WATT weekly

2019-12-02

Solving of Inverse Problems Enables Non-destructive Battery Inspection (2 of 2)

Based on an interview with Kenjiro Kimura (Professor, Kobe University) at Kobe University Incubation Center (Kobe, Hyogo Prefecture, Japan) on September 24, 2019)

Difference Between the Two Main Battery Inspection Devices

There are two types of major non-destructive battery image inspection devices in the market. An X-ray inspection method that detects metallic foreign objects using X-rays and an infrared inspection method that analyzes heat generation using infrared rays. The former is mainly used for in-line inspection, and the latter is used for failure analysis. However, both methods have drawbacks. Looking at the X-ray inspection method, since batteries contain many metal elements such as current collectors and active materials, it is not easy to find metal foreign matter on the scale of several tens of μm, such as metal fine particles that can cause a short circuit. Meanwhile, the infrared inspection method requires a large current of 100 to 1,000 times that of our magnetic field detection method. There is no track record of in-line inspection.

Market Size of In-line Battery Inspection System

According to a trial calculation by Integral Geometry Science, by multiplying the global sales forecast of electric vehicles (BEV / HEV / PHEV) by the capacity of batteries installed in each electric vehicle, the battery capacity required in 2025 will be about 215 GWh. The market size is calculated to be 170 billion Japanese yen in 2025 based on the in-line inspection equipment’s price of 200 million Japanese yen / line and its lifespan of 5 years.

Market Is Full of Faulty Batteries

However, the pace of sales of in-battery non-destructive diagnostic imaging systems is slower than what we expected. At first, we thought these systems would penetrate the market faster. The reason became clear immediately after starting the business. This is because especially foreign-made batteries that are on the market are poorly manufactured so the non-destructive imaging diagnosis identifies nearly all of them as faulty. In particular, the cell in most batteries of overseas manufacturers that we expected as a promising market are cluttered inside so there is no chance for nondestructive imaging diagnosis. As the number of electric vehicles increases, battery inspection equipment must be built into the production line. If the quality of products shipped is not 100%, automakers will make battery suppliers liable for any accidents caused by batteries.

Many Defects Caused by Rapid Charging

By the way, many of the batteries brought to us for testing are thought to have become defective in a short period of time due to rapid charging. Quick charging is not good for batteries in the first place. The battery is repeatedly charged and discharged as ions move inside the battery and adhere to the electrode. At that time, even a good quality battery has a weak part from which electricity can leak. The place where electricity leaks is like a needle inside the battery. Rapid charging means applying a strong electric field to the needle making the needle longer and longer. When the needle is broken and released, lithium crystals precipitate and battery capacity decreases. When the needle grows further and reaches the opposite side, plus and minus becomes connected causing an internal short circuit. As a result, electricity will circulate inside the battery which generates heat eventually causing the battery to ignite and explode. In other words, to make the battery last longer, it should be charged slowly. Fast charging is very convenient, but it should be avoided when considering battery life. Lithium crystals are likely to precipitate at low temperatures, so if you charge quickly in a cold region like Northern Europe battery capacity will decrease that much faster. It is ironic that BEV penetration rate is high in Northern Europe.

Actual Structure of the Battery Can Be Determined Analytically

The inverse problem of deriving a current distribution from a static magnetic field cannot generally be solved. To that end, many researchers have tried to solve the model assumptions and forward problem repeatedly to achieve convergence. We have succeeded in deriving the analytical solution of the inverse problem of the three-dimensional current and magnetic field in the storage battery, reflecting the actual structure that the battery is sufficiently thin compared to the size of the electrode surface. Specifically, the following delta function δ (z−z0) was placed into the calculation. By this delta function, information that the thickness of the battery is sufficiently smaller than the electrode surface is incorporated in the calculation. As the calculation proceeds, a Fourier transform image of the potential distribution is obtained. When this is inverse Fourier transformed to obtain φ (x, y), the conductivity distribution σ (x, y) inside the battery is obtained as follows. Since the mathematical formula derived in this way is expressed using boundary conditions (observation results), the calculation results can be output instantaneously even on a commercially available personal computer by substituting the observation results.

The Inverse Problem Can also Be Applied to Autonomous Driving

By applying the inverse problem, there is a possibility that other issues in the automobile industry, for example, Level 5 self-driving can be easily realized. Most of the automatic driving developments currently being carried out around the world are integrating (fusion) sensor information such as LiDAR and cameras, and determining the driving behavior based on surrounding conditions using artificial intelligence (AI). Although, this AI method can be a major source of income for major semiconductor manufacturers, everyone is starting to understand that the hurdles to realize it are high. It is vulnerable to bad weather and there have been many misperceptions as well. Accidents on public roads during testing have been also reported. Even if the system recognizes the white line and the car runs straight, it is still not 100% safe. Level 5 on ordinary roads is considered to be difficult, and it is considered that there is no other way than limiting the Operational Design Domain (ODD). Then, it is faster to embed a magnetic marker (magnetic tag) in the road surface of a specific route from the beginning. This method embeds magnetic markers in the road surface at intervals of several meters and determines the course by reading the markets with a magnetic sensor. This magnetic marker method is less expensive than the complicated AI method. The problem can be solved by installing magnetic sensors on the sides of the car and having only automatic start / stop and automatic steering functions. The magnetic marker method was adopted by Toyota for the new traffic system Intelligent Multimode Transit System (IMTS) at the 2005 Aichi Expo. Aichi Steel, which cooperated in development at that time, is still conducting demonstration experiments. Generally, there is a problem with magnetic marker systems. Magnetic sensors malfunction due to manholes and nails. Therefore, we are developing a system that reads a magnetic marker by attaching a multi-array magnetic sensor to the bottom of the car. The calculation algorithm has already been completed, and magnetic markers can be detected in real time even at a high speed of 200km / h. Actually it is assumed that the vehicle will travel at around 35km / h in consideration of safety. However, it is possible to run at a higher speed than the demonstration experiments of the automatic driving service currently being carried out all over the world. It is suitable for practical use and for Level 5. In this way, we are developing with confidence that the era of magnetic marker method will come. However, we have succeeded in developing the world’s first technology that can contribute to the AI method which is currently mainstream in the automobile industry. For example, all focus is on technology. We have mathematically figured out how to look at the 3D position of an object with diffused light, rather than scanning a beam focused on the 3D structure of the object. When pulsed light diverges from a point light source, the light bounces around. By analyzing the ripples of the returned light, you can theoretically measure the image of what you want to see. This is what we call the all-point focus. The surprising point of this method is that if you can observe the scattered light behind the object, you can visualize it. If it is not completely sealed, in principle there will be no shadow. All-point focus can diffuse light. It is not necessary to narrow down the light source and apply strong light to one place like a laser. That is, there is no risk of blindness or burns even if it hits someone. All three-dimensional structures of objects can be visualized without incurring such risks. This is much more than technologies such as LiDAR for grasping the three-dimensional shape of conventional objects. We will continue to meet all the challenges of the world by making full use of the world’s top technology in image reconstruction technology that solves inverse problems and high-precision measurement. (Naofumi HIGASHI)