Have you tried explicit (manually written) constraints instead? The constraint documentation shows how to write them. A large series of atom-pair constraints (say, holding the extant hydrogen bonds in place) ought to have the effect you want.
There ought to be an automatic way to do what you want but I don’t know what it is. You could also try using the flag you tried, but passing flags to increase the weights on the constraint score terms (see the constraint documentation).
Do you see constraint score terms in the output anywhere? (Do they appear/disappear as you turn the flag on/off?) If you don’t see them as part of the score function, then the flag probably has no effect rather than little effect.
.
.