Impact of dc-dc Converters on Li-ion Batteries

摘要

This work investigates the interaction between dc-dc converters on the o ne hand and batteries on the other hand. As far as battery research is c oncerned, the focus is on phenomena with relatively large time-constants ranging from seconds to hours and small current ripples. These phenomen a include State of Charge estimation and terminal voltage prediction in simulations. Conversely, the frequency at which dc-dc converters operate ranges from a few to tens or even hundreds of kHz. The current ripples originating from dc-dc converters are large compared to those used in ba ttery research and have a peak-to-peak value ranging between 0.1 to 2 ti mes the dc-value of the battery current. The goal of this research is to investigate the impact of these large, high-frequency current ripples o n the parameters of the battery. These battery parameters include the ba ttery capacity, the energy content, the State of Charge dependent charge and discharge resistance as well as the State of Charge dependent Disch arge and Regen power. In order to allow the investigation of the impact of the current ripple on the battery parameters, the battery is subjecte d to both small and large current ripples. For this purpose three differ ent 30 kW, two-quadrant, single-phase dc-dc converters are designed, two hard-switching converters and one soft-switching converter. The maximum dc-current and the switching frequency of all converters is identical a t 100 A and 8 kHz respectively. The converters have a primary inductor o f respectively 350, 220 and 105 uH. When these primary inductors are com bined with a capacitor, a conventional LC- lter is obtained. The battery is now exposed to a rather large current ripple with an 8 kHz fundament al harmonic which has an amplitude of respectively 16, 30 and 60 A. The addition of a relatively small secondary inductor is suffcient to obtain a tenfold reduction of the fundamental harmonic, while the higher harmo nics all but disappear. The obtained LCL-filter thus results in a batter y current with only a small current ripple as the battery current decrea ses at 60 dB/dec above the 2 kHz resonance frequency. The LCL- lter is d esigned such that any problems concerning the controllability and observ ability can be avoided, due to the low ratio between resonance and switc hing frequency and small secondary inductance. As the current ripple is very small, the core losses of the secondary inductor are negligible, li miting the additional losses of the LCL- lter in comparison with the LC- lter to the cable losses of the secondary inductor. At the same time, t he losses in the internal ac-resistance of the battery are avoided as th e current and voltage ripple disappear. Consequently, as both the cable losses and avoided ac-resistor losses are of the same order of magnitude and relatively small, the LCL-filter does not have a significant impact on the effciency of the dc-dc converter. The dc-dc converters with an L C-filter or LCL-filter both have a PI-controller with feedforward action for the current regulation. This current controller is the inner loop o f a cascade system with outer voltage control loop. The voltage controll er is a PI-controller used to regulate either the dc-bus voltage or the battery voltage. A special case-study is included for a fast-switching d c-dc converter with relatively large dead-time. Two 300 V , 12 kWh Li-io n batteries are subjected to the Charge-Depleting and Charge-Sustaining Cycle Life Test as described by the Battery Test Manual For Plug-In Hybr id Electric Vehicles. These tests simulate the discharging of the batter y of a Plug-in Hybrid Electric Vehicle while driving in pure EV mode and hybrid mode. Each battery is tested under identical environmental circu mstances and with the same charge and discharge pro le during one month, the only exception being that one battery is tested with a large curren t ripple, while the other battery is tested with almost no current rippl e. After each one month period, the battery's energy content, capacity a nd internal resistance are determined in order to measure any deteriatio n of these battery parameters. The results of the tests prove that the c urrent ripple doesn't have a detrimental effect on the battery parameter s. However, the current ripple causes a voltage ripple which does interf ere with the operation of the Battery Management System. As the voltage ripple causes the instantaneous cell voltage to exceed the voltage limit s while the dc-value of the cell voltage is within limits, the maximal a llowable battery voltage must be reduced. As a consequence, the useful e nergy content of the battery is reduced. It remains unclear to which ext ent the battery is damaged when the voltage ripple is allowed to raise t he cell voltage beyond the voltage limits. The double-layer capacitor mi ght provide part of the answer to this question, as it provides a low-pa ss filter for the current ripple to which the charge transfer react ion is subjected.