Polaris Energy Storage Network News: 2017 Urban Energy Internet Development (Beijing) Forum and Energy Internet Demonstration Project Construction and Cooperation Seminar was held on December 1, 2017 in Beijing. In the afternoon of the technical forum, Jiang Jiuchun, director of the National Energy Active Distribution Network Technology R&D Center, delivered a speech on the theme: key technologies of lithium battery energy storage systems.
Jiang Jiuchun, Director of National Energy Active Distribution Network Technology R&D Center:
I am talking about battery energy storage. Our Jiaotong University has been doing energy storage, from power systems and electric vehicles to rail transit. Today we are talking about some of the things we are doing in power system applications.
Our main research directions: one is micro-grid and one is battery application. In battery application, the earliest electric cars we used used energy storage in the power system.
Regarding the most important issue of battery energy storage, the first issue is safety; the second is longevity, and then high efficiency.
For energy storage systems, the first thing to consider is safety, and then efficiency. Adherence to efficiency, the rate of transformers, and lifespan, as well as energy utilization after battery decline, may not be a quantified problem in many cases. Indicators to describe it, but it should be very important for energy storage. We hope that through several things, we can solve the problem of safe life and high efficiency. A standardized energy storage system and a carding analysis system for battery status are used in electric vehicles and public transportation systems.
At present, the use of energy storage systems, node controllers and intelligent distribution boxes that everyone is using, improves the overall economy and stability of the system, enhances the core value of system integrators, and can be friendly access to the back-end cloud platform.
This is a centralized energy scheduling system. This hierarchical structure has been made very clear this morning, and we can achieve a long-term optimal scheduling of coordinated multi-energy storage power plants and microgrids through multi-node controllers.
Now it is made into a standard intelligent power distribution cabinet. This is the basic feature of the power distribution cabinet. It contains various functions, such as charging and discharging functions, automatic protection, and interface functions. This is standard equipment.
The node controller implements local energy management core equipment, main data collection functions, monitoring, storage, execution management strategies and uploading. There is a problem here that requires serious and in-depth research on the data sampling rate and the time of data sampling when data is uploaded. In this way, the analysis of battery data in the background of the battery is implemented, and the maintenance of the battery is turned into intelligent maintenance. Do some work, in the end, how large the number of samples is, or how fast is the storage, to fully describe the current state of this battery.
If I drive an electric car, you will find that many electric cars are in a state that often changes and jumps. In fact, energy storage faces the same problem in power system energy storage applications. We hope to solve it through data. We have a BMS sample size that is appropriate.
Let me talk about flexible energy storage. Everyone says that I can do it 6,000 times, and it can be used a thousand times in a car. It’s hard to tell. You can help it as an energy storage system, claiming to be 5,000 times. How much is the utilization rate, because the battery itself has a big problem, the decline of the battery is random during the recession process, each battery declines differently, and the difference between the single cells becomes more and more different The inconsistency of the manufacturer’s battery decline is also different. How much energy can this group of batteries use and the energy is available? This is a problem that requires careful analysis. For example, when electric vehicles are currently used, they are used from 10 to 90%, and the recession can only use 60% to 70% to a certain extent, which poses a big challenge to energy storage.
Can we use the grouping according to the law of decay to make a compromise, how big is the right choice to get better performance and better efficiency, we hope to group it according to the law of battery decay, 20 branches as a node is Whether it is more appropriate or 40 is more appropriate, which makes a balance between efficiency and power electronics. So we do something about flexible energy storage, which is also our project to do this thing. Of course, there is a better place to use it in cascades. I think that cascade utilization has certain value in the past two years, but it is worth using in the future, but also think about the efficiency of charging and discharging, once the price of the battery drops, There are some problems with cascading. Flexible grouping can solve big problems. Another kind of high modularity reduces the cost of the entire system. The biggest one can improve the utilization rate.
Like a battery used in a car three years later, the decline is less than 8%, and the utilization rate is only 60%. It is due to its difference. If you make 5 sets of utilization rate, you can achieve 70%, which can improve the utilization rate. Stringing battery modules together can also improve battery utilization. After maintenance, energy storage increased by 33%.
Looking at this example, after balancing, it can be increased by 7%, after flexible grouping, I increased by 3.5%, and balancing can increase by 7%. Flexible grouping can bring a benefit. In fact, the reason for different manufacturers’ battery decline is different. It is necessary to know in advance what this group of batteries will become or what the parameter distribution will be, and then you will make a targeted optimization.
This is a scheme adopted, the module full power independent current control, which is not suitable for high power applications.
Part of the power of the module is independently controlled by current. This circuit is suitable for medium and high voltage and repeated use. This is the MMC battery energy storage solution suitable for high voltage and high power.
Also about battery status analysis. I have always said that the battery capacity is inconsistent, the decline is random, the battery aging is inconsistent, and the capacity and internal resistance are very reduced. Using this parameter to characterize, the more you use is the capacity and the internal resistance. If you want to find a way to maintain consistency, you need to evaluate the SOC difference of each battery, how to evaluate the SOC of this single cell, and then you can say how this battery is inconsistent and how much the maximum power can be. How to get a single SOC by maintaining the battery through SOC? The current approach is to put the BMS on the battery system and estimate this SOC online in real time. We want to describe it in another way. We hope to run the sampled data to the background. We analyze the battery SOC and the battery through the background data. SOH, optimize the battery on this basis. Therefore, we hope that car battery data, not big data, is a data platform. Through machine learning and mining, the SOH estimation model is extended, and a management strategy for full charge and discharge of the battery system is given based on the estimation results.
After the data comes up, there is another advantage, I can make an early warning of the battery health status. Battery fires still happen frequently, and the energy storage system must be safe. We hope to do a real-time information and medium and long-term early warning through background data analysis, find short-term and long-term online warning methods for potential safety hazards, and finally improve the safety and reliability of the entire system.
Through this, I can achieve several aspects in a large scale, one is to increase the energy utilization rate of the system, the second is to extend the battery life, and the third is to ensure safety, and this energy storage system can work reliably.
How much data do I need to upload to meet my requirements? I need to find the smallest battery that meets the running state of the battery. These data can support the analysis behind, the data can not be too large, a large amount of data is actually very large for the entire network A load. Dozens of milliseconds, you take the voltage and current of each battery, which is unrealizable when you pass it to the background. We have found a way now, we can tell you, what sampling frequency should be, what characteristic data do you need to pass We simply compress these data, and then pass it to the network. The battery curve parameter is one millisecond, which is enough to meet the needs of battery evaluation. Our data records are very, very few.
The last one, we say BMS, the cost of energy storage becomes more important than the cost of batteries. If you add all the functions to the BMS, you can’t reduce the cost of this BMS. Since the data can be sent, there can be a powerful analysis platform behind me. I can simplify it in the front. There is only data sampling or simple protection in the front. Do a very simple SOC calculation, other data are sent from the background, this is what we are doing now, the entire state estimation and sampling of the BMS below, we pass the energy storage node controller, and finally pass to the network, energy storage The node controller will have a certain algorithm, the following is basically detection and equalization. The final calculation is performed on the background network. This is the entire system architecture.
Let’s take a look at the effectiveness and simplicity of the bottom layer change, which is equalization, low voltage acquisition and equalization acquisition to current acquisition. The energy storage node controller tells the following how to deal with it, including the SOC is performed here, and the background works again. This is the smart sensor, battery management unit, and intelligent node controller that we are already working on, which greatly reduces the cost of energy storage.
Post time: Jul-08-2020