Lithium-Ion Battery Pack Design Simulation for Medical Devices

Introduction

When a phone battery dies, you lose a text. When a medical device battery dies, a patient may lose therapy, monitoring or time.

Portable ventilators, infusion pumps, patient monitors, portable ultrasound, defibrillators and diagnostic carts all depend on lithium-ion packs that must behave predictably. They are expected to report their remaining charge accurately, tolerate years of hospital cleaning and charging routines, and fail safely when something does go wrong, ideally with a clear alarm well before shutdown.

Simulation helps medical device teams answer many design questions before building hardware. The IST Li-ion Battery Simulator is a browser-based workbench that takes a pack from application targets through sizing, cell modelling, thermal analysis, BMS logic, ageing and a design report, with nothing to install. This post walks through how that workflow applies to medical device packs, and where its limits lie.

Where These Packs Are Used

• Life-support and respiratory devices: Portable ventilators and oxygen concentrators, where runtime must be dependable.
• Infusion and syringe pumps: Long, steady, low-power runtimes.
• Patient monitors and transport monitors: Moderate power, frequent moves between charging and use.
• Defibrillators and surgical tools: Short, high-power pulses.
• Portable diagnostics: Ultrasound, ECG carts and point-of-care analysers.
• Mobile medical carts and robots: Larger packs on wheels in hospital corridors.

What Makes Medical Packs Different

The main design pressures are:

• Reliability over peak performance: Predictable runtime matters more than the last few minutes of capacity.
• Accurate fuel gauging: Clinicians must trust the remaining time on screen.
• Safe failure modes: A fault should alarm and protect, ideally leaving time to switch to another power source.
• Surface and enclosure temperature: Devices touching patients or staff face tight temperature limits.
• Long service life: Hospitals expect packs to last for years and maintain predictable capacity.
• Cleaning and environment: Disinfection and spills drive enclosure and ingress-protection choices.

Step 1: Define the Application

The simulator begins with Define Application: voltage window, energy capacity, continuous and peak power, peak duration, charge power, ambient range, IP target, cooling method, weight budget, envelope and standards. The guide lists application types such as 2-Wheeler, 3-Wheeler, BEV and ESS, which select default usage profiles and standards. There is no medical-specific type described, so choose the closest available option and set your targets manually.

For a medical device, translate the use case into these fields. Continuous power represents normal operation, while peak power and duration represent events such as a defibrillator pulse or a pump motor start. As the guide advises, fill the voltage window and power targets carefully, because the most useful output is the list of targets your chosen cell and BMS cannot meet.

Step 2: Choose the Cell and BMS

The database holds cylindrical, prismatic and pouch cells in several chemistries, plus BMS boards, busbars, fuses, connectors, NTCs and other parts. You can import your own parts as .dat files. For medical packs, the main selection criteria are:

• Chemistry: The guide describes LFP as the safest and longest-lived common option, while NMC and NCA offer higher energy. The trade-off between energy density and thermal stability is central, and your risk assessment should guide the choice.
• Format and parallel count: Small cylindrical cells in several parallel strings give flexibility and some fault tolerance, while larger cells mean simpler pack assembly.
• DCIR and temperature limits: Lower DCIR means less heat and less voltage sag, and charge temperature limits matter in cold storage or transport.
• Cycle life and data quality: Prefer datasheet values and note any estimates, which the simulator flags.

On the BMS side, the simulator checks that the series range covers your Ns, that OVP and UVP sit inside the cell window, that continuous and peak current limits fit the load, that the over-temperature trip is inside the cell’s limits, and that enough NTC inputs, balancing current and the right communication protocol are available.

Step 3: Size the Pack

In Pack Configuration you enter series cells, parallel cells and module count, or ask the simulator to suggest them from your targets. It returns the cell model, a cell-to-BMS check, pack sizing, a pass/warn/fail check against your targets and a standards checklist.

An illustrative example. The numbers below use a hypothetical 3.5 Ah, 3.6 V, 30 mΩ, 48 g cylindrical cell and an assumed load. They are my own back-of-envelope calculations with the guide’s formulas, not simulator output, so substitute your own cell and device data.

• Configuration: 4S3P, giving 12 cells
• Nominal voltage: 4 × 3.6 = 14.4 V
• Capacity and energy: 3 × 3.5 Ah = 10.5 Ah, so about 151 Wh, or roughly 136 Wh usable at 90%
• Pack resistance: 4 × 30 mΩ ÷ 3 × 1.10 ≈ 44 mΩ
• Mass: 12 × 48 g = 576 g of cells; dividing by the guide’s 0.65 cylindrical packing ratio gives roughly 0.89 kg
• Runtime at an assumed 40 W average load: about 3.4 hours
• Heat at that load (about 2.8 A): only around 0.35 W

Now the peak check. A 120 W peak at a minimum operating voltage of about 10 V (4 × 2.5 V) needs roughly 12 A, or about 4 A per cell. The pack’s peak heat at that current is about 6 W, which is still modest, but the voltage sag and low-voltage cutoff behaviour matter, because a device that browns out during a motor start is a real clinical problem. The simulator’s target check would flag it if the cell, parallel count or BMS cannot supply that current at minimum voltage.

Step 4: Model the Cell

The BDS Cell Modeling module builds the models behind later analysis:

• HPPC-based equivalent-circuit extraction gives R0 and two RC pairs, which determine how voltage sags under load and recovers afterwards. That matters for low-battery alarm timing.
• EIS-based fitting and a reduced-order P2D (SPMe-type) electrochemical solver, benchmarked against PyBaMM’s DFN model within about 2 mV RMS up to roughly 6C.
• Joint parameter identification from a measured voltage trace.

When test data is not available, the simulator auto-generates inputs from the selected cell, labelled as synthetic. For a device where state-of-charge accuracy is a safety feature, replacing these with measured data for your exact cell is among the most valuable validation steps.

Step 5: Thermal Analysis

Medical packs generate little heat in normal use, but thermal behaviour still matters for three reasons: enclosed housings with no airflow, charging heat, and strict temperature limits on anything touching a patient or user.

The thermal tools include:

• Conjugate heat transfer and coolant flow and pressure drop calculations, useful for larger carts or liquid-cooled equipment.
• 3-D cooling simulation for natural convection, forced air through cell gaps or a liquid cold plate. It reports maximum temperature, cell-to-cell ΔT and energy balance. The guide suggests targets of Tmax at or below 45 °C and ΔT at or below 5 K.
• Thermal imaging live view for a staged cooling controller across modules.

Two honest limits apply. First, the 3-D cooling tool is not a Navier-Stokes solution. Second, it estimates pack temperatures, not the surface temperature of a finished device enclosure, so skin-contact limits need device-level thermal testing.

Step 6: Thermal Runaway and Propagation

In a medical device, a cell failure may occur inches from a patient or clinician, so containment is a priority. The propagation module models a module as a network of cells with per-cell abuse kinetics, internal short after separator melt, venting, and heat transfer through gaps, barriers, busbars and radiation.

You choose the trigger, test SOC, cooling state and barrier (none, 2 mm air, 1 mm mica, 2 mm aerogel or 3 mm ceramic fibre). Results show how many cells enter runaway, the time between the first and second event, and an automatic barrier comparison, so you can find the thinnest and lightest barrier that stops propagation or buys the time you need. An ARC-style runaway analysis, auto-generated from the selected cell, also shows self-heating onset and the margin to the BMS over-temperature trip. The guide’s 5-minute proxy comes from road-vehicle practice, so treat it as a screening reference rather than a medical requirement, and confirm the final design with physical testing.

Step 7: BMS Design and Fault Testing

For medical devices, the BMS is a safety component, so verification carries extra weight:

• Protection and balancing simulation covers OV, UV, over-current, short circuit, over-temperature and sensor faults.
• SOC and SOH estimation (EKF SOC with resistance tracking) supports accurate remaining-runtime display. The guide’s validation example reports an EKF SOC RMS error of 0.4% on a specific drive excerpt, which is a result for that test, not a guarantee for your device.
• Fault-injection matrix tests over-charge, over-discharge, over-current, external short, cooling failure, cold charging and open sensors, with detection latency against requirement, contactor action and a diagnostic code. These help you design alarm behaviour and fault reporting.
• Code generation with SIL and HIL produces C code, CAN messages and test vectors. A SIL pass means the generated C reproduces the simulator’s logic. The guide is clear that this is template-based, MISRA-oriented, not certified and has no ISO 26262 tool qualification, and that short-circuit protection must also exist in hardware. Medical software lifecycle requirements are a separate matter that this tool does not claim to cover.

One more limitation: the built-in drive cycles are road-vehicle profiles, and the duty-cycle card runs an EV day or an ESS day. Neither describes a ventilator or infusion pump, so you represent your device through power targets, peak duration and your own duty assumptions.

Step 8: Electrical Design

The Circuit Design tab derives design current, a temperature-derated fuse, busbar cross-section, pre-charge sizing and minimum isolation, and checks fuse breaking capacity against short-circuit current. The DFM card builds module geometry from the real cell size with a tolerance stack-up, helpful when the pack must fit a slim handheld or a sealed instrument bay. Module wiring diagrams show series-parallel connections, balancing taps and NTC positions, with SVG export. For lower-voltage packs, note that the simulator’s class A and class B distinction (class B above 60 V) determines whether contactors and pre-charge circuits appear in the schematic.

Step 9: Life, Reliability and Replacement Planning

Hospitals plan battery replacement as a maintenance activity, so predicting life has direct value.

• Calendar and cycle ageing models SEI growth, particle cracking and lithium plating, with the capacity knee and resistance growth. Calendar ageing is faster at high state of charge and temperature, which describes a device that sits on a charger at full charge for most of its life. The model is calibrated to the cell’s datasheet cycle-life rating.
• Warranty and life prediction accepts cycles per day and depth of discharge for stationary-style use, and estimates SOH at end of warranty and the longest warranty that keeps claims under your target. Its city climate inputs are written for Indian cities, so adjust the assumptions to your own markets.

The guide cautions that ageing results are a screening estimate until fitted to your own storage and cycling tests. For a medical device, that fitting step is especially important before committing to a service-life claim.

Step 10: Safety Checks and the Design Report

The Safety & Compliance tab covers capacity, resistance, isolation, overcharge, short circuit, drop, shock, vibration and protection response time, then rolls them into a release decision where any failed or un-run gate blocks release. The standards checklist can include UN 38.3 for transport and IEC 62619, which is an industrial standard. Medical-specific requirements are not built into the checklist, so confirm the applicable medical device and battery standards, and your regulatory pathway, separately with your regulatory team and a test lab.

Reports then runs the full chain and produces an A4 design report with the schematic, simulation results, standards checklist and bill of quantities, which can support your design documentation.

Questions Simulation Helps Answer

• Does this cell and parallel count deliver the required runtime, with margin for ageing?
• Will voltage sag during a motor start or pulse trigger a premature low-voltage cutoff?
• How hot does the pack run on the charger and in a sealed housing?
• What is the thinnest barrier that stops a single failing cell from spreading?
• Does the BMS detect each fault quickly enough, and what alarm does the device see?
• How much capacity will the pack retain after several years sitting on a charger?
• Does the pack fit the envelope with room for tolerances?

What the Simulator Is, and Isn’t

It is a design and screening tool. Pack arithmetic follows standard formulas, while thermal, abuse, ageing and system models are reduced-order. It does not replace the physical tests and documentation needed for medical device approval, and it does not model enclosure surface temperature or medical software requirements directly. Ageing results are a screening estimate until fitted to test data.

A sensible validation path:

1. Replace synthetic HPPC, EIS and ARC data with measured data for your chosen cell.
2. Test a prototype pack for capacity, DCIR and temperature at your real load profile.
3. Run SIL, then HIL, then hardware protection tests on the BMS, and verify alarm behaviour.
4. Complete cell and pack safety testing, and meet your applicable medical device and battery standards, with accredited labs.
5. Start calendar-ageing and cycling tests early, then refit the ageing model before making service-life claims.

Conclusion

Medical device packs are judged by what they do on the worst day: when the load is highest, the pack is oldest, or something has gone wrong. Simulation lets teams stress those moments in advance, comparing cells, parallel counts, barriers and BMS settings before committing to hardware, and building a documented basis for the design.

Explore pack design at Academy of EV Technology and follow the step-by-step guide from mission targets to a full design report.