BMS Key Functions: Cell BalancingOvercharging, and over discharging, of Li-ion cells causes permanent damage, also is a significant safety issue. Therefore, reliable control of each cell voltage is mandatory. Battery cells are connected in parallel and in series to obtain greater energy capacity and voltage. The parallel connections are simple to operate since the cells emerge as a full single cell, and cell management is simple. The current divides corresponding to the internal resistance of the individual cell, and the terminal voltage of each cell is matched (Einhorn et al.

,2011). In serially connected cells there is a characteristic between cells in charging and discharging. During applying load, the voltage of the smallest cell capacity is the early reaches discharge voltage limit (DVL). The capacity of the whole battery back is thus restricted by the narrowest cell in the module. Charging of the battery module cannot be extended when the cell with the smallest capacity is fully charged despite some cells are not completed.

Discharging of the battery module stops when also the cell with the smallest capacity is empty, despite the others cells still charged (Einhorn et al. ,2011).One further vital influence when individual cells are formed together to a battery module, each cell has a specific thermal behavior even with a proper-designed cooling strategy, thus cells age differently from each other especially after frequently charging and discharging. Performance of a battery module with various single-cell capacities can further be developed when the charge from the cells is balanced with an electronic circuit. They mainly use two concepts to balance battery module. First approach is passive cell balancing which uses a resistor to discharge the cell with the largest cell voltage so that charging can be maintained till all cells are fully charged. This technique is applicable during the charging process and not efficient due to power dissipation and energy lost. secondly is the active cell balancing, it can transfer the charge between the cells in a battery module using a small-time storage component, which can be a capacitor or an inductor (Einhorn et al. ,2011). A more in-depth insight with experimented results and simulated structural analysis design of the battery balancing system can be found in literatures carried out by Einhorn and by Roessler and Fleig. Battery balancing design is very significant as it is maintaining all battery cells in the same state of charge. Battery ModelingBMS has a significant function in HEVs, as the battery pack is the heart of the vehicle and must work in an energy awareness way. Accordingly, it expects a proper monitoring and supervision of the battery pack. In order to achieve the required efficiency, an exact calculation of the battery state of charge (SOC), state of health (SOH), and remaining useful life (RUL) is required. Since battery parameters cannot be precisely measured with a sensor, a numerical model along with a robust estimation algorithm is necessary. In request to implement the model according to battery chemical characteristics, there are too many parameters to consider when applying them in real time models. A trade-off between model accuracy and complexity, thus models must be specific to read the system energetics while being easy to be implemented in available embedded microprocessors (Ahmed, 2014). According to (Farag, 2013), the models can be arranged ascendingly into four categories corresponding to their complexity and the number of presented parameters. These models are Ideal, Behavioral, Equivalent circuit, and Electrochemical. Ideal ModelIdeal model neglects the internal parameters of the battery thus it is extremely simple to model. Furthermore, this model is considering the battery as purely unlimited voltage source (Farag, 2013). Behavioral ModelBehavioral model or Black-box Model does not extend in battery specific electrochemical parameters. However, it simulates terminal voltage performance. Model working method is based on phenomenological functions, which require specific data measures. Factor measures could be obtained from neural networks, empirical functions or data tables (Farag, 2013). Peukert’s law formula is frequently applied in behavioral models for batteries. An empirical function is adopted to represent the dependency of the battery’s remaining capacity on the discharge rate. I^PC t=Constant (1,1)I is the discharge current, t is the maximum discharge time, PC is the Peukert’s Coefficient. Thus, battery capacity calculated as follow: cn_1=cn*(In/In1 )^(Pc-1) (1,2)Battery remaining capacity is represented by Cn1 at discharging current I (Farag, 2013).Other methods for a behavioral model were presented by Shepherd and Plett to forecast the terminal voltage during charging/discharging conditions, can be found in the literature carried out by Farag. These models can explain cell hysteresis, polarization time constants, and ohmic loss effects. Equivalent-Circuit ModelResistors and capacitors are the major components of the lumped-element equivalent-circuit to simulate battery cell performance. RC circuits are frequently applied in BMS for its clarity, a few parameters to optimize, and simple implementation. Furthermore, RC circuits implemented as first-order, second-order, or third-order corresponding to model complexity and accuracy besides the hysteresis effect. The third-order model is as shown below in Fig 2. However, this model still cannot provide data about the internal electrochemical reactions in the unit cell, thus it is incapable of predicting electrochemical phenomena like cell degradation, capacity fading, and power fading (Farag, 2013).Fig (2) Third-Order RC Battery Model. R0 act in place of the battery internal resistance, and the equivalent circuit applies RC branches to model battery dynamics. The parallel RC components show the dynamic order of the circuit (Ahmed, 2014). Electro-Chemical ModelElectro-Chemical Model (ECM) or physics-based model studies battery electrochemistry parameters in details, what makes it very complex is it can clarify the electrochemical reactions using partial differential equations (PDE) with consistently many unknown parameters to solve. PDE allows a compromise analysis and high accuracy in parameters data. To implement ECM in actual BMS, it requires model reduction calculations. Several techniques for ECM reduction have been introduced in literature. Author noticed that plenty of the computational complexity associated with ECMs comes from solving PDEs for the lithium concentration in the solid particles of the electrodes (Spherical Diffusion). A typical approach is to do approximations and simplifications for this calculation. ECM also needed to generate battery aging by a state of health estimation (Farag, 2013). A full-Order electrochemical model, and a reduced-order electrochemical model were established, as an improvement approaches to ECM. During the battery design stage, the full-order model can predict the physical interactions inside the cell such as the potential distribution and electrochemical species diffusion. To implement the electrochemical model for real-time applications on a BMS for a state of charge and state of health estimation, a reduced-order form has to be performed. Electrode average model, and the states value model are two simple forms of reduced order electrochemical models (Ahmed, 2014). Artificial Neural Network Model (ANN)Artificial Neural Networks (ANNs) shown in fig 3, are a mathematical model similar to the human brain in learning and observing patterns. Through a set of neurons that are connected with weights, ANNs can map input to output datasets. Furthermore, ANNs can apply to battery modeling and for state of charge estimation applications. Nerveless they require training of battery data specific behavior to perform with proper estimations and techniques (Ahmed, 2014).Fig (3) Artificial Neural Network Diagram, as (x1, x2, x3,.., xn) represent neuron inputs, their weights are (w1j, w2j, w3j ,.., wnj), b is bias and denotes a nonlinear activation function (Farag, 2013).netj=X1w1j+X2w2j+X3w3j+‹Xnw2j+b (1,3) Oi=(netj) (1,4) Thermal Management SystemsThermal management of lithium ion batteries is a safety critical consideration, as it could lead to unsafe operating conditions because of battery overheating. To operate the battery within the safe range cells require to be monitored. Since cell temperature is complicated to measure so it must be estimated. Thermal models are combined with battery models to obtain detailed data of battery performance characterization (Farag, 2013). Battery Aging MechanismsBattery aging is complex and influenced by the battery operating conditions. the aging mainly described as a capacity or power fading to a pre-determined limit, where capacity fade is described as a loss of capacity and power fade is described as a rise in the battery internal resistance. Aging could appear because of operating the battery under excessive conditions such as high temperature, high charging rates, or great state of charge levels. Likewise, could take place due to battery storage (calendar aging) or usage (cycling aging). It is declared to result from several transforms where performance degradation happens due to irreversible chemical reactions (Farag, 2013). ConclusionAs lithium-ion battery has a small efficient working range in charging, discharging and thermal efficiency. Battery design is multipart, as battery cells connected in serial and parallel to gain the most capacity and efficiency with keeping battery size small as possible. To solve those conflicts a battery management system was introduced, through which the lithium-ion batteries can be controlled and maintained effectively, and every individual cell would be working under proper conditions, consequently every cell should be operated within the lithium-ion battery safety operating window. The main points in the design and management of a battery for an electric vehicle have been presented. After the explanation of BMS key functions. An overview of the main techniques for charge balancing and state of charge estimation has been pointed out. Battery models and their types were introduced, as a trade-off between the complexity of the model and the accuracy must be taken in consideration, thus increasing the computation and memory requirements of the model could compromise its suitability for real-time BMS applications. The push for efficient and large driving range hybrid electric vehicle will include a future with high energy lithium ion batteries.