Introduction Water electrolysis is expected a key device to introduce large-scale renewable electricity under management of power grid and electrification of non-electric sector. While alkaline water electrolysis (AWE) systems are well-developed large system, degradation under fluctuated operation with start and stop operation is significant issue to combine photovoltaic and/or wind turbine generation is significant issue. In this study, we have been investigated reverse current, which is leak current through manifold of bipolar alkaline water electrolyzers, and electrode potential behavior of stop operation, and proposed accelerated durability test (ADT) protocol for start and stop operation. Experimental and modeling Figure 1 shows configuration of 4-cells bipolar alkaline water electrolyzer that was consisted with end plates of EP(-) and EP(+), bipolar plates of BP1 to BP3, anodes, separators, and cathodes with principle of reverse current. The end and bipolar plates were made of nickel. Anodes and cathodes were commercially available oxygen and hydrogen electrocatalyst coated nickel mesh electrodes (De Nora Permelec Ltd) with 27.8 cm2 of projected area. Zirfon Perl UTP500 (Agfa) was used for separators. Manifolds were made of Teflon tubes and 15 mL/min of 7 M (= mol/dm3) was circulated for each anolyte and catholyte chambers during measurements. An anode and a cathode set on a bipolar plate. During operation anodes and cathodes are oxidized and reduces, respectively. After stop, the anode and the cathode on a bipolar plate connects both electronically and ionically, so oxidized anode and reduced cathode surface discharges to same potential. In this study, we measured electrode potential and reverse current after 1 h water electrolysis of 80oC at 0.6 A/cm2. The reverse current was measured ionic current through communicating tube with D. C. clamp meter (KEW2510, Kyoritsu).Reverse current behavior was analyzed with COMSOL Multiphysics version 5.5 based 2-dimensional model of stack and height direction using experimentally anode and cathode potentials as functions of discharge for anode and cathodes. Results and discussion Figure 2 shows cell performances in the stacks and a picture of the lab-scale zero-gap configuration electrolyzer. All cells in a stack showed almost the same performance. The cell voltage was 2 V at 400 mA/cm2 at 30oC and was 1.8 V at 500 mA/cm2 at 80oC. Therefore, we think all cells and parts work well.Figure 3 shows reverse currents and electrode potentials as a function of time after 1 h electrolysis under 600 mA/cm2 at 30 or 80 oC. Here, the dashed lines in Fig. 3-a) were simulated value and were almost same after 20 s. The measured reverse current increased around 20 s. At this moment, outlet manifolds filled with electrolyte to increase ionic conduction among the chambers, which was not considered in the simulation. After degassed of manifolds, the reverse current decreased with time with the largest reverse current for the BP2. The reverse current at 80oC was significantly larger than that at 30oC. This difference could be explained the dependence of ionic resistance of manifold on temperature. At same moment, anode and cathode potentials on the end plates were almost constant. The anode cathode potentials on the bipolar plates decreased and increased with time, respectively. Both anode and cathode showed significantly potential change region. The final potentials of all electrodes were around 0.9 V vs. RHE. The potential change regions of the electrodes on the BP2 were earlier than others. Here, the anode and cathode potentials as functions of discharge were almost same for the electrode on all bipolar plates. Therefore, the average discharge functions for anode and cathode were treated as characteristics of electrodes of this study. Using this function, reverse current as a function of time could be expressed accurately using the developed model. From this model, reverse current, and electrode potentials as a function of time would be expected with discharge function of each electrode and ionic resistance of manifold.Figure 4 shows that a start & stop operation simulated ADT protocol based on bipolar electrolyzer measurements. We propose the combination of constant current electrolysis, potential sweep, and chronoamperometry as the inset of illustration in Fig. 4, because constant current measurement is easier to get reproductivity in high current that need accurate iR correction for constant potential measurement and current control measurement never simulate stop operation. Acknowledgements This study was based on results obtained from the Development of Fundamental Technology for Advancement of Water Electrolysis Hydrogen Production in Advancement of Hydrogen Technologies and Utilization Project (P14021) commissioned by the New Energy and Industrial Technology Development Organization (NEDO). Figure 1