1. Analysis of the Effects of a Flexible Ramping Ancillary Service Product on Power System Operations
Ibrahim Krad
Eduardo Ibanez
Erik Ela

2. Motivation
Renewable energy resources, particularly wind and solar generation, are becoming increasingly more common in today’s power systems
These resources provide unique operating challenges due to their intermittent nature
Manifests as increased system variability and uncertainty
As a result, operators must revisit traditional operating techniques in order to accommodate this evolving power system

3. Motivation
Operators typically withhold capacity in the form of operating reserves to account for power system variability and uncertainty
New operating reserve methods are being studied and proposed
These operating reserves are designed around the variability and uncertainty in the net load profile (load minus VG)
They are aimed to help operators ride through extreme ramping events in the net load profile

4. Flex Reserves
This new class of reserves are known in general as flexibility reserves (flex reserves) or ramping reserves
Many different types of methods proposed in literature
Variability based methods
Net load based methods
Load based methods
Operators can implement their own versions of this flex reserve as appropriate for their system

5. Flex Reserves
Flex reserve in this analysis inspired by California-ISO’s Flexiramp product*
Implemented within the optimization as a flexibility reserve demand curve (FRDC) with dynamic break points
* L. Xu and D. Tretheway, Flexible Ramping Products: Incorporating FMM and EIM. Draft final proposal. Folsom, CA: California Independent System Operator. Dec <

6. Flex Reserves FRMIN represents the expected ramp need of the system.
FRMIN has an associated penalty cost for insufficiency
The demand curve is designed so as to purchase ramping capability when marginal cost of providing the service is less than the penalty cost
Additional steps are used to increase the flexibility requirement with decreasing associated penalty costs
Final step is the maximum flexibility reserve requirement

7. Flex Reserves Hour-ahead flexibility requirements
Day-ahead flexibility requirements
calculated based on the hourly difference in net load.
FRMIN is calculated based on the difference in day-ahead forecasts for each hour
FRMAX is calculated as the 97.5th and 2.5th percentiles for net load hourly ramps for each month and hour of the day
Hour-ahead flexibility requirements
Calculated every 15 minutes
FRMIN is calculated as the difference between the forecast for each of the 5-minute RTSCED steps that correspond to each RTSCUC solution
FRMAX is calculated as the 95% confidence interval for FRMIN for each hour of the day within a month
Real-time flexibility requirements
based on the difference of each consecutive 5-minute forecasts for net load
FRMIN values are calculated to cover 95% of the expected 5-minute ramps in the net load forecasts
DASCUC
RTSCUC
RTSCED
Step Width [MW]
250 50
Penalty Costs, Up Direction [$/MW]
250, 24, 15, 8, 2.5
Penalty Costs, Down Direction [$/MW]
250, 3.6, 2.25, 1.2, 0.375

8. Maximum reserve demand curve requirements in October
Flex Reserves
Maximum reserve demand curve requirements in October

9. Simulation Testbed: Data
Analysis performed on a modified IEEE 118 bus system*
Generation and transmission data updated to better capture available operating cost data
Approximately 34% annual VG penetration added to the system split 50/50 between wind and solar
Sited to maximize access to transmission
Based on available data for northern California
Simulated for 4 separate weeks (Jan, Apr, Jul, Oct) to capture seasonal trends in VG and load data
* G. Stark, Study on Integration Costs. NREL/TP-5D Golden, CO: National Renewable Energy Laboratory.

10. Simulation Testbed: FESTIV
Simulated with the Flexible Energy Scheduling Tool for Integrating Variable generation (FESTIV)
Developed by researchers at the National Renewable Energy Laboratory (NREL)
Steady-state power system operations simulation tool
Captures all scheduling horizons starting from the day-ahead commitment decisions through automatic generation control occurring at sub-minute temporal resolutions
Simulates the scheduling and deployment of operating reserves

11. Simulation Testbed: FESTIV
All models are interconnected such that the output of any one model serves as the input into subsequent models
Commitment problems are formulated as mixed-integer linear programming (MILP) security constrained unit commitment problems
Economic dispatch problem is formulated as a linear programming (LP) security constrained economic dispatch problem
FESTIV provides both economic and reliability metrics

12. Simulation Testbed: FESTIV
FESTIV simulation flow chart

13. Results Case 1 : Base case without the flexibility reserve product
Case 2 : Includes the flexibility reserve product in all operational time frames
Case 3 : Remove the flexibility reserve product from the day-ahead time frame
Case 4 : Remove the flexibility reserve product from the day-ahead and real-time dispatch time frames
Case 5 : Base case without transmission constraints nor the flexibility reserve product
Case 6 : Base case without transmission constraints but with the flexibility reserve product

14. Results
Real time market infeasibilities occur during times with low available ramping capacity
Removing the flex reserve from the DA reduces the amount of available ramping capacity

15. Conclusions
Although the impacts on production cost are relatively small, there could be significant implications on reliability
Could offer more value in managing the uncertainty of VG than the variability of VG
Important to include this reserve product in the day-ahead commitment because committing slower thermal generators can significantly help reduce the imbalance during operation
The flexibility reserve product was able to reduce the number of real time market infeasibilities

16. Thank you for your time!!
Questions?