Our invention is particularly interesting for critical emergency current batteries like for instance in a UPS-plant. (UPS = Uninterruptable Power Supply.) In this case it will be quite essential to discover potentially poor single cells in the battery, so as to be sure to have a properly operating battery when a critical need arises.

Synopsis regarding battery operation It is a common feature of these applications, that many single cells are connected together in a series connection. The total characteristics of the battery will always be limited by the condition of the"poorest"cell. A series connection of single cells to provide a battery, is based on the absolute premise that all cells are identical, and respond equally to every influence. The same current will always run through all cells. It is a fact that when one cell in a large battery fails, then the complete battery is defective. The battery is not stronger than the weakest cell.

Further, it is known from experience that about 95-99% of the cells in a battery will stay inside the specifications from the battery supplier, and the expected lifetime estimate. It is accepted that it is no longer possible to replace single cells after about 1/2 to 1 year after start of operation for a battery. At that time, the cells have changed considerably from when they were new, so that a new cell cannot be used in this series connection.

During the later years, measurement systems have appeared that in a cost effective way are able to collect data from all single cells in the battery via a very limited number of wires. The present invention is based on the availability of such a measurement system.

The main field of use of the invention The battery cells of the largest interest to the present invention, are in principle gas-tight and liquid-tight. When a superpressure arises inside the cell, a valve will open so as to provide pressure equalization. Henceforth the name"valve controlled batteries". However, it appears that valve controlled lead batteries actually have a considerably shorter lifetime than"traditional"batteries. They also exhibit a tendency for quite sudden and apparently unpredictable breakdowns. It is known from experience that valve controlled cells with a nominal lifetime of for instance 5 years, may actually turn out to have a lifetime of only 3-6 months.

The present invention deals primarily with safeguarding against such apparently unpredictable, sudden breakdowns of valve controlled battery cells. Of course it will be of great importance not only knowing whether a lifetime as stated by the supplier can be expected, but additionally it is important to receive a warning as soon as possible, when a single cell is about to fail.

Regarding gas production in batteries When a battery is recharged, the same charging current will pass through all cells. In every cell a chemical process will take place, in which electrical energy is converted and stored as chemical energy.

During the recharging process, there will be liberated gas in every cell. The pressure inside the cell rises, until an equilibrium is reached, where the gas liberated is recovered in the process. If the pressure should rise above a certain value, a valve will be opened, and gas will be released. (Henceforth the expression valve controlled batteries.) The chemical process is not 100% efficient, so that some gas will always escape through the valve. The normal degree of efficiency, with regard to gas recovery, is about 97-99%. This entails that even for a quite optimal cell, small amounts of gas (substantially hydrogen gas) will escape through the valve.

Norwegian patent application 1997 5165 This application describes a technique in which the absolute amount of gas is recorded and used as a measure of the qualitative characteristics of the battery cell, and for calculating remaining lifetime and remaining capacity. This technique

makes demands on the general tolerances in the battery production, as well as a considerable empirical material around the battery that is actually monitored. For example when used as part of a battery operation system in e. g. forklift trucks, this is of great interest, because one wants to optimize the battery operation and get a maximum from every recharging, and between every battery replacement. In such an application it is not a catastrophe if the battery should suddenly fail. This patent application uses absolute gas amount produced as a quality measure. It is also suggested that an average over all cells should be used as a reference regarding "normal release of gas".

In laboratory setups it is possible to collect the gas that has been liberated, and measure volume and chemical composition. This is a costly procedure, and such a procedure of course cannot be used as a practical method of monitoring an emergency current battery. But in principle the liberated gas could be a measure of the quality of a cell, as well as remaining lifetime expectancy. A measurement methodology based on the number of times the valve opens, as a measure of gas amount liberated, may be a simple manner for estimating escaped amount of gas.

The idea of the present invention: As a point of departure, lifetime data from the battery manufacturer are made the basis for calculating expected remaining lifetime, and available capacity.

However, if it is realized that a single cell continuously, and during a period of time, has a liberation of gas that is larger than the last period of time, then the cell is about to die. In other words, the trend has changed. Further, it is also possible to assess whether single cells have a trend that is substantially different from the average trend for all cells.

In order to use liberated gas volume as an estimator, in a simple manner, and independently of battery and valve parameters, and at a cost that is acceptable in the market, it is possible to make an analysis which automatically makes corrections for all these natural variations.

Referring to the enclosed diagrams: Fig. 1 shows liberated gas volume per week as a function of time. Every curve, for instance for 24 cells, will normally be found in between the envelope

curves. Fig. 2 illustrates in a corresponding manner the total volume of gas liberated during the same period.

Sub-curve"b"indicates that a cell is about to run outside the envelope curve, and consequently end up outside a certain deviation from the average overall cells. It will always be difficult to determine a limit value where an alarm regarding a defective battery shall be set up.

An alternative method, which is the method that this invention is concerning, is to look at the trend of gas liberation from every cell, in relation to the spread of measurement values. Sub-curve"a"indicates a cell that suddenly starts to change, but is still well inside a possible area of acceptance. Hence, our methodology of analysis aims at setting up an alarm for battery replacement when such a sudden and monotonous change arises.

The difference between Norwegian patent application 1997 5165 and the present invention Patent application-5165 describes a statistical analysis based on instantaneous measurement values, and is based on stable and reproducible quality of valves as well as battery encapsulations etc., for estimating a battery lifetime.

The present invention is a method based on trend analysis, for providing, as early as possible, a warning regarding a potentially poor single cell, so that the battery can be replaced and breakdown avoided.