Cell Selection Starts With the Pack – Not the Datasheet

Most cell selection still begins with datasheets, because they are the only information engineers can easily access early on. That is understandable, but it is also why teams spend months testing “promising” cells that later fail in a real pack. The root issue is simple: a cell is not a battery pack. Effective selection requires balancing four competing dimensions that datasheets rarely capture together: energy density, power density, total cost of ownership, and capital cost. Higher energy density reduces mass and volume but often limits power capability. High power cells sacrifice energy. Total cost of ownership includes lifetime, safety margins, thermal derating, and reliability, while capital cost extends beyond cell price to pack materials, cooling, and integration complexity.

These dimensions inherently trade off against one another. There is no universally “best” cell, only the best compromise for a specific application. Different industries weight these trade-offs very differently, which is why cell selection cannot be a datasheet exercise. It must start with pack-level targets and assess how candidates perform once integrated into a real pack, under real power profiles and operating constraints.

The real objective: meet pack-level targets, not cell-level headlines

The right starting point is pack-level metrics and limits. Before arguing about chemistry or form factor, define the system envelope:

• Volume and mass limits (pack packaging and mass budget)
• Voltage ceiling (maximum fully charged voltage)
• Energy target (range or runtime)
• Peak and sustained power targets (take-off, acceleration, or fast charge)

These constraints determine what is physically feasible and they strongly influence the optimal series and parallel configuration. Different cells can produce different optimal pack architectures even under the same headline requirements, so “like-for-like” comparison requires building the pack around each candidate and then comparing packs.

For a drone example this exercise would involve specifying the following:

Choose the optimal cell based on performance and economics

Once the pack envelope is defined, you can filter a broad set of candidates, then evaluate them on the metrics that matter commercially:

• Pack-level usable energy delivered under the duty cycle
• Pack-level power capability with voltage and temperature constraints
• Thermal response and heat generation that drives cooling cost and derating risk

Then translate outcomes into economics: fewer cells, lower thermal management burden, and less oversizing typically reduces cost and complexity. This is where the “best” cell on a datasheet often loses to a cell that integrates better into the pack.

Use simulation to calculate usable energy under the real power profile

Usable energy is not nominal capacity. It is the energy you can actually extract before hitting cut-offs. In practice there are multiple cut-offs and they interact:

• Electrical cut-offs (voltage limits under load, especially during pulses)
• Thermal cut-offs (temperature rise and allowable maximum temperature)
• Electrochemical limits (rate capability, polarisation, SOC window effects)

Simulating pack candidates under the application power profile lets you quantify accessible energy, temperature rise, and pulse power performance across SOC and temperature ranges, without waiting for months of lab work.

About:Energy’s The Voltt offers webapplications that function like “virtual battery cyclers” where one can run simulations in the Cloud to quantify the pack-level metrics that matter when selecting cells.

The Simulations below were created using The Voltt and show the voltage and temperature response of 2 different pack configurations for 2 different cells (Amprius SA88 and SA08). Each is subject to a representative drone power profile with a take-off, cruise, and landing section that repeats. Power profiles where scaled according to the mass of the battery pack.

By comparing the voltage results for pack 1 against those for pack 2 we can see the impact of having 4 additional cells in the pack, allowing us to decide whether they are worth the added cost, volume, or added mass. We can also see that the SA08 seems to outperform the SA88 in terms of available energy, but the SA88 performs better thermally.

A quick 3-step method

1. Define pack envelope: mass, volume, voltage limit, energy target, peak power, operating temperatures.
2. Design candidate packs: select plausible cells, compute feasible series and parallel counts under the envelope.
3. Simulate the duty cycle: rank candidates by usable energy, thermal headroom, and pulse power margin, not datasheet claims.

Beyond performance metrics

Cell selection does not end with energy, power, and cost modelling. Practical factors such as supply availability, manufacturing consistency, lead times, supplier technical support materially affect programme risk. These factors often decide whether a design can scale, be certified, and be supported over its intended lifetime, and should be evaluated alongside performance and economics, not after them.

What comes next in pack design

Once a cell is downselected, the next steps are pack architecture (series parallel finalisation), thermal strategy, mechanical integration, BMS limits, validation testing, and eventually lifetime and safety qualification. The best teams pull forward pack-level truth early, then spend physical resource proving a design rather than discovering basic feasibility.

About:Energy offers it’s cell models as Simulink implementations where one can combine several cell models into module and pack-level simulations for detailed battery pack design or BMS development.