Use Nodal Commitment
This method makes use of nodal information to inform the standard zonal commitment routines. The logic takes information from the linear program (LP) solution to the nodal formulation to adjust the costs/prices seen in the economic (self) and pool-based commitment logic. The nodal commitment logic works differently for pool and economic commitment. (See Commitment Logic for an explanation on zonal commitment methods.)
NOTE: This switch is only applied when using the Traditional Commitment logic. When using the Commitment Optimization logic, the model will use the nodal information directly in the mixed-integer program solution and will not employ the heuristics described below in this article.
With Pool Commitment
When Use Nodal Commitment is selected in conjunction with pool commitment (as specified in the Operating Pools table), the model uses nodal generator shadow prices to inform the hourly commitment decisions. The linear program which finds the optimal nodal solution each hour returns shadow prices for each generator, giving information about the impact of running that generator on the total system cost. The shadow prices will be influenced by what is happening in the nodal system, including congestion, binding corridors, active contingencies, and each generator’s location. The core pool commitment routines are unchanged by the use of nodal commitment. What the model does change is the resource costs seen by the commitment algorithm as well as the order in which units are prioritized for commitment. For example, the zonal logic alone might see no difference between two similarly priced generators, whereas the nodal feedback (i.e., shadow prices) might inform the commitment algorithm that one of them is especially needed to avoid congestion, thus giving it a higher priority of being committed.
The model performs the following general steps to make use of these shadow prices in the commitment decision:
- The first day of the nodal simulation is run twice in order to get the initial shadow price estimates by hour of the day for each generator. The first time the day is run the commitment decisions are based only on zonal information, whereas the second time the model uses the nodal information.
- The model creates a dynamic rolling (24 hour) forecast of the generator shadow prices in order to adjust the cost seen by the pool commitment logic. The commitment algorithm will consider the estimated impact on system cost for each generator throughout the hours of the commitment window in making the decisions. More specifically, when the model calculates the average cost for each unit (a metric involving the incremental cost of all resource segments as well as start costs), a portion of the shadow price will be added to (or subtracted from) the true operating costs so that nodal congestion will implicitly reorder the units when necessary. These shadow prices are considered in conjunction with the standard resource commitment parameters such as minimum up and down times, startup costs, incremental and average costs, etc. that are core to the pool commitment algorithm. Once the costs have been adjusted to take into account the expected nodal congestion, the actual commitment routines are the same as the base pool commitment logic.
- Each hour as the model makes the next hour’s commitment decisions, it takes into account not only the forecasted shadow prices across all hours of the day but also the results of the most previously solved nodal hour. This allows the commitment decision to dynamically adjust for the most recent signs of congestion on the system when the congestion merits an immediate change in the commitment decision.
With Economic Commitment
When using Nodal Commitment in conjunction with one of the economic/self commitment options (i.e., pool commitment is not selected), the model will base the price pre-forecast on the nodal information. The standard zonal logic creates a 168-hour rolling price forecast for each zone against which commitment decisions are made for each individual commitment unit. The pre-forecast is based upon the zonal demand as well as the expected price/demand relationship which is continually updated as the run progresses. With Use Nodal Commitment selected, the model creates a pre-forecast for each nodal LoadFlow Area and Aggregate Area based upon the nodal system. The pre-forecast is developed in the same way as the zonal pre-forecast, however nodal area LMPs and nodal area demand are used instead of zonal prices and zonal demand.
Each commitment unit will base its commitment decisions either on the pre-forecast for its LoadFlow Area or its Aggregate Area. The latter is used only when the LoadFlow Area’s average demand is less than 1% of the system demand. The model will also develop an expected forecast of differences between generator LMPs and the area LMPs, which will be used further to refine the pre-forecast seen by that resource before it is evaluated for commitment/decommitment. These expected deltas between the bus level and area level LMPs are based upon the observed relationship for the most recent week dispatched, and they will be unique for each hour of the day. The inclusion of the bus level LMPs in the algorithm allows the model to give essentially a unique pre-forecast to each commitment unit which is affected by the location of the generator and the congestion caused by branch limits, corridors, and contingencies.
Once the pre-forecast is updated for the hour for a given generator, the normal economic commitment routines are employed to make the commitment decision. When the Dynamic price commit method is selected, the model will dynamically adjust the area level price forecast as units are committed/decommitted within the hour, just as the zonal price forecast is adjusted in the regular zonal logic. When units are not committed and a Non-Commitment Penalty (NCP) is applied, the NCP will be based upon the expected shortfall for the unit as seen from the nodal pre-forecast.
Just as with pool commitment, when Use Nodal Commitment is selected to be used in conjunction with economic commitment, the first day of the simulation will be run twice. The first time will use zonal information to make the commitment decisions and develop the needed nodal information, and then the day will be run again, this time using the nodal developed pre-forecast of prices.
Use Nodal Commitment
