Investigations into methods of measuring the state of health of a nickel-cadmium Industrial Battery Anthony Green, SAFT, France AUTHOR BIOGRAPHICAL NOTES Anthony Green graduated from the University of Aston in 1967 with an B.Sc. (Hons) in Physics and in 1969 with an M.Sc. in Physical methods of Analysis. He joined Alcad Ltd, in the UK, in 1976, where he was Technical Manager, before moving to the Saft Technical Centre at Bordeaux, France, as Product Development Manager in 1988. In 1993 he moved to the Marketing Department in Paris where he was responsible for technical support. He is currently Marketing and Sales Development Manager for the Saft Standby Battery Division. ABSTRACT The health of a flooded type lead acid battery can be established by observing the level of plate corrosion and measuring the electrolyte density. However, due to its electrochemistry and plate structure, this is not possible with nickel-cadmium industrial battery. With the advent of the valve regulated lead acid (VRLA) battery, these traditional methods are no longer available and this has led to the development of impedance/capacitance methods. Tests have been carried out to establish if such equipment can be used to give useful information for the flooded nickel-cadmium battery. These experiments have involved introducing known «faulty» cells into a battery and observing the effect. The results have been compared with simple voltage and current measurements. Parallel to this another approach is being evaluated. This involves modeling the discharge curves of the battery under various conditions and establishing the criteria whereby the voltage level achieved during a partial discharge will indicate the performance available in the battery. 1 INTRODUCTION The nickel-cadmium flooded industrial battery has a number of features which contribute to its long life and good performance. Amongst these are an electrolyte which does not take part in the chemical reaction and a steel plate support structure. The electrolyte is an aqueous solution of potassium hydroxide containing small quantities of lithium hydroxide to improve cycle life and high temperature operation. The electrolyte is only used for ionic transfer, it is not chemically changed or degraded during the charge/discharge cycle. In the case of the lead acid battery the positive and negative active materials chemically react with the sulphuric acid electrolyte with a resulting chemical change and degradation. As a result of this difference, it has always been possible to measure the state of charge of a flooded lead acid battery by measuring the electrolyte density which changes as the chemical reaction proceeds, but this has never been possible with the nickel-cadmium battery as the electrolyte is unchanged. The support structure of both the positive and negative plates in the nickel-cadmium cell is steel. This is unaffected by the electrochemistry, retains its characteristics throughout the life 2.3.1
of the cell and is physically unchanged. In the case of the lead acid battery, the basic structure of both plates are lead and/or lead alloys, which play a part in the electrochemistry of the process, and are naturally corroded during the life of the battery. Thus, with a transparent container, it is possible to visually see the state of health in a lead acid cell by the level of corrosion in the plates and the terminals. With the introduction of the valve regulated lead acid (VRLA) battery the situation changed for the lead acid battery. It was no longer possible to measure the electrolyte density, as there was no free electrolyte, and, due to the use of opaque sealed plastic containers, it was no longer possible to see the level of corrosion. Due to the inherent advantages in having the VRLA low maintenance concept there has been a large swing towards its use. However, the user s were left with the uncomfortable feeling of not knowing what was happening within this opaque box and efforts have been made to provide a means of testing the integrity of the VRLA battery while in service. The most successful methods of achieving this to date are systems based on conductance or impedance measurements. These are claimed to give a reliable indication of battery condition, capacity and state of health, to find worn out or defective cells and to predict the end of battery life. It is not surprising that they are now widely used. As these products are now in regular use, the question arises of whether such a technique can be used with nickel-cadmium cells. 2 CONDUCTANCE/INDUCTANCE MEASUREMENTS The AC impedance of a cell can be determined by forcing a specified AC current (I) through a battery and measuring the AC voltage (E) developed across the terminals. The impedance is given by the expression Z = E/I. Conductance is the reciprocal of resistance portion of impedance but, in fact, a conductance test is similar in effect to an impedance test. In the case of the VRLA battery, changes in the components of the cell, such as grid corrosion, shedding of active material, strap corrosion, or drying out of the limited supply of electrolyte, will be reflected in an increase in the internal resistance of the battery and hence the conductance or impedance. In the case of the nickel-cadmium battery, the situation is a little different. Figure 1, shows a typical discharge curve for a valve regulated nickel-cadmium battery. As discussed earlier, the structural parts of the nickel cadmium cell, the plate structure, the bus bars and the terminals, are made from steel. Thus the resistive path is metallic and this does not change significantly with electrochemical changes in the cell or with time. Thus the curve is relatively flat until the cell reaches a fully discharged state and so differences in state of charge cannot be seen by a measure related to internal resistance. In practice, over time, there is a change in the voltage level of the curve, but this does not significantly change the gradient and so differences are difficult to detect. However, some testing was carried out to see what can be detected by conductance measuring equipment and, if this information was significantly more effective than other simpler methods. 2.3.2
Figure 1 Typical discharge curve for a nickel-cadmium battery. 2.1 State of Charge A number of experiments were carried to compare the state of charge of the battery with the conductance measured. A test result is given in Figure 2 which shows the results from 15 cells which have been progressively discharged until there is no capacity remaining. Table 1 gives the average results for the open circuit voltage and the conductance for the same test. State of Charge Voltage Variance Conductance(Siemens) Variance 100% 1.349 0.29% 1254 1.59% 70% 1.286 0.47% 1256 1.19% 30% 1.259 0.24% 1217 1.07% 0% 1.230 0 1145 0.96% Table 1 Change in conductance and voltage with state of charge Figure 2 would tend to indicate that there is change in conductance with the state of charge. However, evaluation of the date shows that there is no significant difference between 100% and 70% state of charge and there is also no significant difference between 30% and 100% state of charge. Evaluation of Table 1 indicates that measuring the voltage is as useful an indication of the state of charge as the measure of conductance, but, with the small differences involved, this has never been considered a reliable method. 2.3.3
Figure 2 Change in conductance with state of charge. However, the results do indicate that there is a measurable drop in the value for the conductance when the battery is completely discharged. In terms of ageing of the battery and the consequent fall in available capacity, it would appear, therefore, that it is not possible for the conductance or impedance test to give a value which would be useful. 1.2 Faulty Cells To establish if the measurement of conductance could detect bad cells, a test was set up where cells, known to be faulty, were introduced into a battery. It is claimed that the conductance measurement has a value which increases if there is a fault in the cell. In this test we have taken the conductance value for cells we know to be correct to be the standard value. Thus higher values of conductance should be an indication of faulty cells. The faulty cells were of a particular failure mode. They had a non-standard, thin, felted separator and had been charged continuously at high current to cause metalisation on the surface of the separator. This causes the cells to exhibit a low voltage when placed in a battery. However, a low voltage on a cell does not necessarily mean that a fault of this type has occurred. In all, a battery of 20 cells was tested, of which, 7 cells were known to be faulty in the way described. The results are shown if Figure 3. 2.3.4
Figure 3 Change in conductance with voltage The cells can be placed into three groups. Group A are cells which are known to be good cells. They have a low value of conductance and a charge voltage value above 1.4 volts per cell. Both conductance and voltage measurements indicate that they are good cells and so both methods give a correct result. Group B are cells which are known to be good cells. The charge voltage would indicate are good cells but the high value of the conductance would indicate are not good cells cells. Thus the conductance is not giving a correct result. Group C are cells which both the charge voltage and the conductance would indicate that there is a problem with the cells and they are, in fact, the seven faulty cells. Other tests were carried out where cells had missing separators or the electrolyte level was well below the plates. In these cases the conductance measurements would indicate the bad cells but would also indicate good cells as bad. In general the result of these tests were that if the conductance was correct then the cell was correct. If the conductance was bad, then this indicated that the cell could be good or bad. Use of the conductance measurement coupled with voltage measurements has a potential for identifying good cells. 1.3 Conclusion The measurement of impedance or conductance does not yield a method for identifying the state of charge or the state of health of a nickel-cadmium battery. However, it does appear to give an aid in identifying good cells in a battery. 2.3.5
2 CHARACTERISATION OF DISCHARGE CURVE During the gradual ageing of the nickel-cadmium battery, the voltage level of the discharge curve is lowered. Thus, if there is a specific cut-off point in the application, the available autonomy will be reduced, even though the battery has not failed. The nickel-cadmium cell discharge curve typified in Figure 1 can be represented by the equation : Y = ( a + cx + ex 2 ) / (1 + bx + dx 2 ) Where y = cell voltage X = percentage discharged a,b,c,d,e are empirical constants The constants depend on the discharge rate and, they themselves can be plotted to give any discharge curve for any cell type. 1 The most reliable method of establishing the integrity of the battery is to carry out a full discharge. However, in many cases the user cannot do this as he will no longer have a backup system until the battery has recharged. Thus, there is a need for the user to have a simple partial discharge which allows him to decide if he has the full autonomy required for his application. Figure 4 Discharge curve before and after ageing 2.3.6
Figure 4 shows the principle of the test. The top curve shows the performance of the cell when new. The lower curve shows the discharge curve after the battery has aged. In the example shown, the premis is taken that the performance required by the user, under his particular conditions, is such that the battery has to be designed to give 60% of its capacity at the discharge chosen for the application when the average voltage per cell reaches 1.1 volts. Thus, in designing the battery extra capacity is used at the start to allow for the battery ageing. Since the discharge curve is uniform, there is a consistency between the voltage level at a partial discharge and that fully discharged. Thus, the concept is that, a partial discharge of 10% can be made. This discharge will not put the application at risk as the user still has 90% of his autonomy remaining and this will be quickly recharged. From the voltage level obtained for this 10% discharge it is possible to judge if the full autonomy is available and the battery is in good health. In the case shown above, a 10% discharge would arrive at 1.23 average volts per cell when the battery was new, falling to 1.2 average volts per cell when the minimum autonomy was reached. Even below 1.2 average volts per cell, it should be possible to predict what autonomy was available. This concept is currently being evaluated with a major user and discharge curves have been generated for all product ranges uses the above curve equation. 4 CONCLUSIONS Changes in the components of the VRLA battery, such as grid corrosion, shedding of active material, strap corrosion, or drying out of the limited supply of electrolyte caused by overcharge or high temperatures, are reflected in a change in the internal resistance of the battery. This has led to the development of devices based on conductance and impedance techniques to monitor the health of the VRLA battery. Testing has shown that due to its electrochemistry and the steel structure of its plate, it is not possible to obtain the same level of information for the nickel-cadmium cell as can be found with a VRLA product. However, there are indications that such measurements can detect bad cells and that a fully discharged battery can be identified. In order to meet the needs of users, a method is being developed, based on the predictability of the nickel-cadmium discharge curve, which will establish the autonomy available and the health of the battery, from a partial discharge. 2.3.7