Counteractivations Compromise proposal for counteractivations in the m

Counter-activations Compromise proposal for counter-activations in the m. FRR- and a. FRR-Platforms 1

What are counter-activations? • Counter-activations are equal to activations of both a upward and downward bids at exactly the same time. Common Merit Order List (CMOL) • • In case of price inversion between the upward and downward CMOL, the optimisation algorithm will naturally tend to activate in both directions if this is not explicitly prevented by an additional constraint. Price [EUR/MWh] Volume [MW] Counter-activations in this context is not related to situations where the system direction changes within the ISP, which may also cause activations in both directions. 2

Why are counter-activations controversial? Prevention of CA is hard to achieve together for instance with indivisible bids or unforeseen rejection of divisible bids. Common Merit Order List (CMOL) Price [EUR/MWh] From algorithmic perspective, it is therefore beneficial to allow CA Volume [MW] But there is a side effect for the market: counter-activating these bids is creating an additional economic surplus, like the preceding markets (DA, ID) The perception of this side effect is very different • Some NRAs/TSOs and BSPs/BRPs believe that the side effect is very negative: not the role of TSOs, threat for ID… • Some NRAs/TSOs and BSPs/BRPs believe that the side effect is positive by itself thanks to the additional surplus created 3

Overview Can the side effect be avoided / reduced? PICASSO: • The a. FRR market design does not include indivisible bids and the optimisation problem is linear blocking CA is possible • Allowing CA would require a change of approach for the order of the optimisation algorithm compare to what has been previously decided (IN should then be before a. FRR) or to have more complex algorithm From algorithmic point of view, it is easier to block the counter activations, due to the integration with IGCC MARI: • Lessons learned from the TERRE project clearly stated that it is very difficult to avoid counter-activations. • The MARI project wanted a second opinion on that, and engaged with algorithmic specialists and asked them to analyse the problem. From algorithmic point of view, it is extremely difficult to block the counter activations, no commercial solution available, and not obvious that a custom solution will be possible 4

Summary of the analyses in PICASSO Algorithm perspective Feasibility of blocking CAs Interactions between IN and a. FRR • PICASSO TSOs have concluded that in terms of algorithm, all options are possible (allowing CAs, allowing partially CAs, blocking CAs). The difference with MARI arises from the absence of indivisible and link bids. • Having implicit netting within a. FRR (where it applies) before IN contributes to increase efficiency (see INIF: optimisation region) • But having IN after a. FRR without specific tweaking of the algorithm will lead not to activate the bids for CAs • • • BSP-to-BSP exchange and role of the TSOs Economic efficiency Risks for mark-ups Efficient use of cross-zonal capacity Transparency or understandability of the results for BSPs Importance of: repartition of surplus between TSOs, non-monotonic behaviour of price, non-intuitive flows & negative rent Legal requirements (REMIT, EBGL) Other considerations Mixed views… • Conclusion PICASSO TSOs propose to block the possibility of counter-activations in a first stage 5

Summary of the analyses in MARI Algorithm starting point « 1 -step» vs. « 2 -step» • As a starting point, MARI TSOs, comforted by an external analysis (N-Side), believe that the 1 -step algorithm is the most efficient in terms of optimisation for the balancing, and use of CZC • Based on the current design decisions in the MARI project, the use of extended features such as indivisible bids or linking between bids are necessary for technical reasons and also for TSOs to be able to operate their balancing processes • Extended features generate cases of acceptance / rejection of divisible and indivisible bids that MARI has to handle • According to the external analysis (N-Side), the handling of those cases with the reduction/blocking of counter-activations is not known in terms of algorithmic implementation, and it is unknown if a solution can be found MARI project context Foreseen use of indivisible bids Implementation concerns Risk on the go-live date due to complexity of handling CA Conclusion MARI TSOs propose to secure (technically and timely) the implementation of the m. FRR platform based on a 1 -step algorithm, acknowledging the possibility of counter-activations as a side effect in a first stage 6

The perception of the side effects are different TSOs evaluate the impact of the side effects differently In general TSOs that are more reactive (high a. FRR share) are against counter activations, where the other TSOs are more positive towards counter activations. This is all based on expected impact on the market, there is no data on the table – just expectations. The future bidding strategy by BSPs is unknown 7

Common proposal from MARI and PICASSO Starting point shall be: To fully allow counter activations in MARI & to fully block them in PICASSO (Package Deal) And then commit in the IFs to monitor the side effects 8

Benefits of the compromise An agreement today increases the probability for the final IF approval by MC in December 2018 Partly satisfies all TSO’s concerns • “Reactive concerns” mitigated in a. FRR • “Proactive concerns” mitigated in m. FRR The risk of not finding a solver for the MARI AOF is reduced The MARI solution is more similar to the TERRE solution IT can be re-used The joint implementation of PICASSO and IGCC is straightforward Similar approach than TERRE 9