With my best understanding (mostly from the practice of MISO), the goal of Day-ahead SCUC is to provide commitment decisions for the next day while the commitment decisions will also guarantee there are enough "average" power/energy with sufficient ramping for each of the considered discretized time-window. (See [1] for the manual reference)
In other words, we can also say SCUC is not perfect or ideal because:
- it's considering a moving "average" in general "sum" but not every couple of minutes at each generator & bus level -> you may miss the "peaks" (aka worse case situations like temporarily loss of winds for parts of your grid in 10 minutes, then back in another 10 minutes),
- checking all possible contingencies are difficult due to computational complexity -> in practice not all transmission constraints/limits & contingency limits are implemented -> in practice you can still miss some violations/contingencies,
- Same with 2), the transmission network may be omitted/aggregated/reduced to trade fidelity & accuracy for speed,
- there are usually no explicit models on reactive power there -> it's at most a DC solution but not an AC solution -> there is no guarantee on whether there will be voltage collapse plus unsure what are the reactive reserves needed (i.e. you may actually need to correct your decisions to start up more generators due to just the need to provide reactive reserves but not active reserves.)
To tackle at least some of the issues above, individual Day-Ahead SCED will be run separately, where these SCEDs will also use system state-estimation data to try to model extra detail with the transmission network(s). The idea is that hopefully these Day-Ahead SCEDs will capture some of these violations early on in the Day-Ahead market. In practice, we still have to run Future/Intra-Day/Look-Ahead Reliability Assessment Commitment Process (FRAC/Intra-Day RAC/LAC) to adjust many of the issues (e.g. a wrong prediction), plus the real-time market to correct the final bits.
Theoretically speaking I do agree we can always defined one super large Day-Ahead model merging the existing SCUC and all subsequent SCEDs together mathematically in full details (AC, all network constraints, all contingency constraints, etc), and there are research work putting AC equations on SCUC too. Computationally speaking, due to its mixed-integer complexities (and AC non linearity), decomposition methods are still widely used and the current industry seems fond of the current SCUC and SCED(s) decomposed architecture. Of course, it is forever evolving (see [2])
References:
[1] BPM 002. https://www.misoenergy.org/legal/business-practice-manuals/
[2] https://www.ferc.gov/sites/default/files/2020-06/T3-1_Pan_et_al_0.pdf