Prediction of the service life of accumulators

Technical solution proposal G2013-012
Corporate entity: Diehl Aerospace GmbH

Description:

This proposed technical solution is a method or algorithm for predicting the service life of accumulators.

Motivation

The importance of electrical energy storage has increased rapidly in recent years. This results from the increasing demand in a wide range of application areas. These include, for example, the storage of electricity from renewable energy sources, the development of increasingly powerful portable devices and research in the field of electric vehicles. In this context, the accumulator - also referred to as "battery" in the following - which was invented almost 200 years ago, is also experiencing a major surge in development. One of the best-known and technically most interesting types of rechargeable batteries is a lithium-based accumulator. This is characterised by a high degree of efficiency, a high energy density and a short charging time. Unfortunately, these charge storage devices do not only have advantages: Constant monitoring of voltage and current is necessary to prevent overcharging or deep discharging, which can destroy individual cells. Incorrect handling of the charge accumulator can endanger the environment due to its high flammability. The ageing of the cells is also a problem.

To get a grip on these difficulties, many companies produce charge management systems for lithium accumulators that measure current, voltage, temperature, etc. and can switch off the accumulator in an emergency. One product of this kind is the IC "bq20z65" from the company Texas Instruments (TI). This is a complete accumulator management system which, in addition to the above-mentioned variables, can measure the internal resistance, the existing capacity, the remaining capacity and much more on the respective accumulator. Together with the actual accumulator, this creates a complete accumulator "pack".

In the aviation industry, lithium-ion accumulator packs are also used for the emergency power supply of aircraft. For this purpose, the company DIEHL Aerospace produces so-called

"Emergency Power Supply Units" (EPSU). Their task is to supply energy for the emergency lighting in the aircraft in an emergency. According to the specifications, the supply must be guaranteed for 12 minutes, which is why the accumulator must not fall below a minimum capacity. In addition, the accumulator is required to deliver a power of 72W during this period, which means that there are also maximum limits for a permissible internal resistance of the accumulator. In order to comply with the above-mentioned requirements, all relevant accumulators of an aircraft must currently be removed and checked by maintenance personnel every 6 months. After 36 months, the accumulator is replaced and disposed of as a precaution.

This procedure obviously entails a very high level of battery wear. For this reason, DIEHL Aerospace is striving to develop a procedure within the CleanSky project that can predict the service life condition of an accumulator. This makes it possible to replace the accumulators only according to actual need and not according to rigid intervals. This saves costs for maintenance, new equipment and disposal of old equipment.

In order to be able to make a prediction for the service life of an accumulator, it is necessary to develop an algorithm that can predict future events from existing data.

Der Algorithmus

The aim is to predict how many charging and discharging cycles remain for an accumulator - here also representing an accumulator pack - until it no longer meets the above-mentioned specifications and must be replaced. For this purpose, an algorithm must be implemented that can determine the presumably remaining cycles as precisely as possible or can indicate in which of the cycles the battery is likely to fail. The presented algorithm is based on respective measurements of the internal resistance Ri of the accumulator, since the remaining number of cycles decreases with increasing internal resistance.

The calculations of the algorithm can be divided into several steps:

[...] for more information see download