1) From what I understand, the elemental recharge gets aggregated to the nodes based on some sort of "area of influence" (likely a Thiessen polygon). That's ok in general, but introduces a small error to sub-watershed water budgets. Nodes right on the sub-watershed boundary add recharge from the neighbouring elements outside the current sub-watershed! To verify this, compare the sum of recharge of all individual sub-watersheds with the total recharge of the model. The same issue applies to all other BCs on a polygon boundary (luckily, rivers and lakes rarely fall onto a sub-watershed boundary ;D)
2) I had to think about this one for a moment, but it does make sense. If you look at the file documentation (file_formats.pdf), a DAC file is basically the FEM file plus the model results. For a transient run this would be
Ini Cond (as in FEM file) plus
Results TS 1
Results TS 2
Results TS 3
etc ....
For individual TimeSteps, only the results get exported, but not the other model parameters. This is a good thing though, since the DAC file would get really really big for a transient run otherwise. Since you are already working with IFMs, Peter's suggestion is the way to go! Two more thoughts on that:
a) The API function for calculating the flow components (IfmBudgetComponentsQueryFlowAtNode) are SLOW! I normally call them only for nodes that actually have a BC, and avoid the call for nodes with no BCs. The check
IfmGetBcFlowType(pDoc, node) != IfmBC_NO
takes computational time as well, but likely less then IfmBudgetComponentsQueryFlowAtNode. You might wanna check this in your particular situation.
b) Using an IFM module gives you the chance to calculate the recharge to an area more precisely than the water budget functions. Rather than using nodal recharge values, you can use:
elementalRecharge = elementArea * IfmGetMatFlowRechargeValue(pDoc, element)
This overcomes the issue mentioned above (see 1), however, you need to calculate the element area "by hand", since there is no API function available for this (if there is, it's not documented at least)
Fun stuff, Chris