Some survey I have done by myself and more analysis with engine assistance. According to the machine it was only with 49.Ra3 that we went wrong. We just didn't dare to play more agressive and activate the Rook, like some of our members have suggested. Even in the finale position, the engines (Stockfish 10 and Komodo 13) give an advantage to white although it is not clear to me how it should be converted.
A remark: I have kept the evaluation of the engine where it was generated but you shouldn't pay too much attention to that. I did eliminatie the accuracy at the end of the game. It was about 42% on both sides (Stockfish 10 on Chessbase 15) but that percentage means not much to me. I do know that when I analyse my own finished games, I usually end up about 10% higher. Probably I spend a bit more time at these games.
Some survey I have done by myself and more analysis with engine assistance. According to the machine it was only with 49.Ra3 that we went wrong. We just didn't dare to play more agressive and activate the Rook, like some of our members have suggested. Even in the finale position, the engines (Stockfish 10 and Komodo 13) give an advantage to white although it is not clear to me how it should be converted.