Resistance growth

What it is (precise)

Resistance growth is a trend, not a value: the ratio of a cell's internal resistance today to the same measurement at beginning of life. The measurement itself — DCIR, its variants and its conditions — belongs to the Internal resistance entry; what matters here is that the comparison is only meaningful when both numbers share the same state of charge, temperature, pulse duration and current direction.

Internal resistance moves with every one of those variables, and it moves a long way — cold alone can multiply it — so an unmatched pair of measurements produces apparent growth that is really weather. Serious tracking states its reference conditions once and then holds them, or corrects to them, for twenty years.

The baseline deserves more care than it usually gets. The cell datasheet's DCIR line is a beginning-of-life figure at lab reference conditions and a vendor-chosen pulse; the project's own anchor is what was actually recorded — cell-sample DCIR at factory acceptance, rack- and string-level impedance signatures at commissioning.

That record is the denominator of every growth figure the plant will ever compute, and it is what a resistance-based State of Health definition is referenced to: where capacity-based SOH compares present capacity to the beginning-of-life value, the resistance-based counterpart compares present resistance to the beginning-of-life value at matched conditions, quoted as a ratio or a percent growth. Lose the baseline and the plant can still measure resistance — it just can no longer prove growth.

Where it comes from and how it trends

The mechanisms overlap with capacity fade without being identical to it. The solid-electrolyte interphase on the anode thickens through life, and a thicker SEI is both a lithium consumer — capacity fade — and an added impedance in the ion's path — resistance growth. Around it, electrolyte is slowly consumed and its transport properties degrade, raising the ionic resistance of the pore network; and the mechanical members of the current path age too — particle contacts, binder, welds, terminals — raising the electronic contribution.

Heat and time at high state of charge accelerate the chemistry, the same accelerants that drive calendar fade. The full mechanism story is told in the degradation article; the point that matters here is the weighting: the same processes feed both trends in different proportions, which is why the resistance curve and the capacity curve are correlated and yet refuse to be proportional.

The shape of the trend is what catches projects out. Through early and mid life, resistance creeps — slowly enough that design margins absorb it without anyone noticing. Late in life the growth commonly steepens, and the percent rise in resistance can overtake the percent loss in capacity: this is why a cell can fade in power faster than it fades in energy, and why a plant can reach a practical power or efficiency end of life before its energy end of life at the contractual 65-70% capacity floor.

The compounding runs one way — every downstream consequence is computed on the aged value, so the years in which the plant has the least capacity to spare are also the years it pays the most for resistance.

Why it matters in a real grid-scale project

Heat first, because the law is linear in exactly the variable that grows. Ohmic heat per cell is I²R — quadratic in current, linear in resistance — so at the same dispatch, heat tracks resistance growth one for one: 30% more resistance is 30% more heat from the identical megawatt profile. The Internal resistance entry works the beginning-of-life arithmetic at an assumed 0.25 mΩ: on the order of 30 kW of ohmic heat in a 5 MWh-class container at 0.5C.

Grow the fleet's resistance 30% — an illustration, not a warranty figure — and the same container on the same duty is rejecting roughly 40 kW instead, landed on a cooling system sized against beginning-of-life heat tables. The knock-ons are quiet but real: more auxiliary energy spent pumping and chilling, warmer average cell temperatures, and — since calendar aging roughly doubles per +10 °C of sustained cell temperature — a thermal feedback in which the aged fleet ages faster unless the cooling holds the line.

Efficiency is the same growth read in the energy account. The ohmic loss fraction per direction is roughly I × R over cell voltage — linear in resistance at a given current, and paid twice per cycle — so round-trip efficiency drifts down as resistance grows, typically a point or two from beginning of life to end of life.

A point of RTE is not cosmetic: it is charging energy bought and not resold, every cycle, for the rest of the project. It is also where resistance growth first becomes contractually visible, because an RTE retest will show the drift while the energy capacity test can still pass.

The third reading is voltage. Sag under load is I × R, so an aged cell drops further at the same current and the limiting rack meets its voltage floor earlier in the discharge — earlier still when cold raises resistance further, with near-empty state of charge stacked on top.

The system-level consequence — how many MW the plant can hold, for how long, in which season and which year — is Power fade, which has its own entry; what resistance growth contributes is the mechanism: the same amps, a bigger voltage drop, a limit reached sooner. It also stresses the battery-PCS pairing over life: a DC-bus margin that cleared the PCS voltage window comfortably at year one can fall out of it by year ten as deeper sag drags the bus down on a hard discharge.

How it shows up in specs, tests and contracts

The uncomfortable fact is that the plant's contractual instruments are written on the other axis. Capacity warranties and degradation reports track retained MWh; End of Life is defined at a capacity floor; the annual capacity test discharges the window and measures energy. Resistance appears on the cell datasheet as a beginning-of-life DCIR line and then largely vanishes from the paperwork — warranties that carry a resistance or power-capability metric are the exception.

So the growth surfaces indirectly: in RTE retest results, in heat-generation tables that stop matching the cooling plant's reality, in a late-life shortfall against duration at rated power. When delivered efficiency or peak power drifts across the years, resistance growth is usually the physics doing the work, argued under clauses that never name it.

What a well-run project does about it: record a proper baseline — factory-acceptance DCIR sampling on cells, commissioning impedance records at rack and string level — then trend against it at matched or corrected conditions, using the per-group resistance estimates the BMS computes anyway.

Secure owner access to that BMS history in the contract; it is the earliest evidence available in a warranty dispute and the input a resistance-aware degradation model needs. And where the revenue case depends on rated MW for the full duration late in life, write the test that checks it — a duration-at-rated-power or power-capability test in the later-year regime, explicitly, because the energy-denominated regime does not watch power by default.

Common pitfalls

The classic false alarm is condition mismatch dressed up as aging. A February rack measurement compared against an August commissioning baseline shows dramatic apparent growth that is temperature, not age — cold raises resistance reversibly, aging raises it permanently, and only matched or temperature-corrected comparisons separate the two. The same trap runs through state of charge and pulse duration: trend data assembled from measurements taken wherever the plant happened to be sitting is noise with a slope.

The opposite error is deriving resistance from capacity. There is no fixed exchange rate between the two trends — 90% capacity SOH does not imply any particular resistance growth, and a degradation model that scales resistance off the capacity curve is inventing the number that decides late-life heat, efficiency and power. The two axes need separate measurements and separate trend lines, anchored to the same commissioning baseline.

And fleet averages hide the cell that matters. Growth is never uniform: the group aging fastest runs hottest under the same current and sags first, so the limiting rack — not the average — decides when the discharge terminates. A widening resistance spread across the fleet is a finding in its own right even while the mean still looks acceptable; average-only reporting files that early warning under noise.