Resistance

What it is (precise)

For a metallic conductor, resistance is geometry and material: R = ρL/A, resistivity times length over cross-sectional area. Copper's resistivity is about 1.7 × 10⁻⁸ Ω·m at 20 °C, which is why every practical BESS conductor lands in the microohm-to-milliohm range — a 300 mm² copper cable works out to roughly 0.06 mΩ per metre, a laser weld to tens of microohms.

Because the DC current path from cell to PCS is a single series chain, these small values add: the loop resistance a string presents is the sum of every cell, weld, busbar segment, contactor contact, fuse element and cable run in the circuit. The arithmetic that follows belongs to its own entries — Ohm's law for the voltage a resistance drops, I²R losses for the heat it makes, voltage drop for what the sag does to the operating window. This page is about where the ohms are and how they behave.

The chain contains two species of resistance, and they behave differently. Bulk resistance — cable lengths, busbar runs, transformer windings — is calculable from the drawings and stays where the design put it. Contact resistance at every bolted, sprung or plugged interface is different physics: current crosses a joint through microscopic contact spots whose total area depends on clamping force and surface condition, so the ohms depend on assembly torque, plating and cleanliness rather than on anything a drawing shows.

A joint can be mechanically tight and electrically poor. Both species share one coupling: copper's resistivity rises roughly 0.4% per degree Celsius — about 40% between 20 °C and 120 °C — so a resistance that makes heat raises its own value as it does.

Inside a cell the current path is electrochemical rather than metallic — current collectors, electrolyte, electrode interfaces — and it follows its own rules: strongly dependent on state of charge, temperature and the measurement method. That figure, its DCIR and ACIR measurement families and their conditions, is the internal resistance entry's subject, and its rise through life is resistance growth. What this entry adds is the rest of the plant: the cell's few tenths of a milliohm are one link in a chain the project must account for from electrode to point of interconnection.

Where the ohms live in a plant

Follow one ampere from an electrode toward the grid. It crosses the cell — a few tenths of a milliohm of DCIR for a modern 280-314 Ah LFP cell — then the terminal and tab welds at tens of microohms each, hundreds of them per module, each dissipating roughly a watt at the ~157 A a 314 Ah string carries at 0.5C.

Then module busbars, inter-module links, the rack busbar up to the high-voltage box, the contactor contacts and the fuse element, the string cable to the DC bus and the bus to the PCS terminals. It is one conduction path carrying full string current through every element, so a single degraded joint anywhere heats at the whole string's current squared, and each extra milliohm in the loop costs about 25 W and 0.16 V at that current — losses and sag the design never budgeted.

The AC side plays the same game at higher current. A 3,600 kVA PCS block at 690 V draws about 3,000 A per phase, so every milliohm per phase on the low-voltage run dissipates on the order of 27 kW across the three conductors (total conductor loss is 3 × I²R) — which is why the LV connection between PCS and step-up transformer is kept to metres of busbar and why the transformer sits beside the PCS rather than across the yard.

The transformer's copper loss is its winding resistance doing the same thing under another name. Beyond it, the medium-voltage collection system runs at a fraction of the current: for fixed power, current falls as 1/V and conductor loss as 1/V², the same argument that moved DC architectures from 1000 to 1500 VDC. Every voltage level in the one-line diagram is, among other things, a decision about how much current the resistances at that level will be asked to carry.

Measured, not assumed

The bulk ohms can be computed; the contact ohms have to be measured. That split is visible in the records a well-run project generates. On the production line, weld resistance is sampled in microohms — a bad weld is caught electrically long before it is caught thermally — and joint resistance measurements and torque checks appear in factory and commissioning records for exactly this reason.

At commissioning, a four-wire micro-ohmmeter (a ductor test) is put across bolted joints and contactor contacts, because a two-wire meter cannot resolve tens of microohms through its own lead resistance. The value of the baseline is the future: each recorded number is the denominator against which every later measurement decides whether a joint has degraded or was always like that.

In service, contacts drift one way. Clamped interfaces relax as conductors creep under bolt pressure and thermal cycling works the joint; surfaces oxidize; vibration frets plated contacts. The drift is self-reinforcing — a higher-resistance joint runs hotter at the same current, and heat raises resistivity and accelerates the surface degradation that caused it — which is why joint problems tend to accelerate rather than creep.

The practical instrument for finding them is an infrared camera: at equal current, a joint running warmer than its siblings is a higher resistance announcing itself, measurable from a distance with the plant on load. Periodic thermographic surveys and torque audits against the commissioning record exist to catch exactly this. On the DC side the BMS keeps its own version of the watch: a step change in one cell group's resistance estimate points at a connection or weld, not at chemistry.

How it shows up in specs, studies and contracts

Resistance enters the paperwork under several names. On the cell datasheet it is the DCIR line and the conditions behind it. In the factory acceptance record it is weld and joint resistance sampling; at commissioning it is the micro-ohmmeter log and the torque sheet. In the thermal design it is the heat-generation tables — the cooling plant is sized against the fleet's I²R at the design duty, so an ohm that was not in the table is heat the HVAC was never asked to remove.

And in the efficiency guarantee it is embedded rather than named: the round-trip figure at the point of interconnection includes every conductor, joint, winding and cell between the meter and the electrodes, which is one reason the measurement boundary decides whose ohms count against whose number.

Two studies read the same ohms in opposite directions. The cable-sizing and voltage-drop calculation asks whether resistance is low enough to hold voltage at the far end under load — the voltage drop entry carries the DC-window consequence. The DC short-circuit study asks the reverse: loop resistance is part of what bounds the prospective fault current the string fuses must clear, so a loop value that was assumed rather than measured props up protection settings too.

Contractually, the useful clauses are unglamorous — periodic thermography and torque audits in the O&M scope, owner access to the commissioning baselines and the BMS resistance histories — because when delivered efficiency drifts years later, those records are what separate cell aging from a neglected joint.

Common pitfalls

The first trap is dismissing the numbers for being small. Microohms and milliohms read as rounding errors until they are multiplied by the square of the current and the size of the fleet: roughly a watt per weld is hundreds of watts per module before the cells contribute anything; a stray milliohm in a string loop is ~25 W of unbudgeted heat; a milliohm per phase on a kiloamp LV run is tens of kilowatts. The unit of judgement is never the single joint — it is the population of joints carrying the plant's current all day, inside enclosures whose cooling was sized against the design tables.

The second family is measurement carelessness. Resistance moves with temperature — roughly 0.4% per degree for copper, far more for a cell between summer and winter — so a February reading compared against an August baseline reports weather as degradation; comparisons hold only at matched or corrected conditions, and the trend discipline itself belongs to resistance growth.

A cell datasheet DCIR is not the loop resistance of an installed string, so quoting one where a study needs the other understates every consequence downstream. And a passed commissioning test settles nothing permanently: the bulk conductors will measure the same in year fifteen, the contacts will not — which is exactly why the plant keeps measuring.