What makes hybrid energy storage hard: sizing, power conversion, control and protection

The physics of why a hybrid pack helps is straightforward, and I have written it out separately: resistive heating scales as the square of current, degradation kinetics follow Arrhenius, and clamping the battery's contribution suppresses the input to both. That argument takes about a page and it is not in dispute.

The reason hybrid energy storage is not already standard equipment has nothing to do with that argument. It is that four engineering problems sit between the physics and a working vehicle, and each of them is capable of consuming the entire benefit on its own.

First, the duty cycle is not a number

Consider what a traction battery actually experiences over a working day.

Average power draw across a city bus route might sit at 30–40 kW. Peak demand during acceleration can reach 200 kW or more. Regenerative braking sends a comparable spike back the other way. The pack spends the overwhelming majority of its time far below its rating, punctuated by brief, violent excursions.

Size that pack for average power and it fails on the first hill. Size it for peak and you are carrying — and paying for, and hauling the mass of — a great deal of cell capacity that exists purely to survive transients it sees for a small fraction of the duty cycle.

Both options are bad. That is the design tension hybrid storage resolves — and it is also why every number in this field is conditional on a duty cycle. A figure quoted without one is not an engineering number.

Problem one: sizing

The capacitor bank must be large enough to absorb the transients that matter and small enough not to dominate cost and volume.

That requires a real statistical characterisation of the target duty cycle — not a peak figure but a distribution. How often does a transient of each magnitude occur? How long does it last? What is the interval between them, and is the bank recovered by the time the next one arrives? A bank sized against a peak that occurs twice a day is mostly dead mass; a bank sized against the median is saturated when it is needed.

Get it wrong in either direction and the economics collapse. This is the least glamorous part of the work and the one that most often decides whether an installation is worth having.

Problem two: power conversion

Battery and capacitor operate at different voltages, and those voltages diverge continuously as the capacitor charges and discharges — a capacitor's voltage is its state of charge, which is the price of its electrostatic storage mechanism.

Coupling them usually needs a bidirectional DC-DC converter. That converter adds cost, mass, and losses of its own. Real converters run in the mid-90s percent, and every joule routed through the buffer pays that toll. A poorly chosen converter can consume more than the architecture saves.

Passive and semi-active topologies avoid the converter at the cost of control authority: the split between sources is then set by the relative impedances rather than by a decision. That is a genuine engineering trade with different answers for different applications, not a solved question with one right topology.

Problem three: control

The split between sources must be decided in real time, faster than the transient it is responding to.

Rule-based state-of-charge thresholds are robust, cheap to compute, and easy to certify. They are also leaving performance on the table, because they respond to the present instant rather than the shape of the demand.

Filter-based strategies split demand by frequency — high-frequency content to the capacitor, low-frequency to the battery — which matches the physical division of labour well but needs the corner frequency tuned to the duty cycle.

Model-predictive strategies perform better still and demand more computation and better system identification: they are only as good as the model of the pack they carry, and pack models drift as the pack ages.

The control law is where much of the achievable benefit is won or lost, and it is the substance of what the claims in this family actually cover.

Problem four: protection

Deep-discharging either component damages it, and the two fail differently.

Any serious architecture needs a state-of-charge floor with a disconnect — a mundane requirement that turns out to be central to whether the system survives real-world use rather than a test bench. It is also the requirement most likely to be omitted from a proof of concept and then discovered in the field.

What has actually been measured here

Two independent evaluations have been carried out on this hardware.

CIRT Pune tested a cranking configuration to 10,000 cycles at 60 °C (report No. B34000, 20 March 2020, Model HPP35A). DRDO's R&DE (Engineers) establishment conducted an environmental discharge evaluation at +70 °C with sustained 450 A draw (report No. 020623ESG/ETF, 21 June 2023).

Both establish something specific and useful: that the hardware holds up under sustained high-current draw at elevated temperature. Those are demanding conditions and surviving them is a real result.

They are also endurance and environmental evaluations of a single configuration. No baseline was run alongside either, so neither supports a comparative claim, and both are test reports rather than certifications. The next stage is comparative type-testing to IEC 62619 and IEC 62133-2 against a battery-only control on a defined duty cycle, which is where improvement figures would come from.

I set it out this way deliberately. Percentage improvements quoted without duty cycles or baselines are why OEM engineers discount this category before reading past the headline. Saying precisely what a test established, and what the next test will establish, is what earns a technical reader's attention — and it is the standard I would want applied to anyone else's numbers before I took them seriously.

The state of the field

Hybrid storage has moved from laboratory curiosity to active engineering across several fronts — heavy commercial vehicles with stop-start-dominated duty cycles, rail traction and regenerative braking recovery, marine and port equipment, grid frequency response, and uninterruptible power supplies where the transient is the entire specification.

Work is progressing in parallel across academic and industrial groups internationally, in India as much as anywhere. That breadth is good for the field. This is not a problem one architecture solves; it is a design space with genuinely different optima for different duty cycles, and the more configurations that get built and honestly measured, the faster the whole category matures.

What the field needs now is not more concepts. It is comparative data: identical duty cycles, hybrid versus battery-only controls, full instrumentation, published conditions. Until that body of evidence exists, every claim in this space — mine included — remains a physically-grounded expectation rather than a demonstrated result.

The physics is not in doubt. What remains to be established, application by application, is how much of the theoretical benefit survives contact with a real vehicle, a real converter and a real control loop. That is a solvable problem. It just requires the discipline to measure honestly and publish what the measurement says.

Measure everything. Validate everything. Improve continuously.

Cite this note
A referenced edition of this note, with a permanent DOI, is deposited on Zenodo: Rohera, H. K. (2026). What makes hybrid energy storage hard: sizing, power conversion, control and protection [Technical note]. Zenodo. https://doi.org/10.5281/zenodo.22700056

Hemant K. Rohera is an inventor and engineer in Pune, India, sole named inventor on three granted patent families — bioelectronic medical devices, hybrid energy storage and vehicle power electronics. The Hybrid Power Pack family is granted and in force as IN 301517 · US 10,523,019 B2 · EP 3 320 595 B1 (validated in Germany, the United Kingdom and Switzerland) · JP 6644883 · CA 2,991,527 C · MX 377263 B · VN 40455 · ZA 2018/00747 · EA 035682 (in force in Russia), priority 10 July 2015. The independent claim is chemistry- and voltage-agnostic. Available for licensing, assignment or joint development. ORCID: 0009-0005-3275-1743

References. Burke, A. (2007), Proceedings of the IEEE, 95(4), 806–820. de Vries, H., Nguyen, T. T. & Op het Veld, B. (2015), Microelectronics Reliability, 55(11), 2247–2253. Kötz, R. & Carlen, M. (2000), Electrochimica Acta, 45(15–16), 2483–2498. Vetter, J. et al. (2005), Journal of Power Sources, 147(1–2), 269–281. CIRT Pune, test report No. B34000, 20 March 2020. DRDO R&DE(E) Pune, environmental test report No. 020623ESG/ETF, 21 June 2023.

The complete patent register, with a grant number and grant date for every entry, and the research record, with abstracts and DOIs for the deposited notes, are published at hemant-rohera.vercel.app/patent-register.html and hemant-rohera.vercel.app/research.html.

Search This Blog

Hemant Rohera — Bioelectronics, Energy Storage and Power Electronics