WO2003067735A1

WO2003067735A1 - Method and apparatus using a circuit model to evaluate cell/battery parameters

Info

Publication number: WO2003067735A1
Application number: PCT/US2003/003606
Authority: WO
Inventors: Keith S. Champlin
Original assignee: Champlin Keith S
Priority date: 2002-02-08
Filing date: 2003-02-07
Publication date: 2003-08-14
Also published as: AU2003207861A1; GB0417678D0; US6737831B2; GB2401192B; US20020109504A1; GB2401192A

Abstract

Testing apparatus (5) senses the time-varying electrical response of an electrochemical cell/battery (20) to time-varying electrical excitation. Computation circuitry (30) responsive to the time-varying electrical response evaluates elements of a unique circuit model representation of the cell/battery (20). Performance parameters and physical parameters are computed from these element values. Computed performance parameters include, but are not limited to, 'total storage capacity', 'absolute stored charge', 'state-of-charge', 'absolute cranking current', 'fully charged cranking current', and 'state-of-health'. Computed physical parameters include, but are not limited to, 'exchange current', 'maximum exchange current', 'charge transfer conductance', 'maximum charge transfer conductance', 'double layer capacitance', and 'maximum double layer capacitance'.

Description

ETHOD AND APPARATUS USING A CIRCUIT MODEL TO EVALUATE CELL/BATTERY PARAMETERS

BACKGROUND OF THE INVENTION

"Total storage capacity" (TSC) , "absolute stored charge" (ASC) , "state-of-charge" (SOC) , "absolute cranking current" (ACC) , "fully charged cranking current" (FCCC) and "state-of-health" (SOH) are important performance parameters of an electrochemical cell/battery. These six parameters are assumed herein to have the following definitions:

• "Total storage capacity" (TSC) denotes- the total amount of charge that a fully charged battery can supply under specified discharge conditions. TSC is

. usually expressed in ampere-hours or in reserve capacity minutes.

» "Absolute stored charge" (ASC) — also expressed in ampere-hours or reserve capacity minutes — denotes the amount of charge that a battery can supply in its current charge state. As a battery is discharged, its ASC decreases — much like the level of liquid in a fuel tank.

• "State-of-charge" (SOC) , or "relative stored charge", is the ratio of a battery' s ASC to its TSC — generally expressed as a percentage. A battery's SOC indicates whether charging is advisable and identifies the point at which charging should be discontinued.

• "Absolute cranking current" (ACC) denotes the high- rate discharge current in amperes that a battery can sustain at a specified voltage for a specified time in its present charge state. As a battery discharges, its ACC decreases. β "Fully charged cranking current" (FCCC) denotes the value that the ACC would assume if the battery were fully charged. • "State-of-health" (SOH) describes a battery's full charge capability, either its TSC or its FCCC, vis-avis its rated specifications . SOH identifies the point at which battery replacement is advisable.

Both ASC and TSC have traditionally been measured by performing timed-discharge tests on batteries that are partially or • fully charged, respectively. -Because of the time and expense involved in performing complete discharge tests, other techniques for determining ASC and TSC have been proposed. In U.S. Patent 6,255,801, Chalasani claims to determine battery capacity from observations of the coup de fouet effect. 0' Sullivan, in U.S. Patent 6,211,654, discloses a method for predicting battery capacity from the discharge characteristics over a relatively short time period at the beginning of a full discharge.

, Techniques employing time-varying signals have ^' also been proposed. Sharaf, in U.S. Patent 3,808,522, purportedly determines the ampere-hour capacity of a lead-acid battery from ac measurements of its internal resistance. Yang, in U.S. Patent 5,126,675, also uses single-frequency internal resistance measurements to predict battery capacity. Muramatsu reports in U.S. Patent 4,678,998 that he can determine both the remaining amp-hour capacity and the remaining service life of a battery from measurements of the magnitude of the ac impedance at two different frequencies. Fang, in U.S. Patent 5,241,275, teaches a method for determining remaining capacity from complex impedance measured at two or three frequencies in the range from 0.001 to 1.0 Hz. Hampson, et al., in U.K. Patent Application GB 2,175,700A, report determining battery capacity from the frequency of the maximum value of capacitive reactance in the "impedance characteristic curve". Yoon et al., in U.S. Patents 6,208,147 and 6,160,382, claim that a battery' s capacity can be found by analyzing the complete impedance spectrum over a wide frequency range. Presumably, any of these techniques, if effective, could also be used to determine SOH by comparing the TSC thus determined with a rated value.

Champlin, in U.S. Patent 5,140,269, shows that the percent capacity of a standby battery — and hence its SOH — can be determined from its ac conductance measured at a single frequency if the ac conductance of a reference, fully charged, identically constructed, new battery is known. This method, although quite effective, requires that such ac conductance data be available, apriori . ^<λAbsolute cranking current" (ACC) and "fully charged cranking current" (FCCC) have been traditionally measured with timed, high-rate, discharge tests. Such tests have many disadvantages, however. They require heavy and cumbersome equipment, cause dangerous sparking, give imprecise results, and leave the battery in a significantly worse condition than existed before the test was performed. In response to the need for a better method, Champlin pioneered a testing technique based upon single-frequency ac conductance measurements. Various aspects of this well-accepted methodology have been disclosed in U.S. Patents 3,873,911, 3,909,708, 4,816,768, 4,825,170, 4,881,038, 4,912,416, 5,572,136, 5,585,728, 5,598,098, and 5,821,756.

With lead-acid batteries, SOC has been traditionally evaluated by observing the battery' s open- circuit voltage or the specific gravity of its electrolyte. However, neither of these quantities provides information about the battery' s TSC, ASC, ACC, FCCC, or SOH. Furthermore, specific gravity measurements are messy and impossible to perform on sealed cells. Moreover, open-circuit voltage cannot be measured under load conditions and, at any rate, is imprecisely related to SOC because both "surface charge" and temperature affect it. Because of these drawbacks, several techniques for correcting voltage of lead-acid batteries to obtain SOC have been proposed. These include techniques described by Christianson et al . in U.S. Patent 3,946,299, by Reni et al. in U.S. Patent 5,352,968, and by Hirzel in U-. S. Patent 5,381,096. However, such voltage correction methods are not very accurate. i. J. Luci-ilL a-C ^Xι _ αj.c 1 J.±L J.C n x W-I-L-IJ. electrochemical systems other than lead-acid in which voltage may bear little relationship to SOC. Due to these and other problems, techniques based upon ac or time-varying signals have been proposed for determining SOC. For example, atner claims to determine SOC of NiCd batteries from ac bridge measurements of farad capacitance in U.S. Patent 3,562,634. U.S. Patent 3,984,762 to Dowgiallo purports to determine SOC from the phase angle of the complex impedance at a single frequency. In U.S. Patent 4,743,855, Randin et al. assert that SOC can be determined from the argument (i.e., phase angle) of the difference between complex impedances measured at two different frequencies. Bounaga, in U.S. Patent 5,650,937, reportedly determines SOC from measurements of the imaginary part of the complex impedance at a single frequency. Basell et al. purport to determine SOC from the rate of change of impedance with frequency in U.S. Patent 5,717,336. Ding et al., in U.S. Patent 6,094,033, broadly assert that SOC can be determined from a battery' s "impedance response, which can include series and parallel equivalent circuit parameters, i.e., resistance, capacitance, and phase angle, among others". Finally, techniques purporting to determine SOC from the transient response to an applied pulsed voltage and/or current are disclosed by Andrieu and Poignant in U. S. Patent 5,530,361 and by Simon in French Patent Application FR 2,749,396A. The fact that none of these metiods has g^' m wide acceptance, ±iowever, suggests that they may not be altogether satisfactory methods for determining SOC. SUMMARY OF THE INVENTION Testing apparatus senses the time-varying electrical response of an electrochemical cell/battery to time-varying electrical excitation. The cell/battery may, or may not, be in service. Computation circuitry responsive to the time-varying electrical response evaluates elements of a unique circuit model representation of the cell/battery. Performance parameters and physical parameters are computed from these element values . Computed performance parameters include, but are not limited to, ^' "total storage capacity", "absolute stored charge", "state-of-charge", "absolute cranking current", "fully charged cranking current", and "state-of-health". Computed physical parameters include, but are not limited to, "exchange current", "maximum exchange current", "charge transfer conductance", "maximum charge transfer conductance", "double layer capacitance", and "maximum double layer capacitance". Computed parameters are either displayed to the user, employed to initiate an alarm, or used to control a process such as charging the cell/battery. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of apparatus for evaluating performance parameters and physical parameters of an electrochemical cell or battery in accordance with one embodiment of the present invention. FIG. 2 depicts a generic 2n-element small signal circuit model representation of an electrochemical cell or battery employed in the present invention. FIG. 3 depicts the six-element small signal circuit model representation of a particular fully ■ charged 12V VRA battery determined from complex immittance measurements at 5 Hz, 50 Hz, and 500 Hz. FIG. 4 is a plot of the variation of the three subcircuit time-constants defined in FIG. 3 as charge is removed from the battery.

/ FIG. 5 is a plot of values of the. three conductances defined in FIG. 3 as charge is removed from the battery.

FIG. 6 is plot of values of capacitances C2 and C3 defined in FIG. 3 as charge is removed from the battery.

FIG. 7 is a plot of the "exchange current" I_Q derived from "charge transfer conductance" G3 of FIG. 5 as charge is removed from the battery.

FIG. 8 is a plot of the "state-of-charge" SOC derived from "charge transfer conductance" G3 of FIG. 5 and "double layer capacitance" C2 of FIG. 6 as charge is removed from the battery.

FIG. 9 is a plot of the "absolute stored charge" ASC determined from "double layer capacitance" C2 of FIG. 6 as charge is removed from the battery.

FIG. 10 is a plot of the "total storage capacity" TSC derived from "charge transfer conductance" G3 of FIG. 5 and "double layer capacitance" C2 of FIG. 6 as charge is removed from the battery. FIG. 11 is a plot of the "absolute cranking current" ACC derived from series conductance G1 =1/R1 of FIG. 5 as charge is removed from the battery.

FIG. 12 is a plot of the function used to correct the "absolute cranking current" ACC for "state- of-charge" SOC to obtain the "fully charged cranking current" FCCC of FIG 13.

FIG. 13 is a plot of the corrected "fully charged cranking current" FCCC as charge is removed from the battery.

FIG. 14 is a diagram of the "commonly accepted" circuit model showing the placement of a "charge transfer resistance" and "double layer capacitance" in parallel with one another. FIG. 15 is a diagram of the n = 3 circuit model according to the present invention showing the placement of the "charge transfer conductance" and the "double layer capacitance" in two separate G-C subcircuits that are actually in series with one another.

FIG. 16 is a block diagram of apparatus for evaluating performance and physical parameters of an electrochemical cell or battery wherein an external source produces time-varying electrical excitation. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

Method and apparatus for quickly and accurately determining performance parameters and physical parameters that does not discharge the battery, does not require the battery to be fully charged, and tests batteries while "on line" would be of great value. The present invention addresses this need.

FIG. 1 discloses a block diagram of apparatus 5 for evaluating performance parameters and/or physical parameters according to one embodiment of the present invention. Measuring circuitry 10 electrically couples to cell/battery 20 at positive terminal 15 and negative terminal 25 by means of current-carrying contacts A and B and voltage-sensing contacts C and D. Cell/battery 20 may, or may not, be in service. Under control of microcontroller circuitry 30 via control path 35, measuring circuitry 10 passes periodic time-varying excitation current i(t) through contacts A and B and senses periodic time-varying response voltage v(t) across contacts C and D. Amplification and analog to digital conversion circuitry contained within measuring circuitry 10 formulates digital representations of i(t) and v(t) samples and communicates them to microcontroller circuitry 30 via data path 40. By appropriately processing these digital representations, microcontroller circuitry 30 computes real and imaginary parts of complex immittance either, impedance Z or admittance Y — at a measuring frequency f^ ; where f_k is a discrete frequency contained in the periodic waveforms of i(t) and v(t) . Microcontroller circuitry 30 commands measuring circuitry 10 to repeat these measurements at each one of n discrete measuring frequencies, where n is an integer number equal to or greater than 3. This action defines 2n experimental quantities: the values of the n real parts and the n imaginary parts of complex immittance at each of the n measuring frequencies. Microcontroller circuitry 30 can employ algorithms disclosed in U. S. Patents 6,037,777 and 6,222,369 to numerically combine the 2n experimental quantities to evaluate the 2n elements of the circuit model representation of the cell/battery disclosed in FIG. 2. This unique circuit model comprises a single series R-L subcircuit in series with a plurality, n-1, of parallel G-C subcircuits.

Microcontroller circuitry 30 computes one or more performance and/or physical parameters of cell/battery 20 from values of one or more elements of this circuit model representation of FIG. 2. Additionally, microcontroller circuitry 30 can compare one or more computed parameters with battery rating information inputted by a user on input device 45 to determine the battery's "state-of-health" SOH. Finally, microcontroller circuitry 30 can output appropriate performance or physical parameters to a user on display 50, and/or apply one or more computed parameters to initiate an alarm 55 or to control a process 60 such as charging of the battery.

FIG. 3 discloses the six-element circuit model of a fully charged 12V battery as determined with the apparatus of FIG. 1 using n = 3. The experimental battery was a VR A (valve-regulated lead-acid) automotive battery of spiral-grid construction that was rated 700 CCA, 95 minutes reserve capacity, and 50 Ah capacity @ 20 hour rate. The three measurement frequencies were 5, 50, and 500 Hz.

One notes from FIG. 3 that the n=3 circuit model comprises three two-element subcircuits in series:

• A series Rl- l subcircuit .

• A parallel G2—C2 subcircuit.

• A parallel G3-C3 subcircuit.

One notes further that the three subcircuits are characterized by three very different time constants. The shortest -time constant, τ_x = L1/R1 = 215.4μS , belongs to the series Rl-Ll subcircuit. The next longest time constant, τ₂ =C2/G2 = 3.33mS , belongs to the parallel G2-C2 subcircuit; and the longest time-constant, τ₃ = C3/G3 = 51.3mS , belongs to the parallel G3-C3 subcircuit. Accordingly, the three subcircuits represent quite different physical processes that can be differentiated from one another by their time constants . By analyzing similar three-frequency measurements on many lead-acid batteries, a consistent physical interpretation of the circuit model of FIG. 3 has emerged. The theoretical basis for this interpretation can be found in the paper by Champlin and Bertness, "Results of Discrete Frequency Immittance Spectroscopy (DFIS) Measurements of Lead Acid Batteries^-"-', Proceedings of the 23^rd International

Telecommunications Energy Conference (INTELEC) ,

Edinburgh Scotland, paper 53, (October 2001) . According to this interpretation, series resistance Rl represents the current-induced drop in electric potential across the metal connectors, plates, separators, active material, and electrolyte - with the majority of Rl associated with the active material and electrolyte. Accordingly, conductance Gl =l/Rl correlates directly with high-rate load-test current such as cold-cranking amps (CCA) . Series inductance LI describes energy storage in the battery' s surrounding magnetic field, with no electrochemical component to LI yet observed.

Both the parallel G2-C2 subcircuit and the parallel G3—C3 subcircuit are believed to be linearized representations of nonlinear electrochemical processes occurring at the negative electrode surfaces. These two subcircuits together describe the response of negative electrode surfaces to changes in an externally applied electrochemical overpotential .

Electrochemical overpotential at an electrode surface consists of two parts: an electrical overpotential and a chemical overpotential. The electrical part arises from electrical forces on charged particles, and the chemical part arises from concentration differences. The faster G2-C2 subcircuit responds to the electrical part of an electrochemical overpotential and the slower G3-C3 subcircuit responds to the chemical part. These two subcircuits add in series in the circuit model of FIG. 3 because the electrical overpotential and the chemical overpotential add together to form the electrochemical overpotential. Voltages (potentials) add in series.

FIGS. 4, 5, and 6 display the results of three-frequency immittance measurements performed on the battery whose model is depicted in FIG. 3 during a discharge sequence. The battery .was initially charged, allowed to reach equilibrium, and was then repeatedly discharged at an approximate 8-hr rate in increments of approximately 3 ampere-hours. The battery was permitted to rest for 2.5 hours after each discharge period before the three-frequency immittance measurements were taken.

FIG. 4 is a plot of the variation of the three time constants defined in FIG. 3 as charge is removed from the battery. One notes that the time constants remain widely separated as charge is removed, and that the longest of the three, τ₃, is nearly independent of state-of-charge .

FIG. 5 discloses the variation of the three conductances, ^" Gl =l/Rl, G2 , and G3 as charge is removed from the battery. Of particular interest is the fact that G3 becomes small near full charge as well as near full discharge. This remarkable behavior sets G3 apart from Gl and G2 , both of which reach maxima at full charge. Such unusual behavior can be explained by assuming that G3 describes faradaic processes occurring at the negative electrode surfaces and considering the kinetics of such processes.

The Butler-Volmer equation for faradaic processes at the negative electrode surfaces can be written: =¹._f-¹._b=¹.oleanfv^" ₃ ³-e -{^ll-aj*nfv ³) (1)

In equation (1) , i is the current in the external circuit (assumed positive in the "charge" direction) while if and ij, are "forward" and "backward" ionic "reaction" currents, respectively, in the electrolyte. Physical parameter Ϊ is the^' "exchange current"; is a "charge transfer coefficient" (less than unity) ; n is the number of electrons transferred per molecule reactant (two); and f = (F/RT), with F being Faraday's constant, R the universal gas constant, and T the absolute (Kelvin) temperature. Voltage V3 in equation (1) is believed to be the chemical component of the electrochemical overpotential (positive for "charge" and negative for "discharge") at the negative electrode surfaces .

The chemical overpotential, V3 , drives the two ionic reaction currents. Under charging conditions (V₃>0), the forward ionic current, if =iø exp(αnfV3) , is larger than iø ; while the backward ionic current, i_b =i₀ exp(-{l-α]nfV3) is less than iø . Accordingly, their difference yields a positive ("charging") current in the external circuit. Under discharging conditions (V₃<0), the opposite is true. In equilibrium (V3=0), both forward and backward ionic currents are equal to the exchange current i₀ , and the current in the external circuit is zero. Note that even with i=0, there are still two equal ionic "exchange currents" flowing in opposite directions on the electrolyte side of the electrode/electrolyte interface.

For small variations in V₃ , the equilibrium small-signal faradaic conductance, or "charge transfer conductance", follows by differentiating equation (1) with respect to (N_CV₃) , where Nc is the number of battery cells in series, and evaluating this derivative at V₃=0. The result is

One sees that at any given temperature, the "charge transfer conductance" G3 is proportional to the "exchange current" Ϊ_Q . The "charge transfer conductance" and electrode "exchange current" are both physical parameters of great importance.

.Equation (2) can be inverted to yield the

"exchange current" ig in terms of "charge transfer conductance" G3. For a 12V (6-cell) battery at room temperature (T = 233° K) , equation (2) yields

i₀ = (N_cRT/nF)-G3 = (.07574)-G3 amperes (3) Fig. 7 displays i₀ calculated from equation (3) using the experimental G3 data of Fig. 5. One sees that iø approaches zero at both full charge and full discharge and reaches a maximum value of iø_max=27.5 amperes at approximately 50% state-of-charge.

The remarkable variation of G3 disclosed in FIG. 5 can now be explained. As the battery approaches full charge, the forward, or "charge", component of reaction current goes to zero because the number of reaction sites available for the charge reaction (PbS0₄ sites) approaches zero. Since forward and backward ionic currents are equal in equilibrium, the equilibrium backward current likewise goes to zero. Similarly, as the _.battery approaches full discharge, the backward, or "discharge", component of reaction current goes to zero because the number of sites available for the discharge reaction (Pb sites) approaches zero. Again, since forward and backward ionic currents are equal in equilibrium, the equilibrium forward current likewise goes to zero. Accordingly, the "exchange current" i goes to zero in both cases. From equation (2), the "charge transfer conductance" G3 — which is simply proportional to i

— likewise approaches zero at full charge and at full discharge, reaching .maximum at 50% state-of-charge.

By reasoning in accordance with this mechanism, one can write "charge transfer conductance" G3 as - G3 = 4 G3_max •(Np_b/N-)-(Np_fcSθ4 /N^") (4)

where N^~ =Np_b +Np_bso₄ i-^{s the} total number of negative-

electrode reaction sites, Np^~ _b is the number of such

sites occupied by Pb molecules, and Np_so. is the number occupied by PbSO₄ molecules. The factor 4 in equation (4) arises because G3 equals G3_max when

I now define the performance parameter "state-of-charge" SOC as the fraction of negative- electrode reaction sites occupied by Pb molecules. That is,

Equation (4) can therefore be written

G3 = 4 •G3_max •(SOC)•(1-SOC) (6)

which obviously approaches zero when SOC approaches unity as well as zero.

FIG. 6 discloses the variation of capacitance C2 and capacitance C3 as charge is removed from the battery. Capacitance C3 is seen to mimic the behavior of "charge transfer conductance" G3 over the complete range of Ah removed. This variation can be understood by noting that C3 =τ₃-G3, and that τ^ varies little as charge is removed (see FIG. 4) . Time constant τ₃ is believed closely related to the reaction time of faradaic processes at the negative electrode surfaces (See U.S. Patents 6,137,269 and 6,294,697). Capacitance C2 in FIG. 6 is observed to vary in an entirely different manner from C3 as charge ■ is removed. Instead of increasing and then decreasing, C2 assumes its largest value with zero amp-hours removed and simply decreases monotonically as charge is removed from the battery. Such behavior suggests that the G2-C2 subcircuit is responsive to the nonfaradaic or electrical component of an electrochemical overpotential at the negative-electrode surfaces. In this interpretation, capacitance C2 represents the negative-electrode's ionic "double layer capacitance" _. - a physical parameter of considerable interest.

An ionic double layer can only form on metallic Pb molecules. It cannot form on nonconducting PbSO₄ molecules. It follows that C2 is proportional to Np , the total number of reaction sites that are occupied by Pb molecules. Accordingly, "double layer capacitance" C2 can be written

where C2jø is "maximum double layer capacitance' the value that would occur at 100% state-of-charge. Equation (7) indicates that physical parameter C2_j relates to battery size and condition

by being proportional to N^_ , the total number of reaction sites available for charge/discharge reactions. Similarly, equation . (4) indicates that physical parameter G3_max relates to size and condition

by being proportional to (N ) . Accordingly, I define the following useful quantity:

ξ ≡ ^(C2l0o)2 (8)

4-G3

which, for any particular class of cell/battery, is independent of battery size, condition and "state-of- charge". I have found that ξ = 1.368 -10 for batteries of the type under consideration herein.

Equations (6) , (7) , and (8) can be combined algebraically to provide separate expressions for SOC , C2_jø , and G3_max in terms of ξ , C2 , and G3. The results are-

SOC = 7—- (9) l+ξ-(G3/C2²)

and By combining equation (3) and equation (11) , "maximum exchange current" iø_max can be written as -

{N_cRT/nF}-{C2+ξ-(G3/C2}² ^ϊOmax —; 4⁾

This physical parameter is of considerable theoretical interest since it describes the level of maximum chemical activity at the electrode/electrolyte interface.

FIG. 8 displays performance parameter SOC (in percent) calculated from equation (9) by using the experimental G3 and C2 data disclosed in FIGS. 5 and 6. Note that combining G3 and C2 data to determine SOC resolves the multi-value ambiguity of equation (6) . The "state-of-charge" determined in this manner varies from a maximum value of 93.6%, with zero ampere- hours removed from the battery, to a minimum value of 4.1% with 45.2 ampere-hours removed. The observed, nearly linear, variation of SOC with ampere-hours removed provides excellent corroboration of the theory.

The performance parameters "absolute stored charge" ASC and "total storage capacity" TSC are proportional to C2 and C2ι ₀ , respectively. Accordingly, I write -

ASC =K_CS C2 (13) and, from equation (10) -

TSC = K_cs - {C2 + ξ - (G3/C2)} (14)

where _cs is a "charge storage" constant that depends upon the battery type, and upon the units of charge storage (i.e., ampere-hours @ an 8 hr rate, ampere- hours @ a 20 hr 'rate, reserve capacity minutes, coulombs, etc.) I find that for batteries of the type under consideration herein, and with ASC and TSC expressed in ampere-hours @ an 8 hr rate, K_cssll.3.

Taking the ratio of equation (13) to equation (14) shows that -

Thus, the performance parameter "state-of-charge" SOC, that is defined physically in terms of numbers of reaction sites by equation (5) , is actually the ratio of the two performance parameters "absolute stored charge" ASC and "total storage capacity" TSC .

FIG. 9 displays performance parameter "absolute stored charge" ASC (in ampere-hours) calculated from equation (13) using the experimental C2 data disclosed in FIG. 6. The "absolute stored charge" determined in this manner varies from a maximum value of 47.2 ampere-hours, with zero ampere-hours removed from the battery, to a minimum value of 1.9 ampere-hours with 45.2 ampere-hours removed. Note that "absolute stored charge" ASC decreases as the battery discharges - much like the level of liquid in a fuel tank. Accordingly, ASC would comprise an excellent "electric fuel gauge" readout.

FIG. 10 displays the performance parameter "total storage capacity" TSC (in ampere-hours) calculated from equation (14) using the experimental G3 and C2 data disclosed in FIGS. 5 and 6. Note that the observed TSC varies very little as the battery is discharged. Ideally, TSC would not vary at all. The small variations observed are believed caused by minor inaccuracies in the n=3 model. Performing immittance measurements at four or _. five frequencies and then employing an eight- or ten-element circuit model in the analysis greatly reduces such variations.

Performance parameter "absolute cranking current" ACC has been found experimentally to be proportional to Gl = l/Rl. Thus -

ACC=K_CC G1 (16)

where K_cc is a "cranking current" coefficient that depends upon the type of battery and the definition of "cranking current". For batteries of the type under consideration and with "cranking current" defined to be SAE "cold cranking amperes", K_cc = 2.7. FIG. 11 displays performance parameter ACC determined from equation (16) by using the Gl data of FIG. 5. One sees that ACC decreases significantly as charge is removed from the battery.

For batteries that are less than fully charged, the performance parameter "fully charged cranking current" FCCC can be . determined by correcting the ACC according to -

FCCC = ACC •F(SOC) (17)

where F(SOC) is a "correction factor" that depends upon "state-of-charge". FIG. 12 discloses an e perically determined F(SOC) for correcting the ACC of batteries of the type under consideration. This function can be expressed mathematically as -

F(SOC) = (18)

1-0.5 •(SOC-1) -1.0 •(SOC-1)² + 0.4 •(SOC - l)^δ

FIG. 13 displays performance parameter FCCC obtained by correcting the data of FIG. 11 according to equation (17) using the correction factor defined by equation (18) (FIG. 12) . One sees that the corrected FCCC varies by only ±2.7% from 695 amperes over the entire range of discharge down to SOC = 20%.

One determines the performance parameter "state-of-health" SOH by either of two methods. The first method is to compare the computed "total storage capacity" TSC with an appropriate user-inputted capacity rating such as the battery's rated "ampere- hour capacity" or rated "reserve capacity" in minutes. The second method is to compare the computed "fully charged cranking current" FCCC with an appropriate user-inputted cranking rating such as the battery's rated "cold cranking amperes" (CCA) . In either case, the battery is assumed to have "failed" the SOH test if the calculated parameter value is less than a predetermined fraction, say 75%, of the rated parameter value. Either SOH method, or both methods, may be employed — depending upon the type of service that the battery is expected to experience.

This completes the disclosure of my invention. FIGS. 14 and 15, however, will serve to place the innovative nature of my invention in perspective. FIG. 14 illustrates the "commonly accepted" equivalent circuit model of a cell/battery at the present time. (See, e.g., U.S. Patent 6,167,349 to Alvarez. See also E. Karden, et al., "A method for measurement and interpretation of impedance spectra for industrial batteries", J Power Sources 85 ^'(2000), pp. 72 - 78.) The "commonly accepted" model, also known as the "Randies equivalent circuit", places a "charge transfer resistance" and a "double layer capacitance" in parallel with one another. In my opinion, this model is fundamentally wrong and cannot possibly lead to correct evaluation of any of the • performance parameters or physical parameters evaluated herein.

My measurements and analysis have shown that the "charge transfer conductance" (the reciprocal of "charge transfer resistance") and the "double layer capacitance" are elements of two separate G-C subcircuits that are actually in series with one another in the correct circuit model (see FIG. 15) . Furthermore, the "double layer capacitance" of the correct model varies directly with the number of reaction sites available for the discharge reaction (equation (7) ) ; while the "charge transfer conductance" of the correct model varies directly with both the number of sites available for the discharge reaction and also the number available for the charge reaction (equation (4) ) . These' important new discoveries set my work totally apart from all of the prior art.

To summarize, I have herein disclosed new methods and apparatus for evaluating the following six performance parameters : 1. "Absolute stored charge" ASC (equation (13)) .

2. "Total storage capacity" TSC (equation (14)).

3. "State-of-charge" SOC = ASC/TSC (equation (9)).

4. "Absolute cranking current" ACC (equation (16)) . 5. "Fully Charged Cranking Current" FCCC (equation (17) ) . 6. "State-Of-Health" SOH. Furthermore, in the course of this disclosure I have also revealed new methods and apparatus for evaluating the following six important physical parameters:

1. "Exchange current" iø (equation (3)).

2. "Maximum exchange current" iø_max (equation (12)).

3. "Charge transfer conductance" G3. 4. "Maximum charge transfer conductance" G3_max (equation (11) ) .

5. "Double layer capacitance" C2.

6. "Maximum double layer capacitance" C2₁₀ø (equation (10) ) .

My invention, however, is not limited to evaluating only these twelve parameters.

Although my disclosure has referred to particular apparatus and algorithms described in U.S. Patents 6,002,238, 6,037,777, 6,172,483, 6,222,369, and 6,262,563, other methods will be apparent to those skilled in the art. For example, one could evaluate elements of the correct circuit model by using pulsed or stepped excitation such as that described in U.S. Patent 6,118,275 to Yoon or in U.S. Patent 6,167,349 to Alvarez; or by using random or pseudo-random excitation such as that described in PCT Application WO 99/18448 to Scott. One could also employ bridges, frequency response analyzers, or other types of ac instrumentation to measure complex admittance or complex impedance at multiple frequencies . The range of measurement frequencies is unrestricted and could include frequencies in the millihertz and microhertz region

(£<1) as well as frequencies in the range of hertz (£>1) .

The embodiment disclosed in FIG. 1 utilizes time-varying electrical excitation i(t) generated internally to apparatus 5 by internal measuring circuitry 10. Alternatively however, the requisite time-varying electrical excitation could be developed by an external source 65 as disclosed in FIG. 16. Such an external source could simply comprise normal "noise" currents flowing through a cell/battery during service, such as ^' taught in PCT Application WO 93/22666 to Robinson. It could also comprise "ripple" signals arising from an unfiltered battery charger during bench- charging of the cell/battery or charging by an alternator in a vehicle. Although, time-varying current i(t) has been presented as an electrical excitation and time-varying voltage v(t) as an electrical response, the roles of i(t) and v(t) could be reversed. Time-varying voltage v(t) could be the electrical excitation and time-varying current i(t) the electrical response. All such modifications and variations are included in my invention.

It is well known that elements of any circuit model can be readily subdivided and rearranged to obtain a derivative circuit that appears to be quite different, but is in effect electrically equivalent to the original circuit. My invention includes all such derivative circuits that are electrically equivalent to the models of FIG. 2 and/or FIG. 3. One could use a circuit model containing, e.g., eight or ten elements rather than the six elements treated herein. This straightforward modification is likewise included in my invention.

Instead of the algorithms set forth herein to evaluate elements of the circuit model, one could employ a conventional "complex least-squares curve fitting" procedure. If accuracy is not essential, one could even take advantage of the fact that the time constants are widely separated from one another and simply assume that the subcircuits are not coupled. Thus, with some batteries, it may be possible to obtain a reasonably accurate circuit model from a very simple computational analysis of measured time-varying electrical response.

The invention is not limited to the particular circuitry implementations disclosed herein. Measuring circuitry, electrical excitation circuitry, electrical response sensing circuitry, computational circuitry, etc., can be implemented in hardware, software, or combinations thereof. Such circuitry can share common elements or be implemented with common components such as a common microprocessor, microcontroller, or personal computer. Workers skilled in the art will recognize that these and other variations may be made in form and detail without departing from the true spirit and scope of my invention.

Claims

WHAT IS CLAIMED IS:

1. Apparatus for determining one or more parameters of an electrochemical ^• cell or battery comprising: response sensing circuitry coupled to said cell or battery and adapted to sense time-varying electrical response of said cell or battery generated as a result of time-varying electrical excitation of said cell or battery; and, computation circuitry coupled to said response sensing circuitry and adapted to utilize said response to evaluate elements of a circuit model comprising the series combination of a series R-L subcircuit and a plurality of parallel G-C subcircuits, said computation circuitry further adapted to evaluate one or more said parameters by combining values of two or more said elements.

2. A method for determining one or more parameters of an electrochemical cell or battery comprising: sensing time-varying electrical response to time-varying electrical excitation of said cell or battery; processing said time-varying electrical response to evaluate elements of a circuit model of said cell or battery comprising the series combination of a series R-L subcircuit and a plurality of parallel G-C subcircuits; and, determining values of one or more said parameters by combining values of two or more said elements.

3. The apparatus or method of claim 1 or claim 2 wherein one said parameter is state-of-charge.

4. The apparatus or method of claim 1 or claim 2 wherein one said parameter is total storage capacity.

5. The apparatus or method of claim 1 or claim 2 wherein one said parameter is maximum exchange current.

6. The apparatus or method of claim 1 or claim 2 wherein one said parameter is maximum charge transfer conductance.

7. The apparatus or method of claim 1 or claim 2 wherein one said parameter is maximum double layer capacitance.

8. The apparatus or method of claim 1 or claim 2 wherein one said parameter is- fully charged cranking current .

9. The apparatus or method of claim 1 or claim 2 wherein one said parameter is state-of-health.

10. Apparatus for determining one or more of the parameters exchange current, maximum exchange current, charge transfer conductance, maximum charge transfer conductance, double layer capacitance or maximum double layer capacitance of an_. electrochemical cell or battery comprising: response sensing circuitry coupled to said cell or battery and adapted to sense time-varying electrical response of said cell or battery generated as a result of time-varying electrical excitation of said cell or battery; and, computation circuitry coupled to said response sensing circuitry and adapted to utilize said response to evaluate elements of a circuit model comprising the series combination of a. series R-L subcircuit and a plurality of parallel G-C subcircuits, said computation circuitry further adapted to evaluate one or more said parameters from values of one or more said elements.

11. A method for determining one or more of the parameters exchange current, maximum exchange current, charge transfer conductance, maximum charge transfer conductance, double layer capacitance or maximum double layer capacitance of an electrochemical cell or battery comprising: sensing time-varying electrical response to time-varying electrical excitation of said cell or battery; processing said time-varying electrical response to evaluate elements of a circuit model of said cell or battery comprising the series combination of a series R-L subcircuit and a plurality of parallel G-C subcircuits; and, determining values of one or more said parameters from values of one or more said elements.