An industrial-scale Vanadium redox flow battery test facility (IS-VRFB) in operation at the Electrochemical Energy Storage and Conversion Laboratory of University of Padua aims at exploring the technological developments from VRFB laboratory testbeds to grid scale VRFB systems. The 40-cell 600-cm2 active-area stack and two 550-L tanks provide a power of 9 kW and an energy 27 kW h. An AC/DC bidirectional static converter that controlled by the battery management system (BMS) provides the electric power conditioning during charge and discharge with currents up to 75 A and a variable passive load allows high-current discharges up to 600 A. The plant is fully instrumented with electrical and thermo-fluid dynamic probes and the in-house LabVIEW-based BMS provides acquisition, processing and control functions. Since VRFBs are capable of fast response times and can provide fast grid services, such as frequency regulation, it is quite important characterizing their transient behavior. Nerveless, extensive experimental campaigns for investigating the fast response of a large-scale VRFB systems, are largely missing, especially in term of fast time-domain analysis. This presentation is an early contribution in such regard, in view of real large-scale applications of the VRFB technology. Two different time intervals involving different phenomena occurring in the cell have been analysed. Initially, the first 20 ms after battery powering were acquired using an oscilloscope with a sample rate of 2 GHz. Secondly, the first 120 s were explored by means of the BMS acquisition system a sampling rate of 1.25 Hz. In order to widely explore the fast response capability, different conditions were examined by varying state of charge (SOC), flow rate (Q), and current (I). The oscilloscope analyses revealed a “peak transient” time interval during which both current and voltage undergo large variations, up to 50%. During this interval, lasting few milliseconds, no reliable current and voltage value can be provided by the battery. In the longer timescale, different behaviors occurred depending on the Q, SOC, and I. The higher Q and SOC, the shorter the transient before a steady-state condition was achieved, with both current and voltage constant in time (Fig.1a). Preliminary analysis demonstrated the suitability of the model of Fig.1b to predict the stack dynamic response. Modelling resorted to experimental data obtained from a multichannel Electrochemical Impedance Spectroscopy (EIS) analyzer, tailored for IS-VRFB and capable of operating in the range 0─10 kHz with bias currents up to 400 A.
Read full abstract