A mine is often described by a headline throughput, but throughput does not define the operating system that produces metal. Nordgold's Roman design in Yakutia separated ore preparation, heap leaching and adsorption across three industrial sites, then planned to send loaded carbon to the neighbouring Gross plant for recovery. The project was therefore a network of capacities, interfaces and shared dependencies rather than one self-contained factory.

The announcement described a design, not an operating mine

On 24 June 2025, Interfax reported that Nordgold had completed the first design stage for the future Roman processing complex. The planned throughput was nine million tonnes of ore per year.

That number was a project parameter. It was not commissioned capacity, achieved production or a forecast of annual gold output.

The article did not provide a recovery rate, mine life, operating cost or Roman commissioning date. Any business case using those variables must identify them as assumptions rather than restore facts that were not published.

The valuable information was architectural: three crushing stages, three industrial areas, heap leaching and a planned downstream connection to an existing neighbouring plant.

Nine million tonnes was a balance equation

Nameplate tonnes become useful only when every stage can accept the same sustained flow. The mine, primary crusher, secondary and tertiary crushing, agglomeration if required, stacking, leach pad, solution system and adsorption circuit must be balanced.

A crusher capable of nine million tonnes does not create nine million tonnes of leached ore if conveyors, stacking windows or pad area are constrained. Local peaks can hide a lower annual system rate.

Design should therefore begin with calendars, not only hourly capacities. Weather, maintenance, relocation, lift construction, laboratory feedback and inventory between stages reduce available time.

The production promise belongs to the slowest repeatable interface. Extra capacity elsewhere is valuable only as recovery from variability or planned growth.

Three-stage crushing was metallurgical preparation

Interfax specified primary, secondary and tertiary crushing. Each stage consumes power, wear parts and maintenance while changing the size distribution delivered to the heap.

Smaller particles can expose more mineral surface, but excessive fines may reduce permeability and create uneven solution flow. The correct product size is an economic and metallurgical compromise.

Control cannot stop at an average size. Variability by ore domain affects crusher load, moisture, fines and leach behaviour. Blending decisions made before crushing can influence recovery months later.

The crushing circuit should be governed by the performance of the downstream heap rather than rewarded solely for tonnes processed.

Three sites created interface risk

The planned layout included ore preparation, heap leaching and a gold-recovery plant across about 200 hectares. Physical separation can place each facility on suitable terrain and isolate hazards.

It also creates roads, conveyors, pipelines, power, communications and handoffs. Every boundary needs a design basis, owner and response when the upstream flow differs from specification.

A distributed layout can reduce congestion and make phased construction easier. Conversely, long interfaces add pumping energy, heat loss, inspection effort and exposure to weather.

Project controls should treat connective infrastructure as production equipment. A road or solution pipe is not an auxiliary item when its outage stops the value stream.

The heap was a long-cycle process asset

Heap leaching places prepared ore on an engineered liner and applies solution through an irrigation network. Metal recovery develops over time rather than at one instantaneous plant pass.

This changes working capital. Mining and stacking costs occur before recovered value arrives. Inventory exists physically on the pad and metallurgically in solution and adsorbent.

Lift height, compaction, permeability, irrigation uniformity and solution chemistry influence the recovery curve. A pad can hold the planned tonnes while delivering value more slowly than the financial model expects.

Management therefore needs cohort accounting by stacking period and ore type. Total tonnes on the pad do not explain whether recovery is on schedule.

A bright aerial view of a remote Yakutia mining project shows separated crushing, heap-leach and adsorption areas connected across taiga terrain to a distant neighbouring plant
Distributed facilities can use terrain and shared assets efficiently, but every road, pipe and transfer becomes part of production reliability.

Water balance linked process and environment

A leach system must manage solution inventory through precipitation, evaporation, ore moisture, pond capacity and seasonal conditions. Too little water constrains irrigation; too much can exceed storage and treatment capacity.

Northern temperature cycles affect pipes, pumps, access and solution behaviour. Design margins need to reflect realistic operating seasons rather than annual averages.

Liner integrity, leak detection and storm capacity are both environmental protections and continuity controls. A containment event can stop production and damage the project's licence to operate.

The water model should therefore be updated with measured conditions and connected to daily operating decisions, not preserved only in the original approval package.

Loaded carbon crossed the project boundary

The source said gold was planned to be recovered from loaded carbon at the neighbouring Gross mine's plant. Roman's internal process would therefore end before final metal production.

Shared recovery can avoid duplicate capital and use existing expertise. Its benefit depends on spare capacity, compatible specifications, secure transport and aligned schedules.

Loaded carbon is valuable inventory. Sampling, custody, moisture, transport, reconciliation and security must preserve both physical material and accounting evidence.

The transfer price or service agreement should make losses, delays, reprocessing and priority explicit. Informal cooperation between neighbouring assets is not enough for a bankable project.

Gross capacity became a hidden constraint

A neighbouring plant can look available at design stage while its own ore plan, maintenance and improvement projects change over time. Roman's growth could compete with Gross's base operation for recovery slots.

The shared facility needs a joint capacity model covering normal throughput, peaks, planned shutdowns and incident recovery. A yearly total can conceal simultaneous demand.

Quality compatibility matters as much as volume. Carbon loading, contaminants and elution behaviour can affect cycle time and downstream performance.

Capacity rights should survive organisational change. The dependency remains even if budgets, managers or ownership structures later differ.

Shared infrastructure changed capital efficiency

A fully standalone Roman complex would require more equipment, utilities, staff and commissioning. Reusing an existing recovery plant can lower initial capital and shorten the capability build.

The saving is not simply the avoided purchase price. It includes duplicated laboratories, workshops, spares, specialist teams and supporting infrastructure.

Against that saving sit interface capital, transport, service charges and dependency risk. The correct comparison is the lifecycle cost and resilience of two architectures.

Shared assets are most valuable when their capacity is genuinely spare and the connection is simple. A cheap interface that creates chronic queuing can destroy more value than the duplicated plant would have cost.

Roman and Tokko needed separate accounting

Government information previously put combined investment in the Roman and Tokko heap-leach projects at 83.1 billion rubles excluding value-added tax. That was not a Roman-only budget.

The source also said Tokko commissioning was planned for late 2025 to early 2026. It did not assign that schedule to Roman.

Portfolio reporting must preserve this distinction. Combining projects can show shared strategy, while asset-level decisions still require separate scope, commitments, progress and risk.

Otherwise, a portfolio total can obscure which project has consumed capital and which has delivered operating capacity.

Forty-nine tonnes was a resource reference point

Interfax reported that Nordgold placed 49 tonnes of gold at Roman on the balance in 2022 after approval of temporary exploration conditions.

Metal in a resource statement is not equivalent to recoverable production or revenue. Mining loss, dilution, metallurgical recovery, schedule, price and cost stand between geology and cash.

The design process progressively converts geological confidence into mineable shapes, process feed and economic reserves. Each conversion has assumptions that need reconciliation.

Management should resist dividing 49 tonnes by an imagined annual output. The source did not provide the inputs needed for that calculation.

Remote logistics needed a reliability inventory

A project in Russia's Yakutia operates across long distances, seasonal constraints and limited alternative infrastructure. A missing wear part can affect the entire balanced chain.

Criticality analysis should connect probability, lead time and production consequence. Not every expensive component deserves a local spare, while a small unique sensor may stop a crusher.

Fuel, reagents, liners, pipes and tyres also need seasonal plans. Buffer stocks must be sized against replenishment uncertainty, shelf life and storage conditions.

Digital monitoring helps only when communications, power and local decision rights remain available during disruption.

Phasing could turn uncertainty into evidence

A distributed design supports staged construction: enabling infrastructure, one process area, initial pad capacity and later lifts can proceed through controlled gates.

Phasing is valuable when early operation tests ore behaviour, stacking productivity and shared-plant interfaces before the full nine-million-tonne system is committed.

But a partial system must still be operable. Starting crushers without adequate pads or recovery slots creates inventory rather than cash.

Each phase needs a complete value path, acceptance criteria and funding for the next constraint. Construction progress alone is not evidence of commercial readiness.

Commissioning had to cross organisational boundaries

Individual equipment tests cannot demonstrate the distributed system. Commissioning must move representative ore through crushing, stacking, irrigation, adsorption, carbon transfer and recovery.

Performance testing should define feed characteristics, duration, availability, product quality and accounting reconciliation. Short runs on favourable ore can overstate sustainable capacity.

Gross personnel participate in Roman's value test even though the recovery plant belongs to another operation. Responsibilities for troubleshooting and acceptance need advance agreement.

Ramp-up curves should include learning and seasonal conditions. Immediate nameplate performance is rarely a prudent financing assumption.

The control model followed the material

A distributed-project readiness gate

  • Balance mine, three crushing stages, stacking, pad area, solution and adsorption by calendar.
  • Define specifications and ownership at every material, solution, data and custody transfer.
  • Reserve Gross recovery capacity under normal, peak and shutdown scenarios.
  • Reconcile ore cohorts, solution inventory, loaded carbon and recovered metal.
  • Test water, weather, road, power, communications and critical-spares resilience.
  • Release each expansion phase only after a complete value path meets acceptance criteria.

These controls turn the diagram into an operating contract among sites.

One plant was not the project

Roman's design illustrates why a mining asset cannot be judged by a crusher number or a single facility photograph. Value moves across geography and changes physical form at every stage.

Shared recovery can produce disciplined capital efficiency, while also creating a dependency that must be priced, contracted and tested.

The nine-million-tonne objective becomes credible only when the slowest interface, long-cycle heap and neighbouring plant support the same calendar.

A distributed project succeeds when every separate site behaves like part of one operating system—and when management can see, reconcile and govern the material between them.

Commissioning must follow the material, not the ribbon-cutting calendar

A conventional opening ceremony can make a mine look finished when its value chain is still fragmented. Roman needs a different commissioning logic. Dry crushing, agglomeration, stacking, irrigation, solution collection, carbon loading, transport and recovery should each have an acceptance test, but none should be declared operational in isolation. The decisive test is an accountable parcel of ore moving from the mine face to recovered metal while mass, moisture, solution and carbon balances remain reconcilable.

This approach changes the sequence of readiness reviews. Operators should first prove stable mechanical flow at conservative rates, then establish solution circulation without unacceptable leakage or inventory uncertainty, and only afterward increase leaching intensity. Loaded-carbon dispatches to Gross need their own trial programme: sealed containers, sampled batches, custody documents, transport windows, receiving capacity and assay reconciliation. A defect at the final interface can hide behind apparently successful crushing statistics for weeks.

Ramp-up targets therefore need linked gates rather than one headline date. A sensible gate might require a defined number of continuous operating hours, acceptable size distribution after tertiary crushing, stable pad permeability, verified solution recovery and a closed carbon balance. Management then sees whether the system is genuinely learning or merely accumulating work-in-progress between disconnected units.

Contracts should allocate interface risk before operations inherit it

Distributed infrastructure often crosses the boundaries of construction packages, operating teams and legal entities. If those boundaries are vague, every delay becomes somebody else's problem. Roman's commercial architecture should identify who controls the ore handover, pad availability, reagent supply, carbon containers, road maintenance, Gross processing slots and final assay. Each service level needs a measurable definition, a response time and a remedy that reflects the real economic consequence.

The recovery arrangement deserves particular attention because the neighbouring plant is both an efficiency and a concentration of risk. Capacity reservations should distinguish planned campaigns from emergency access. The parties need rules for off-spec carbon, contaminated batches, assay disputes, metal-accounting adjustments and temporary storage when transport or recovery is interrupted. Pricing should expose the cost of congestion rather than disguise it in a broad annual charge.

Good contracts do not remove operational collaboration. They give collaboration a shared language. When the dispatcher, metallurgist and finance controller use the same batch identifier and the same cut-off time, a physical handover becomes a visible economic event. That visibility is essential when working capital is tied up in material travelling between sites.

Winter readiness is a design input, not a seasonal checklist

Yakutia turns ordinary interfaces into temperature-sensitive ones. Conveyors, crushers, water lines, pumps, instrumentation, roads and sampling stations must work as a connected cold-weather system. A winter plan written after construction cannot compensate for inaccessible valves, exposed lines or maintenance bays that are too small for gloved work. Equipment selection, insulation, heat tracing, spares and shelter belong in the design basis.

Heap leaching also requires deliberate inventory management across seasons. Operators need to understand where solution can freeze, how irrigation distribution changes, how snow and meltwater affect the water balance, and which pad areas can be isolated safely. Transport to Gross needs alternative windows and storage capacity sized for credible road closures, not average weather. The objective is not uninterrupted maximum output; it is controlled degradation without losing custody, environmental protection or restart capability.

A small set of measures can reveal whether integration is working

Roman's dashboard should avoid drowning the operating model in local equipment data. A compact hierarchy can connect throughput to value: mined tonnes released to crushing, tertiary product within specification, tonnes stacked, irrigation availability, solution inventory, carbon loading, dispatch age, Gross recovery turnaround, reconciliation variance and gold recovered per tonne placed. These measures should be viewed together, because improvement at one station can damage the next.

Leading indicators matter as much as monthly production. Growing stockpiles before a constrained crusher, declining pad permeability, delayed carbon lots or widening assay differences warn of future lost output. A control room that only records finished ounces reacts too late. The distributed design becomes manageable when its queues, inventories and obligations are visible early enough for intervention.