Assessment of spent EV batteries for second-

life application
M. Muhammad, M. Ahmeid, P.S. Attidekou, Z. Milojevic, S. Lambert and P. Das
School of Engineering, Newcastle University
Newcastle upon Tyne, United Kingdom
[email protected]; [email protected]

Abstract—Electric vehicle (EV) batteries typically retained diagnostics data to confirm the state of health (SoH) and
around 70- 80% of their initial capacity at the end of their absence of faulty cells in modules/battery packs before their
automotive life. The first batches EV batteries are hitting their deployment in second-life is, therefore, key to repurposing.
retirement age and around 3 million used battery packs will be Since the volume of end of life (EoL) batteries is likely to
coming out from EV annually which represent around 108 increase as penetration of EV/HEV increases.
GWh. Automotive companies and OEM have announced
numerous projects and business models for second-life batteries, Whilst, the most promising application for spent EV
such as the stationary energy storage system (ESS) to extract batteries is the ESS and perhaps power system. To enable the
additional services and revenue in post automotive life with second-use battery market, it is necessary to demonstrate the
cross-cutting benefits. The potential of second-life batteries lies capability of the retired batteries to adequately provide these
in their state of health (SoH) or remaining useful life. To enable services. In particular quantifying system response in real-
this market for second use batteries, it is necessary to world scenarios and validating the longevity of these batteries
demonstrate the capability of such batteries in real-world in post automotive application is critical. To this end, we
scenarios and validating the longevity of these batteries in this acquired an EoL 24 KW Nissan Leaf battery pack that has
application is critical. This paper characterises three selected been disassembled to the module level. Module from this pack
Nissan Leaf Cells/Modules disassemble from 24 KW Nissan are being tested individually to provide insight into the
Leaf pack that reach an end of life (EoL). The modules are tested
variation in degradation or SoH across a single battery pack
individually to provide insight into the variation in degradation
and to ensure that the retired batteries meet the EoL
across a single battery pack and also establish the available
energy at EoL via Hybrid pulse power characterization (HPPC).
characteristics of ESS performance targets for EVs [5] - [7].
The result shows three data points of capacity and Ohmic It is necessary to present a SoH or RUL data for a value
resistance. Furthermore, the batteries achieved the power chain analysis from the lifecycle perspective to help
density of 700 W/Kg/300 W/Kg during discharge/regen in the stakeholders identify potential value opportunities. This
working regions from 100% to 30% and /80% to 20% DOD would be achieved via a rapid testing protocol for gateway
respectively. Besides, the HPPC micro cycle discriminates testing and sorting battery at the end of their automotive life
between the strong and weak battery in 120 s.
prior to second-use. This will enable accurate identification of
Keywords—EoL Lithium-ion batteries, Assessment,
a battery's value and its viability for second use service. A
Sorting, capacity, Electrochemical impedance spectroscopy, single weak cell in a module would affect the output power
Hybrid pulse power characterization, reuse, second-life and comprises the performance of the whole module.
Furthermore, the failure of a cell in a battery-powered system
I. INTRODUCTION can lead to performance deterioration, costly replacement and
serious hazards [8]. Therefore, the gateway testing and sorting
Battery reuse strategies in which a battery pack first serves of second-life batteries must be rapid and accurate.
in an automotive application, then once deemed appropriate is
redeployed into a secondary market could help overcome Literature classifies testing of lithium-ion batteries into
lithium-ion battery cost barriers to the deployment of both three phases i.e., characterisation, life and reference
electric vehicles (EVs) and grid-connected energy storage. performance test (RPT). Characterisation test establishes the
Allocating retired automotive batteries in applications that are baseline performance. Life testing establishes behaviour over
less power demanding such as energy storage systems (ESS) time of both cycle and calendar life at various temperature,
to mitigate the intermittency of renewables, utility peak load state of charge (SoC) as well as other stress conditions; whilst
reduction, smart grid, grid stabilisation, low energy buildings RPT track the evolution of performance over cell life and are
and utility reliability. The prospect of ESS via second use performed periodically during the battery life from the
batteries would improve materials efficiency, bring beginning of life (BoL) and EoL [5]-[7]. Typical parameters
tremendous value to the wide range of stakeholders across the denoting the degradation of lithium-ion cells are capacity
automotive and energy sectors, as well as society and depletion and resistance growth. Essentially, the capacity
environment. [1]-[4]. measures the amount of electric charge stored in a cell and
defines its maximum capacity in the application under a given
Although, the economic margins that make reuse batteries load. Whereas, the resistance characterises the cell dynamics.
viable are often small, and thus several factors could affect this
conclusion. The major impediment being the complexities of Batteries can be sorted optically or electrically. Optical
repurposing heterogeneous lithium-ion batteries, collection sorting involves sorting the batteries based on their shape or
and technical labour are the primary cost element. appearance. Electrical measurement requires sensors to
Furthermore, non-availability of on-board diagnostics data establish contact with the battery and measure certain
and accurate assessments of automotive and second use parameters, which is difficult to automate in a real-time
battery degradation stand out in particular. Use of vehicle process. Besides, most of the existing electrical methods are

978-1-7281-3153-5/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 23:20:01 UTC from IEEE Xplore. Restrictions apply.
not rapid in separating good cells from bad ones. For example, 10 -4
4
capacity measurement involves cycling the battery and takes
several hours to complete. These techniques are unsuitable in 2
a recycling facility where tens of thousands of batteries need
to be processed every day. 0
---------- Cell A
Many approaches are being used to determine the SoH, of -2 ---------- Cell C
a battery e.g. Pseudo open-circuit voltage (pOCV), ---------- Cell B
Incremental capacity-Differential voltage (IC-DV), -4
electrochemical impedance spectroscopy (EIS) and 0.9 1 1.1 1.2 1.3 1.4
differential thermal voltammetry (DTV), each approach has
Real Z (Ohm) 10 -3
its strategy and advantages.
Fig. 2. EIS measurement.
Outside these different choices, hybrid pulse power
characterisation is another useful tool for determining the Where Qt1 , Qt0 are the initial and the current capacities of
pulse power capability of battery in both discharge and regen the battery respectively. Fig. 1 shows the measured capacity
regimes. EoL characteristics of energy storage system of the three cells A, B, C. The test is executed base on
performance goals for EVs [5]- [7], targets a 700 W/Kg peak manufacturers cycling protocol i.e. C/1 constant current -
specific discharge power lasting 30 s and a 300 W/Kg peak constant voltage (CC-CV) to full charge state Vmax (4.2 V).
specific regen power lasting 10 s. A minimum of 300 Wh/Kg Discharge begins following a default rest from a fully charge
of energy must be available within the SoC operating range state and is terminated on discharge voltage limit Vmax (4.2 V).
capable of meeting the power goals. The goals dictate the use 57 Ah, 54.5 Ah and 49.45 Ah are the measured capacities
of HPPC to quantify the remaining available energy with corresponding to cells A, B and C. The capacity fade of the
obvious drawbacks of requiring tens many hours of testing. batteries stands at 86.36%, 82.57% and 74.92%.
This paper demonstrates the assessment and sorting of the B. Impedance test
EoL EV batteries with unknown history to meet the
requirements of repurposing in ESS. This demonstration is EIS is considered as a powerful method for investigating
accomplished via the HPPC, capacity and EIS techniques. the operational, ageing mechanism and thermal degradation of
Section II provides an overview of the characterisation a cell [8], [9]. EIS technique is a frequency response method
methods employed and sections III discussed the results which typically measures cell impedance along a range of
obtained. Section IV outlines the conclusions drawn from the frequencies (i.e. KHz to mHz). The test can be performed in
study. either potentiostatic or galvanostatic mode by simply
perturbing the cell voltage and measuring the cell current or
II. CHARACTERISATION vice versa. The cell impedance is computed as in (2)
Three commercial modules A, B, C, disassembled from
Nissan Leaf battery pack that reach EoL in an automotive ∆ ·
= = (2)
application were selected. Each module comprises four ∆ · ( +Ф)
identical pouch cells. The cells are electrically configured as A Shift in Nyquist spectra from left to the right indicate an
2X2 arrangement (2 in series and 2 in parallel). The cell aged or degraded cell. In other words EIS is characterisation
chemistry is lithium-manganese-oxide with nickel oxide tool used to infer ageing mechanism of a cell. Fig 2. Illustrate
widely known as (LMO). The cells are characterized as the Nyquist spectra of the three cells. As can be seen, the
follows: Ohmic resistance of the cells are 0.95 Ω, 0.98Ω and 0.86Ω.
A. Capacity Fade C. HPPC test
This test is a time domain method that measure the battery Fundamentally, HPPC is a measure of battery’s dynamic
capacity in ampere-hours. For the battery subjected to power capability over its useable voltage ranges incorporating
automotive use, the change in capacity from BoL to some later a discharge and regeneration current pulses. The HPPC
point in time is regarded as capacity fade and expressed as a procedure determines this ability by applying current
percentage of the original capacity as in (1) waveforms as shown in Fig. 1 to the battery under test, the
response to this stimulus is used to determine power
(%) = 100 ∗ 1 − ⁄ (1) capability. The test comprises of 30s discharge, 40s rest and
immediately followed by 10s regen; together these make up
one complete HPPC micro cycle. Between each pair of the
micro cycle, the battery is discharged to the next 10% depth
of discharge (DoD) level based on the operating capacity. It is
Voltage (V)

worth noting that the current values are relative and the micro
cycle is not charged balance.
A HPPC is executed on each module at 0.1 SoC interval
(constant current C/3 discharge segments) starting from 1.0 to
0.1, each interval is followed by one hour rest to allow the
battery to attain electrochemical and thermal equilibrium
before discharging to the next SoC. The HPPC micro cycle
and complete current profile is shown in Fig. 3a. To achieve
Fig. 1. Capacity fade.
conservative estimate of pulse resistances a low current HPPC

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 23:20:01 UTC from IEEE Xplore. Restrictions apply.
 (a) Cell current vs time during HPPC cycling (b) Cell voltage vs time during HPPC cycling

(c) OCV as a function of depth of discharge DoD (d) Apparent resitance as a function of DoD

(e) Pulse power capability as a function of DoD (f) Pulse power capability as a function of enrgy removed

Fig. 3. HPPC cycling and analysis of second-life batteries

[ ] is adopted during the test. Fig. 3b illustrate the voltage vs discharges the resistance increases. From these resistance
time of the three cells. As can be seen, the 10s regen pulse values and the OCVs, the maximum pulse power capability
appeared as a spike in the waveform whilst the 30s discharge of each cell can be determined using equation (4) and (5).
pulse is clear. A close up of the HPPC microcycle is also ( − )
shown. Following each HPPC microcycle is the 10 % capacity = (4)
removal followed by the one-hour default rest prior to the
application of the subsequent HPPC micro cycle. Fig. 3c ( − )
illustrate the OCV of the cells as a function of the individual = (5)
cell DoD. OCV between each 10% increment can be estimated
by straight-line interpolation. The shift in OCV between the Where and are the OCVs of the
cells is apparent and clearer towards the nominal voltage of battery immediately prior to applying either the discharge or
the batteries. The pulse resistance of each cell is calculated regen pulse. , are the minimum and maximum
during discharge/regen pulses corresponding to the DoD by terminal voltages limits within which the battery is cycled.
dividing the change in voltage during the 30s discharge and Fig. 3e shows the discharge and regen pulse power
10s regen periods by the applied current using (3). capability of the three cells as a function of DoD of the cells,
Fig. 3d shows the plot of the discharge and regen and Fig. 3f shows the discharge and regen pulse power
resistance calculated using (3) as a function of the DOD. As capability of the cells as a function of the cumulative energy
can be seen, the 10s regen resistances are slightly shifted to removed from the cells. This is achieved with the assumption
that the capacity removed from the constant power discharge
the right compared to the 30s discharge resistances due to the
and subsequent HPPC test are equivalent.
inclusion of the 30s discharge regime during the HPPC micro
Having obtained the pulse power capabilities, the total
cycle when determining the percentage of the rated capacity
cell useable energy as a function of power level the device is
removed. Furthermore, there are minor differences between
being operated can be determined. The useable energy is the
the discharge and regen resistances. It is apparent as the cell
amount of energy that a cell can accept or deliver when it is
∆ being cycled at a certain power level. The battery size factor
/ = (3) (BSF) of a target application can then be determined using
/
the useable energy as a function of power.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 23:20:01 UTC from IEEE Xplore. Restrictions apply.
 III. DISCUSSION There is a clear distinction between the discharge and
Most battery performance analysis tools use Coulomb regen pulse power capability despite the similarity in values
counting to estimate the SoH of a battery. The energy flowing of discharge and regen pulse resistances. This could be as a
in-and-out of a battery is measured and stored in an accessible result of overpotential due to driving high current during
register. The content of the register is then compared with cycling. Table I summarises the selected SoH determination
subsequent measurements to predict the SoH of the battery. techniques and duration required to perform each test.
The major shortcoming is losing accuracy when the battery is
TABLE I. SUMMARY OF SELECTED SOH METHODS AND DURATION
randomly cycled and integral errors accumulating over time.
The leading SoH indicator of a battery is its capacity. S/N SoH Methods Duration (m)
Estimating the capacity of a battery on the fly is complex. The 1 Capacity 120/C1
best approach is to read the capacity through the 2 AC Impedance EIS 15
charge/discharge cycle. The result obtained is accurate and 3 HPPC microcycle 2
noninvasive. The downside is the battery must often be
removed from service to determine the capacity and the IV. CONCLUSION
cycling takes 2 - 3 h to complete for the commercial Nissan
leaf module tested in this case. In this research, three different battery SoH measurement
EIS test measures Ohmic internal battery resistance to techniques were employed to measure the SoH of three LMO
verify the loading characteristics. The resistance reading does 66 Ah module for rapid EoL assessment/sorting. From
not correlate with capacity. EIS performs rapid testing on a comparison of the results, it has been shown that each
battery by scanning with multiple frequencies to generate technique provide different insight into a particular aspect of
Nyquist spectra. The horizontal axis reveals the real battery degradation across module/cell. The leading SoH indicator is
Ohmic impedance and the vertical represent the imaginary the capacity measurement which account for capacity fade as
impedance. Movement of impedance spectra from left to the battery age. On one hand the EIS display all the individual
right denotes ageing. Fig. 4 illustrate the closed up of the resistance contribution and phase transformation in an
three batteries impedance spectra showing the two domains electrochemical cell. On the other hand HPPC provides
of Nyquist plot entitled migration and charge transfer information of resistance degradation during discharge/regen
corresponding to high and mid-range frequency scales pulses. The pulse resistance allows pulse power capibility
respectively. The mid-frequency range called charge transfer during HPPC microcycle and available energy of the battery
displays valuable characteristic explaining the kinetics of the to be determined as a performance indicator of ESS at EoL. In
battery that tells the SoH reference. As can be seen cell B addition, Both EIS and HPPC can be parametirised into an
denote higher Ohmic resitance than all the remaining cells. A equivalent circuit model (ECM) and have potential to be used
rapid test shouldn’t take a long time i.e. from few seconds to for rapid testing using algorithms with parameters serving as
no more than 5 minutes, but scanning all the three frequency lookup tables.
regions (i.e 3 KHZ - 10 mHz) in this case takes about 15 m.
ACKNOWLEDGMENT
Test duration can be shortened by scanning only the high and
mid-frequency regions. The authors acknowledge the financial support of the Faraday
HPPC measures the mobility of ion-flow between cathode Institution: ISCF Faraday Challenge Fast Start Project on
and anode at diffrent DoD by applying discharge/regen pulses “Recycling and Re-use of lithium ion batteries” made
and evaluating the voltage response. As can be seen from Fig. available through “EP/FIRG005”.
3b battery A appears firm and has quick recovery than battery
B and C. Battery C shows softness and sluggish recovery in REFERENCES
comparison to battery A and B. Furthermore, all the three [1] G. Eason, B. Noble, and I. N. Sneddon, “Battery second life: Hype,
batteries meets the requirement of EoL ESS performance hope or reality? A critical review of the state of the art,” J. Ren.&
Sustain. Energy Rev., vol. 93, pp. 701-718, Oct 2018.
target by achieving the power density of 700 W/Kg during
[2] Pinsky N, Gaillac L, Mendoza A, Argueta J, Knipe T. Performance of
discharge in the working regions from 100% to 30% DOD. advanced electric vehicle batteries in stationary applications. In:
Similarly, the cells meet the regen target by achieving 300 Proceedings of the 24th annual international telecommunications
W/Kg from 80% to 20% DoD. energy conference; 2002. p. 366–72
The technique involves measuring the DC resistance of [3] Pinsky N. Electric Vehicle Battery 2nd Use Study. Tech. rep., Argonne
the battery due to current perturbation and consequent drop National Laboratory, Electrochemical Technology Program; 1998.
in voltage. The voltage drop is largely due to contribution of [4] Cready E, Lippert J, Pihl J, Weinstock I, Symons P, Jungst RG.
pure Ohmic and bulk electrolyte resistance of he battery, Technical and economic Feasibility of applying used EV batteries in
double layer capacitance, charge transfer resistance and stationary applications. Tech. rep.,andia National Laboratory; 2003.
polarisation resistance. [5] J. P. Christopsen, “Battery Test Manual for Electric Vehicles,” INL
Rev. 3, vol. 93, pp. 1-50, June 2015
[6] Battery Test Manual for Plug-In Hybrid Electric Vehicles, INL pp. 1-
54, March 2008.
[7] FreedomCAR 42V Battery Test Manual, INL pp. 1-54, April 2003.
[8] S. Lambert, M. Armstrong P.S. Attidekou, P.A. Christensen,, J. D.
Widmer, C. Wang, and K. Scott, “Rapid Nondestructive-Testing
Technique for In-Line Quality Control of Li-Ion Batteries,” IEEE
Trans. Ind. .Electron, vol. 64, pp. 4017–4026, May 2017.
[9] P.S. Attidekou, S. Lambert, M. Armstrong, J. Widmer, K. Scott, and
P.A. Christensen, “A study of 40 Ah lithium ion batteries at zero
percent state of charge as a function of temperature,” J. Power Source,
vol. 269, pp. 694–703, Dec 2014.
Fig. 4. Close up of migration and charge transfer regions.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 23:20:01 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 23:20:01 UTC from IEEE Xplore. Restrictions apply.