BetaWA — Battery Pack Optimizer

How to Build a Balanced Battery Pack from Used Lithium Cells

A practical 14S4P example: matching 56 second-life lithium cells by measured capacity.

Reusing lithium-ion cells can extend their useful life and reduce battery waste. However, cells recovered from an aging pack often have different remaining capacities. When those cells are grouped into a new series-parallel battery, the distribution matters: a lower-capacity series group can reach its operating limit before the others and restrict usable pack capacity.

Capacity matching aims to make the total measured capacities of parallel groups as similar as possible. Below, we show a real BetaWA optimization result from a synthetic 56-cell dataset, including an important limitation: the requested tolerance was not fully achieved.

1. What Does 14S4P Mean?

A 14S4P pack has 14 groups connected in series (14S), with four compatible cells connected in parallel in each group (4P). The total inventory needed for one complete assembly is:

14 × 4 = 56 cells

With cells having a nominal voltage of approximately 3.7 V, the nominal pack voltage is about 51.8 V. Actual operating voltage limits depend on chemistry and the equipment used; “48 V class” does not by itself define a safe charging or discharging range.

For compatible parallel cells, the group's measured capacity is approximately the sum of the constituent cell capacities, under comparable test and operating conditions. As an illustration, 48.5 + 50.2 + 51.1 + 49.7 Ah gives a 199.5 Ah parallel group. In a series string, each group carries the same current, so a lower-capacity group can become the limiting group.

2. Why Arbitrary Grouping Can Waste Usable Capacity

Suppose 56 tested cells have capacities between approximately 46 and 54 Ah. If four consecutive inventory entries are assigned to each group, some groups may contain mostly stronger cells while others contain mostly weaker cells. All cells are used, but the 14 group totals can differ considerably.

Matching seeks combinations in which stronger and weaker capacities compensate for each other. This is a mathematical grouping problem, not a substitute for cell qualification or electrical design.

3. Demonstration Dataset: 56 Cells

We prepared a synthetic dataset of 56 cells with varied capacities. These are simulated values, not measurements from a real battery. We then ran the operational BetaWA Pascal optimization engine on that dataset.

Input parameterValue
Cells in inventory56
Minimum measured capacity46.11 Ah
Maximum measured capacity53.97 Ah
Mean cell capacity49.578 Ah
Sample standard deviation2.262 Ah
Target configuration14S4P
Requested spread tolerance0.10%

To reproduce the baseline, group the input cells in the order they appear in the CSV: rows 1–4 form Group 1, rows 5–8 form Group 2, and so on.

4. Before Optimization: 6.681% Capacity Spread

Grouping four consecutive cells at a time produced 14 groups with total capacities ranging from 192.03 Ah to 205.28 Ah. The mean group capacity was 198.3114 Ah.

We calculate capacity spread as:

Capacity Spread (%) = (Maximum Group Capacity − Minimum Group Capacity) ÷ Mean Group Capacity × 100

For this baseline: (205.28 − 192.03) ÷ 198.3114 × 100 = 6.681%.

This baseline is simply consecutive grouping in the CSV input order. It is not a comparison against another optimization algorithm, and another initial order could produce a different baseline.

5. After BetaWA Optimization: 0.1412% Capacity Spread

We uploaded the same 56-cell inventory to BetaWA, selected NMC/NCA, 14S4P, one requested battery, and a requested capacity-spread tolerance of 0.10%. The Pascal engine returned one complete assembly using all 56 cells.

Before (CSV-order baseline)6.681%
After BetaWA optimization0.1412%
Optimization metricResult
Complete assemblies1
Cells used56
Unused cells0
Parallel groups14
Minimum group capacity198.13 Ah
Maximum group capacity198.41 Ah
Mean group capacity198.3114 Ah
Actual capacity spread0.1412%
Requested tolerance met?No

The spread decreased by approximately 47 times compared with the consecutive-grouping baseline, without discarding any cells.

Important result: The optimization produced a complete 14S4P assembly, but 0.1412% is above the requested 0.10% tolerance. BetaWA reported that the requested accuracy was not achieved. A completed assembly should not be confused with one that meets every requested acceptance criterion.

6. What Better Capacity Matching Can — and Cannot — Do

More uniform measured group capacities can reduce capacity-related imbalance between series groups. However, the usable energy and performance of an actual battery also depend on discharge current, temperature, voltage limits, internal resistance, aging, and BMS behavior. A small capacity spread alone does not guarantee a reliable or safe pack.

7. Safety Considerations for Second-Life Lithium Cells

Capacity matching is not a complete safety assessment.

Before considering reuse, qualified personnel must evaluate cell chemistry and voltage compatibility, internal resistance, self-discharge, physical condition, temperature behavior, protection, electrical connections, and the BMS. Cells at different voltages can produce dangerous equalization currents when connected in parallel.

This example demonstrates mathematical grouping only. It is not an assembly procedure, an electrical design, or a safety certification.

8. Try Capacity Matching with BetaWA

BetaWA helps organize measured second-life lithium cells or compatible modules into series-parallel groups based on capacity. Upload a CSV inventory with unique cell identifiers and measured capacities, choose a configuration, and review the proposed assembly map, group-capacity spread, and tolerance warnings.

The current Simple Mode focuses on capacity-based grouping. Additional cell qualification and safety engineering remain necessary.

Try BetaWA Optimizer

Conclusion

In this synthetic 56-cell demonstration, BetaWA's operational Pascal engine reduced the capacity spread of a 14S4P assembly from 6.681% to 0.1412%, while using every cell. The requested 0.10% tolerance was not met, illustrating why transparent reporting matters as much as optimization itself.

Capacity matching is a useful engineering aid for second-life battery projects, but it must be combined with cell qualification, protection, and validated battery design.

Demonstration note: Input capacities are synthetic; the optimized assignments and output metrics were generated by the BetaWA Pascal calculation engine. Results vary with the inventory and configuration.

Related reading: Battery Cell Matching by Capacity.

Related article: How to reuse Nissan Leaf battery cells for 24V and 48V battery packs