Battery system life/reliability evaluation is a relatively difficult topic, involving many aspects, especially to conduct product research rather than scientific research, need to stand on the basis of science to solve the problem of implementation engineering.

 

 

 

To better understand and evaluate the life/reliability of a battery system, you need to know four things :(1) the basic principles of cells, mainly aging and capacity decay; (2) basic knowledge of reliability mathematics, mainly weibull equal life function, repairable/repairable product life modeling; (3) battery package structure, thermal and electrical design; (4) automobile reliability engineering.

 

 

 

Lack of any one, are easy to go astray. For example, the study of electrochemical mechanism of the cell; Those who do math focus on mathematical models; Lack of first-hand materials for front-end basic data and back-end application conditions; Make cars rarely go deep into the basic battery pack as a whole parts; Cooperation is not so smooth, mutual understanding.

 

 

 

Every Angle is correct, but because of the lack of many common "languages" in various fields, relatively comprehensive and effective methods cannot be formed.

 

 

 

Moreover, with the increasing popularity of big data, new trends have emerged in reliability assessment.

 

 

 

 

 

one

 

 

 

From the point of view of a cell, its reliability/life is largely equivalent to two aspects: capacity and safety. Capacity mainly refers to the attenuation of capacity, which affects the performance of the whole vehicle and the quality assurance of the whole vehicle. Therefore, a large part of the work or research methods are targeted at the attenuation of capacity. In the field of cell, a series of studies on capacity attenuation have been established, which can be basically summarized as some kind of accelerated life model, similar to the following.

 

 

 

Different factors such as time, temperature and electrochemical system are taken into account by the parameters of the accelerated life model. The essence is an electrochemical based data fit that can explain what you see (test data) or predict the future based on existing data (when capacity falls to 70%). Note that the data fitting here is not simply "digital fitting", but has its electrochemical basis. Therefore, this model can guide the cell design improvement, so that it can know the specific influence of a specific factor.

 

 

 

Safety is easy to understand, mainly to study how the inner core causes fire and explosion, namely failure/failure mechanism.

 

 

 

Make the battery friends need to understand the fact that the battery is a non-maintainable products! And battery pack, the car is a repairable product! The life of the cell is one time, failure, failure, must be replaced as a whole, can not be used again; And battery pack, the car in quite a few fault cases, after maintenance, or can continue to work. For batteries, there is no MTBF (meantimebetween failures) concept, only MTTF (meantimebetween failures).

 

 

 

So you can't use a mathematical model of a cell to put a battery pack or a car. The model of a cell is just one part of it.

 

 

 

In many cases, the battery cell model can be used to predict PACK life based on the assumption that the battery cell has the shortest life of all parts and components in the PACK, and once the battery cell fails, it cannot be replaced, resulting in the whole package being scrapped. However, this is not inevitable. In the parallel design of Tesla, when the individual electric core fails, its aluminum wire welding breaks off and the electric core is isolated, the whole vehicle can "work with disease" according to the basis, but the performance is damaged.

 

 

 

The most reasonable value of the battery life/reliability lies in: external, demonstrate that the attenuation of capacity during the warranty period meets the requirements of vehicle enterprises; Identify the safety mechanism of the cell and assist the vehicle in functional safety protection. The scientific basis and method of continuous improvement of cell are provided.

 

 

 

 

 

 second

 

 

 

From a battery PACK /PACK perspective, more reliability means less breakdowns throughout the product. In addition to the cell, the whole package has many mechanical structures, electrical components, thermal management components. They all have their own failure mechanism and mode, and they interact with each other. When modeling, there will be coupling problems.

 

 

 

The mathematical modeling of battery packs is difficult to be based on the mechanism of each component, which is often not available. At this point, there are two ways to deal with it: one is to simplify the model of the battery pack and model the key components; Second, the battery pack is directly regarded as a black box, and only the failure data of the battery pack are looked at. The "digital" fitting is simple. This method is easy to obtain data, and can predict the probability and times of battery pack failure and maintenance.

 

 

 

The problem with pure digital modeling is that you may know how many faults have occurred, but you cannot predict which parts have failed, and the vehicle enterprises cannot stock up. This statistical method can only be used to further mine the faults through big data in the future, so that it can play a better role.

 

 

 

The reliability evaluation at the battery package level is more valuable at the reliability engineering level, that is, to solve the failure probability of parts and improve the design.

 

 

 

Battery pack personnel need to do: first, determine the failure distribution of each zero part of the battery pack; Secondly, determine the mechanism of the failure of each part; Third, develop a design optimization plan for each component to reduce the failure to the target level; Fourth, determine the mechanism that may lead to the electric cell safety accidents, and provide corresponding solutions.

 

 

 

It should be noted that most battery pack reliability problems are engineering problems to be solved by the battery pack side before the battery life problem occurs. This stage is more of a technical problem than a "mathematical model problem".

 

 

 

 

 

three

 

 

 

From the perspective of the vehicle, the reliability of the battery pack means: 1. Second, the battery pack level of the fault. It just happens to be a collection of core and PACK personnel's work priorities. In more cases, the reliability problem of the complete vehicle belongs to the non-battery PACK problem, and their understanding of PACK reliability is more similar.

 

 

 

However, vehicle companies have more advantages and perspectives: they have a more thorough understanding of the use of vehicles, a more comprehensive grasp of data, and a more macro perspective. The whole vehicle has the ability and resources to reverse mining from the top (data of fault phenomena) to the bottom (fault mechanism), which benefits from big data technology, as shown in the following figure. This kind of top-down exploration requires the cooperation of the upstream, otherwise it requires the establishment of a strong research and development team and a huge amount of money.

 

 

 

four

 

 

 

From the perspective of reliability personnel, battery package reliability is often just another "repairable product". They have many models, engineering methods and test means to guide research. They have more theoretical knowledge and methodology on how to model, how to do reliability engineering and how to increase reliability. This is their strength, and their weakness is obvious: it's hard to understand products and technologies in depth.

 

 

 

In the real work, which side is strong, which side to dominate the whole reliability work mode and direction. The reliability problem is first of all a system problem that needs to be coordinated by a person who can mobilize resources from all sides. Ultimately, it needs to solve the actual failure problem of the product, rather than how to better report: reliability is not false.