THE EFFECT OF PHEV AND HEV DUTY CYCLES ON BATTERY AND BATTERY PACK PERFORMANCE Lars Ole Valena

and Mark I. Shoesmithb

b

Milj Innovasjon AS, Nystrandvegen 59, N-3944 Porsgrunn, Norway -!ne Moli nergy "#anada$ %td&, '(,((( Stewart #res)ent, Ma*le +idge, ,#, -'. 9 /

ABSTRACT Comparing different batteries or different battery chemistries for real life duty cycles is an onerous process and requires each battery to undergo cycling for each specific duty cycle. he present paper is a first step to!ards simplifying this process by de"eloping a metric for describing ho! a duty cycle current profile de"iates from the constant current state. his quantity# the duty cycle eccentricity# is then utili$ed to compare the efficiencies of commercial %iCd# %iM& and Li'ion po!er tool batteries. Li'ion battery efficiency !as found to be higher than that of %iM& and %iCd batteries. KEYWORDS Li'ion# batteries# duty cycles# efficiency# &(V# )&(V INTRODUCTION Coulombic efficiency can be seen as a measure of ho! much of the electric charge that is used for the intended purpose of an electrical de"ice. %on'aqueous battery chemistries tend to run "ery small side reaction currents# hence achie"ing near *++, coulombic efficiency# e"en for real life duty cycles as seen in po!er tools or electric "ehicles. he energy efficiency is the fraction of the total stored energy in a cell that is measurable as electrical energy. %o battery chemistry can obtain *++, energy efficiency# for the simple reason that there !ill al!ays be energy dissipated as heat through internal impedances. -enerally# the efficiency drops as the current is increased because the rate dependent component of electrode o"erpotentials and the formation of concentration gradients !ill consume energy. In order to effecti"ely design and utili$e batteries for real life# po!er' demanding applications# some understanding of ho! the heat to electrical energy ratio "aries !ith current and duty cycle is required. he aim of the present paper is to demonstrate a methodology for adequately comparing different batteries or different battery chemistries for real life type of duty cycles !ithout ha"ing to deal !ith the e.act details of such duty cycles /uty cycles are often defined in terms of their a"erage current I 0 ho!e"er# this is insufficient to characteri$e pulsed duty cycles that may contain numerous steps. he standard de"iation gi"es a representation of the departure from the constant current state. o make it more uni"ersally applicable and to enable scaling !e define the duty cycle eccentricity 1/C(2 as a unitless quantity gi"en by
1 I I 0# = = I I
3 3

If !e then choose duty cycles such that the mean current remains constant in all cases# !e can measure the cell efficiency as the duty cycle eccentricity gro!s. Ignoring concentration gradients and assuming constant temperature# the cell !ill e.hibit ohmic beha"ior4 the "oltage drop !ill be e.actly proportional to the current. 5e use this theoretical beha"iour to estimate the net contribution from temperature changes and concentration gradients and to determine !hether they are positi"e or negati"e by e.trapolating a least squares fit of energy efficiency "s /C( at lo! currents. Selecting current r!"ile# 6or pulsed current testing in"ol"ing different cell chemistries# selection of the current profile becomes non'tri"ial since the coulombic efficiency must be kept as close to unity as possible for the charge step !hen characteri$ing the discharge step and "ice "ersa !hen characteri$ing the charge step. his is relati"ely straightfor!ard for Li'ion batteries as long as the temperature is kept around room temperature or lo!er# and pro"ided the recommended "oltage range is not e.ceeded. his is possible because side reactions are "irtually absent for the aforementioned conditions. 6or aqueous chemistries# the situation is quite different# particularly for the charging reaction since the re"ersible hydrogen e"olution potential is *.37V 18yl!ard and 6indlay *9:;2. his means that at any "oltage e.ceeding *.37V# some hydrogen e"olution !ill occur. he rate of the hydrogen e"olution !ill be dependent on the "oltage difference bet!een the electrodes. E$PERIMENTAL Standard ('One Moli (nergy IM<3=:++8 cells using LiMn3O; based positi"e electrodes !ere employed for all tests !ith Li'ion batteries. %ickel'Cadmium 1%iCd2 batteries !ere obtained from an *>V 3.;8h po!er tool pack. %ickel'Metal hydride 1%iM&2 cells !ere obtained from a *9.3V# 3.=8h po!er tool pack. &igh current measurements !ere performed using an ('One Moli (nergy charger system or a Maccor Series ;+++ battery cycler. Calorimetric measurements !ere performed using a Columbia Scientific Industries 8ccelerating <ate Calorimeter 18<C2 and a ?ikusui )L@ :++5 electronic load. 6or characteri$ing pulsed discharge# Li'ion batteries !ere charged using a constant currentAconstant "oltage regime# starting !ith a current of *C# s!itching to a constant "oltage once ;.3V !as reached. Charge !as terminated !hen the combined charge time reached 9+min. his *++, state'of'charge 1SOC2 !as chosen as a reference state for the Li'ion cells. %iM& batteries !ere first tested for coulombic efficiency using a charge current of CA*+. It !as found that the coulombic efficiency !as near *++, up to a SOC of 9+, using this current. he fully charged# reference state !as therefore defined as 9+, SOC for the

%iM& batteries. 6or the %iCd batteries# the charge efficiency !as some!hat lo!er using CA*+ as a charge current and >+, SOC !as chosen as the fully charged# reference state. 6or all cells# a charge appro.imately equal to *+, of the battery capacity !as remo"ed during the pulsed discharge step. he cells !ere then charged up replacing an equal amount of charge compared to !hat !as remo"ed during the discharge step. 6or the %iM& and the %iCd batteries# a charge current of CA*+ !as selected !hereas the Li'ion batteries could sustain *C charge current !ithout the coulombic efficiency dropping. o compensate for limitations in the testing equipment# some of the tests !ere conducted using %iM& and %iCd 7 cell series packs. 6or characteri$ing pulsed charge# a lo!er cutoff "oltage !as selected and used to define the discharged state. 6or %iCd and %iM&# discharged state !as defined as a CA*+ discharge to +.9V. 6or Li'ion# the cells !ere discharged to a lo!er cutoff "oltage of 3.BV using a C'rate current. his !as defined as the fully discharged reference state. 8fter e"ery pulsed charge an equal amount of charge !as remo"ed before the ne.t charge !as initiated.

RESULTS AND DISCUSSION 6igures *# 3# and 7 sho! the coulombic efficiencies for the three battery types using a CA*+ constant charge current.

2igure 34 NiM5 +ound tri* )oulo6bi) e77i)ien)y and dis)8arge )a*a)ity as a 7un)tion o7 )8arged )a*a)ity 7or #93( )urrents&

2igure '4 Ni#d +ound tri* )oulo6bi) e77i)ien)y and dis)8arge )a*a)ity as a 7un)tion o7 )8arged )a*a)ity 7or #93( )urrents&

It can be obser"ed that both full efficiency and full discharge capacity are unobtainable e"en for this relati"ely lo! current. In particular for %iM& 16igure *2 and %iCd 16igure 32 cells the discharge efficiency drops rapidly as the cell approaches its fully charged state. he onset of this efficiency loss !as chosen as the fully charged reference state for energy efficiency tests. his corresponds to appro.imately 9+, of the ma.imum discharge capacity for the %iM& cell and >+, for the %iCd cells.

2igure 34 %i-ion +ound tri* )oulo6bi) e77i)ien)y and dis)8arge )a*a)ity as a 7un)tion o7 )8arge voltage 7or #93( )urrents& :nli;e NiM5 and Ni#d %i-ion batteries are )8arged using a ##-#- )8arging routine& <8is 6a;es it easier to avoid )ells being over)8arged using a si6*le )8arger wit8 no te6*erature 6onitoring

6or the lithium ion cells# the coulombic efficiency is "oltage determined. his is due to the constant "oltage charge method. he fully charged state for a Li'ion cell is typically ;.3V# and the coulombic efficiency is "ery close to *++, at this point. 8s the "oltage increases beyond ;.:V the efficiency does start to drop since electrolyte decomposition starts to take effect. here !ill be some "oltage threshold for all battery chemistries beyond !hich this electrolyte decomposition starts to happen. &o!e"er# there is a fundamental difference bet!een %iC/A%iM& and Li'ion since this electrolyte decomposition is mostly re"ersible for the former# !hereas for the latter electrolyte decomposition is irre"ersible 1Cesenhard# *99>2.

In 6igure ;# the measured round'trip efficiency as a function of the duty cycle eccentricity is sho!n. In all tests the a"erage current !as kept to 7.7C 1*+82. 6or the tests !ith "ariable current on charge# the discharge current !as kept constant at *C and for the tests !ith "ariable current on discharge0 the charge current !as kept constant at *C.

2igure 44 Measured round-tri* )8arge and dis)8arge e77i)ien)ies as a 7un)tion o7 duty )y)le e))entri)ity 7or 3&3# "3(A$ dis)8arges& 6or constant current# Li'ion sho!s a higher efficiency than %iM& and %iCd. 8s the duty cycle eccentricity increases the energy efficiency drops for all chemistries tested. he %iM& packs drops off faster than %iCd and Li'ion. 8s pre"iously described# !e can predict an ideal cur"e for efficiency "s. duty cycle eccentricity# denoted DohmicE in the plot. It can also be obser"ed that the discharge efficiency de"iates from predicted ohmic beha"ior once the eccentricity is greater than one. his is an indication that non'ohmic processes# such as diffusion or reaction kinetics# start to significantly influence the cell performance abo"e that point. 6rom 6igure ;# it can be obser"ed that as the duty cycle eccentricity of the current profile increases# the efficiency decreases for all the battery systems tested. he efficiency for the Li'ion batteries is higher than for the %iM& and the %iCd batteries for the constant current discharge. he %iM& batteries sho!ed a higher decline in efficiency compared to the Li'ion and the %iCd batteries !hereas the latter !ere comparable. It is also

!orth!hile to obser"e that the measurements performed using 7 cell series packs for %iCd and %iM& cells yielded the same efficiencies as the measurements performed for single cell tests. Since the coulombic efficiency is "ery close to * for Li'ion batteries# it is also possible to analy$e the efficiency as a function of duty cycle eccentricity for the charge current in a "ery straightfor!ard# simple !ay. his is sho!n in 6igure B.

2igure 54 Measured )8arge e77i)ien)ies as a 7un)tion o7 ratio o7 duty )y)le e))entri)ity 7or Ni#d, NiM5 and %i-ion batteries 6rom 6igure B it can be obser"ed that the beha"ior is fundamentally different for the different battery chemistries. Li'ion has a "ery high charge efficiency# decreasing slightly as the duty cycle eccentricity increases. 6or the %iCd batteries# the charge efficiency is substantially lo!er. he %iCd efficiency sho!s a peak for an intermediate duty cycle eccentricity. 6or constant current# the coulombic efficiency is quite lo!# increasing !ith increasing duty cycle eccentricity of the current profile. &o!e"er# as the duty cycle eccentricity increases# the heat production increases and offsets the gain in coulombic efficiency. he importance of the findings presented in 6igure B are of less importance for a battery charged using abundant po!er from the grid than for a battery charged from a limited energy source such as recaptured energy from regenerati"e braking. 6or a plug'in hybrid or electric "ehicle# this !ill impact the all'electric'range 18(<2 and dependent on the

8(< e.tension pro"ided by the regenerati"e braking significantly contribute to !hich battery chemistry !ould constitute the ideal selection for such "ehicles. In 6igure =# the efficiency as a function of the depth of discharge 1/O/2 is gi"en. he heat production is measured using the accelerated rate calorimetry 18<C2 technique.

2igure =4 5eat *rodu)tion as a 7un)tion o7 dis)8arged )a*a)ity 7or a =&/#"'(A$ dis)8arge 7or a %i-ion battery

8s can be seen from 6igure =# the heat production starts out lo! and increases !hile internal concentration gradients are building. his peaks !hen appro.imately 3+, of the cell capacity has been utili$ed. he heat production then decreases and stays lo! until the end of discharge !here it increases rapidly as heat is produced from entropy effects to!ards the end of the discharge.

>

CONCLUSION In the present paper# a methodology for comparing different batteries# including different battery chemistries for real life parameters is presented. 8"erage current !as found to be insufficient for characteri$ing real life duty cycles. he performance of any battery is highly dependent on the duty cycle and kno!ledge of the po!er profile is crucial for battery and battery pack design. /e"elopment of standard tools for characteri$ing and grouping together duty cycles is important to be able to design optimal batteries and battery packs. It is also sho!n that the charge and discharge efficiencies decrease as the duty cycle eccentricity of the current profile is increasing. his holds for all three battery chemistries tested. he energy efficiency of Moli IM<3=:++8 LiMn3O; based batteries remained remarkably high under all test conditions. his indicates the e.cellent "iability of Li'ion batteries in applications requiring high efficiency. &eat production for a constant current discharge !as found to increase rapidly in the beginning of the discharge# and then decrease as the cell temperature escalated due to self'heating. 8t the end of the discharge# increased entropy production again increases the heat production.

Re"erence# 8yl!ard# -.&. and 6indlay# .F.V. 1*9:;2. SI chemical data# 3nd ed# 5iley Cesenhard# F.O. 1*99>2. &andbook of battery materials# 5iley'VC&