By Dr. Warwick Powell, Substack, 2/17/26

Warwick Powell is an Adjunct Professor at Queensland University of Professor working at the intersection of China, digital technologies, supply chains, financial flows and global political economy & governance.

Preface: as we near the fourth anniversary of the formal initiation of Russia’s Special Military Operation in Ukraine, I reflect on the state of the war, its trajectory and likely cadence through a quantitative lens. It’s something I have been doing on-and-off for a few years now, which led me to conclude a while ago that the collective west has already experienced a debilitating strategic defeat, even as the fighting continued. Wars are system-on-system propositions, and the west’s fragilities in terms of repair, replenishment and replacement capabilities has been fully exposed. My core evaluation remains, and this essay explains some of the reasoning for this focusing on the raw ledger of the situation as it presents itself in Ukraine. I won’t be the least surprised to see the war transition to what some call a ‘dirty war’ with increased frequency of assassinations, sabotage and general terror replacing formal battlefield engagement. Meanwhile, representatives of the protagonists arrive in Geneva to continue talks.

---

Introduction

Attritional warfare, where victory emerges not from decisive manoeuvres but from the sustained erosion of an opponent’s capacity to fight, lends itself to mathematical modelling. In the Russian-Ukrainian War, now entering its fifth year as of February 2026, the dynamics have shifted decisively toward this mode since mid-2022. The conflict’s outcome hinges on the relative rates at which each side can replenish losses in personnel, equipment, and munitions compared to the damage inflicted by the adversary.

This essay synthesises key analytical findings from open-source data, outlines the methodology used to derive estimates of collapse timelines, and presents the raw data ranges underpinning these projections. The goal is not to forecast an exact endpoint – warfare defies such precision – but to demonstrate how known parameters allow us to sketch reasonable trajectories and cadences. By aggregating disparate estimates into a coherent framework, we can discern patterns: gradual depletion accelerating into non-linear collapse, with a plausible window of 6-9 months from now before Ukraine’s defensive sustainability falters irretrievably.

This approach draws on historical precedents, such as the World War I models of Frederick Lanchester, adapted to modern data. It reveals a ledger tilted against Ukraine, driven by Russia’s superior replenishment and fire dominance, exacerbated by recent Western aid constraints. Yet, the analysis underscores some degree of uncertainty: data biases, doctrinal adaptations and external variables like aid surges could alter the cadence. What follows is a dispassionate examination, grounded in numbers, to illustrate how such estimations emerge.

Key Analytical Findings

The core insight from quantifying the war’s attritional phase is that Ukraine’s effective combat power – a composite of manpower, machinery and munitions – is depleting at a net rate that outpaces its replenishment, while Russia’s holds steady or grows marginally. This imbalance, compounded by recent reductions in Western support, points to a tipping point where Ukrainian force density thins below viability, triggering rapid territorial losses and operational collapse.

Based on integrated projections from November 2025 to February 2026 data, the estimated window for this tipping point is 3-6 months from now (May-August 2026), followed by a 3-4 month cascade to functional exhaustion. Overall, this yields a 6-9 month horizon to “floodgates opening,” where advances accelerate from the current 0.3-1 km/day to 5-10 km/day, as seen in historical breakthroughs like the 2022 Kherson retreat. The cadence is non-linear: initial depletion appears stalemated, with monthly losses of 10,000-20,000 lethal units (a term normalising soldiers at 1 unit, tanks at 10, etc.), but once below 73% of peak strength (around 400,000 units), losses surge 2-3 times due to exposed flanks and reduced fire support.

Trajectories vary by data optimism / pessimism. Using Western-leaning estimates (lower Ukrainian losses, higher aid inflows), the threshold arrives in July 2026, with collapse by October. Russian-sourced figures (higher inflicted damage) accelerate this to April-May, with endpoint by August. Recent developments sharpen the grim end: US missile stockpiles, depleted to 25% of Pentagon requirements, have prompted prioritisation of domestic needs under the Trump administration, reducing deliveries. Germany’s February 2026 announcement of exhausted stockpiles further cuts munitions inflows by 20-30%, equivalent to an additional 100-200 daily unit losses for Ukraine.

These findings highlight sustainability as the decisive factor. Russia’s net daily gain of 700-900 units sustains its force at 680,000-700,000, enabling methodical pressure without overextension. Russia also has a reserve army of similar scale not yet mobilised. Ukraine, starting February at 450,000-500,000 units (down from 550,000 in November), nets -100 to -900 units/day, a trajectory that compounds subtly until critical. Aid like the Prioritised Ukraine Requirements List (PURL, ~$15 billion in 2026 from Europe) might add 50-100 units/day temporarily, extending the window by 1-2 months, but production lags (e.g., 60 Patriot missiles/month globally) cannot offset Russia’s 10:1 fire advantage.

The broader implication: the war’s cadence follows a predictable arc in attritional models: slow grind to threshold followed by exponential decay. This allows estimating not just endpoints but inflection points, such as when air defence failures (current stocks cover dozens of salvos against 450 monthly threats) amplify Russian strikes by 10-20%. Absent major escalations, like full NATO intervention, the math suggests Ukraine crosses the ledger’s wrong side by late 2026, with territorial concessions becoming inevitable to preserve residual forces.

This will open the pathway for Russia to successfully push towards, and claim Odessa and perhaps even being in a position to hasten a change of regime in Kiev. My own estimation is that the war will only come to a formal end when these two conditions are met, which will enable formal negotiations to proceed to not just surrender terms but more significantly from Russia’s point of view, a number of critical treaties that go to regional security architecture re-engineering. (See my essay from this time last year, explaining these.)

Methodology and Caveats

The methodology employs a modified Lanchester model, a framework from early 20th-century operations research, to simulate combat dynamics. At its heart, it tracks two variables: each side’s effective force level over time, influenced by replenishment rates and the opponent’s inflicted losses. Forces are aggregated into “lethal units” for comparability – personnel count as 1 unit each, armoured vehicles as 10 (reflecting firepower), and munitions as 0.01 per shell (approximating strike equivalence). This normalisation allows modelling the war as a system of differential equations, where Ukraine’s force U(t) changes as: net replenishment minus losses proportional to Russia’s force R(t), and vice versa.

For transparency, the base model assumes linear attrition until a threshold, then introduces non-linearity. Replenishment (r) includes recruitment, repairs and aid inflows minus decay (e.g., equipment wear). Effectiveness coefficients (α for Ukraine’s impact on Russia, β vice versa) embed doctrinal factors: Russia’s mass artillery yields higher β (0.0012-0.0025 losses per Russian unit-day), while Ukraine’s precision strikes give α around 0.0018-0.0020. Initial conditions are set from theater estimates (U_0 ≈ 550,000 in November 2025, adjusted downward by observed attrition).

Numerical integration – discretising time in daily steps – projects forward: U_{t+1} = U_t + r_U – β R_t, iterated until U hits θ U_0 (θ ≈ 0.73, based on historical density for coherent defence). Post-threshold, a multiplier γ (2.5) amplifies losses, simulating breakthroughs. This yields time to tipping (t*) and full collapse (to 50% force, proxy for operational failure).

Data inputs draw from diverse sources: Western (ISW, CSIS, Oryx) for optimistic ranges, Russian (MoD briefings, mil-blogs) for pessimistic, balanced per stakeholder representation. Recent updates incorporate February 2026 reports on US/German aid constraints, adjusting r_U downward.

Caveats abound, underscoring that this is estimation, not prediction. Data biases: Western sources may underreport Ukrainian losses (500-700/day) to sustain support, while Russian claims (1,200-1,800) inflate for propaganda; truth likely middles. Unmodelled variables include morale (potentially accelerating collapse), weather (winter slows cadences), or black swans like drone surges or some successful third party mediation. Aid is volatile: PURL could ramp, adding months; full cutoff subtracts them. The model assumes constant parameters, but adaptations (e.g., Ukrainian drones flipping local α) could localise deviations. Threshold θ is empirical, drawn from past phases (e.g., Pokrovsk encirclements at ~70% density), but varies by terrain. Finally, aggregation into lethal units simplifies: a tank’s value isn’t fixed at 10, and munitions efficacy depends on targeting.

These limitations mean trajectories are probabilistic ranges, not fixed paths. The value lies in sensitivity: tweaking r_U by +100 units/day (e.g., via Japanese Patriot backfill) delays t* by 30-60 days, showing how data parameters inform cadence adjustments. This framework thus enables reasoned estimation, revealing the war’s underlying arithmetic without claiming omniscience.

Raw Data Ranges

The projections rest on raw data compiled from open sources as of February 2026. Ranges reflect divergences: Western estimates (e.g., Ukrainian General Staff, CSIS, Kiel Institute) tend lower on losses/higher on aid; Russian (MoD, Rybar mil-blogs) higher on inflicted damage. Aggregates focus on theatre (frontline) capacities; global stocks are larger but delivery-constrained.

Manpower (Theatre Active Strength):

• Ukraine: 450,000-550,000 (Western: ~500,000 frontline, rotations strained; Russian: ~450,000 effective, implying higher cumulative depletion). Cumulative losses since 2022: 400,000-1,200,000 (Western ~500,000; Russian ~1.2M).
• Russia: 600,000-700,000 (consistent across sources; total committed ~1.2M). Cumulative: 800,000-1,200,000 (Western ~1.16M; Russian lower, ~600,000).

Daily casualties: Ukraine 500-1,800 personnel (Western 500-700; Russian 1,200-1,800); Russia 900-1,200 (Western higher; Russian ~1,000).

Recruitment: Ukraine 5,000-10,000/month (net ~0 due to losses); Russia 25,000-30,000/month (net +700-900/day).

Machinery (Operational Tanks/AFV/Artillery):

• Ukraine: 2,000-2,500 (losses ~10,000 cumulative; aid ~2,000 delivered). Refurbishment: 50-100/month (net negative from attrition).
• Russia: 7,000-8,500 (losses ~30,000-35,000 cumulative, but ~1,000/month refurbished from stocks).

Daily equipment losses: Ukraine 20-50 (Western lower); Russia 30-60.

Munitions (Artillery Shells, Missiles):

• Ukraine: Stock 500,000-1,000,000 (low; daily use 2,000-3,000). Production/Delivery: 20,000-40,000/month (US/EU ramp to 100,000 delayed; PURL adds ~50,000/year but queued). Air defence: Patriot stocks ~200-300 interceptors (covers 20-30 salvos vs. 450 threats/month); production diversion via PURL delays replenishment.
• Russia: Stock 4M-6M; production 4M-5M/year (~12,000-15,000/day). Fire ratio: 5:1-10:1 advantage.

Recent constraints: US inventories at 25% (1,000-1,500 Patriots total; production 740/year, 75% to Ukraine via Europe). Germany: €20B+ exhausted, no more direct transfers; contingent ~35 PAC-3 (~1 week’s defence).

Net Replenishment and Effectiveness:

• Ukraine r_U: -100 to +100 units/day (pre-February; now -100 to 0 from aid cuts). β (Russian effectiveness): 0.0012-0.0025 (up 10% from air gaps).
• Russia r_R: +700-900/day. α (Ukrainian effectiveness): 0.0018-0.0020.

Threshold: ~400,000 units (73% of Nov 2025 peak). Post-threshold multiplier: 2-3x losses.

These ranges, when integrated, produce the 6-9 month window: base depletion 900-1,500/day to threshold in 50-170 days, then 50-90 days to halving. Cadences emerge from sensitivities – e.g., +50 units/day from PURL extends by 30 days – showing how parameters bound trajectories without dictating them.

Conclusion

This quantitative lens illuminates the Russia-Ukrainian War’s attritional logic: a slow, compounding imbalance leading to abrupt shifts. Key findings sketch a 6-9 month collapse window; the methodology provides the tools for such estimates, and raw data ranges highlight the evidential foundation. By focusing on flows over events, we gain insight into cadences – gradual until non-linear – enabling informed trajectories amid uncertainty. The math doesn’t predict; it frames possibilities, reminding us that wars end when ledgers demand it.