If requirements were not clear in the RFP, and especially if the construction was allowed to start before specs were approved - you really have no recourse other than to pay the Consulting bill.
However, if something was learned - it was a cheap lesson. On the other hand, if your organization is willing to repeat the mistake, the lesson will become increasingly expensive and painful (as you will still not have requirements met, as well as the money wasted).
Someone needs to Manage the Consulting.
Functional Requirements can be a deliverable from the Consultants, if they are still not clearly documented - it would be worth some Consulting time to make them clear - make sure that you are clear about requiring approvel before continuing to Technical (Deliverable) Specifications.
Again, Consultants may be used to document clear and detailed Technical Specifications (and generally should include screen mockups, report mockups, input specifications, workflow documentation). Again, be clear that these must be approved before any Construction is to start.
Then you should have a Test Plan, even prior to actual construction - exactly what criteria will define success must again be documented very clearly, perhaps even to the point of hypothetical test cases and how they should process for the system to be considered "Acceptable".
If controlling Consultants is an issue, bring in another Consulting group to handle acceptance testing (make Consultants evaluate work of other Consultants and document success and deficiencies). Most often, once the developers are aware that they are going to be evaluated by clear criteria - they will fulfill that criteria (or will also realize that the clarity is sufficient to define outcome of any disputes about developing what was required.
Best way to handle deadlines is completion bonuses based on equally clear 'milestones', such bonuses might call for lower overall compensation for Consulting (to manage your total cost to be about the same as it would have been on an 'uncontrolled' fixed bid). Bonus sounds more like a 'carrot' than a 'stick' - but can serve both functions. Make darn sure that they are not going to be 'automatic', but rather are based on clearly demonstrated completion of major sets of tasks (like screen prints from a test, for example).
Management might not like the idea of bringing in multiple Consulting groups (one to 'control' the other), but point out the savings that can be realized vs paying for an uncontrolled development that may not only fail to deliver what was required - but may actually hobble the efforts by delivering something that may require more effort to run than if it was never developed in the first place. Mis-management can be much more expensive than no management at all...