It seems odd that the amount essentially gets counted twice. Assuming that makes sense, however, I would recommend something along the line of the previous suggestion. I wouldn't necessarily use the surrogate key, but I would create an artificial key if there is not a natural key to use that would be different for all the accounts, even those of the same name. The ETL process would take the single value for the repeating accounts and ensure that "both" accounts are populated with the proper single value. SSAS would see two accounts, but it would still function as you've described.

What do you think?