India oil intelligence: demand, refining and the economics of the next barrel

The commercial question: which barrel earns its place?

An oil business earns an advantage when it knows which decisions are being improved by a dataset. A national consumption series can support infrastructure planning, but it does not by itself choose a crude cargo, a product slate or a depot allocation. The first question is therefore not simply whether India’s oil demand grows. It is where the incremental tonne appears, which assets can serve it, and how much contribution survives after delivery, conversion and financing. This report connects those questions without presenting a public-data exercise as a proprietary refinery optimisation model.

The evidence combines PPAC’s FY2024-25 national demand baseline, the ministry’s published statistical compendium and IndianOil’s FY2025-26 company disclosures. These have different dates and boundaries. They are deliberately identified rather than spliced into one apparently current time series. The IEA’s February 2024 oil outlook and EIA’s February 2025 country review provide historical forecast context, not confirmation of what subsequently occurred. No November 2026 spot price, freight quote, crude discount or refinery yield is asserted here. [S1] [S2] [S3] [S5] [S7]

Read the demand stack before choosing a growth story

The historical anchor is 239.17 MMT of petroleum-product consumption in FY2024-25. Within that total, high-speed diesel was approximately 91.4 MMT, motor spirit 40.0 MMT, LPG 31.3 MMT and aviation turbine fuel around 9.0 MMT. Those four products represented roughly 71.8% of the total using rounded inputs. Their operating channels differ: logistics and industry, passenger mobility, cooking and feedstock, and aviation. Product tonnes should not be converted into one precise barrels-per-day number without product-specific density assumptions. [S1] [S2]

The analytical implication is concentration with heterogeneity. A small percentage change in the largest product can move more tonnes than rapid growth in a smaller one. For an organisation choosing storage, distribution or customer-acquisition priorities, market size and growth velocity need separate columns. Demand also has a geographical dimension: a national product balance does not tell a marketer whether the next shipment should enter through a western port or an inland terminal. Start with state and channel exposure, then overlay delivery constraints and realised netbacks.

A second distinction is domestic consumption versus refinery production. Refiners may serve exports, hold inventory or adjust intermediate streams; marketers may source from more than their own refineries. Neither national product growth nor an individual company’s total sales can be used as a one-to-one proxy for refining throughput. A disciplined database keeps product, geography, reporting period, source revision and corporate perimeter attached to every observation. That discipline prevents an attractive dashboard from becoming a collection of incompatible denominators.

A transparent FY2029-30 demand envelope

Long-horizon planning needs an explicit range, not a point estimate dressed as certainty. The following calculation starts at 239.17 MMT in FY2024-25 and compounds three assumed annual growth rates for five fiscal years. The rates are analytical stress assumptions, not econometric estimates, management guidance or an update of an external agency forecast. The lower case represents slower end-use expansion or faster efficiency and substitution. The upper case represents stronger activity with slower substitution. Individual product paths could diverge sharply even when the aggregate reaches the same endpoint.

Use the range to test capacity commitments. A project that needs the upper case to fill its capacity deserves different governance from one that remains useful at the lower case. Stress the timing as well: a five-year endpoint can look attractive while the first three years produce poor utilisation. Where a project serves several products, allocate the scenario by customer segment and substitution exposure rather than distributing the total in fixed historical proportions. A central sensitivity is a comparison case, not the most likely outcome by definition.

The landed crude bridge: why a larger discount can lose

Procurement comparisons should use a common delivery point, time basis and quality basis. The table uses a deliberately round $75/bbl benchmark and invented commercial assumptions. Cargo B carries a $5 discount rather than $3, but also costs more to transport and handle. The result is $75.40/bbl landed versus $74.75/bbl for cargo A. This example does not describe any origin, sanctioned trade, supplier, current route or executable cargo. It shows why ranking offers on the differential alone can reverse the economically correct order.

Before calculating, inspect the contract. If the offer is delivered, an additional full freight charge would double-count logistics. If cargo quantity is measured at loading but refinery receipts are measured after transit, reconcile the loss allowance. If the price averages over a future window while the benchmark comparison uses today’s close, identify the timing basis separately. PPAC’s public crude and product price series is a useful reference structure, but an Indian basket reference is not an executable invoice price for every refinery and grade. [S6]

Quality is equally important. Two cargoes at the same landed price can produce different value because of the refinery’s configuration, unit availability and product specifications. The marginal barrel should be evaluated against the actual constrained system. A refinery with limited hydrogen, sulphur handling or conversion capacity may value a particular grade differently from a less constrained site. These are model-design considerations; the report does not claim access to plant assays, proprietary linear-programming models or current refinery operating conditions.

Translate procurement into a contribution sensitivity

A cost comparison becomes useful when connected to saleable product value. In the next simplified model, cargo A produces an assumed $88/bbl of gross product value, incurs $3/bbl of variable conversion cost and $1/bbl of product logistics, and costs $74.75/bbl landed. Its illustrative contribution is $9.25/bbl. Cargo B produces an assumed $87.20/bbl, incurs $3.20/bbl conversion and the same $1 logistics, leaving $7.60/bbl. The $1.65 difference is larger than the landed-cost difference because yield value and conversion also move.

The stress shows why an outright crude-price forecast is insufficient. Product prices may move with crude, partly offsetting the change; product cracks can also compress while crude is stable. A rigorous operating view therefore tracks product netbacks, feedstock cost and site costs independently before aggregating. Inventory valuation and accounting treatment can create a further difference between current replacement economics and reported financial performance. Keep the spot operating model separate from the accounting reconciliation instead of treating either as an error.

At company level, IndianOil reported a FY2024-25 GRM of $4.80/bbl in its investor presentation. That historical reported measure should not be compared directly with our hypothetical contribution without reconciling boundaries and accounting conventions. It illustrates the importance of metadata, not a claim that our assumed refinery is outperforming an actual business. Our model excludes the complexities required for a real operating benchmark. [S4]

A cheap barrel can still be expensive to finance

Cargo economics and liquidity planning should share the same assumptions. Suppose one million barrels cost $74.75 each and a delivery change extends the financed holding period by fifteen days. At an assumed 10% annual funding rate, the incremental simple financing cost is about $307,192, or $0.307/bbl. At an assumed INR 85/$ conversion, that equals approximately INR 2.61 crore. This is not a quote for any company’s borrowing cost; it is a reproducible cash-cycle stress that can be replaced with treasury’s actual rate and contractual payment dates.

The financing clock may start before physical receipt and stop only after customer collection. A longer voyage, tighter supplier payment term or slower product receivable cycle can outweigh a nominal purchase saving. The appropriate comparison includes letters of credit, collateral, credit limits and inventory buffers where material. Stress adverse combinations, rather than assuming the cheapest freight, shortest cycle and best selling price occur together. Commercial teams should escalate an offer that looks attractive only by assuming unused liquidity is free.

A practical decision memo should show the contribution and liquidity consequences side by side. That makes trade-offs visible: a slightly more expensive cargo may release working capital or reduce the risk of a plant interruption. The value of reliability is site-specific and should be estimated from avoided contribution loss, not asserted as a universal premium. Operational resilience is therefore a measurable commercial variable, provided the assumptions and constraints are written down.

What public company data can, and cannot, establish

IndianOil’s performance table reports refining throughput of 75.451 MMT in FY2025-26 against 71.564 MMT in FY2024-25, a calculated increase of about 5.4%. This is evidence of physical throughput growth within the stated company perimeter. It does not reveal a marginal cargo’s profit, a site-specific yield curve or the profitability of every incremental tonne. The chairman’s discussion also gives operating context, but management explanations should remain distinguishable from independent causal analysis. [S3] [S8]

A useful benchmarking dataset separates company-wide sales, standalone refinery throughput, group throughput, pipelines including gas, and liquid-pipeline throughput. Similar labels can conceal different activity sets. Comparison must also address acquisitions, commissioning, maintenance and nameplate-capacity revisions. A high utilisation ratio can be legitimate when effective capability exceeds an older design denominator; it is not sufficient evidence to infer unsafe operation or a permanent productivity advantage. Those conclusions would require engineering and asset-level evidence beyond the public numbers.

For a buyer of intelligence, this is a valuable test of research quality: can the analyst show exactly how each ratio was constructed? A report should provide its source period, unit, revision and corporate boundary, then explain what remains unknown. That is more decision-useful than a league table in which every company’s numbers appear comparable but the definitions are not. Enalyt’s sample deliberately exposes these limitations so a commercial team can challenge the calculation before using it.

An operating watchlist for the next planning cycle

The strongest next step is a controlled pilot, not an indiscriminate data purchase. Choose one recurring decision, define a baseline process, identify the external and internal observations it needs, and record how new evidence changes the decision. For a crude desk that may be delivered cargo ranking; for a marketer it may be replenishment risk; for a strategy team it may be downside utilisation of a proposed asset. Measure improvement against the original decision question rather than the number of charts produced.

The forward view is conditional. Strong aggregate consumption can coexist with weaker economics for a particular product, route or asset. Conversely, slower national growth can still create local opportunities where logistics, specification or customer service are constrained. The research task is to connect observable drivers to a bounded commercial exposure. The scenarios and cost bridges in this report are starting points that can be audited and replaced, not conclusions that should survive contradictory evidence.