BACKGROUND OF THE INVENTION The recent development of the telecommunication market is at the origin of the parallel growth of portable electrical energy sources. The diversification of the chemical nature of the portable energy sources is due to the requirements of more efficient sources of portabte electrical energy in order to satisfy applications like mobile telephones, portable computers and associated equipment.

At least four types of battery power packs are on the market : nickel- cadmium, metal-hydrides, lithium-ion and zinc-alkaline. Several other systems are announced in a near future : zinc-air, lithium-PEM, lithium polymer solid state...

In most European countries, the EEC directive on the collection of spent batteries will be enforced by local legistation. The collection of electrical and electronic equipment will start and many of the portable electrical power sources will come back to electronic waste collection centres.

It is known that spent batteries have to be separated when they are collected. However, the methodology applied to the identification and sorting of individual cells cannot be used for the identification of spent battery packs.

Therefore, it would be useful to develop a new method for the separation of packs according to their chemistry.

Many technologies have been considered for the identification of spent battery packs like X-Ray analysis, colour sensors and magnetic sensors.

Technologies using image analysis can be used but they are limited by the fact that many spent battery power packs do not have identification labels or marks.

Also it has been observed that lithium polymer and lithium ion batteries give an identical signal when their bulk ferromagnetic properties are analysed.

Finally, due to the large variety of different geometry of power packs, it is difficult to present them in a reproducible manner in front of or facing a measuring device. Consequently, one must find an identification technology which is independent of the position of the pack within the sensor system's physical detection field.

SUMMARY OF THE INVENTION It is an object of the invention to provide a technical solution to the above- identified problem, based on the fundamental physical and chemical properties of spent battery packs.

The method consists in the measurement of the apparent density of the battery power pack and in the establishment of a correlation between the apparent density and at least one basic fundamental physical property of the power pack.

Before analysing the method proposed below, one has to realise that the geometry of the battery power pack is a characteristic of the application for which it has been used. Indeed while individual primary and secondary batteries have a well defined and standardised geometry, the variety of shapes and geometry of power pack assembly is infinite.

The power packs consist either in the assembly of individual standardised elements or in the assembly of elements of new geometry for new chemical systems introduced on the market like the lithium-ion and lithium polymer batteries.

It has been found that the apparent density is a criteria for selectivity which is simple to measure. It is a fundamental criteria that can be used for large scale identification of power packs. By apparent density, it is meaned a measured weight to volume ratio or any equivalent correlation such as the volume to weight ratio.

In a first step, the apparent density is measured on an apparatus consisting for example of a conveyor belt moving at fixed speed. If the moving belt is installed on a scale (weight detector), then the weight of the object is measured (Data ? 1). The system can be equipped with a commercially available laser

beam analyser (or any other device measuring the volume of the object) which detects the profile area of the object (Info 1), namely a power pack to be sorted.

The combination of the profile area with the speed of the belt and/or the time during which the object is crossing the beam (Info 2) gives access to the volume of the object (Data N°2) The combination of the weight of the object (Data N°1) with the volume of the object (Data N°2) gives the apparent density of the object (Info 3).

When a correlation is made between the apparent density of the object and the chemical nature of the power pack, it is observed that a distribution exists which gives already the possibility to separate several power packs from the other types. Such a correlation will be further described below in the table 2.

It has been found that some systems based on nickel-cadmium and nickel- metal hydride chemistry cannot be separated by this method due to the overlapping of their apparent density values. Also several other possibilities of overlapping exist between several chemistry systems.

A better separation of the power packs appears when the Info No. 3 information is correlated to one of the basic physical parameters of the measured object : for example the length or the projected area of the power pack.

When this correlation is made, it becomes possible to take into consideration the manufacturing design of a power pack which is strongly related to its application : for instance the geometry of the portable phone or the geometry of the portable computer in which it has been installed.

This method has been tested on several hundreds of power packs and has given satisfactory results for the separation of all the power packs measured. One exception remains : the separation of nickel-cadmium and nickel-metal hydride power packs of identical apparent density and identical geometry. Those are representing a fraction of a few percent of the total pack market.

Such power packs are separated from the other streams and are clearly identified. They are sorted by one of the methods developed recently : by example, the separation according to their ferromagnetic properties as described in TRIMAG I (Patent Publication N° WO 91/15036) and FIRSTSORT (Patent Publication N° WO 94/25992). Indeed, in the case of a nickel-cadmium battery,

the ferromagnetic elements are the envelope made of nickel plated steel and the current collector's spiral which is made of nickel or nickel-plated steel. The active mass of one electrode (the cadmium electrode) has no fundamental ferromagnetic properties. In case of nickel-metal hydrides batteries, the basic construction remains identical for the jacket and the nickel-electrode, but the metal hydride electrode is compose of nickel alloys with basic ferromagnetic properties (nickel-vanadium or nickel-misch-metals alloys).

Consequently it is possible to distinguish between those types of cells when they are of identical size. In a country like Japan, it will also be possible to separate those specific elements by a colour identification system. Such a system has been introduced recently, by the battery manufacturer.

BRIEF DESCRIPTION OF THE DRAWINGS The Figure 1 is a schematic representation of a power pack identification installation.

DETAILED DESCRIPTION OF THE DRAWING Figure 1 shows a power pack sorting installation according to the invention comprising : several transfer belts in order to carry the battery packs through or in front of the measuring devices, from which the belt A is a conveyor belt where a scale-weight detector is installed, belt B is installed under a laser-bream dimension detector, belt C is linked to several automatic ejection/sorting mechanical devices, belt D is transferring the pack through a ferro-magnetic sensor and belt E is linked to a complementary automatic ejection/sorting mechanical device.

- a scale weight detector device in 1 - a laser-based or equivalent dimension detector in 2 - a computer for data acquisition and analysis in 3 - several mechanical sorting devices for battery pack sorting from the line in 4 - a ferro-magnetic properties detector or colour detector in 5 - a final mechanical sorting device in 6 A proposed scheme for the separation of power packs, using for example the installation of Figure 1, is summarized step-by-step in Table I.

EXAMPLE A battery pack from a portable telephone is introduced on a belt conveyor either manually or by an automatic distribution machine. The belt has a constant speed of 60 cm per second, by example. The battery pack crosses a beam projected by a laser beam of a 3-D laser profile sensor camera such as one constructed by the company Baumer Electric or the Fanuc C02 HF - Laser.

The invention will be further described in the following example. This camera will measure the cross section of the object crossing the laser beam, in this case a battery pack. The integration of the values measured by the camera and the time during which it has been measured gives an information on the length of the object crossing the beam. From those measurements, the volume of the object can be calculated.

Table I - Mixed battery power pack/ Example of logic path for identification and sorting Step 1 Distribution Individual distribution on belt conveyor Step 2 Weight measurement Individual weight measurement on belt conveyor -"Data 1" Step 3 Volume detection Individual measurement on belt conveyor of profile area"Info 1"and speed (length per second)"Info 2". The computing of Info 1 and Info 2 gives the apparent volume which is"Data 2" Step 4 Mathematical analysis Step 5 Apparent density "Data 1"/"Data 2"="info 3"

Step 6 Mathematical analysis Correlation between"Info 3"and"Info 2"or basic physical parameter Step 7 Sorting criteria "Data 3"= Correlation Factor = Info 3/Info 2 Step 8 In one specific category Complementary sorting criteria. Analysis of Ferro-magnetic property, or colour recognition : Measurement of"Data 4" Step 9 Final selection Combination of"Data 4"with"Data 3"to obtain"Info 4" Step 10 Final sorting between elements distinguished according to Info 4 This camera will measure the cross section of the object crossing the laser beam, in this case a battery pack. The integration of the values measured by the camera and the time during which it has been measured gives an information on the length of the object crossing the beam. From those measurements, the volume of the object can be calculated.

Immediately before or after (or simultaneously on the same belt) the laser beam analysis of the pack dimensions, the battery weight is detected by a scale installed on a belt. The relation between the weight measured and the volume of the pack gives an information which is calculated by computer as an apparent density (Info 3) of the pack tested. The definition of apparent density is critical to the field of invention because many power packs are equipped with small electronic control devices installed inside the packaging volume of the power pack. This confers to the power pack an apparent density which is only indirectly related to the chemicals and metals used for the manufacture of the power pack.

This is the reason why the apparent density is a characteristic of the manufacturing and the application of power packs.

Finally, the calculation of the Correlation Factor establishes the relation between the pack's apparent density and its length or any of its intrinsic physical property. Consequently, one obtains a discrete value (the Correlation Factor) which is a characteristic of the pack crossing the beam. Table II gives a presentation of the calculation made to obtain discrete values for pack identification. Such a discrete value serves to identify the pack and to direct it to the appropriate exit of an automatic sorting machine.

Table II - Mixed battery power pack/<BR> Identification of the Power Pack according to the Apparent Density (D) Correlation Factor (L/D) Type Manufac- Serial N° Length Width Height Volume Weight Apparent (L / D) Sample turer (L) DensityCorretationPack (D) mm mm mm Cm3 g g/cm3 Factor N° Ni-Cad Motorola SNN 4467 A 120 59 10 70. 80 79 1. 12 107. 14 1 Ni-MH Ericsson BKB 193 079 110 47 20 103.40 130 1.26 87.30 2 Li-lon Sony AM 65 27 27 47. 39 82 1. 73 37. 57 3 Li-S02 Saft NP 1. 2-6 94 64 33 198. 53 226 1. 14 82. 45 4 Pb-Acid Yuasa BA 5847 U 96 24 50 115.20 272 2.36 40.67 5 Zn-HgO Varta 7450 80 56 38 170.24 405 2.38 33.61 6 Zn-Alk Crompton NS-10295 78 60 36 168. 48 353 2. 10 37. 14 7 Li-SOCI2 Cypres Airtec 54 50 25 67.50 106 1.57 34.39 8 Ni-MH Motorola SNN4249A 120 59 10 70.80 77 1.09 110.10 9 Ni-Cad Motorola SNN4211A 193 40 25 193. 00 214 1. 11 173. 87 10 Li-lon Sony AM 50 42 21 44.10 79 1.79 27.93 11 Ni-MH OKI 64-300004 100 54 20 108.00 191 1.77 56.49 12

Representative values the Correlation Factor obtained during the testing of the method according to the invention for sorting various commercially-available power packs using an installation are given in Table II.

From Table II it can be seen that each individual type of power pack is identified by a unique Correlation Factor which corresponds in this example to the length divided by the apparent density.

As shown in Table II, for samples N° 9 and 10, the measured apparent density has a value which is different by less than two percent while the calculated Correlation Factor clearly allows to distinguish between the amples.

In a similar manner, the apparent density of the samples N° 11 and 12 has a value which is very similar but their Correlation Factor opens the route to a clear dsitinction between those two packs.