Looks like the weather still was way ahead of its time at May15th at 0810Z. Notice the cloud formation well east of the adratic coast near the black sea, where irl it probably will reach in the evening.
And here’s what was loaded May15th between 0845Z and 0901Z :
Look familiar? Maybe because it’s nearly the same that was loaded yesterday at around 2010Z:
And it’s definitely the actual situation and was pretty accurate even yesterday evening. (as there’s actually a bit more high cloud coverage overhead EDDH in the SAT-image I marked the frontal convection in northern Denmark as it is probably hard to notice in the screenshot. And notice again the clouds over the adriatic sea and their actaul position at this time.
CONCLUSION
I guess it’s safe to assume that 12 hours of weather is missing completely in the sim.
The weather model from 0900Z onwards is obviously loaded twice, once around 2000Z the evening before, and again - correctly - around 0900Z. Now the only thing to do for ASOBO is to find the missing hours and inject them at 2000Z.