Global Practice Library

Defining operational flexibility in planning studies

In recent years, planners have started to consider operational flexibility explicitly in studies and decision making. This includes a number of different approaches, ranging from detailed consideration based on simulation tools, to more simplistic approaches that screen the flexibility available in the system as a starting point to further potential studies. Generally, the overall aim has been to ensure ramping capability is considered in planning, and that systems have a sufficient amount to effectively operate in the future. Examples of ongoing processes include the following.

California’s Flexible Resource Adequacy Must Offer Obligation (FRAC-MOO): CAISO, under mandate from the California Public Utilities Commission (CPUC), developed a new resource adequacy related procurement target [4], which has been in place since 2014. Scheduling coordinators are required to not only procure sufficient capacity to meet forecasted peak load as they currently do, but also to meet additional flexibility requirements with their capacity. The objective is that the system should have sufficient flexible capacity available to meet forecasted system needs. The method for determining the need and the contribution of different resources to the need has evolved over the past few years. The need is determined based on a minute-by-minute dataset of actual load from the previous year, together with minute-by-minute data for variable generation; projected future variable generation installations are included by scaling wind and load using 1-minute historical data. The requirement is then calculated based on the largest 3-hour ramp in each month (since 3 hours is the period of most concern to CAISO), plus the maximum of either the largest contingency or 3.5% of the peak demand in that month, plus an error term to adjust for load forecast error. The three hour ramps are separated into the maximum primary and maximum secondary daily ramps of each month, with the primary being the timeframe where the maximum ramps occur and the secondary ramps being the timeframe which does not coincide with the primary timeframe where the second largest ramps occur.

Once the total procurement requirement has been calculated, the next step is to distribute the total requirements among each LSE. This is done for each of the three individual components (maximum ramp, contingency or 3.5% of peak demand and error term), based on the contribution of the LSE to that component. They must procure sufficient effective flexible capacity to meet their seasonal requirement. The Effective Flexible Capacity (EFC) is calculated for each resource based on their capability to ramp in 3 hours. The contribution from quick start units is their maximum ramp from cold start in 3 hours, whereas for longer start units, the EFC is calculated as the 3-hour ramping capability when the resources are online and at minimum generation (even if these are not online when ramps happen, it is assumed they would be if the ramps were extreme cases). While California has identified 3-hour ramping issues as the key issue for its particular circumstances, this may not be the same for all regions at all times. EFC resources must offer into the day ahead and real time markets at certain times of the day depending on which type of ramp it has cleared for (primary or secondary) – hence the must offer terminology.

The FRAC-MOO process provides a good overview of a potential mechanism to provide flexibility in the planning time frame, however it is expected that it will need to evolve. This may include more detailed consideration of minimum generation levels, over-generation issues when solar power is high and understanding what a generator is likely to be doing when its flexibility is needed.

Eirgrid (Ireland) System Services: EirGrid is the transmission system operator for the island of Ireland where renewable generation currently meets over 22% of annual energy demand and is set to increase further to 41% in the near future. The majority of the renewable generation consists of wind generation, with future growth in the sector all linked to wind and a small amount of solar generation. The Irish system does not have synchronous connections with other systems, but does have two HVDC links to the UK (250 MW and 500 MW, each) and another 750 MW HVDC link to France in the design stage.

In order to address the future operational issues with high penetrations of renewable generation, EirGrid launched the DS3 initiative in 2012. This project was aimed at developing the system services and capabilities that would be needed by the system operator to increase the system non-synchronous generation penetration limit from 50% to 75% in operations. To date the limit on non-synchronous generation penetration has been raised to 60%, effective from November 2016. The main focus of the new system services was on three areas: frequency control, voltage control and ramping requirements. Through various studies, a set of 14 ancillary services were proposed including synchronous inertial response, fast frequency response, operating reserves dynamic reactive reserve and ramping capability.

The ramping services have been specified to manage the expected variability and uncertainty in the upward direction that will arise at the renewable energy penetration that is expected to materialize. Three services are being implemented, differentiated by the deployment time frames: 1, 3 and 8 hours. Each of the products has an associated duration for which the response should be maintained. The durations are currently set at 2, 5 and 8 hours, respectively. Resources are contracted on an annual basis and remunerated for a payment in respect of the additional margin that they can provide in the horizon and sustain for the duration. For example, a 100 MW unit at an output of 50 MW and with a 25 MW/h ramp rate can recover the payment for 25 MWh, whereas the same unit dispatched to 90 MW can only receive payment for 10 MWh of the service. If a resource cannot technically sustain the response for the duration period, the payment is limited to the minimum potential capability.

The products are not mutually exclusive and payment for availability for multiple services can be received in the same hour. The ramping services are currently remunerated at the rates shown in the following table for each MW of capability in each hour (settlement units are in MWh).

Table 1‑1
EirGrid DS3 tariff arrangements for ramping margin products for Oct 2017 to April 2018

Service	1 Hour Ramp	3 Hour Ramp	8 Hour Ramp
Rate	0.11 €/MWh	0.17 €/MWh	0.15€/MWh

The maximum payment that can be achieved is by a resource that remains offline but can start and reach maximum output in an hour and maintain output for 16 hours. Such a resource can earn a maximum of €72.24 per megawatt of capacity per week from ramping payments. For a 100 MW fast start resource that remains offline and provides perfect service when called upon, this equates to €72.24/MW·week×100MW×52week/year = €375,648 per annum.

Flexibility has also begun to be considered in various Integrated Resource Plans. The following is a selection of different plans:

The Oregon PUC has required investor owned utilities in the state of Oregon to consider flexibility as part of their Integrated Resource Plans (IRP) [6]. It is still up to each utility to determine how to quantify requirements and whether their current resources have the ability to meet those requirements. If there are insufficient resources, then all resources (including institutional/operational reforms as well as supply- or demand-side resources) must be considered. Based on this requirement, utilities in Oregon are considering flexibility in their IRPs. To the authors’ understanding, some have proposed detailed study methods using production cost tools, while most are still using relatively simple screening methods that look at large ramps over specified time horizons and compares to system resource capabilities. As an example, Portland General Electric (PGE) now include flexibility metrics in their IRP, which resulted in the need for 400 MW of flexibility in the 2016 IRP, to avoid real time imbalances. This used the REFLEX tool, described later, and similar concepts to some of the EPRI flexibility metrics Oregon also has a storage mandate, providing flexibility to meet the needs of the system to manage variability and uncertainty.

Similarly, Public Service of New Mexico (PNM) have recently begun to include the concept of flexibility in their IRP process. Here, traditional resource expansion tools identified least cost mix of resources to meet future demand, under various policy and other preferences. The SERVM tool, described later, was then employed to provide additional analysis as was carried out in the California CES-21 effort described later. This probabilistic production cost tool was used to examine the intra- and multi-hour flexibility needs of the system using a 5-minute dispatch model incorporating variability and uncertainty of wind and solar power. Metrics related to flexibility shortfalls with and between hours were then used to ensure the resource mixes developed by the resource expansion tool could maintain reliability. One of the challenges associated with this approach, as discussed later in the report, is to determine the level of reliability which must be maintained. Traditional planning standards such as loss of load expectation (LOLE) due to capacity shortfalls do not consider ramping or flexibility issues. Therefore, if new methods determine loss of load due to ramping shortfalls, one should not continue to just use the same reliability level (e.g. 1 day in 10 years); instead the appropriate standards need to be determined first, based on a combination of detailed studies as described in this report, and determining a baseline from current systems.

The Tennessee Valley Authority, in their 2019 IRP, included two flexibility metrics: the Flexible Resource Coverage Ratio and the Flexibility Turn Down Factor. These metrics are defined in the IRP as follows:

Flexibility Resource Coverage Ratio: the ratio of flexible capacity available to meet the maximum 3 hour ramp in demand in 2038

Flexibility Turn Down Factor: the ability of the system to serve low load periods as measured by the percent of must-run and non-dispatchable generation to sales

For the scenarios considered in the 2019 IRP, coverage ratios ranged from 0.98 to 2.22 and turn down factors ranged from ~32% to ~66%.

The importance of flexibility is given prominence in the report with substantial reporting of the performance of each portfolio under the 6 different scenario and 5 strategies evaluated. Broadly speaking, the portfolio based metrics reflected highest flexibility in cases where inflexible or must-run generation is replaced with more flexible generation. Cases with higher levels of solar are seen to have lower coverage ratios.

While it did not significantly impact on the specific plans being developed, it was calculated based on the ability of the fleet to follow load swings, and each scenario in the IRP was assessed against the flexibility metric. This was calculated based on the annual system regulating capacity (regulating reserve, demand response and quick start resources) expressed as a percentage of peak demand. While for this round of IRP scenarios, there was not a significant change in flexibility across scenarios, TVA will continue to monitor as they observe greater shares of variable renewable energy resources in their footprint. As well as this flexibility related metric, two other related metrics were examined though not considered part of the scoring for different scenarios. These included the variable energy resource penetration, which measures the amount of variable included in the plans; and a flexibility turn-down factor to measure the ability of the system to serve low load periods. These measures are consistent with other areas that also include renewable penetration and turn down ability in their planning processes.

Finally, the North American Electric Reliability Corporation (NERC) recently released developed a set of measures related to Essential Reliability Services, in light of the changing generation mix expected in the coming years [7]. One of the key reliability services identified was the Essential Reliability Services Ramping Measure. Here, the purpose was to develop a measure to help identify when ramping and balancing are becoming more challenging for system operations. The measure involved identifying if system ramps are changing sufficiently such that existing operational ramping capability could be exhausted. A number of levels are proposed for the measure, with both a screening method and a more detailed method based on the Control Performance Standard (CPS1) score, which reflects the Area Control Error deviations in a balancing area.

The screening method consists of two steps – in the first step, the minimum load is compared to non-dispatchable generation (which may nuclear, geothermal and renewables without dispatch capability – note all of these resources may have the capability to ramp to some extent at present or in the future if so enabled). If non-dispatchable resources are greater than a certain percentage of minimum load, then the more detailed screen is used. The percentage is chosen by the user, but should be between 30% to 50% based on NERC recommendations (this is subjective based on end user).

If this first step shows high percentage of non-dispatchable generation during minimum net load, then in the second step the upwards and downwards ramping capability for critical hours is compared to the regulation, load following and load increase in the hour. If both of these levels fail, then more analysis is needed. This is not a prescribed analysis, but one potential option described in the report is based on the EPRI framework described in this report, where simulations of future system are carried out to assess flexibility. This second step should be done for future years, though the method to determine ramping needs and resources if left to the user- it is not proscribed to do a full simulation of what wind/solar resources may look like, or what resources may be available during critical operating periods.

The other analysis proposed by the ERSWG is based on CPS1 scores. Here, historical ramps are assessed against CPS1 performance on an hourly basis. A number of analyses are performed, similar in nature to the flexibility requirements proposed by EPRI discussed below. These include analyzing by hour and month, and also analyzing for consecutive hours of CPS shortfall. The aim is to determine whether the balancing authority is seeing an increase in total shortfalls or consecutive shortfalls across the year, or in particular months or hours. If so, then more detailed analysis should be performed, with a recommendation to work with NERC’s Resources Subcommittee. More details on the NERC measures can be found in the “Sufficiency Guidelines” report finalized in December 2016 [8].

The above examples show that flexibility assessment methods have evolved significantly, and now span everything from inclusion in IRPs to reliability assessment. However, methods are still evolving, and generally try to balance data and effort intensive simulation approaches with simple to use screens. Additionally, they balance economic decisions with reliability related approaches, and in the long run will need to evolve to where they can better inform planning decisions.