block comes out at 1.61 percent copper equivalent, compared with 1.40 for the copper block. So based on that number, the polymetallic block looks about 15 percent better. You would probably send it to the plant first. But now, let’s include recovery. And suddenly — the result flips. The copper block becomes the better block. And if we look at the actual recovered value, using the prices in this model, the copper block is worth about 149 dollars per tonne, while the polymetallic block is around So we’re talking about almost six dollars per tonne difference. Same two blocks. Same prices. Same day. But depending on the formula, you get the opposite answer to a very basic planning question: which block has actually better ore? And that brings us to an important point. Copper equivalent is not something we physically measure. Nobody goes into the field and measures 1.4 percent CuEq. It is a calculated value. And behind that value is a set of assumptions — about prices, recoveries, and which metals we decide to count. So whenever we use an equivalent grade, those assumptions need to travel with the number.

Now, I want to be fair to equivalent grade — because it would be easy to make it sound like the problem itself. And it isn’t. A well-built equivalent grade can be perfectly accurate. In our case, the recovery-adjusted CuEq is 1.3728 percent. If we convert that back into value, we get 146 dollars and 6 cents per tonne. And that is exactly the revenue of one tonne of this ore. Not approximately. Exactly. Because that is what the formula was designed to do. So the problem is not that the number is wrong. The number is right. The problem is that it compresses everything. And once five metals become one number, you lose the detail behind it. If I tell you we have 74.8 million tonnes at 1.38 percent CuEq, you still don’t know how much zinc is there, how much lead, how much gold, or how much silver. You can’t size the flotation circuit from that. You can’t properly discuss the product mix with a trader. And you can’t prepare a metal-by-metal reserve statement from CuEq alone. And there’s one more problem. You also can’t answer one of the simplest questions a board is going to ask. And that’s what we’ll look at next.

Now we get to the question that really matters for project risk. What happens if the copper price drops by 30 percent? We tested that in two ways — same deposit, same software, same assumptions. The only difference was how we valued the blocks. In the first model, everything was expressed through copper equivalent. In the second, we kept all five products separate. Now look at the chart. At today’s prices, both models land at almost the same NPV. So at the base case, they agree. But as soon as we start changing the copper price — the picture changes. The CuEq model shows NPV moving from about 3 billion dollars to more than 6 billion. The five-product model moves much less — from roughly 3.9 to 5.3 billion. So according to the equivalent model, NPV changes by about 35 percent. According to the product model — only about 16 percent. That’s more than a two-times difference in the answer to exactly the same question. And the reason is actually very simple. In the CuEq model, everything has been converted into copper. So when you change the copper price, the model behaves as if 100 percent of the project’s value moves with copper. But in the real deposit, copper represents only about 45 percent of the value. And this is an important distinction: nobody did anything wrong. The CuEq calculation is correct. The sensitivity calculation is correct. But if you reduce the deposit to one economic variable, the model can only show you one-variable risk. And that is the limitation we need to understand.

Now let’s look at the full picture — metal by metal. For copper, the equivalent model shows about 35 percent sensitivity. The five-product model shows closer to 16 percent. Now look at gold. In the equivalent model, there is no gold sensitivity at all. And I don’t mean it gives us a bad answer. I mean the question simply doesn’t exist inside that model. Gold has already been folded into copper. The same is true for silver, zinc, and lead. So instead of five separate risk numbers, we’re left with one. Now look at the five-product model. Here, every metal gets its own sensitivity line. Copper is the steepest. But gold is not far behind — which is interesting, because gold is often treated as a by-product. Then we have silver and zinc, which behave quite similarly. And finally lead — which is almost flat. You could move the lead price significantly and barely change the project NPV. And this is what the five-product model gives us: a much clearer picture of where the risk actually sits. Now look at the bottom of the table, because this is the part I find really interesting. If you add up the sensitivities from all five metals, you get roughly 35 percent. Which is almost exactly what the equivalent model gave us for copper alone. So the CuEq model is not necessarily wrong about the total amount of price risk. It’s wrong about where that risk comes from. The equivalent model effectively tells management: this is a copper project, so copper is the price you need to worry about. The five-product model tells a very different story. About 45 percent of the risk comes from copper. Around 32 percent comes from gold. Then silver, zinc, and lead make up the rest. And that matters. Because those are two very different risk profiles. They may lead to different hedging strategies, different offtake discussions, and different answers when an investor asks: “What happens if copper weakens next year?” a project where a third of your price risk comes from gold is simply not the same investment as a project where you believe all of the risk is copper. Same rock. Same optimizer. Same day. But a very different understanding of the project’s risk — simply because of how the block was valued.

There are two other things that came out of this analysis that are worth pointing out. They’re not the main story — but they’re interesting. The first is cut-off grade. On this particular deposit, the cut-off sensitivity is basically flat. We can move the cut-off grade up or down, and the NPV barely responds. Why? Because most of the ore is already around thirteen times above break-even. So there’s very little marginal material sitting near the cut-off that can move in or out of the mine plan. And that’s worth remembering, because testing different cut-off grades is something we do almost automatically. But on a deposit like this, it may tell us very little. The second observation is at the bottom of the slide — ore losses. Here, ore losses have a bigger impact on the project than changing the prices of zinc and lead combined. And I think that’s important, because loss and dilution are often entered early in a study as fairly standard assumptions — and then we don’t spend much time looking at them again. But on this deposit, those operational assumptions matter more than two of the five metals. So sometimes, the biggest economic sensitivity isn’t where we instinctively expect it to be.

This one is short — but it comes up all the time. A copper equivalent grade is never just a universal measurement. The coefficients behind it come from the prices and recoveries used for one specific project at one specific point in time. So two deposits can both report 1.2 percent CuEq — and still have very different real revenue per tonne. And the same issue can happen inside your own project. If your equivalent formula was built three years ago, but the price deck has changed since then, then this year’s CuEq and last year’s CuEq are not necessarily measuring the same economic thing. And we’ve already seen why — the price ratios themselves can move significantly over time. That’s exactly why CIM guidance and Canadian securities requirements expect the formula to be disclosed together with the equivalent grade. Because without the formula, the number is missing its context. And the simplest way to put it is this: a number without the formula is not really data. It’s an impression.

And now, let’s talk about when an equivalent grade actually works perfectly well. Because there are cases where it is absolutely the right tool. Four things need to be true. One saleable product. One processing route. Recovery that stays essentially constant across the deposit. And relatively stable price ratios. If all four conditions hold, then the equivalent grade is not really a shortcut. It gives you the same economic answer as the full multi-product calculation. In that situation, there is no reason to avoid it. It is simple, convenient, and correct. The real question is just this: does your deposit actually meet those four conditions?

Before we move into the demonstration, there’s one modelling decision I want to make very clear — because you’re about to see it happen inside the software. When a block can go to more than one destination — for example, the mill, the heap leach, or the dump — those destinations are alternatives. The block can only go to one of them. So economically, we choose the destination that gives the highest value. That’s the left side of the slide. But final products are different. If one polymetallic block goes through the same processing route, it may produce copper concentrate, zinc concentrate, and other products at the same time. Those products are not alternatives. They are additive. So instead of taking the maximum, we need to take the sum. And that distinction matters. Because if we use max where we should use sum, the model keeps only the single most profitable product — and the rest of the value quietly disappears. And the blocks that suffer most are usually the balanced, truly polymetallic blocks — the very blocks that make this kind of deposit valuable. So in the demonstration, you’ll see us define five final products and place them into one group. And what you’re seeing there is exactly this economic logic translated into the software.

So, what are you about to see? About ten minutes, and six key steps. First — the model attributes. Then, we’ll define the five final products and group them together. Next — we’ll set the recoveries for each product. Then we’ll calculate the block value and determine the destination. After that, we’ll run the optimization and generate the nested pit series. And finally — we’ll put the two runs side by side and compare the results. So if you’re watching the recording later, these six steps can also serve as your navigation through the demonstration.

Now let’s see how this works inside the software. We’ll start with the block model — because that is the foundation for everything that comes next. In this model, we have a material class attribute that separates ore from waste, based on the confidence category. That gives us our starting point. We also have a copper equivalent grade — because in this deposit, copper contributes the largest share of the economic value.

We load the prepared project and walk through the parameters that will be used in the optimization. The first thing to check is how the model itself is configured. The input fields need to have the correct format — because that directly affects how the final product mass and the block value are calculated. Here we can also see the range of values for each parameter.

Next, we go to the rocks tab. Here we select the model type — either percentage or material class. Then we tell the software what should be treated as ore and what should be treated as waste. This step is important, because every quality parameter we want to use later needs to be connected to the ore here. If it isn’t included here, it won’t be available in the calculation later. We also specify which field contains the density — and then we move on.

For this example, we’re keeping the geotechnical model simple. We use one overall slope angle for the entire deposit — 50 degrees.

Now we move into the economic model. For the revenue factor, we use a range from 0.1 to 1.5, with steps of 0.05. That gives us enough resolution to see the full nested pit series without creating an unnecessary number of shells. And now we get to the part that really matters for today’s demonstration. We have five final products. All five can come from the same processed block. In other words, the components exist together inside the same feed. To calculate how much of each final product a block contains, we use the content field. Think of it as an additional multiplier. And again, the field type has to be configured correctly for the calculation to mean what we think it means.

Now we enter the processing cost. And this is worth slowing down for. We process one block — but we may recover several products from it. So we have one processing cost, but multiple revenue streams. How do we distribute that cost? The practical approach is to allocate it in proportion to each product’s contribution to the total value. The exact arithmetic is not the important part here. The important part is the logic. We need to understand what each parameter represents in the real plant — and reproduce that logic correctly inside the model.

Next is the cut-off grade. For this deposit, the break-even equivalent grade is about 0.11 percent. Anything above that is treated as ore. And then we define the recovery factor. This represents how efficiently run-of-mine ore is converted into a saleable product.

Now we come to one of the most important settings. There are two ways the software can treat these final products. If we do not create a group, the products compete with each other. For every block, the software chooses the single most profitable product. There is one winner. And that competition is repeated block by block. But if we place the products into one group, the logic changes. Now all five products contribute to the same block value. Their value is added together. And which approach is correct depends on the flowsheet — not on preference. Processing routes are alternatives. A block might go to the mill, or the heap leach, or the dump. Those destinations compete. But final products from the same processing route do not compete. A polymetallic block can produce copper concentrate and zinc concentrate from the same feed. So here, their values need to be combined. If we accidentally use competition where we should use a sum, the software keeps only the most profitable product — and the rest of the block’s value simply disappears. So the rule is simple: configure the model the way the plant actually works.

From here, we move through the remaining inputs: mining cost, ore loss, dilution, haulage, and the general calculation parameters. One useful setting here is the minimum reasonable block cost. You can think of it as a financial cut-off. If a block does not reach the required profitability level, it is treated as waste.

Once the optimization is complete, we move into the results. We have the results table, prepared charts, visualization of the block groups by revenue factor, and the generated pit surfaces.

By displaying the block groups one after another, we can see how the pit outline responds as the final product price changes. We can also see how stable the optimization is and which direction the deposit may be worth developing.

The results table is really the main analytical output. This is where we get the data for evaluation, comparison, and ultimately pit shell selection. How you choose the final shell is up to you. You might select based on the highest NPV, the highest profit, the ultimate pit, or another profitability criterion. The software gives you the tools — the decision remains with the engineer. The table itself is live. You can restructure it, filter the information, and build charts directly from it. And every revenue factor has its own unique data set behind it. We can also see the cash coming in and the costs broken down by line item. Everything we entered earlier is reflected here — and the system also calculates additional metrics. Life of mine, for example, is derived from the amount of ore in the shell and the annual production rate. The name column also identifies which factor corresponds to the highest NPV, the highest profit, and the ultimate pit. So you don’t have to find them manually. And we also get tonnages for all rock types, together with their quality parameters — mean, minimum, and maximum.

For a quick first look, we also have prepared charts. The main ones show ore tonnage, waste tonnage, and profitability — either profit or NPV. We can also look at life of mine, revenue, marginal profit, and other parameters. And both tables and charts can be viewed in cumulative, incremental, or differential form.

Now let’s run the second scenario. This time, instead of valuing the block through five separate products, we value it through one — the equivalent grade of the dominant metal. Everything else stays exactly the same. Same prices. Same mining cost. Same ore loss. Same dilution. Same haulage. Same cut-off logic. Same deposit. That is intentional. We want one thing to change: how the block value is calculated. Because if several assumptions change at the same time, we’re no longer comparing the same problem.

And here is where the result gets interesting. The final pit shells are almost identical. The number of blocks inside the selected shells matches almost everywhere. So let me say it very plainly: the pit barely moved. At the shell we would actually select, ore tonnage matches to within about one hundredth of a percent. And the total value is effectively identical relative to a project worth around 9.5 billion dollars. Why? Because this particular deposit is very rich. The final pit boundary is usually controlled by marginal blocks — blocks sitting close to zero value, where a small change can move them from ore to waste. But here, there are very few blocks like that. Revenue per tonne is around seven times the cost. And the average equivalent grade is roughly thirteen times the cut-off. So there simply isn’t much to flip. And that is a characteristic of this dataset. This model is synthetic, and it was deliberately built with strong economics. On a real deposit where revenue is only two or three times cost, the response could look very different. But even though the final outline barely changed, several other things did. First — waste. Essentially the same ore. Essentially the same value. But about one million tonnes less waste. The strip ratio moves from 1.83 to 1.82. And at the maximum-NPV shell, the difference is closer to two and a half million tonnes. That means fuel, fleet hours, and road time — for essentially the same ore. Second — the starter pit. This is the smallest shell, the highest-value core, and usually the material you mine first. In the multi-product run, the starter pit contains about 19 percent more ore, requires around 22 percent less stripping, and carries roughly 13 percent more value at the same prices. So the final hole may look the same. But the early years do not. And third — revenue by zone. In the copper-gold zone, the equivalent method understates the value per tonne by almost 4 percent. In the zinc-lead-silver zone, it overstates it by around 3.5 percent. Across the entire deposit, those errors largely cancel. The total difference falls below one percent. And that is exactly why the issue can be hard to notice. But a mine schedule does not process the average deposit every year. It moves through different zones in sequence. So those errors do not necessarily cancel inside a single year. And the main point still remains: when the products are kept separate, we can see the contribution of each one. We can compare them. We can test sensitivity to each one. We can understand where the revenue is actually coming from. The equivalent model gives us the total — but not what sits underneath it. So the pit outline may actually be the most robust number in the entire plan. And that means using only the final pit to validate your assumptions can be misleading. The important differences often appear after the outline is defined.

Next, we move to sensitivity analysis. Again, the settings are identical between the two runs. We vary each parameter from minus 30 to plus 30 percent of its base value. Parameters that are not defined are excluded. And some parameters that are extremely difficult or expensive to change — like certain recovery assumptions — may also be left out. The most interesting variable here is product price. And the result is intuitive: the larger the share of total revenue a product contributes, the more strongly the project NPV responds when its price moves. We can look at this either in profit terms or in NPV terms. Lead is a good example. Changing the lead price moves the project less than changing ore loss or mining cost. And that makes sense, because lead contributes only around 2.5 percent of the ore value.

Now let’s look at the cost variables. And here, the picture is very different from the price side. A 30 percent change in most cost parameters moves NPV by only around one percent. Ore loss — about 1.5. Mining cost — about 1.1. Haulage — around 0.7. Dilution — roughly 0.2. Now compare that with prices. A 30 percent change in the copper price moves NPV by around 16 percent in the multi-product model. So we’re talking about roughly an order of magnitude greater sensitivity. And there are two practical observations here. The first: ore loss moves the project result more than the prices of lead and zinc combined. That is important, because ore loss is often entered very early in a study — and then barely revisited. But on this deposit, that operating assumption matters more than two of the five metals. The second: the cut-off grade sensitivity is completely flat. We can move the cut-off up and down, and NPV barely changes. And again, the reason is the lack of marginal material. There just isn’t much ore sitting close enough to break-even for the cut-off to change the result. So on a deposit like this, simply nudging the cut-off to see what happens doesn’t tell us much. And the broader conclusion from this sensitivity tab is very clear: on this deposit, the economics are driven much more by prices than by costs. Cost savings still matter — but they do not change the overall project picture the way a major price movement does.

The final step is to visualize the optimal pit and the ore inside it. This lets us immediately see how completely the mineralization is being captured. We can also look specifically at the inferred material — how much of it sits inside the main pit and how much remains outside for future drilling and re-evaluation. For this optimization, only measured and indicated resources were valued as ore. Inferred material was treated as waste in both scenarios. That is the conservative approach — and because we apply it the same way in both runs, it does not affect the comparison itself. So in the end, we have the same block model valued in two different ways. Almost the same final pit outline. But a different waste tonnage. A different starter pit. A different distribution of revenue across the deposit. And most importantly — a very different picture of project risk.

Before we look at the results, I want to make the comparison itself very clear. Because this is usually where these kinds of tests can become misleading. So here are the conditions. Both runs use the same block model. The same mining costs: 1 dollar 80 per tonne for potential ore, and 1 dollar 50 for waste. The same haulage cost — 1 dollar per tonne. The same processing cost — 11 dollars 50 per tonne of ore. The same price deck — and you can see those prices directly on the slide. The same geotechnical assumption — a 50-degree slope angle. The same shell selection rule — maximum profit. And in both runs, inferred material is treated as waste — never as ore. We also use the same cut-off grade. That matters. Because if we changed the cut-off between the two runs, we would no longer be testing the same thing. So there is only one difference between these scenarios: how the value of each block is calculated. And one important disclosure. This is a demonstration model. It is not a client deposit, and it is not intended to reproduce one. Also, the prices are used on a gross basis — without payability, treatment charges, refining charges, or penalty elements. That affects the absolute dollar values. So those numbers should be viewed as an upper estimate. But because the same assumptions are applied to both runs, the comparison between them still holds.

And now, the first result. And I want to be very clear about this — because it is not the result we expected going in. The pit barely changed. At the maximum-profit shell — the one we would actually select — the ore tonnage differs by less than 0.01 percent. And the total value differs by only about ten thousand dollars on a project worth roughly 9.5 billion. So essentially — the same pit. Why? Because this deposit is very far above break-even. The average ore grade sits at roughly thirteen times the cut-off. And remember — a pit boundary is controlled by marginal blocks. Those are the blocks sitting right around zero value — where even a small economic change can push them from ore to waste, or the other way around. In this model, there are almost no blocks sitting in that range. So there simply isn’t much for the optimizer to change. And this leads to one of the most important takeaways from today. We are not saying that using an equivalent grade means your pit outline is wrong. In many cases, it may be almost exactly the same. In fact, the final pit can be one of the most robust parts of the entire mine plan. But that creates another problem. If you use the pit outline as your test of whether the underlying assumptions are correct — you may completely miss what those assumptions are changing everywhere else.

But while the final pit barely changed — almost everything else did. First — waste. We recover essentially the same ore, for essentially the same value, but we move about one million tonnes less waste. The strip ratio changes from 1.83 to 1.82. And at the maximum-NPV shell, the difference grows to about 2.4 million tonnes. That’s 2.4 million tonnes you don’t have to drill, blast, load, and haul. Which means less fuel, fewer fleet hours, and less road time. Second — the starter pit. This is the highest-value core — the material you mine first, and the part that has a huge impact on your NPV. In the multi-product model, the starter pit contains about 19 percent more ore, with 22 percent less stripping, and about 13 percent more value at the same prices. So yes — the final pit may look almost identical. But the first few years can look very different. Third — revenue by zone. In the copper-gold zone, the equivalent model understates the value by almost 4 percent. In the zinc-lead-silver zone, it overstates it by around 3.5 percent. Across the whole deposit, those differences almost cancel out. The net difference falls below one percent. And that is exactly why they’re so easy to miss. But a mining schedule doesn’t process the average deposit every year. It moves through different zones at different times. So in any single year, those errors may not cancel. And fourth — the point we’ve spent most of today discussing: the risk profile. So the takeaway is simple. Same pit. But a different plan. A different annual budget. And a very different understanding of the project’s risk.

And before we finish, I want to be very clear about the limitations. Because it’s better to understand them up front. First — the pit optimization gives us an unconstrained optimum. It does not account for things like plant capacity, blending requirements, or contract volumes. Those constraints come later, during scheduling. That is not a flaw — it’s simply a different stage of the mine planning process. So the pit shell is not the same thing as a final mine plan. Second — each shell is based on one price deck. Which means sensitivity analysis is absolutely essential. Third — recovery is still a model. It is an assumption about how the plant will perform. And if that assumption is wrong, that error flows directly into the block value and potentially into the pit outline. And finally — in the current version, payability, treatment and refining charges, and penalty elements are not modelled separately. Instead, commercial terms are represented through an effective net price for each product. So if you’re working with a deposit where penalties are significant, you should treat the resulting dollar value as an upper estimate. The main point is — the model is powerful, but only if we understand exactly what it does and what it doesn’t do. And that’s it for today! Thank you for spending time with us today!