The Fuel You Design For

Years ago I owned an AMG GLC 63S, the twin-turbocharged V8. On delivery day the dealer passed on one instruction with some weight: run it on 98 or 100 octane, and do not make a habit of 95. None of it was new to me. I had spent time at AMG performance driving events, where the relationship between a fuel and an engine is the whole point, so the instruction landed as confirmation rather than news. And the reason for it was design, not loyalty to premium fuel. The engine had been calibrated around a particular envelope of energy and knock resistance, and below that envelope the ECU quietly protects itself by retarding the ignition timing. The car survives. It just stops being the machine it was built to be.

The road bore this out. Moving from 98 to 100 changed nothing I could feel. Dropping from 98 to 95 put a softness into the response, most of all under load, as though the engine were holding something back. It was. It was protecting itself from a fuel it had not been designed to burn.

We are now doing a version of this to an entire national fleet. India has changed the fuel diet of every petrol car on its roads, and the official message is that nothing really changes. The physics says otherwise, and so does every engineer who has ever calibrated an engine to a fuel.

Two numbers, moving in opposite directions

Two properties of petrol matter here, and it helps to keep them separate because they do not move together.

The first is energy density, the usable energy packed into each litre. It sets, more than anything else, how many kilometres a given engine gets from a tank. The second is octane, measured as a RON number, which is the fuel’s resistance to knock. Higher octane lets an engine run more aggressive spark timing and higher compression without combustion running away into damaging detonation.

Ethanol moves these two in opposite directions. It carries only about two-thirds of petrol’s energy per litre,¹ so as its share of the blend rises from E0 to E10 to E20, the energy in each litre falls and mileage falls with it. But neat ethanol has a high octane rating, well above the petrol it is blended into,² so the same move raises the blend’s resistance to knock. E20 therefore does two things at once. It hurts energy per litre, and it helps knock resistance. Whether that is good news or bad depends entirely on what the engine was designed to expect. Handed to an engine built to exploit it, the extra octane is an opportunity. Poured into an engine built for weaker fuel, only the energy loss shows up, and the octane gift goes unspent. In plain terms, E20 gives an engine more knock safety and less distance per litre.

The energy loss is not a matter of opinion. When the US Department of Energy ran sixteen unmodified legacy vehicles on rising ethanol blends with no other change, fuel economy fell by 3.7 per cent at E10 and 7.7 per cent at E20, tracking the energy content of the fuel almost exactly.³ That is the curve below. Public debate in India has circled for months around a number that lab work settled years ago.

What it means to design for a fuel

Treat E20 as the fuel an engine is built for, rather than one it is asked to tolerate, and the designer has to change the engine in several places, none of them a firmware tweak.

Start with the combustion. To cash in the extra knock margin the higher octane provides, the designer raises the compression ratio or reshapes the combustion chamber, then retunes the valve timing and the exhaust-gas recirculation to ethanol’s different flame behaviour. The control logic follows: the calibrator advances the spark maps toward the timing that yields maximum torque, and adjusts the mixture and the cold-start enrichment for the way ethanol evaporates. Then the materials. Ethanol is a more aggressive solvent than petrol and it holds water, so certain legacy elastomers and metals, particular nitrile rubbers, terne-plated steel, some die-cast zinc alloys, corrode or swell faster in higher blends, and the designer has to specify them out of the fuel system.⁴ This part needs care, because most materials show no meaningful difference between E10 and E20; the vulnerability sits in specific older parts, not the whole car. Finally the gearing, which the engineer picks to hold the engine in the band where the E20 tune runs most efficiently and to win back some of the mileage that lower energy density gave away.

The engineering literature is blunt. Run an engine across its full operating map on E20 and the higher octane does let you recover torque, but fuel consumption still rises unless you also raise the compression ratio and re-optimise the gearing.⁵ Designing for a fuel means making a chain of mechanical and control decisions at the design stage. It is not a sticker on the fuel cap.

What happens when you pour E20 into an E10 engine

Most of India’s petrol fleet was conceived for E0 or E10, and for those engines the outcome of a switch to E20 is not mysterious.

Mileage falls, and energy density alone guarantees it. The DOE road-load figure for uncalibrated legacy cars is about 7.7 per cent at E20. Indian carmakers report less for their newer models: Maruti Suzuki’s Rahul Bharti told a government press event that E20 cuts mileage by roughly 3 to 3.5 per cent, about 0.6 kilometres per litre on a car that does twenty, and attributed it plainly to the lower calorific value of the fuel.⁶ Owners of older cars report far worse. A June 2026 crowd survey of tens of thousands of owners had two-thirds of pre-2023 drivers claiming losses above ten per cent.⁷ That last number should be held at arm’s length: it is self-reported, gathered on a polling app in the middle of a public backlash, and it is not a controlled test, where the regulator’s own range tops out nearer six per cent. The honest reading is the chart above: the lab value sits between India’s official figure and what owners actually report, and both sides are arguing around a number settled in a laboratory years ago.

Durability is a slower story. Modern BS4 and BS6 cars tolerate E20 without sudden failure. Ethanol has been in Indian petrol since 2003, and from April 2023 every new vehicle was required to be E20-compatible, with the calibration and the fuel-system materials to match.⁸ But tolerate is not the same word as designed for. A car built around one envelope of energy and knock and then run on another spends its whole life a little off its design point, and that mismatch compounds slowly through wear, deposits and corrosion. In the terms I started with, the engine runs, the way my V8 ran on 95, but it lives in a permanent state of self-protection.

There is one more system-level vulnerability, and it sits in the ground rather than the engine. Ethanol is hygroscopic. It pulls water out of the air and holds it in solution, and where straight petrol lets water settle into a layer you can drain from the bottom of a tank, an ethanol blend carries that water along until it passes a limit and drops out as an ethanol-rich layer that misfires and corrodes.⁹ That is a fuel-handling and infrastructure problem rather than a verdict on ethanol, but it grows more consequential the faster a whole fleet is moved to higher blends through storage and pumps of uneven quality. Put crudely, bad tanks and sloppy storage can turn E20 into a mix that misfires and rusts parts instead of burning cleanly.

There is a deeper problem still, and it undoes the tidy notion of simply adapting to E20. A design fuel has to be a known quantity. My AMG’s calibration assumed 98 octane because 98 was what the pump reliably delivered. Moving an entire country from E10 to E20 is not a switch thrown on a single morning. It is a logistical migration across tens of thousands of outlets, with old stock and new stock in circulation together, regional lag, blend-accuracy tolerances, and the ordinary variance of a fuel supply chain the size of India’s. For a long stretch of that migration, and arguably well past its official end, a driver has no dependable way to know what is actually in the tank. You cannot calibrate an engine, or your own expectations, to a fuel you cannot identify. It is not only that the cost of E20 is narrated rather than shown; the fuel itself is not a stable, verifiable input at the point of sale. The owner who does everything right, buys the E20-native car, reads the manual, accepts the mileage, still cannot be certain that the blend they pump this week is the one their engine was tuned to burn.

A design choice narrated as an announcement

India made E20 the default petrol nationwide from April 2026, with E21 and E25 floated, not yet mandated, for 2027 and 2029.¹⁰ Parts of the official case are correct. On E20-ready hardware, engines do not die. Calibration can claw back some of the loss. The emissions and oil-import arguments are real and worth having. The problem is not the direction. It is that a change of design fuel is being presented as a change of nothing.

Three things sit underneath the messaging. The first is a drop-in mindset: a fleet designed for E0 and E10 is being asked to absorb a new design fuel quietly, and the vehicles caught in the middle, with E20-capable materials but E10-era calibration, are the ones asked to carry the mismatch. The second is an opaque trade-off surface. The curve exists, it is measured, and rather than show it the messaging narrates it. The reassuring “1 to 2 per cent” that circulated applies only to E20-certified cars built after 2023; the oil ministry’s own headline figure for the wider fleet has since become “up to 5 per cent,” and a senior minister has called the damage claims “overblown” and a “false narrative,” daring critics to “show me a single car” that E20 has harmed.¹¹ The trade-off is not being shown to people. It is being told to them, and the telling changes with the audience.

The third is cohort-blindness, and it is the one that matters most. The policy does not sharply distinguish a legacy E10 car, a transition car with E20-ready materials but E10-era tuning, and a future car designed for E20 from the start. E20 arrives for all three at once, and the entirely predictable differences between them get reframed as misinformation or as individual owners neglecting maintenance. From a decision-making standpoint, this is a category error: treating a change of design fuel like a minor tweak.

It is worth being fair about why India is doing this at all. The country imports roughly 88 per cent of the crude oil it burns,¹³ and every litre of domestic ethanol blended into petrol is a litre of imported oil it does not have to buy. The government puts the saving at around ₹1.9 lakh crore of foreign exchange since 2014, set beside higher farm incomes and lower emissions.¹⁴ Those are real benefits, and they are the honest case for blending.

But the benefit and the cost do not fall on the same people. The benefit is national: it lands on the oil-import bill, on the exchequer, on the ethanol producer and the sugarcane farmer. The cost is personal, and it lands on the driver, because E20 is sold at exactly the same price as the E10 it replaced. The pump price did not drop to pass on the cheaper domestic ethanol inside the blend, and it did not drop to offset the three to four per cent fewer kilometres each litre now returns. On a fleet burning some forty million tonnes of petrol a year, that unadvertised penalty is worth roughly ₹16,000 to ₹22,000 crore a year,¹⁵ paid by motorists who were told nothing had changed. The transition is real and defensible. It is simply being funded, quietly, out of the fuel economy of people who never saw the bill.

Brazil, and the axis nobody wants to spend

Brazil is the example everyone reaches for, and it is a good one, as long as it is read properly. Blending there began with the Proálcool programme in 1975, in the wake of the oil shocks.¹² Over the decades that followed, fuel, engine architecture and driver expectation were allowed to move together. The mandatory blend settled in a band of roughly 20 to 27 per cent, reaching E27 in 2015. Flex-fuel cars arrived in 2003 and passed nine in ten new sales within a few years, and to this day a Brazilian driver chooses at the pump between high-blend petrol and pure ethanol, in a fleet explicitly built for that choice.

The lesson is not that high blends are easy. Brazil succeeded because it let the fuel, the hardware, the calibration and the public adjust to one another over a long horizon, with feedback at every step. India is compressing that half-century of adjustment into a few years, and compressing the design work along with it. A transition that took Brazil fifty years is being asked of India’s fleet in a single mandate cycle. The friction it produces is the price of skipping the time axis, and no amount of engineering talent removes it.

A more honest roadmap

Take fuel transitions seriously as design problems and the sensible path looks different. Segment the fleet honestly, into a legacy cohort certified only to E10, a transition cohort with E20 materials but E10 calibration, and a native cohort designed for E20 and beyond. Keep an E10 stream available for the legacy cars until ordinary scrappage shrinks that cohort to a remainder. Signpost the blends plainly, at the pump and in the manual, the way Brazil does. Require carmakers to publish the E0, E10 and E20 performance and economy curves rather than leave owners to reverse-engineer them from their own fuel bills, and to state the expected mileage delta in the owner’s manual. Tighten the standards for water content, blend accuracy and storage, and publish the compliance data. And let the timeline for E25 bend to what that data shows rather than to a fixed political date.

None of this is an argument against ethanol. It is an argument for treating a change of operating fluid across a whole national fleet as the design problem it is, and for showing people the curve instead of narrating it.

Closing

My AMG on 95 was fine, in the everyday sense of the word. It started, it drove, it did not explode. It was simply never quite the engine it had been built to be, because the fuel and the calibration were misaligned and the ECU spent its life compensating for the gap.

A national fleet on E20 without E20-native engineering is in the same position. The engines start, the vehicles move, and success is declared, while the whole system runs a little way off its design point and carries the difference in fuel economy, in slow wear, and in trust, none of which appears in any account anyone is keeping. Design-centred thinking would treat E20 not as an ideological victory to announce but as a new environment to design for, with time, iteration, attention to which cohort a car belongs to, and honest trade-offs shown in full. For a country that prides itself on engineering talent and frugal innovation, that is not too much to ask.

Sources

[1] Ethanol carries roughly 67% of petrol’s energy on a volumetric basis (about 21 MJ/L against 32 MJ/L), the figure the DOE testing itself uses. IEA-AMF fuel-properties data (ethanol); corroborated in NREL/TP-540-43543 §E.4.1.

[2] Neat ethanol has a research octane number around 108, with a blending octane quoted higher still, against roughly 91 to 92 for regular petrol. IEA-AMF fuel properties (ethanol).

[3] Knoll et al., Effects of Intermediate Ethanol Blends on Legacy Vehicles and Small Non-Road Engines, Report 1, NREL/TP-540-43543 / ORNL/TM-2008/117 (February 2009). Sixteen legacy vehicles, LA92 cycle, no recalibration: economy loss of 3.68% at E10 and 7.71% at E20 relative to E0.

[4] Kass, Theiss et al., Intermediate Ethanol Blends Infrastructure Materials Compatibility Study, ORNL/TM-2010/326 (2010-11). Certain legacy elastomers and metals degrade faster at higher ethanol fractions; many materials show no meaningful E10-to-E20 difference, so the effect is material-specific, not universal.

[5] Study of Effect of Ethanol Blending on Performance and Fuel Economy of a Naturally Aspirated Gasoline Engine, SAE Technical Paper 2022-28-0024. E20’s higher octane enables a full-load torque benefit, but brake-specific fuel consumption still rises (~6 to 7%) unless compression ratio and gearing are changed.

[6] Rahul Bharti, Maruti Suzuki, at a Ministry of Petroleum and Natural Gas media interaction (early July 2026): E20 reduces mileage by about 3 to 3.5%, roughly 0.6 km/l on a 20 km/l car, owing to lower calorific value. Business Standard, 5 July 2026.

[7] LocalCircles survey (June 2026, 44,000+ responses): about two-thirds of pre-2023 petrol-vehicle owners reported efficiency drops above 10%. Self-reported crowd poll, not a controlled test; the regulator ARAI’s controlled range runs to roughly 1 to 6%. Present as reported/perceived, concentrated in older non-certified vehicles.

[8] India’s ethanol blending began at 5% in 2003 and scaled through the 2010s; BS6 Phase 2 (from April 2023) required all new vehicles to be E20-compatible. Vehicles from before that are broadly E20-tolerant rather than E20-optimised.

[9] Ethanol is hygroscopic and holds water in solution up to a temperature-dependent limit (around 0.5 vol%), beyond which an ethanol-rich aqueous layer separates, causing misfire and corrosion while leaving the upper layer octane-depleted. NREL water-uptake study (2016); US Grains Council phase-separation brief. (Cold-start difficulty is a high-blend, E85-scale effect driven by latent heat of vaporisation, and is modest at E20; do not attribute it to blend vapour pressure, which a small ethanol fraction actually raises.)

[10] E20 became the mandated nationwide petrol grade from 1 April 2026 (pulled forward from 2030); E21 by 2027 and E25 by 2029 are under evaluation, not legislated, with E25 described as a likely cap. Autocar India; ChiniMandi; Business Today (7 July 2026).

[11] Ministry of Petroleum and Natural Gas framing of a “scientific, phased rollout”; the “up to 5%” full-fleet figure (Business Standard and The Week, 10 July 2026); Union Minister Nitin Gadkari calling damage claims “overblown” and a “false narrative” and challenging critics to “show me a single car” (Scroll.in). The “1 to 2%” figure applies specifically to E20-certified post-2023 vehicles.

[12] Proálcool launched 1975; mandatory blend in the 20 to 27% band, reaching E27 in 2015; flex-fuel from 2003, exceeding 90% of new light-vehicle sales by the late 2000s; drivers choose between high-blend petrol and E100. History of ethanol fuel in Brazil (overview); Green Car Reports.

[13] India’s crude-oil import dependence was 88.3 per cent in FY2024-25, and 88.6 per cent for April 2025 to January 2026 (PPAC, Snapshot of India’s Oil and Gas Data, January 2026).

[14] Ministry of Petroleum and Natural Gas statement (July 2026): ethanol blending credited with roughly ₹1.9 lakh crore in cumulative foreign-exchange savings since 2014-15 and about 31 million tonnes of crude displaced, alongside farmer income and emissions benefits. This is a government claim, cumulative over eleven years and not independently audited; it should not be set directly against a single year’s consumer cost.

[15] Author’s estimate of the annual consumer cost. About 40.0 million tonnes of petrol were consumed in FY2024-25 (PPAC), roughly 54 billion litres at 1,350 litres per tonne, multiplied by a conservative 3 to 4 per cent E10-to-E20 economy loss and a pump price near ₹100 per litre, gives ₹16,000 to ₹22,000 crore per year. This is consumer cost only, the extra fuel bought to cover the same distance, and is not netted against national import savings. The 40.0 MMT petrol figure is confirmed from PPAC’s Snapshot of India’s Oil and Gas Data (January 2026), and refers to the petrol pool including blended ethanol.