I believe a partial answer might be the following.
The average premium of coin value over the value of the metal in them was actually relatively high. The reason for melting coins was the variance in this premium.
At issuance time the premium was set by the mint (it was a sum of minting costs and the implicit tax on coinage set by the ruler).
While coins were in circulation, the amount of metal in them gradually fell due wear and tear, and due to (illegal) scraping of the metal. This reduced the value of the metal in the coins, so the premium got even higher up. (Unless, that is, the coins were accepted by weight rather than by face value, as they often were in wholesale trade; in that case, the premium didn't change.)
From time to time, the mint had to replace old coins with new ones (otherwise, people would stop accepting coins due to the extreme wear). To avoid losses from such an exchange, the mint reduced the quantity of the metal in the new coins to match the average amount of metal in the old (worn) coins.
However, the wear was not uniform. When the exchange time arrived, whoever could get hold of coins that were relatively unworn, may find it profitable to melt them instead of exchanging them for new coins.
Example. Suppose a coin was minted with 5 g of silver. After 30 years, the mint wanted to issue new coins. Perhaps by then average coin silver content dropped to 4.5 g, so the new coins would be minted with 4.5 g of silver. Suppose that the mint charged the public 0.2 g to convert bullion to coins. If someone held well-preserved coins with 4.8 g of silver, they would melt them, and buy new coins from the mint (profiting 4.8 - 0.2 - 4.5 = 0.1 g of silver per coin).
Sources:
Jim Bolton, Money in the Medieval English Economy: 973-1489.
John Munro, The Technology and Economics of Coinage Debasements in
Medieval and Early Modern Europe: with special reference to
the Low Countries and England, p. 18.