Abstract
Nowadays, arising from the environmental issue, demands of electrical vehicles and green energy storage are growing. Among all of the energy storage devices, Li-ion battery is more preferred because of its specific peculiar characteristics. Generally, the common ways to model Li-ion batteries can be divided in two: empirical models and electrochemical models. Electrochemical models were developed on the basis of physical mechanism and chemical process which can quantatively describe both the external characteristic and the inner state of the battery with high accuracy. The inner state of lithium ion batteries can only be obtained by simulation instead of monitoring outside. By knowing the inner information, efficient management can be realized to reach the demand of long life and the choice of material can be also evaluated. However, numerous material parameters are required in the simulation and it takes great efforts to obtain all of them. Besides, sophisticated governing equations that result in time-consuming computations limit its practical use in real-time applications like Battery Management Systems. Now in real-time applications, empirical models such as equivalent circuit models or neural network models are extensively used. These models are determined by the experimental data and less computational load is required. In this study, we propose an experimentally-based DNA power model which can rapidly and accurately predict the charging/discharging characteristic of the Li-ion battery. This model is developed mainly for practical applications because it is easy to implement with low cost and no complex background knowledge is involved. Similar to the Rint model [1], our empirical model comes from the “power” concept within the battery as in Eq. (1), where I is the output current when charging/discharging, V is the battery terminal voltage, V OCV is the open-circuit voltage which is SOC-dependent, and R int is the internal resistance. In real practice, the measurement of V OCV takes time. Thus, for getting a better empirical model, let P =IV OCV, α=R int/(V OCV)2, and then Eq. (1) can be rewritten as Eq. (2), where P is the nominal power, α is the nominal power reciprocal, and they are design manipulating factors in our model. The computing process of this model is shown as in Fig. 1. Each dotted rectangle is under a certain C-rate. As the name DNA implies, we need to find the genotype P and the genotype α . First, we select an arbitrary C-rate(s) and a suitable constant α , and input experimental data Iexp , Vexp to find PDNA . This PDNA is used to “shape” the nominal power P in other C-rate conditions through the relation Pk =PDNA /ak , where ak=C-rate (s)/C-rate(k). Next, we use the information of the other C-rate(2), put in P2 , αDNA can be determined. Then we translate αDNA in other C-rate conditions to fit the experimental data and find α for each C-rate. Finally, we can find the rule between α and C-rate to predict the charging/discharging properties at any desired C-rate. In order to validate the model, the experiment was performed on 2.4Ah NMC at room temperature (25°C). We use 0.5C discharge as the standard to get the genotype P and use 2 C discharge to get the genotype α , then translate α to 1C discharge and 3C discharge. As shown in Fig. 2, the simulation results agree well with the experimental results, demonstrating the great capability of this model. Experimentally-based DNA power model provides an excellent tool for obtaining the charging/discharging characteristics of Li-ion battery in an empirical way, especially for real applications. The effect of temperature rise and SOH can be further included. A more comprehensive and detailed simulation will be presented and discussed in this work. REFERENCES [1] H. He, R. Xiong, and J. Fan, Energies, 4, 582 (2011). Figure 1
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have