Missing in Action-the Day Ahead Schedule
- Jul 14, 2015 12:00 pm GMTJul 20, 2015 8:19 pm GMT
- 2728 views
Today’s electricity networks were designed on the presumption that load and generation were both predictable and slow-moving. Because of this presumption, control room operators had time to wait for the Energy Management System to identify the size of any change and available resources to respond to the change and to manually dispatch the generation resources necessary to bring the system back into balance.
With the advent of Variable Energy Resources (VERs) that is no longer the case. As larger amounts of VERs such as wind and solar PV are deployed, the underlying design basis of those control room networks, now contained in RTO/ISO environments, is increasingly invalidated.
For example, scheduling of load and generation is critical to the reliable operation of electric grids because it ensures that there is adequate transmission and energy available to balance the two. Scheduling has worked well in the past because the output of conventional generation resources has the same characteristics as load. Except on very rare occasions, both generation and load were highly predictable and changed slowly with time.
Energy production from wind and solar PV is notoriously prone to unpredictable swings. Using publically available data, our analysis has calculated the wind forecast errors for the CAISO system for most of 2014. The results are shown in the figure below and clearly indicate the difficulty of maintaining a schedule for wind resources.
The prevailing response to this variability is to enhance the ability of Balancing Authorities to reactively mitigate these forecast deviations by increasing the pool of “flexible” gas-fired generation, either by building out capacity or by enlarging Energy Imbalance Markets. While there is merit in this approach, it raises a few questions in our minds:
- - Does it mean that the industry should abandon the schedule as the primary tool for balancing load and demand and ensuring grid reliability?
- - Even with more generation at its disposal, will BAs be able to meet NERC compliance requirements by reactively managing schedule (forecast) errors with centralized dispatch, especially as the level of renewable generation increases?
- - Are we willing to accept the cost of this approach with its heavy reliance on the Energy Imbalance Markets and capacity contracts?
The Schedule is Still the Best Tool for Ensuring Reliability
Grid reliability would never be an issue if all generation schedules were met. Problems arise when demand is more or less than scheduled but these situations, with the exception of rare summer days have been routinely handled by operators, e.g., by dispatching peaking plants. Until the last decade about the only time actual generation deviated from scheduled generation was when a power plant experienced an unexpected, forced outage.
The industry should not give up on using scheduling as a reliability tool. Our team at Onset, Inc. has developed a control system and software to directly couple VERs with such firming resources as a dedicated gas plant and energy storage systems for the purpose of maintaining a schedule in real time. This allows the coupled VER generation to be scheduled in the Day Ahead Market with a minimal risk of schedule deviations.
Onset has codified this approach in a product identified as “UniGen.” Using the CAISO data mentioned above, we have simulated UniGen under a number of conditions, all the while using realistic operating data (meteorological data, forecasts and ramp rates, for example). A typical result is shown below. The application of UniGen reduces schedule deviations by 97% compared to the case of scheduling VERs without UniGen. In our view the UniGen approach makes it possible to schedule VERS in the Day Ahead Market, dramatically reducing the risk of incurring large balancing costs in the real time market.
Integrating Renewables at the Project Level
Currently, grid managers rely on an Energy Management System (EMS) to identify deviations from the scheduled generation, after which operators make adjustments manually and wait for the adjustment to take effect. All this can take 10 to 30 minutes, by which time the disruption caused by the volatility of wind and solar has most likely disappeared or changed. This lengthy control process is referred to by control engineers as a long loop process. Without an ability to correct the swing in power within the 10-minute timeframe, which is becoming an increasingly difficult task, Balancing Authorities run the risk of violating basic NERC reliability standards. Indeed, NERC’s new emphasis on embedding frequency response characteristics into the grid is seemingly an admission of how difficult it will be to continue meeting such standards as CPS-1 and CPS-2.
The UniGen control system is a short loop control system that directly couples renewable resources and any firming resource such as gas-fired power plant and/or storage resource in near real time. This is accomplished by slaving the output of the firming resource to the output of the VERs using proprietary algorithms and software. The UniGen control system operates in real time so that the combined output never deviates (except within a very small range) from the committed schedule. And unlike the centralized approach to offsetting schedule deviations, UniGen is a distributed solution. The problem is automatically and quickly solved at the project level instead of allowing the problem to make its way into the grid manager’s EMS.
A Hybrid Approach is Less Expensive than Relying upon the Energy Imbalance Market
Savings will occur using the UniGen approach. The primary benefit is from using dedicated gas-fired and/or storage resources to firm the generation schedule of a VER project. The costs currently occurring from having to rely on the dispatch of other resources, or an Energy Imbalance Market, can be eliminated or dramatically reduced.
Onset’s simulation model of UniGen contains a financial module that calculates the cost of firming the schedule using an ISO/RTO’s real time information and the cost of using a gas fired generator with or without a storage system. The balancing energy costs, without UniGen, are shown below. It is obvious that the costs are unpredictable and at times extraordinarily high. Comparatively, the cost of using UniGen to couple a VER with, say, a dedicated power plant, especially an underutilized plant with a reduced debt burden, is not only less but the schedule is much more predictable. The simulations demonstrate that the cost differential can be substantial, up to $1M to $2M for every 100 MW of VERs. What this means is that using UniGen, market participants have financial incentives to schedule VERs and thereby lend a hand to the task of integrating renewables.