Battery Engineering


Battery swelling, inaccurate range, poor performance, and inability to claim warranty are some of the bigger problems that E2W users face. Building a performant, safe and reliable battery is extremely important since it forms the core of the electric scooter. The battery mainly consists of:

  1. Cells
  2. BMS and 
  3. Battery casing

Building custom cells means going into the chemistry of the cells, which is currently beyond our scope since it is capital intensive. We are focussed on building a better battery pack by managing the cells better using a proprietary BMS and building a thermally efficient and mechanically tested casing.

Based on our motor controller requirements of 60V input voltage and peak discharge current of 32A, kWh battery with 13S 14P is built using 18650 Li-ion NMC cells which means 13 series 14 parallel, where each parallel string has 14 cells and 13 similar strings are stacked on top of each other that are connected in series. So, a total of 182 cells are used per battery pack. Below is a quick specification of the battery.

  • Nominal Voltage: 48V (converted to 60V using custom DC to DC converter)
  • Pack capacity: 33.6 Ah
  • Battery operating range: 35.75 V to 54.6 V
  • Charging Voltage: 54.6 V

NMC Cells

A battery pack is characterised by the cells that are used to build it. To build a battery pack for E2W we need to select cells that have the single most essential characteristic: High Energy Density. To maximize this characteristic, we need a cell that has a higher capacity, a smaller form factor, and is lighter in weight. We compared various battery chemistries that offer this characteristic, and the most prominent ones were Lithium-Ion (Li-ion) cells.

Considering a small factor, being cost-effective, and being readily available, we selected 18650 form factor for our Li-ion cells. Let's compare Li-ion chemistries:

The NMC cathode chemistry is widely used in electric vehicle applications, purely because of its exceptional power and energy density. Most importantly, NMC offers excellent performance under high temperatures where most other battery chemistry fails. Furthermore, the cycle life offered by NMC is far better and acceptable compared to its counterparts.

This is the reason behind our considering the NMC cells for our battery.

BAK cells

Our powertrain is driven by a 60V Hub motor which peaks at 100 Nm of torque and requires a peak of 32 amperes of current from the battery. To meet the > 100 kilometers range requirements, we performed MATLAB simulations for the urban driving cycle and we observed 22Wh/km energy consumption.

Having studied the cell characteristics and cost of Panasonic, LG Chem, and BAK cells, BAK cells seemed a good choice, as kWh first scooter is a mid-speed range vehicle and the instantaneous current needs are not that high. A Panasonic cell would have increased the cost of the battery by 2x.

BAK Grade A 18650 cells have the following specifications:

  • Nominal voltage: 3.6V
  • Nominal capacity: 2400 mAh
  • Charge cut-off voltage: 4.20 V
  • Discharge cut-off voltage: 2.75 V
  • Max charge current rate: 1C 
  • Max discharge current rate: 3C 
  • Internal resistance  ≤30 mΩ(AC Impedance, 1000 Hz)
  • Weight ≤ 48g
  • Cell Diameter: 18.55 mm (maximum)
  • Cell Height: 65.10 mm (maximum)
  • Average cycle life: 800 to 1000 cycles

Proprietary Battery Lattice Structure

Thermal management is a big consideration for 2 Wheelers. The only way to cool cells is to air cool or manage the thermal temperature of the battery pack in such a way that the operating temperature stays below the safe operating temperature range. 

After collecting the thermal data of the battery pack and analysing thermal air flow using CFD we are currently working on a lattice structure for our cells that dissipates heat in a highly efficient manner. Two factors are in consideration:

  1. A lattice structure of cells that can be compared to a modified cuboid or an elliptical shape to ensure the airflow is regulated.
  2. A non-symmetrical cell spacing in the battery pack so that there is enough space between parts of the battery that get hot during extremes.

The Fusion of Multiple Cells

We have identified that the battery pack can be made thermally better by reducing thermal hotspots. We validated using our data collection from sensors, that the middle portion of the battery is under thermal constraints due to lack of airflow, hence generating a hotspot. We plan to replace the hotspots with very few thermally superior cells that operate at better thermal ranges.

kWh Battery Pack

Cell Testing

For validation of whether the specified information presented by the manufacturer is valid, we will perform the following tests. We test at different discharge rates and operating voltage per cell. Below are the 4 tests to validate a particular cell: 

  1. Charge Discharge Charge (CDC) @ 0.3C: The data generated from this will be used to understand how the cells will behave when the cells are being discharged/charged at a nominal rate or at a rate at which the vehicle will be in use for more than 80%  of the time. Expected data points look like this: 
  1. Cell Voltage: Range: 2.75 V – 4.2 V, Accuracy: ± 2 mV
  2. Cell Current: Range: 0-720 mA, Accuracy: ± 5mA
  3. Cell Temperature: Range: 20-55 ºC, Accuracy: ±0.5 ºC
  4. Cell Capacity: Range: 0 – 2500 mAh

  1. Charge Discharge Charge (CDC) @ 1C: The data generated from this will be used to understand how the cells will behave when the cells are being discharged/charged at a peak rate or at a rate at which the vehicle will be in use for about  20%  of the time. Expected data points look like this: 
  1. Cell Voltage: Range:  2.75 V – 4.2 V, Accuracy: ± 2 mV
  2. Cell Current: Range: 0-2400 mA, Accuracy: ± 5 mA
  3. Cell Temperature: Range: 20-60 ºC, Accuracy: ±0.5 ºC
  4. Cell Capacity: Range: 0 – 2500 mAh

  1. Charge Discharge Charge (CDC) @ 1C: The data generated from this will be used to understand how the cells will behave when the cells are being discharged/charged at a nominal rate or at a rate at which the vehicle will be in use for more than 80%  of the time. The difference here will be to understand the cell behavior when it is being used only up to 80% of its rated capacity. Whether it will have superior performance or longer life or both. Expected data points look like this: 
  1. Cell Voltage: Range:  3 V – 4.1 V, Accuracy: ± 2 mV
  2. Cell Current: Range: 0-2400 mA, Accuracy: ± 5 mA
  3. Cell Temperature: Range: 20-60 ºC, Accuracy: ±0.5 ºC
  4. Cell Capacity: Range: 0 – 2500 mAh

  1. IR Test: Internal resistance test: The data generated from this will be used to understand how the cells can be arranged inside the battery pack. Which cells should go in series and which in parallel. And combining the above data points we can build an arrangement that will have superior performance over all the other possible combinations. Expected data points look like this:
  1. Internal Resistance at 1kHz
  2. Expected accuracy: <1 mΩ
  3. Test method: 4 terminal pair method

Once we have this data, we post-process it and analyse it to validate the battery pack. For explanatory purposes, we are showing the data given by the vendor.

  1. The discharge curve helps determine the discharge current rating of the battery
  2. Temperature curve helps determine operating temperature range of the battery
  3. Cycle life curve helps determine the cycle life aka health of the battery

Curve 1 - Discharge Curve of the cell

Source: BAK H18650CIL-2.4Ah Test Report

This graph indicates the capacity of the cell when it is discharged at various C rates and how the discharge C rates affect the capacity of the cells. For our vehicle, we will be operating the cell around 0.5C as the continuous operating discharge rate is 1.2 A per cell. The maximum discharge rate is 1C at which is 2.4 A. This sufficiently fits our motor’s needs of the peak discharge current of 32 A when 14 cells are connected in parallel thus gives us a total peak discharge current of 2.4 * 14 = 33.6 A.

Curve 2 - Temperature performance of the cells

Source: BAK H18650CIL-2.4Ah Test Report

India with growing metropolitan cities brings different environmental conditions to the table and temperature is a significant factor among all. The Li-ion cell’s aging and degradation largely depend on temperature and higher temperatures can lead to thermal runaway and reduce the battery capacity. Furthermore, in severe cases, abnormal temperature leads to exothermic chemical reactions within the cell which leads to venting and large release of energy and in some cases fire or explosions. Our battery needs to operate between 25° and 60°C thus we are getting maximum capacity from the current cells but we have to ensure its efficient use without damaging the cells.

Curve 3 - Cycle Life

Source: BAK H18650CIL-2.4Ah Test Report

This graph shows us how the capacity degrades over time or as the cell is cycled in the range of 2.75 V - 4.2 V at a charging C rate of 0.5C and a discharging rate of 1C. We are confident of taking out 1,000 cycles from these cells.

BMS Architecture

The Battery Management System aka BMS protects the battery and ensures its efficient use and health. The diagram below shows various components of kWh BMS architecture.

Fig 1: kWh Battery Architecture Diagram

Let’s consider each BMS component in detail:

  1. Battery Monitoring Unit: It is the analog front end that senses critical battery parameters like series string voltage, battery pack voltage, the temperature at various places on the battery, and protects by generating fault codes for various fault scenarios like overvoltage, over-temperature, etc and communicates fault codes to the Battery Control Unit (mentioned in point 2 below).

  1. Battery Control Unit: It is the brain of the Battery Management System. It consists of various subunits like The Information Management Unit, the Energy Management Unit, etc (See Fig). The Information Management unit collects and stores the data from the Battery Monitoring Unit and Battery Gauging Unit (mentioned in point 3 below). This data is shared with the Thermal Management Unit and then the Energy Management Unit takes the necessary action by giving commands to the FET Control Unit for stopping/allowing charging and discharging. Thermal Management Unit runs a thermal model that ensures that the battery pack is always operating within safe operating conditions. The Energy Management Unit communicates with the Information Management Unit and Thermal Unit and in case of a fault, disables incoming signals to the FET control unit to avoid further damage. Apart from this, Battery Control Unit also supports Gauging, i.e. keeping track of available battery capacity. This ensures that the battery pack is never discharged beyond its rated capacity and never charged above its rated capacity.

  1. Battery Gauging Unit: It keeps track of the battery capacity. It ensures that the Battery Control Unit is always aware of the available capacity which can be discharged and how much more the battery can be charged. Thus, the battery is never discharged below the safe discharge limit and never charged above its maximum available capacity.

  1. FET Control Unit: It controls the charging/ discharging of the battery pack. This unit has Field Effect Transistors (FET) which behave as a switch on the charging and discharging path of the battery pack. This switching action is very fast and very efficient and thus it allows us to perform complex functions like dynamic current control.

While building our battery prototype, we wanted to control the charging and discharging path of the battery circuit in a more power-efficient way.  Minimising heat losses is helpful because it increases the battery pack efficiency. We have attached a picture of the FET and the gauging circuit (Fig 2).

For a peak current of 30A, this FET circuit dissipates only 1W of heat ( 10mΩ resistance, where Power = I*I*R ), where an ordinary BMS’s FET circuit can dissipate up to 10W of heat, hence requiring more air cooling to operate efficiently.

Fig 2: Custom PCB for FET control unit and Battery Gauging Unit 

Once we built our battery pack, we deployed our software algorithms to achieve the following:

  1. Thermally Better Pack:  BMS’s Thermal Management Unit’s job is to ensure that the battery pack is always within the safe operable temperature range at any point in time. The goal is to make sure the operating temperature of cells should not exceed the safe operating temperatures. We use NTC thermistors at various positions inside the battery pack. These play a vital role in giving a real-time battery heat profile. The data is fed into a sensor fusion model with multiple inputs: 
  • The temperature of various spots inside the battery pack
  • current vehicle speed
  • acceleration of the vehicle, and the 
  • real-time current value

These inputs give us a complete image of the battery at that instant. Let's dive into more details. The operating conditions of the battery range from -20°C to 60°C but for optimal use, we should use the battery pack within 10°C to 40°C. The goal is to make sure that the battery pack’s cell temperatures should never exceed 60°C to avoid permanent damage. For example, if the vehicle is accelerating, and the battery cells’ temperature in the middle portion of the battery pack is at 58°C and let’s say that the algorithm knows that the temperature is going to rise beyond 60°C because the vehicle is accelerating, then the intelligent algorithm will take care of this condition by limiting the current at the FET (field-effect transistor, that controls charging and discharging path) circuit. 

This level of software control of the BMS is not possible in most of the BMS imported from China since it acts as a black box and is difficult to do any two-way control. Most of them have a set hard limit and thus cannot limit the maximum current flowing through the charging/discharging path. Also in scenarios like this, we would refrain from doing regenerative braking (battery charging), since that would also lead to a spike in battery pack temperature. The vehicle conditions are dynamic and often unpredictable, and such real-time decisions like charging or discharging the battery make a huge impact on the battery life. If this is not taken care of properly, this can impact the health of the battery cells every time a vehicle is operating in such conditions, thereby reducing the battery health.

  1. Failure Prediction & Warranties: BMS’s Information Management Unit’s primary role is to collect the data from our battery pack with minimum latency and maximum accuracy. Data collected from our battery pack, through our own BMS, is very critical and will be used for failure prediction and claiming warranties without a tussle.

  • For example, A scenario could arise where one particular string in the battery pack is not being balanced due to either a bad cell or if it has found a discharge path. This discharge path can be because of a mechanical wear/tear of the battery pack causing the plastic wrapped around the cell body to tear. BMS Information Management Unit collects this data and analyses such spikes of voltages and currents during rides, which allow us to call such batteries back and prevent any damage. Over time, such predictive actions build trust among the users and also improve our manufacturing processes and algorithms for the next battery packs.

  • Currently, many E2W owners are unable to provide charging data (start time of charging, the end time of charging, and the specs of the charging) to OEMs and are hence unable to claim battery warranty easily because most E2W batteries\ do not have a CAN BUS setup (that can share this data) and no way to check if the charging was properly done. This data (charging cycles), if made available, can help owners claim battery warranty easily as they can prove that only the appropriate charger and charging conditions were used for charging. To solve this warranty headache, we plan to build a proprietary algorithm to calculate battery health score and determine if the battery is non-performant. If it is, the fleet owner will be notified on the Fleet Dashboard immediately and can claim battery warranty using 1 Click.

  1. Accurate Estimation Of Charge: The ability to accurately estimate the remaining range is difficult. Imagine being told that your vehicle can still go 20km, but it stops after 7 km. Wouldn’t be a great experience, right. We would be able to predict range better as we control the charging and discharging path of the battery, and also measure the battery parameters in real-time, and as we have more data, we can build our own very accurate SoC estimation algorithm. This gives high confidence to the rider and removes range anxiety.

  1. Proprietary Balancing, Charging, Discharging & Fault Algorithms: kWh BMS is driven by our own well-tuned data acquisition system. We read real-time current, voltage, temperature sensor values accurately and feed this data into complex software algorithms that enable a better state of health, battery protection, and real time fault handling.

  1. Software updates: Cloud connectivity ensures that data that is collected and stored locally is streamed to the cloud server. This helps us run our machine learning algorithms to tune, improve our algorithms and configure parameters while collecting various information like load variations, drive train data, tilt sensor data and much more. This data will be tested and eventually help ship Over The Air Updates (OTA) to the BMS to improve range and power delivery efficiency.

Specifications of kWh BMS

Battery monitoring unit:

  • 13s cell monitor
  • Maximum Open cell voltage for any channel: 5.5V
  • 8 channels for temperature measurement using NTC/other analog sensors in the range of 0-5 V.
  • Passive cell balancing current of 56 mA 
  • Operating temperature –40°C to 105°C
  • 14 bit ADC
  • Undervoltage, Overvoltage, Under-temperature, over-temperature, and various other fault detection capabilities
  • UART Communication Interface with a maximum baud rate of 1 MBPS
  • Dedicated pins for fault signaling

Battery Gauging Unit:

  • Maximum battery capacity: 300 Ah
  • Accuracy of current sensing: ± 10 mA
  • Overcurrent while charging, overcurrent while discharging, short circuit, and other fault detection

Battery Control Unit:

  • STM32F103 MCU with Arm Cortex-M3 core
  • The maximum Clock speed of 72MHz
  • Flash memory of 64KB
  • SRAM: 20KB

kWh Battery Testing

We built our prototype BMS on a Texas Instruments (TI) Board conforming to the architecture mentioned above. This prototype can be further improved by conducting extensive research, testing, collecting the data, and analysing the data. Hence doing a system-level optimization for the battery pack over drive cycles.


We plan to conduct simple battery pack tests for 3 to 5 cycles, to get a more comprehensive picture of the battery's behaviour over its lifetime. Data from tests over 800 cycles will be performed in the future. Tests that we plan to do for our production-ready vehicles are: 

  1. Coulombic Efficiency:

It is the ratio of the total charge drawn from the battery pack to the total charge put into the battery pack. Usually, for Lithium-Ion battery packs, this is 99% or above. We perform charging, rest, and discharging in steps to get the coulombic efficiency of the battery pack.

  1. Capacity Fading Test:

We test the battery rigorously over 800 to 1,000 cycles. Each cycle comprises charging, resting the pack for a short time, and discharging the battery pack again. We note the obtained pack capacity after each cycle and analyze the data. We expect the capacity of the battery to reduce or fade as the battery pack cycles and we stop when we reach 80% of the pack capacity obtained from the first cycle.

  1. Thermal Chamber Testing:

We test the battery pack at various temperatures in steps of 5°C and from 10°C to 60°C to understand the battery pack’s performance over different environmental conditions. This test will simulate the Indian road conditions and will be very close to what the battery pack will go through when the user is riding the vehicle.

These tests will enable us to understand the behaviour of the battery pack during its complete lifetime and these data points will be used by our Vehicle Control Unit to generate insights on the battery.  Apart from this, we will be able to extract metadata as a simulation of the vehicle on road lifetime performance. This metadata will be used to simplify the algorithms on VCU and will be used to predict failures which will make the ride a more improved experience.

Charging of kWh battery

It would be advisable to use a kWh proprietary charger. The reason is that the kWh charger communicates with the BMS to ensure safe and long-lasting battery health. Using any other charger may significantly reduce the battery life. kWh charger does a multi-level step charging using JEITA protocols (a charging standard). It shall ensure over voltage protection and helps maintain the temperature of the battery at all times. Any other charger will not be able to understand the current state of the battery and will always charge in constant current mode. This can provide unnecessary current even if the battery is fully charging leading to over saturation and decreasing battery life. Also, using a proprietary charger can provide dash charging i.e. charging for 15 minutes can get about 15% battery charge equivalent to 15kms.