Circulating Current in Parallel Battery Strings: Why It Happens and How BMS Architecture Prevents It | JKESS

The Expansion Scenario That Goes Wrong

Consider an illustrative commercial and industrial (C&I) expansion: a factory installs a 128kWh battery system in 2024 and needs 256kWh in 2026 as its load grows. Procurement proposes buying another identical system and connecting it in parallel. Three months later, the operations team notices persistent current imbalance and a faster reported SOC decline in one string. The model numbers match, but the batteries no longer have the same operating history.

Circulating current is one possible explanation, alongside unequal load sharing, temperature differences, connection resistance, sensor error, or cell faults. New and aged strings can have different capacities and internal resistances, so their internal voltage and SOC trajectories diverge during operation even if their initial SOC readings match. A higher source voltage can then drive current into the other string. Check signed branch currents and cell data before diagnosing the cause.

Uncontrolled internal transfer creates resistive losses and extra battery throughput. It can increase thermal and cycling stress, cause cell-voltage limits to be reached, and trigger BMS protection; sufficiently high currents may operate a fuse or breaker. Some problems appear at the instant of connection, while others become clearer over weeks of operating data. This guide explains the physics, three engineering countermeasures, and the role of JKESS master and slave hardware in an engineered parallel-string design.

The Physics: Why Circulating Current Happens

Circulating current does not by itself prove poor battery quality. It follows circuit laws when parallel sources have different internal voltages. A useful first estimate for two strings at zero external load is:

I_circulating = ΔV / R_loop

Once connected, the strings share a common bus voltage, apart from interconnect voltage drops. The driving mismatch is between their equivalent sources, not two different voltages at the same ideal bus node. Under external load, each branch current includes load sharing as well as the effect of voltage mismatch. This resistive estimate does not model connection transients, inductance, or electrochemical relaxation.

2V ÷ 0.05Ω = 40A

P_loss = I² × R_loop       = 40² × 0.05 = 80W

Engineering reference: Weng et al., research on current imbalance in dissimilar parallel-connected batteries, develops an equivalent-circuit model and examines how capacity and resistance differences affect current, SOC, and aging. The numerical example above is an illustrative circuit calculation, not a product test result.

Unprotected Parallel vs Independent String Protection

Independent branch protection makes faults easier to detect and isolate. It must be paired with per-string sensing and a coordinated control response; protection devices alone do not eliminate circulating current.

If a string disconnects, the remaining strings can receive more current. The EMS/PCS must reduce power or stop operation when necessary to prevent a cascading shutdown. Independent protection is not permission for live maintenance.

The 3 Engineering Countermeasures

1. Equip every string with independent breakers and fuses

   The purpose of branch protection is selective fault isolation. Specify a DC-rated breaker and fuse for each string, together with BMS-controlled contactors or a suitable automatic trip mechanism. A moderate circulating current can remain below an overcurrent trip threshold, so a fuse is not a substitute for controlled isolation.

   Selection priorities: coordinate the continuous-current rating, time-current curves, DC voltage rating, available fault current, breaking capacity, conductor protection, and permitted operating peaks. Evaluate fault contributions from every connected string. Selective interruption and continued operation must be verified at system level.

   Installation location: use the approved single-line diagram and grounding arrangement to determine which conductors need protection and isolation. There is no universal rule that the breaker belongs on the positive conductor and the fuse on the negative conductor. Confirm the pre-charge and connection sequence for the delivered equipment; a DC-link pre-charge circuit is not automatically rated to equalize mismatched battery strings.

2. Monitor each string in charge, discharge, and standby

   Place an independent bidirectional current-sensing channel on every parallel string and retain its sign. A system with only a total-current sensor cannot identify the internal transfer hidden by summation. For example, with discharge positive, +40A in one string and −40A in another gives approximately 0A at the external connection.

   Log current with cell voltages, string voltage, temperature, SOC, contactor state, and operating mode. Configure thresholds and time delays around sensor accuracy, expected load sharing, string ratings, and the approved control strategy. Respond with alarms, connection inhibition, power limits, or isolation as appropriate; setting external power to zero may not stop internal current.

   For a two-string comparison with currents in the same direction and above the measurement floor, one descriptive imbalance metric is:

   Current imbalance (%) =100 × |I1 − I2| / max(|I1|, |I2|)

   Near standby, use signed amperes and absolute thresholds. Percentages become unstable near zero. Opposite current directions require separate evaluation, and unequal-capacity strings may legitimately carry different currents; compare each branch with its expected allocation and limits.

3. Use master-level supervision to limit SOC and voltage mismatch

   SOC mismatch is one contributor to circulating current, but voltage response also depends on resistance, temperature, cell condition, and recent operation. The string masters and system supervisor should exchange validated measurements and limits through the approved CAN/RS485 protocol, then enforce voltage and SOC matching criteria before connecting a branch.

   During operation, supervisory control can review deviations, reduce system power, inhibit reconnection, or isolate an unsuitable branch. A 5-percentage-point SOC deviation can be an illustrative review trigger only if accepted for the project; it is not a universal connection threshold. On a directly paralleled DC bus, BMS software cannot independently set each string’s continuous current without suitable power hardware, such as a per-string DC/DC stage.

   JKESS describes daisy-chain management of 4–16 slave boxes per master. Combine the appropriate module-management hardware with a defined protection and sensing channel for every parallel string, and confirm the cross-string supervisory functions for the selected firmware and system design. Cell-level active balancing supports consistency within modules; it does not replace inter-string current control.

Architecture reference: Parallel-string application note: protection and topology considerations (PDF). This external engineering reference discusses independent string monitoring, automatic isolation, and topology tradeoffs; it is not evidence of a JKESS product feature or compatibility claim.

How JKESS HV BESS Architecture Manages Parallel Strings

A master box, slave BMUs, and a system-level controller serve different roles. The JKESS master-plus-slave platform provides module measurements, communication, and protection building blocks. A parallel-string project must also define the sensing, isolation, and supervisory control of each complete string.

Configuration matters. The master product page publishes two series-cell ranges, so this guide does not treat them as interchangeable. Energy capacity is determined by the selected cells and electrical configuration, not by box count alone. Match the purchase specification to the exact hardware, firmware, and approved system drawing.

Pre-Expansion Checklist: 8 Items Before You Parallel a New String

1. Verify electrical matching. Confirm identical approved chemistry and series cell count, compatible voltage windows, and correct polarity. Measure rested string OCVs and use the manufacturer-approved connection limit; ≤2V is not universally safe in a low-resistance loop.
2. Review the cells and their condition. Compare model, batch records, capacity, internal resistance, age, state of health, and cell-voltage spread. A matching product label or initial SOC reading does not establish compatibility.
3. Confirm independent branch protection. Verify each string’s breaker, fuse, controlled isolation path, fault-current coordination, and approved pre-charge/connection arrangement.
4. Validate current sensing. Confirm independent bidirectional sensing for every string, including sensor polarity, zero offset, accuracy, data logging, and standby measurement capability.
5. Verify the supervisory strategy. Confirm SOC and voltage acceptance limits, connection permissions, current limits, communication-loss responses, and power reduction after a string disconnects.
6. Check thermal conditions. Compare cell and ambient temperatures and inspect airflow or liquid-cooling performance. A temperature difference of ≤3°C can be a project target, not a universal pass/fail threshold; use the approved thermal criteria.
7. Complete the specified insulation checks. Qualified personnel must use the equipment-approved isolation procedure, test voltage, and acceptance resistance. A criterion such as ≥1 MΩ at 1,000 VDC applies only when explicitly specified; do not apply a high-voltage insulation test across connected electronics unless the procedure allows it.
8. Plan commissioning and follow-up monitoring. Set alarms before connection, verify the approved energization sequence, and log branch current, SOC, cell voltages, and temperatures during an initial 72-hour observation period. Continue trend monitoring afterward.

5 Common Mistakes When Paralleling Battery Strings

1. 1. Directly paralleling strings at different SOCs

   A full new string and an existing string at 80% SOC can exchange a substantial current on connection. Match voltage and SOC using the approved procedure, and check cell limits as well as pack voltage. Do not use the main contactor to equalize an unapproved mismatch.

2. 2. Protecting the entire group with only one breaker

   A shared breaker cannot provide selective isolation for each branch. Specify coordinated per-string breakers and fuses plus a controlled disconnection mechanism. Moderate internal currents may never reach a fuse’s operating threshold.

3. 3. Monitoring only total battery current

   The total can look normal while one string discharges into another. Provide bidirectional per-string measurement and examine current direction, absolute magnitude, and expected load sharing in every operating mode.

4. 4. Ignoring temperature differences

   Uneven cooling changes resistance, available capacity, and SOC behavior. Inspect measured cell temperatures and thermal trends, not just room temperature. Keep the installation within its approved temperature-uniformity limits; 3°C is not a universal acceptance rule.

5. 5. Skipping the first 72 hours of trend review

   A brief startup test may miss behavior that appears at a different SOC, load, or temperature. Review signed branch currents and associated alarms during an initial 72-hour observation period, then continue long-term monitoring. Neither the first connection nor 72 hours without alarms proves lifetime compatibility.

Frequently Asked Questions