Agglomeration, in deontic logic, is the rule that says: OBA, OBB, therefore OB(A&B). There is a quick argument from agglomeration to the nonexistence of at least one kind of irresolvable moral dilemma: such a dilemma appears in the form of OBA, OB~A, therefore OB(A&~A), i.e. you would be obligated to bring about a contradiction. Supposing that ought-implies-can, this would mean that it was possible for a contradiction to be true. But assuming the law of noncontradiction, this must not be the case; ergo...
On the other hand, if you deny agglomeration, then the above kind of irresolvable moral dilemma doesn't manifest.
Alethic trivialism is when every proposition is true. That contradictions would lead to alethic trivialism (AKA the doctrine of logical explosions) is taken for evidence (so to speak) that the law of noncontradiction is true, instead. Deontic trivialism would be when everything is obligated.
I haven't been able to design the argument to give us the stronger conclusion, "If agglomeration is false, then deontic trivialism is actually true." But it seems to me that negating agglomeration leads to, "Deontic trivialism could be(come) true." For suppose you had a list of all actions possible at a given time: a1, a2, etc. Usually, because it is impossible to perform every an at the same time, we would say that not every one of the an could be obligatory. However, if such impossibility is not an issue (due to the lack of agglomeration), then why couldn't all the an be simultaneously obligatory?
I imagine we could just throw in, as a basic premise, "Even though some contradictory sets of actions can be obligatory, it is never true that every such set can be obligatory." How ad hoc of a solution would this be, though? (You might try out, "A is obligatory iff it is deducible in the appropriate system; in some such system, it is possible to infer that ~A is also obligatory; but it is never possible to infer that every A whatsoever is obligatory..." But again, how should this be so?)