Questions & Answers

The Nordic CCM project publish questions (and answers to them). Questions during stakeholder meetings are published as part of the meeting notes, available under Stakeholder meeting material. The other questions, for instance by e-mail, are published here. To post new questions, please email ccm@nordic-rcc.net.

For the abbreviations used, please see the List of abbreviations.

Questions 2023-2024

Please note that the questions and answers below are arranged from newer to older.

CVXPY Clarabel Interior Point solver for the intraday capacity calculation

You recommend CVXPY Clarabel Interior Point solver for the intraday capacity calculation:

IPOPT can be a good option. if you develop the scripts in Python, you may try CVXPY Clarabel Interior Point solver. Please refer to the description here: https://www.cvxpy.org/tutorial/solvers/index.html

This is from the documentation:

It is stated that stated that it is a solver for convex optimization. Thus, the eigenvalues of the hessian should be greater or equal to 0 in case of formulated as a minimization problem, shouldn’t they? Could you please provide the eigenvalues of the hessian for one of the hours of your intraday capacity calculation?

If it is not a big effort, it would be nice get the full optimization problem with constraints and solution for an hour. I have tried set up the problem for a few hours and sometimes it looks to me like your solution slightly violates the constraints, at least the way I calculate them trying to follow the documentation. Could it be so that you have more decimals when you calculate than available for the ptdfs published? It is also a bit tricky to compare since you do not publish values for NO4<->FI and SE4<->SE_BC, due to that these also impact the possible allocations for other boarders.

Answer provided 16 August

The ATCE methodology description document is designed to explain the ATCE concept in a way that non-experts can understand, particularly in relation to its application to Nordic systems. It is not an IT specification document for developing an ATCE tool. In other words, the methodological description focuses on the concepts behind ATCE without delving into implementation details. Regarding your question about achieving consistent ATCE results, you would need the ATCE tooling specification or implementation guide, which we have not share with stakeholders due to IPR.

Regarding the solver recommendation, there are two solvers the CCM project considers suitable for solving the ATCE problem, i.e. IPOPT and Clarabel

The Norcap tool, being the industrial tool facilitating the EPR and at go-live, uses IPOPT, please see the 'short' description below of the IPOPT solver.

The Clarabel solver may use the same steps described below, as some of them are standard optimization techniques. The most distinct difference between Clarabel and this IPOPT algorithm is the former is for convex problems while the latter is for non-convex or convex problems.

Technical content:

The following summarizes the "On the Implementation of an Interior-Point Filter Line-Search Algorithm for Large-scale Nonlinear Programming" paper with additional information.

The IPOPT algorithm applied in NorCap is based on the Primal-Dual Barrier Approach. The primal-dual part is to solve the Stationarity Condition and the Complementary Slackness (KKT condition) directly by applying the Newton-Raphson Method. This ensures primal and dual feasibility: All constraints are valid, and the dual variables are larger than or equal to zero for the inequality constraints. The Barrier part is a method to solve an optimization problem of inequalities by transforming the optimization problem into a new sub problem. Hence, this is the Barrier problem. Each provided inequality constraint is converted to an equality constraint by introducing a slack variable.

The basic principle behind the Primal-Dual Barrier Approach is to solve a sequence of Barrier problems. As stated before, the Barrier problems is solved by the Newton-Raphson method, which means there are several immediate solutions during this method. Not all solutions are leading to a reduction in the objective function or may violate the constraints. Therefore, measures must be taken. Solving a Barrier problem is the inner loop. An outer loop modifies Barrier parameters in the Lagrange function (Lagrangian) before entering the Barrier problem.

Iteration Steps of the Inner Loop

1. The Barrier parameters are modified by the outer loop

2. The Primal-Dual equations are solved by the Newton-Raphson method. The equations are the derivative of the Lagrangian, the KKT condition, and some additional equations. The solution is the Newton step, which is the direction of the starting point of the next iteration in this loop

3. The Hessian matrix of the Lagrangian must be a positive semi-definite to ensure descent properties for the backtracking line search, which is applied at a later stage, . This is verified by calculating the Inertia (the number of positive, negative, and zero eigenvalues) of the system matrix A (Ax = b). The number of positive eigenvalues must equal the number of equality constraints, and the number of negative values must be equal to the number of inequality constraints. No zero eigenvalues

4. If the number deviates, the system matrix is perturbed iteratively to correct the number of eigenvalues, i.e., a small variable is added to the Hessian matrix. Furthermore, the system matrix A must be non-singular. A second perturbation of the matrix, with a second variable, is performed to ensure non-singularity

5. The Newton step does not guarantee a descent in the objective function and may violate internal inequalities in the Barrier problem. Accordingly, a backtracking line search is executed by scaling the Newton step; this does not change the direction. Remark the current solution at this stage is equal to the results from the k-1-th iteration. The following two condition reduces the Newton step at each of the k-th iterative steps if

a. the direction derivative at the current solution in the direction of the Newton step is positive and the objective function is not decreasing

b. the direction derivative at the current solution in the direction of the Newton step is negative and the Barrier objective function is not decreasing

6. The scaling stops when

a. the derivative at the current solution in the direction of the Newton step is positive, and the objective function is decreasing

b. the derivative at the current solution in the direction of the Newton step is negative, and the objective function of the Barrier problem is decreasing

7. In the case the previous step does not lead to an improvement, a Second-Order Corrections measure is applied. This corrects the direction and length of the Newton step

8. The Newton step is accepted if the previous steps are successful. Hence, the current k-th solution is updated and used as the starting point in the next Newton-Raphson iteration

9. The algorithm might fail due to infeasibility. That is, some constraints might be violated at the k-th iteration. The algorithm reverts to the Feasibility Restoration Phase. The algorithm initiates an additional optimization problem to find a new starting point for the next iteration that minimizes the constraints at most and is close to the current solution where it failed

10. When the Barrier problem returns, the Primal-Dual equations are evaluated. If they are under a threshold, the optimum is found: The First-order Necessary Condition (the gradient at the optimum is zero) and Second-order Sufficient Condition (The Hessian matrix of the Lagrangian is positive semi-definite) are fulfilled, and the solution is guaranteed to be a minimum. The outer loop continues with the sequence of Barrier problems if the evaluation is larger than a threshold

Regarding your request for a Hessian matrix, the short answer is no, we cannot provide it. The reasons are as follows:

1. From the IPOPT description above, you will notice that IPOPT internally perturbs the Hessian matrix if it is not positive semi-definite, as described in steps 3 and 4 above.

2. From the Norcap tooling perspective, the NorCap tool does not have the capability to export any Hessian matrices, whether before or after the perturbation.

3. Even if we were able to provide a Hessian matrix, we cannot be certain if the Hessian has been perturbed by IPOPT. This uncertainty would render any comparison of Hessian matrices or their eigenvalues inconclusive.

ID ATCE model and lower risk of overloads

We recognize that the ID capacity allocation model applied by Nordic TSOs today in theory can induce considerable risk of overloads in special situations based on that those capacities, in simplified terms, often show the residual remaining of the in SDAC allocated capacity (today NTC based) after utilization in SDAC. However, that potential risk of extreme utilization of CZC in SIDC appears to have occurred in practice fairly infrequently, and then usually fundamentally driven, thus supporting also power system operation. In any event it appears clear that Nordic TSOs have been able to very well manage those potential risks very well for many years, and therefore we struggle to see the justification why the proposed ID ATCE model have to reduce those potential risks in such a drastic manner. Can the Nordic TSOs together and individually explain the basis and justification for what appears to be very low risk taking on their part (compared to today) in the proposed new ID ATCE model?

Answer provided July

1. It is correct that the historical NTC most likely have led to overloads that have been managed in real-time.

2. Historically, however, there has been no way to quantify the overloads due to too much ID capacity because there was no CGM. Now is the first time that potential overloads from too much ID capacity on dependent borders can be quantified.

3. If we understand Nordpool's question / proposal correctly, you propose to allow for some overloads, given that they result from potentially unlikely situations and with the aim to increase ID capacities. This is something that is being discussed in the CCM project. We are looking into methodologies to (1) assess the likelihood of the overloads caused by ID ATCs and (2) tune the ATC extraction to consider this likelihood. We want to stress, however, that it is not an easy task and requires careful analysis and tuning since this would otherwise result in likely overloads that would need to be managed with a high probability in real-time, at a cost.

4. The fact that the historical overloads caused by too much DA / ID NTC capacities seem to have been managed is not, in itself, a good argument to allow them in the future (now that they can be quantified). It is not clear: (1) what the magnitude of these overloads have been, (2) what risk they actually put on the power system and (3) what costs were incurred because of actions taken to manage them.

NTC final for DK1A-SE3A and DK1A-SE3

Already allocated flow for the DK1-NO2 connection

Sum flow SE3-NO1/DK1

Flows in DK1-NL in week in the period 3/6-16/6

Solver for non-linear objective function

Detailed GSK information

I propose some more information about the GSK. Basically, this is the information I found about GSKs:

So strategies 0, 5 and 6 are used in Nordics:

• Strategy 0 can to my understanding be anything. If so, it is total secret to the market participants. What is the reason this secrecy? Would it possible that you at least give some hints how this is applied?

• Strategy 5 is used in NO5. Thus, load is assumed to be non-price sensitive. On the generation side, NO5 has almost only hydro. This strategy says that generators participate relative to their current generation. This leads 2 questions:

oWhat does “current” mean? Does it mean the point in time at D-2 when calculations starts? Or maybe the same MTU D-2? If so, should I be able to go the ENTSO-E transparency platform and divide the generation of a plant with total generation of NO5 and get the GSK for that plant? If so, which point in time is then the one to use? If it is in some other way, please explain.

o Yesterday (May16), according ENTSO-E transparency platform, Hydro Run-of-river and poundage fluctuated between 526-584 MWh/h, while Hydro Water Reservoir fluctuated between 1398-3297 MWh/h. Still, strategy 5 is applied for all generators according to the table above. Can this really be the case? If it not the case, could describe how it done instead and update the documentation?

• For rest of Norway strategy 6 is applied. Some questions about this:

o The other NO1-NO4 has as well wind power and some smaller thermal units. According to the table there is no exception for wind power from strategy 6. If so, GSKs for wind are calculated in the same way as for run-of-river hydro and reservoir hydro. Is this the case? If not, please explain and update documentation.

o The load in all buses get a factor as their part of the area load at “current” point in time. Is this to be understood as that all load is assumed to be price sensitive? If not, please explain and update documentation.

• Strategy 0 and 6 are used in combination in Finland and Sweden. Some questions about this:

o Strategy 0 is used for nuclear and wind. Does this mean that the TSOs make price forecasts and in case they think prices are higher than the offer price of wind and nuclear, GSKs will be 0 and in case the TSOs think price might be in range of the offer prices of wind and nuclear, the will get GSKs higher than 0? If it is in some other way please explain.

o Strategy 6 is used for hydro and thermal. For those plants, share of market participation might vary quite a bit between days depending demand, wind and to some extent solar situation. For example during last winter, we had situations where D-2 was very cold and with little wind and D was mild and windy. If “current” means a point in time D-2, I would expect the ratio of market participation of some generators to be very different at the current point in time D-2 and D. But, if GSK are calculated in a different way, please explain.

o Same question for loads as for NO1-NO4. Is all load in Finland and Sweden assumed to be price sensitive? With exception of the non-conform load in Finland (which to my understanding is only 2-4% of the total load).

Answer provided 27 June

So strategies 0, 5 and 6 are used in Nordics:

o What does “current” mean? Does it mean the point in time at D-2 when calculations starts? Or maybe the same MTU D-2? If so, should I be able to go the ENTSO-E transparency platform and divide the generation of a plant with total generation of NO5 and get the GSK for that plant? If so, which point in time is then the one to use? If it is in some other way, please explain.

Feedback from Statnett: When it says that the generators/load participate with their "current" production or load it means that they contribute relative to what they are producing in our CGM basecase, which is a forecast. This forecast creates a flow on each CNEC, which are the values on CNEC loading that you can see under "Fref" in the domain. We use forecasts on load and production for all units for each hour to create the 24 CGMs for day ahead calculation. We don’t publish generation per node or bidding area from our CGM data.

Feedback from Statnett: We don’t use strategy 5 for any of the areas at this moment, but I can give a general answer since we use strategy 6 for NO2-NO5 which also scales on production. If we interpret your question correctly you are wondering if we also use the Run of river for scaling, even though it does not fluctuate much. The answer is that we don’t. For Run of River (and wind) we "lock" the production to our prognosis for that power. And the rest of the regulation is done for load and the reservoir hydro. We also use our own forecast of run of river in our CGM creation in the day ahead process, so our values will not be the same as what you can see on the ENTSO-E transparency platform.

• For rest of Norway strategy 6 is applied. Some questions about this:

Feedback from Statnett: We apply strategy 6 as a general strategy, but Statnett has exempted wind power and run of river from participating in the scaling. For Run of River and wind Statnett "lock" the production to our forecast for that power. And the rest of the regulation is done for load and the reservoir hydro. Please note that this approach might not be the same for all Nordic TSOs.

Feedback from Statnett: The GSKs are to represent the change in CNEC loading when the day ahead outcome does not correspond to our forecast for that bidding area. We therefor also scale load but have exempted the non-conform load. We have cases where our CGM base case for bidding areas in Norway generate net positions where all load is covered by import and run of river/wind power. In these cases there is nothing to scale if we would use a production based GSK only.

• Strategy 0 and 6 are used in combination in Finland and Sweden. Some questions about this:

Svk forecast all production types in the D-2 CGM:s. The GSK:s are then used to calculate the ptdf:s to represent a move in net position from the CGM basecase. First all wind and nuclear generators are given a GSK-value of zero (meaning that they will not be scaled in the ptdf calculations i.e. we trust that the forecast for these generators are good). All other production and load units are then scaled according to strategy 6 (proportional to their prod/load in the CGM basecase). GSK:s are not altered based on the difference in NP:s between D and D-2. The D-2 CGM:s are created with D-2 forecast to create the best possible forcast grid model for D-2, the static GSK strategies are then used to shift the NP:s around the CGM bascase NP:s.

Fingrid feedback is similar to Svk. GSK 0 is used for nuclear and wind(and solar), with wind being based of forecast on D-2. All other production and consumption is then scaled to shift NP to match CGM basecase. We are unable to comment on price sensitivity of what/which/how loads have price dependency and how they are modelled.

ENDK feedback: Energinet uses strategy 0, which is applied on all thermal powerplants and offshore windfarms. This strategy is chosen as ENDK finds this strategy as the most suitable for the Danish Power System.

Statnett general feedback: In Norway we use strategy 6 for NO2-NO5, and strategy 7 for NO1. We have changed the strategies used throughout the parallel run, and will continue to do this when we see that it would be better to change a strategy. The GSKs are used to estimate the change in loading on CNECs when the day ahead outcome differs from our forecast. This difference can arise both when our forecasts for production/load on each node are wrong or our market forecast is wrong. We chose the GSK strategy that will most accurately represent the change in CNEC loading.

More capacity allocation on HVDC in flow-based compared to NTC?

Data sets for the graphical reports

External TSO limitations

Differences between the duration curves

Increased and/or Decreased trading space quantified in numbers per BZs

Improve graphs

Publication of GSK values

Representation of UMM restriction in flow-based simulation

What kind of limitations are captured on external borders?

Aligning DA and ID process

Transparency on the parameters used

Differences between SE3-SE4, SE3_AC-SE4_AC and SE3-SWL-SE3_SWL

ATCE results on Norwegian borders

How will the new ATCE proposal change trading in the intraday timeframe compared to today’s situation?

Transparency of key parameters

Remedial actions

Losses on HVDC connections

It would be great to clarify how losses on HVDCs are accounted for.

This is from the documentation:

- Complementary constraint: The HVDC capacity at the sending end of an HVDC cable (i.e. the extracted DA NTC at the sending end of an HVDC cable) should be equal to the capacity at the receiving end of the cable (i.e. the extracted DA NTC at the receiving end of the cable) plus the losses.

So let say that as AAF at sending end (A) has a flow of 1680 MW and the loss factor is 2.9%. Thus, receiving end (B) gets flow of 1631 MW. So the condition is:

Capacity(A to B , receiving) = Capacity (A to B , sending) - Flow(A to B , receiving) * loss_factor

In the other direction the sending AAF is -1631 MW and receiving -1680 MW. Here condition needs to be (at the starting point)

Capacity(B to A , receiving) = Capacity (B to A , sending) + Flow(A to B , receiving) * loss_factor

However, during the optimization Capacity(B to A) might get positive and If I am not mistaken, the sign before the loss-factor in the last equation needs to flip from positive to negative.

First of all, did I misunderstand? If not, how do you handle it?

Answer provided 27 June

Currently, the only internal Nordic HVDC modelled with losses is Skagerrak. The losses are modelled by means of an implicit loss factor (ILF) of 2.9%

This means that the capacity at receiving end will be 2.9% smaller than capacity at sending end, or put in an equation:

NTC(receiving_end) = (1-ILF)* NTC(sending_end)

This linear equation provides a correct model of the losses, when NTCs are positive.

However, EPR has shown us that negative NTCs can occur. And in these cases, the model is wrong.

Negative NTCs could occur in the case when flow on Skagerrak is forced in specific direction by the Day-ahead market to relieve internal AC cnecs that would otherwise be overloaded. This is a particular situation that has been observed from time to time during EPR.

The impact of this error is that in case of negative NTCs, the loss component is added rather than subtracted, leading to other constraints appearing as more constraining than they should be. This can lead to convergence issues, for which the TSOs have implemented a manual workaround of further relaxing the excessively limiting constraints.

Nordic TSOs and NRCC are working towards updating this constraint to be evaluated for ATC instead of NTC. ATC is by definition is a non-negative value in ATC extraction, and the above model would then always be correct. This update is planned to be introduced in operation early 2025. Meanwhile, with the manual workaround in place, the impact of the error is deemed acceptable for go-live in October.

Old ATCE data publication

Fluctuating ID ATCE

ATCE impact on IDAs

Balancing needs

Total trading space and directional space curves

Missing ATC duration curves week 13

Skagerak, Kontiskan and Baltic Cable are missing in the ZIP files

Naming convention in the historical rerun file is different than the naming convention used in the ATCE zip files?

The naming convention in the historical rerun file is completely different than the naming convention used in the ATCE zip files.

• Most of them can be mapped into each other. However, in the rerun file there is several connections that are usually not included in the zip files. For example: Skagerak (DK1<>NO2), Kontiskan (SE3<>DK1), and also all of the “NO2A” connections, SE3_ACDC-SE4_ACDC.. just to mention a few.

• At first I thought you would start using this new naming convention, but then when I checked the W13 files, which are with the new methodology, they have once again the “old” naming convention.

• Could you confirm which naming convention you are planning to use? It creates extra effort when analysis tools must be adapted when there are inconsistencies in the data.

• Another thing that confused me was the presence of the NO2A connections in the rerun files, as I understood it these would not be included in FB.

Answer provided

Indeed, we have a bit of a mix-up of the naming conventions in the published results. I would like to apologize for the inconvenience that this is causing.

We are working to resolve this and will update the ATCE results for week 13 shortly. This is in order to adhere to the naming convention used on transparency platform that we also used in the historical re-run results from week 26 of 2023 to week 12 of 2024.

In the week 13 results Skagerrak and KontiSkan capacitites are labelled NO2-DK1A and SE3-DK1A. But this will change to NO2-DK1 and SE3-DK1 after the update.

BalticCable is not a market border in the intraday market and will not be included in the ATCE results. The same applies for the border NO4-FI.

With the updated topology you will find capacities for the two virtual borders; NO2-NO2A and SE3-SE3A. These virtual areas will continue to be part of the intraday market topology after flow-based go-live in day-ahead market. On the other hand, NO2A and SE3A are not included in the flow-based market topology, where virtual bidding-zones are strictly used for representing end-points of HVDC connections or interconnectors to non-Nordic bidding zones.

Why is not SE4-DE a part of the optimization?

Objective function of SE3-FI

For optimisation variables for SE3-SE4, what is the included objective function?

Areas without PTDF

Negative flow FI-DK2

The CNEC is defined from north to south. Yes, that is correct according the bidding zone information: But, please explain to me how flow injected in FI going to DK2 can be negative, thus going from NO3 to to NO4. I have hard find that to make sense. I as well I find it hard to make sense that flow injected in NO3 to DK2 to go from NO4 to NO3. Could you please explain?

Answer provided

These questions all are about the same issue, so we will answer them all together. Since these CNECs contain a Swedish component, we can't say too much about them to respect Svk's anonymization request. However, we can say that three of these CNECs are CNECs with an outage in NO4, but with components between NO4 and NO3 as well as a component in the Swedish grid. These CNECs are especially difficult to explain, since they are PTCs, and contain several lines in one CNEC. However, here is our best explanation:

In the case of increased NP in Finland, we see that most of the flow from Finland will move south in the strong Swedish grid. However, a small amount of the flow will go through Sweden to Norway, and then further south in Norway before going back to the slack. Thus, because of the direction the CNEC is defined in (which is from Norway to Sweden), the increased NP in Finland gives a slight relieving effect on the CNEC. In the case of increase in NP in NO1235, again, most of the power will flow south to NO1 and then to Sweden and further to the slack. However, a small flow will go up north, to Sweden and then back towards the slack. Again, because of the direction definition of the components in the CNEC (PTC), this gives a slight loading effect on the CNEC (defined in the direction NO->SE).

As mentioned, these PTCs are not easy to explain, because the components that constitutes the CNEC may be defined in different directions (some going south in the grid, and some going east). But the overall effect on the CNEC is either loading (positive PTDF – as with NO3) or relieving (negative PTDF – as with FI).

Remedial action in basecase

Now back to question about remedial action in the basecase. The word “redispatch” indicates to me that there is a basecase dispatch and that dispatch is changed. However, in basecase we have the basecase dispatch. This contradictory to me, since we cannot at the same time have the basecase dispatch and redispatch from the basecase dispatch. So, should I then interpret this as a pre-dispatch, basically the TSO has ordered a specific generator to produce regardless of price outcome? Right/wrong? If there is a pre-dispatch of some generators in basecase, are the GSKs used for calculation this CNEC the same as for all other CNECs in the basecase? I am not expert on UMMs, but I have seen UMMs for example where Energinet has ordered thermal generators to run. If this is the same thing, should I expect to find UMMs in case of pre-dispatch of generators?

Answer provided

The previous answers regarding calculation for contingency might already help to understand this in more detail. But there is no dispatch or pre dispatch. The generators that are running in the basecase (or CGM) are the ones that are running in the TSO’s IGMs which have been created as per the CGMA scenarios. Effects of remedial actions are then provided as specific MW values by TSOs. See above answer regarding current status on counter trade. Answers below are provided around an unimplemented theoretical framework of how countertrade could work. The remedial action (RA) values would still provided per CNEC and would not reflect the planned volume of countertrading or redispatch, but the additional capacity that can be provided on the CNEC. This does not necessarily mean that the countertrade will be applied in practice either, only that the TSO is willing to allow additional flow on a CNEC. If the need for the remedial action does not materialize or if other more efficient remedial actions than the considered countertrade are available after the market results are received, then those will be used instead. For example if a TSO considers countertrade in day ahead, but using non costly remedial actions like HVDC-loops to achieve the same affect, the countertrade will not materialize. In general, market messages for considered counter trade in day ahead are not published, but the RA values are as part of the FB parameters. If, as in the example of ordered thermal generation, TSOs must activate remedial actions subject to requirements of publishing market information, those will be communicated. However, in many cases the decision on which remedial actions to activate will be taken as close to real time as possible in terms of operational security, and will not be communicated via market messages.

Calculation for contingency with remedial action

Calculation for contingency with no remedial action when contigency is a generator

Calculation for contingency with no remedial action when contigency is a line or transformer

Calculation of PTDFs in basecase

Calculation for basecase. In order to calculate the PTDFs for the basecase we need the network configuration without contingencies, the impedances of the network elements and the GSKs, right/wrong? To get the GSKs we need to make an assumption about the dispatch of the generators, right/wrong? Let’s call this dispatch assumption basecase dispatch. With this information, PTDFs for the basecase is calculated for all network elements, but only a fraction is published. Right/wrong?

Answer provided

Correct. Each TSO sends their individual grid model (IGM), without contingencies but reflecting maintenance etc. The flows and netpositions of the IGM are prealigned between the TSOs through a common grid model alignment process (CGMA). This determines the net position each bidding zone should have, and the flows to neighboring areas. Each TSO creates IGMs and assign production and load in their IGMs to be in line with CGMA. IGMs are then merged into a common grid model (CGM) by the Nordic RCC.

To get the GSKs we need to make an assumption about the dispatch of the generators, right/wrong? There are predefined GSK strategies for each bidding zone that describe how different generators in the merged CGM are to be scaled up or down per area in order to calculate PTDFs. From these strategies GSK-factors are determined. In general GSK strategies are defined per production type, but can be applied per general. For example, in “Area-A” it can be defined that the net position should be increased/decreased by increasing the production of only hydro and maintain wind production as per prognosis in the IGM. This is a general strategy maintained for all hours. Different strategies can be defined for different bidding zones. The current strategies are explained on a general level in the JAO handbook.

The generators are Let’s call this dispatch assumption basecase dispatch. With this information, PTDFs for the basecase is calculated for all network elements, but only a fraction is published. Right/wrong?

Nodal PTDFs are calculated for all CNECs defined by TSOs (not all lines in the model). These are translated into zonal PTDFs using the GSK factors. All CNECs are published together with their zonal PTDFs but only those deemed as significant and non-redundant are used in the market optimization.

Imax clarification

In case of unplanned outages, there are remedial actions for some contingencies, is this action also accounted for in PTDFs?

Fmax and voltage

In the domain this voltage is presented per CNEC. As far as I know, the volatge is per bus and a network elemtent connects two buses. So, what does the voltage value mean is this case? Is it the voltage at the substation_from bus or substation_to bus or maybe the average of these two numbers or something else?

Furthermore, could you please elaborate on the how this voltage number is used in the fmax calculation? To my understand fmax is either limited by thermal limits or due to the voltage drop.

Above it is stated that Imax is “Maximum operational current corrsponding to the implicit Fmax value.” What is meant by “implicit fmax value”?

I see that for some network elements Imax seem to be a fixed number, while for others Imax is different from hour to hour. Has this something to do with that fixed Imax value indicates that fmax is limited by the thermal limit, while non-fixed Imax indicates that is the voltage drop at the network element that limits fmax? This is a wild guess from my side, but to me it seems plauseble, since losses that heats the wire are proportional to the currents and above some current level the wire gets too hot. While the lowest accpeted voltage might also be fixed number, it could depend on the operation sitaution how close we are to that limit and thereby the maximum current will not have a fixed value.

Next question is about the reliability margin. According to the table below it is stated at SvK uses a 5% margin for AC-lines.

However, looking at the data it seems like SvK uses 5% for some line and 10% some lines. One example (of many).

Could you please explain this?

Answer provided

• In the domain this voltage is presented per CNEC. As far as I know, the voltage is per bus and a network element connects two buses. So, what does the voltage value mean is this case? Is it the voltage at the substation_from bus or substation_to bus or maybe the average of these two numbers or something else?

It is the average of the two, but also has to be at least minimum 95% of nominal voltage of the element.

• Furthermore, could you please elaborate on the how this voltage number is used in the fmax calculation? To my understand fmax is either limited by thermal limits or due to the voltage drop.

If the limit is due to voltage drop this is based on TSOs own analysis of Fmax, if it is based on the thermal limit then Fmax is calculated based on current limit and voltage.

• Above it is stated that Imax is “Maximum operational current corresponding to the implicit Fmax value.” What is meant by “implicit fmax value”?

Fmax is implicit as in it is not explicitly given as input from the TSOs but calculated.

• I see that for some network elements Imax seem to be a fixed number, while for others Imax is different from hour to hour. Has this something to do with that fixed Imax value indicates that fmax is limited by the thermal limit, while non-fixed Imax indicates that is the voltage drop at the network element that limits fmax? This is a wild guess from my side, but to me it seems plausible, since losses that heats the wire are proportional to the currents and above some current level the wire gets too hot. While the lowest accepted voltage might also be fixed number, it could depend on the operation situation how close we are to that limit and thereby the maximum current will not have a fixed value.

Indeed, some values are static and for some the TSOs apply dynamic line ratings where some limits vary according to weather forecast.

• Next question is about the reliability margin. According to the table below it is stated at SvK uses a 5% margin for AC-lines. However, looking at the data it seems like SvK uses 5% for some line and 10% some lines. One example (of many). Could you please explain this?

Indeed, you are correct. SvK recently changed the RM value to 10% for certain CNECs, and we are currently in the process of updating the handbook accordingly. The updated JAO handbook will be uploaded shortly!

Clarification on flows in FBMC domain

Could you please help me to understand some numbers in the FBMC domain?

It is one of the parallel lines between HVE and ISH in case that BJS-HVE is unavailable.

According the public data at Energinet from 2020 the following network data apply. Could of course have changed after 2020.

I limit this discussion to the 400 kV lines, even though small part of the flow will go at lower voltages. The flow at lower voltage grids does not play a major part here. At least according to my calculations.

The CNE on the red row is out. The PTDF above is for one of the CNEs at orange rows. I guess the lower one, since it has lower reactive impedance and will take more than 50% of the flow. CNEs on yellow rows will have flow in case flow feeds in at either Kontek or Stora Belt to HVE. When I sum the reactive impedances from the CNEs if power feeds infrom Kontek and Stora Belt, I get (considering that there are parallel lines ISH-HVE):

In the first case with Kontek, it looks plausible that most of the flow will go on ISH->HVE, since this impdeance is much lower. Then this flow will be diveded between the ISH->HVE lines and around 55% of the flow might go on one of the lines.

However, for the case with Stora Belt, the impedance is much higher on the branch including ISH->HVE, compared to the direct branch HKS->HVE. Given this, it is hard for me to understand how over 50% can flow on one of the parallel lines ISH->HVE. Could you please explain to me what I have missunderstood?

Answer provided

The Ptdf shown below looks correct it is to be expected that the value for Kontek and Storebælt is almost equal with the physical grid and expected switching status.

The CNEC below represent:

• monitored element, the secund 400 kV connections between HVE and ISH

• contingency, 400 kV connection between BJS and HVE

The Ptdf value is calculated based on the operational grid model with the expected switching status and production/consumption scenario, the operational grid model is used for all time horizon described in CACM and SOGL (capacity calculation, lookahead security calculation and real-time security calculation).

The operational grid model is not public and Energinet have no intension to publicate this grid model.

The provided model was originally developed for educational purposes, such as universities. The model is not maintained, and the included network topology may not necessarily reflect the current status of the transmission grid. This is evident, for example, in the inclusion of a 400 kV connection between nodes HKS and HVE, where such a connection should only be considered a possible future 400 kV connection.

Given the observed topology differences between the provided grid model and the operational grid model used, corresponding differences in the calculated PTDFs should be expected.

Method used on results for week 1-6

Information regarding buses, lines, transformers and network configuration for other countries?

Will the implementation of Nordic FBMC in the region support stronger exports or imports to mainland Europe in the simulations you have run and what will the impact be on the Nordic region?

Why are prices changing in the region?

Missing connections in the ATCE files

Limiting flow on CNECs

Missing min and max net positions for 22/2-2024 at JAO

Inaccurate substation names

RAM values and limits published in JAO

Question regarding the RAM values and limits published in JAO. According to the JAO handbook the RAM is the limit for the flow through a CNEC. There is a definition how the RAM restricts the flow-based domain: zPTDF*NP <= RAM, where NP = net position, and zPTDF = zonal PTDF.

In the JAO data only positive values are given for the RAM and all the examples in the documents related to calculating RAM are done with positive reference flows. Added to the definition above, it seems that the CNECs are only limited to the positive flow direction or that zPTDF*NP is always greater than zero. I don’t think that this is true since there are research papers such as [1], which state: “Positive and negative RAMs are determined. Positive RAMs consider the reference flow on a line in the standard flow direction, while negative RAMs consider the opposite flow.” I do understand that CNECs are essentially AC connections and therefore the RAM essentially defines the limit for both the negative and positive flow.

These definitions therefore seem kind of contradictory and confusing. Is the actual definition for the flow limits in CNECs: -RAM <= zPTDF*NP <= RAM (or with absolute values |zPTDF*NP| <= RAM) or is there something else that I am somehow missing here? If the definition is as I suggested, I think this could be made clearer in the documents related to FBMC, either by explaining how the flow directions are taken into account in the RAM or by just changing the mathematical definition.

[1] Wyrwoll, L.; Blank, A.; Müller, C.; Puffer, R. Determination of Preloading of Transmission Lines for Flow-Based Market Coupling. In Proceedings of the 16th International Conference on the European Energy Market (EEM), Ljubljana, Slovenia, 18–20 September 2019.

Answer provided

Indeed the observations by the requester are correct. The CNECs are defined only in positive direction. This means that if there is a need to limit market coupling flows in both directions of a network element, the TSOs will define two CNECs – one for each direction.

Running JAO auctions for the interconnector capacities between Nordic countries

Aggregated SEW results

Substation column typo

Capacity publication under outage situations

Will CZC be given in 15 min resolution in 12/2024 to MFRR EAM?

Do you compute RAM differently among TSOs? How do you set FRM?

How are DC connections considered in FB (Advanced Hybrid Coupling)?

How are the Generation Shift Keys that you use defined and what process has been used to determine the preferred structure/parameters?

Answer provided 19 February

The Nordics have chosen to implement AHC (some Nordic HVDC-connections are not part of the common European DA auction and they are handled slightly differently).

Methodology and concepts for the Nordic Flow Based Market Coupling (nordic-rsc.net):

"Advanced Hybrid Coupling" The term "hybrid coupling" refers to the integration of the two capacity calculation methodologies, the CNTC and the FB approach.

Power flows on HVDC interconnections are by nature fully manageable, and a radial AC transmission grid has no meshed structure for the power to fan out. Thus, in a pure HVDC network, or in a radial AC transmission grid, both the CNTC and FB perception of the power flows corresponds fully to the real physics of the power system. However, in a meshed AC network, the FB (or nodal) approach is the only one of the two which is able to manage real physical power flows.

In the Nordic countries, all interconnections to adjacent synchronous areas are either HVDC or radial interconnections. These parts of the Nordic transmission grid area by definition a physical embodiment of CNTC, and it doesn't make sense to implement an FB approach on these parts of the transmission grid (an FB approach would anyhow behave as a CNTC approach). With this realization in mind, the Nordic CCM have to apply a hybrid coupling to integrate the HVDC and radial AC interconnections in the meshed AC grid.

The "hybrid coupling" might be either the standard hybrid coupling (SHC) or the advanced hybrid coupling (AHC). Before entering into the explanation of SHC and AHC, it is important to bear in mind that when the power flows from an HVDC or a radial AC interconnection enters the meshed AC transmission grid, the power flow will fan out in the AC transmission grid and use the scarce transmission capacity like all other power flows in the transmission grid.

The distinction between SHC and AHC is the difference in how power flows coming from a radial AC or HVDC interconnection are managed by the market coupling in the meshed AC transmission grid. On a high level, the SHC is granting priority access in the meshed AC transmission grid for power flows coming from a radial AC or a HVDC interconnection, while in the AHC, these power flows are subjected to competition for transmission capacity with all other power flows in the transmission system.

In the rest of this chapter, the term HVDC interconnection means both radial AC and HVDC interconnections. Both SHC and AHC are based on CGMs. In SHC, an expected flow on the HVDC interconnection is at first calculated for the base case net positions. In order to guarantee the estimated power flow on HVDC interconnection, the resulting power flows in the meshed AC grid must be granted priority access on the relevant grid limitations. This can be done by applying the nodal PTDF matrix on all limiting CNEs from the "access point node" of the relevant HVDC interconnection to calculate the amount of MWs the estimated HVDC flow puts on all CNEs in the meshed AC power system. The calculated amount of MW for each CNE is removed from the relevant RAMs to make room for the estimated flow from the HVDC interconnection. The adjusted RAMs are provided for allocation to the market coupling for all other power flows.

If the realized HVDC power flow falls below the estimated power flow, the SHC process might thus leave "unused" transmission capacity on CNEs, even with excess demand for that transmission capacity by other power flows. The SHC is by the same mechanism neither able to optimize the distribution of transmission capacity between different HVDC interconnections or between HVDC interconnections and other potential efficient power flows in the system. Thus, the SHC is clearly not able to ensure optimal use of transmission infrastructure.

In the AHC, the nodal PTDFs from the "access point node" is provided directly to the market coupling for allocation, and the RAMs for the affected CNEs in the AC transmission grid are left intact without reductions caused by the HVDC power flows. The "access point node" is established as a "virtual bidding zone" in the market coupling. This "virtual bidding zone", which is a bidding zone without any orders from market participants, is "only seen" by the market coupling during capacity allocation, in the sense it will obtain a unique price in the market equilibrium, while the actual power traded on the HVDC, will receiving the market price of in the surrounding bidding zone. In the AHC, each HVDC interconnection is provided with its own virtual bidding zone with unique PTDFs.

With the AHC, the power flows from the HVDC interconnections become a part of the FB approach within a CCR, and are thus treated as all other power flows in competing for transmission capacity. Transmission capacity in the meshed AC grid will be assigned for the power flows from each individual HVDC interconnection due to price differences and impact on CNEs in the AC transmission grid based on the competitiveness of the power flows coming from the individual HVDC interconnection.

By utilizing the AHC, there is no priority for HVDC power flows on any interconnection, and by utilizing the market coupling, the allocation of power flows between different HVDC interconnections will be optimized, as will the allocation of power flows between HVDC interconnections and all other power flows in the power system. This leaves no unused transmission capacity with excess demand. The AHC is thus a more flexible approach than the SHC in managing power flows on/from HVDC interconnections in the meshed AC transmission grid, and also the welfare economic more efficient congestion management approach.

How are the GSKs defined?

Process to define intraday ATCs

Processes to define Critical Network Elements and Contingencies

Does prices converge more with FB than today?

Are price spikes being accentuated?

What is the impact on the system price?

Can you provide more insights to price forcasting?

What are the seasonal impacts of FB market results?

Why is Scheduled exchange NO3-SE2 always negative?

Why is the PDF of DK1_DE zero?

What is the "Jutland constraint"?

AAC for IDA

Are there any documents explaining how to calculate the net positions?

What happens for the ID timeframe, if the FB in DAM results in non-intuitive flows for an HVDC interconnector?

We previously thought that due to the implementation of advanced hybrid coupling neighbouring interconnectors can also experience non-intuitive flows and this has also been published accordingly in your reports where interconnectors like Baltic Cable were shown to have some hours where non-intuitive flows occurred. After studying the minutes of the last stakeholder meeting on 26 October, this is not so clear anymore.

The minutes under section 4 read : “The flows on those HVDC borders can change as a result of Nordic FB (but it doesn't mean that there will be non-intuitive flow on those borders as a result of Nordic Fb I believe). I imagine there may be: a virtual area is modelled as a Nordic bidding zone, and from here an "intuitive" flow will be scheduled to the connected Baltic region. But if the price of the virtual area is not use, but instead the price of the "real" Nordic bidding zone is used, the non-intuitive properties that inherently exist under FB may also propagate on the Baltic HVDCs Thanks for the update on the modelling! Has anyone from the TSOs looked into if counter-intuitive flow are present/possible on borders to BZ´s adjacent?
CCM: we will get back to you”

Could you shed some light on this topic and clarify the intended behaviour from go-live? - Can you help me understand what happens for the ID timeframe, if the FB in DAM results in non-intuitive flows for an HVDC interconnector? Will you pull an ID stop in order to avoid trading in the opposite but “bilaterally right” direction? To be very concrete, say you have an interconnector which has an NTC of 600MW and gets allocated a non-intuitive flow of 600MW in the day-ahead auction. What would be the ATC in this case for the opening of intraday in the direction that is opposite to the non-intuitive flow?

Answer provided 29 January

DA: Flow-based can lead to a non-intuitive flow on the borders between a Nordic bidding area and the neighboring area outside the Nordic. This has been observed in EPR; however, most cases with non-intuitive flows on the Hansa/Baltic borders occurs due to ramping or other allocation constraints which is also present today. It’s essential to note that EPR is not a forecast of the future, but rather a comparison of two capacity calculation methodologies – FB and NTC. The CCM project does not anticipate an increase in non-intuitive flow increases compared to EPR after go-live.

Non-intuitive flow will result in a negative congestion income, but ACERs Decision No 16-2023 addresses the issue on congestion income distribution between CCRs.

ID: the ID ATC is computed by the ATCE method that is currently being further enhanced at the TSOs. The direct outcome of the ATCE method is the extracted NTC, denoted as 'NTC_initial'. The 'NTC_initial' is subject for the TSO validation, considering the operational aspects, resulting to the final extracted NTC, denoted as 'NTC_final'. The 'ID ATC' is computed by 'NTC_final' - 'DA AAC'. During the EPR, the 'NTC_initial' is equal to 'NTC_final' due to the 2-week of the grace period. After CCM go-live, the TSOs will further assess the operational needs to adjust the NTC_final where needed, e.g. new outage occurred between the DA market outcome and the ID gate opening and the TSOs can assess the impact during the ID validation window.

Handling of the Viking DC interconnector

New model for ID capacites and recalculated ID capacites for the whole EPR period

Questions on Operational learning points

Questions on CCM EPR market report for week 34

How will the change to FB effect prices in the different zones?

Why does EPR capacity differ from production ATCs?

Questions regarding 14 September

The spreads out of Euphemia for 14 September were quite striking and I was expecting to check the impact of the parallel run. I was expecting especially for SE3 and SE4 significant changes in NEX and (at least visible) deltas in prices… but in fact we can hardly observe any change there.

Checking with the (excellent) tool Netto Position Nordic Flowbased Analysis - Dashboards - Grafana (boerman.dev) one would say that there would have been space to change the NEX of SE3/SE4. So in fact, to go deeper here in the details of the welfare, I would need the bid-ask per bidding zone, which I don’t have access to.

Can you shed some lights on this day in your webinar next week or – if not possible- to publish a short explanation on hot wo interpret this one?
Can you consider changing the way that the bid-ask are published in the Nordics? In Italy there are two way of visualizing the whole merit orders:

- a detailed offer list per powerplant which allows to recognize the behavior of every single unit (I know…. This is a no go for the rest of Europe.)
- The aggregated bid-ask for coupled bidding zone> which means that if all bidding zones in one hour have the same prices, then the bid-ask describe the over all merit order of the country BUT if Sicily+ Calabria decouple in one hour, then GME publishes the bid-ask describing 1) the coupled Sici+Cala area 2) the rest of Ita without those two

Maybe you guys can think of something like this? It would mean a lot in term of transparency.

Answer provided in October

For this period there was several outages in the Swedish grid that effects the capacities and flow. These can be found on the NUCS platform.
FB provides a better SEW result then NTC for this hour. We see a minor increase in the price in SE4, but a significant decrease in both FI and SE3.
The NP in SE3 is only changed with 70 MWh and this results in a price reduction of 30 EUR/MWh. This correspond to our assumption that SE3 is an area with very steep bidding curves. However there is a significant change in the flow in and out of SE3 where FB allows a higher import from SE2 and a higher export mainly to FI and SE3.

All prices, NP and buy/sell volumes on MTU and BZ level are available for both NTC and FB on the RCC website.

Regarding the publication of bidding curves on bidding zone level, that is not something that the TSOs can provide as it is data owned by the NEMOs.

Water values and modelling of North Sea Link

Can you fix the mismatch between CNEC names in JAO and the EPR GCM file?

Data calculations not aligning when using the publication data tool

I have created flow-based calculators for the Nordics but am unable to get my calculations within an acceptable margin of error using the publication tool data. I use a linear algebra solver to solve the final domain for maximizing zones and then cross-reference this against the published max NP. This results in major differences mainly for DK1, which suggests I am missing some form of constraint for the Danish zones. I already identified missing equality constraints for the virtual hubs and overall balance and added those (resulting in quite a decrease in the difference) but cant think of any more. I did notice that you publish a MaxBflow but according to the handbook that is something that is calculated, not an extra constraint. Allocation constraints on the external virtual hubs seem to be included in the final domain itself. Could one of you perhaps help me with understanding what goes wrong and why I see differences between the calculated values using the final domain and the published max boundaries?

Answer provided 4 October 2023

Yes, the MaxBflow published is a calculated value, not a constraint. The reason for the discrepancies is most likely due to the currently published results being a simplified calculation. In the handbook at JAO section 6.5 the calculation of the min/max NP is described as ‘The current version of the calculation is only a simple approach that for each bidding zone takes the minimum and maximum net position among the set of net positions which have been found to yield a minimum or maximum flow on any CNEC according to the calculation described in section 6.4.4.’

https://test-publicationtool.jao.eu/PublicationHandbook/Nordic_PublicationTool_Handbook_v0.2.pdf

This simplified calculation that does not directly work on the objective of maximizing the net position of a bidding zone. But rather it takes the largest and smallest NPs from the decision variables used in calculating max and min flows on CNECs. The simplified method is not guaranteed to return the global extremes because the same maximum flow could be realized with different configurations of bidding zone net positions. We have no explanation why the difference is more outspoken for DK1.

We have an update coming live in December where the maximum and minimum NP calculation should follow the logic given below:

NP_max(n) <- maximize(NP(n))
NP_min(n) <- minimize(NP(n))

Subject to the same following constraints: Constraints

Coupled pairs of virtual bidding zones include:

(DK1_SK, NO2_SK)
(DK1_KS, SE3_KS)
(DK1_SB, DK2_SB)
(SE3_SWL, SE4_SWL)
(FI_FS, SE3_FS)

Bidding zones belonging to the “Jutland” area include all zones with name beginning with “DK1” the rest are placed in Nordics.

So, the currently published min/max net positions does not serve as a particular good reference for your test. But hopefully you can recognize the above logic in your implementation.

Maximum and minimum NP and allocation constraints relating to the Jutland area

Ramping restrictions at Fenno-Skan when FB is implemented

Effects of flow-based market coupling

How is the occurrence of non-intuitive flows counted in Nordic RCC’s report?

Regarding EPR topology

I see that in EPR topology, we no longer have DK1A/NO1A/NO2A.

Are the cumulative capacity limits for NO1-NO3 + NO1-NO5 not needed anymore (or somehow taken care of by PTDF constraints)?
Similarly for DK1-SE3 + DK1-NO2. For the latter set of lines, currently we also have a cumulative ramping limit. How is it handled in NFB configuration?
Similarly for SE3-NO1 + SE3-DK1
Similarly for DE-NO2 + NL-NO2. I see in EPR data a lineset is introduced containing DE-NO2 and NL-NO2 for which cumulative ramping limits are given. These two lines are defined as NL-NO2 (Biddingarea_From = NL, Biddingarea_To = NO2), and NO2-DE (Biddingarea_From = NO2, Biddingarea_To = DE) in database, which means that the up direction of one is towards NO2 and the other is from NO2. However, in table Lineset_members, in the Direction column we have “1” for both lines, while I expect “-1” for one of the lines. Can you help me understand why?

Answer provided 4 October 2023

Please see the answer for each sub-question below.

The cumulative capacity limits (line sets) will not be needed in the day ahead market after transitioning to flow-based market coupling. Indeed, the additional degrees of freedom given by these line sets are now redundant with the even greater degrees of freedom provided by the ptdf constraints
Ramping limits cannot be replaced by flow-based parameters and will still be relevant after transitioning to flow-based. This is to ensure that the changes in import/export of the Nordic synchronous area can be managed with the available balancing power. Group ramping limits will still be used after flow-based go-live.
Same as “b”
The questions suggest that the asker has access to market simulation input data, which is not accessible to the general public. Questions from the NEMOs regarding errors in the topology configurations in the flow-based market simulations is better directed to the simulation and analysis working group for a more swift resolution. However, in this case the SIWG had it confirmed that there indeed was an error in the group ramping configurations. We thank you for the notice which led to the error being corrected for future simulations. The impact of the error on previous is estimated to be minor.

Go-live date?

How is the time-unit set in the MarketResults file for NetPositions set compared to the JAO website?

Will there be ILFs also for HVDC interconnectors with one terminal outside of Nordic CCR, i.e., for single virtual zones?

How will revisions for ILF values be announced?

Modelling of virtual zone pairs

How will linesets be treated within flow-based methodology?

Is there a reason why the present model has been favored over a model with all existing flows between regions?

While looking at JAO's data for the Nordic flow-based model, I couldn't help noticing that the only variables that are considered are net positions of all the Nordic areas and flow variables between Nordic areas and virtual bidding zones. It appears that flows between areas that are not connected by one of the lines that give rise to a virtual bidding zone are discarded. As a consequence, no CNE constraint depend, say, on any flow between Norway and Sweden, because none of these lines are associated to a virtual bidding zone. In particular, no CNE anywhere in the Nordics depends directly on any flow from and to NO1. For instance, there is a CNEC with name "Border_CNEC_SE2_NO3", but the associated PTDF coefficients show an intricate combination of many flows and net positions instead of a more natural constant vector with all its coefficients except one equal to 0. Is there a reason why the present model has been favored over a model with all existing flows between regions?

Answer provided 4 October 2023

We are not sure we understand the question so we explain the difference between a ZoneToZone-PTDF and a ZoneToSlack-PTDF in the hope that it will help. Those are two different things and can be used for different purpose. Both are available on JAO, so you don’t need to calculate them yourself.

What is a ZoneToZone-PTDF and a ZoneToSlack-PTDF?

The PTDF_BZ (ZoneToSlack) describe how the Border CNEC SE2-NO3 are affected by an increased export in BZ. E.g If the export is increased in FI with 100 MW, then it will release the flow on the Border between SE2-> NO3 with 3,286 MW. In practice does this mean that there will be a flow 3,286MW from NO3->SE2.

The PTDF_BZfrom_BZto (ZoneToZone) describe how Border CNEC SE2-NO3 are affected by a flow from BZfrom to BZto. This means an increase in export in BZfrom and a decrease of the same amount in BZto. This is calculated by PTDF_BZfrom minus PTDF_BZto = PTDF_BZfrom_BZto

Just because we increase the export in SE2 and increase the import in NO3 is there no guaranty that all the flow will be on the border SE2-NO3.

Th PTDF-value z2z_SE2-NO3 show that with export of 100MW in SE2 and import of 100 MW in NO3 will 54 MW be transformed on the border SE2-NO3. The rest 45 MW will be transported on other paths. Eg. SE2->SE1->NO4->NO3, SE2->SE3->NO1->NO3 or another way.

The ZoneToZone value is 1 if all the flow will run on that CNE.

Why is there a delay in the data publication?

There are a lot 0 for the RAM for June 8-9 at JAO

Will the error regarding RAM for June 8-9 impact the KPIs in any way?

One question about the ID ACTE for NO2 > NO5.

The number of non-intuitive flows for SE3>SE4 seems to be wrong in the weekly appendix reports, as well as in Figure 22 of the consultation report.

Does aFRR CM process timing create a constraint for publishing the FB domain?

You have published 21 questions on you webpage from the stakeholder meeting in Oslo.

To get straight to the point, as 17/5 is a holiday in Norway and 18-19/5 is practically a holiday in the other Nordic countries, is it possible to extend the deadline to 22/5? Or at least to 21/5?

Can you publish GSK, the actual GSK numbers and GSK strategy at least?

What is the impact of the wrong wind forecast in the CGM and how are they mitigated?

When will the TSOs publish the LT data publication, e.g. LT PTDF? Who is in charge?

Did you perform the SEW impact of backup and/or fallback FB domain?

Is the current data publication, e.g. format and location, the same as after the CCM go-live?

Arbitrage possibilities in ID

Can you elaborate the modelling of Kriegers Flak in the FB?

The Finnish P0 and P1 cuts do not have basecase, but only are associated with contingencies? Why is that?

How will the UMM look like in the FB?

How should the SHs interpret the Norwegian CNECs? E.g. how to understand the naming convention of the Norwegian CNECs?

There is a wish to have more comments/data on long-term effects in the report.

How much of your process will change with 15 minutes?

How will the network tariff change with FB? (not same as congestion income)?

How will Flow-based affect the LT capacities?

How will the communication regarding the NUCS messages be published when we go over to FB?

Is it schedule flow or physical flow in the report when counting the non-intuitive flow?

Is the cut 2 change of the series compensation for both NTC and FB? This is not clear in the report.

Can figures be provided to explain SHs’ comments and are there word limits in SHs’ responses?

What is the FB process timing and can there be discrepancy with NTC, e.g. if forced outage occurs at 9 CET in D-1?

How do you compute congestion income distribution? (updated 12 May)

What is the daily process of FB?

If an document/result file needs to be updated, please make sure to publish the date of the update next to the updated information.

Can you elaborate the modelling of the North Sea Link in FB?

How are the dynamic constraints computed in the FB?

For NO4, increasing 1 MW in net position on NO4 does not add up 1 MW when summing up the border flows of NO4.

Can we expect more data on NO3/4 and SE1/2 due to them both being somewhat erroneous during the 3-month period?

What is the status on the system price calculation?

We need grid models for the future to do accurate forecasts.

What wind prognosis is used in the FB CC?

Are the SHs’ comments and TSOs’ replies to consultation public?

Does aFRR CM process timing create a constraint for publishing the FB domain?

Who makes the FB go-live decision and based on what? Can NTC continue?

The effects of this winter's market situation on the results? Are the results of 3 months EPR representative to evaluate FB methodology?

Price forecasting issues when big changes compared to NTC. SHs need more consistency and understanding to support them in price forecasting.

How to manage reservations for dimensioning fault in flow-based?

Weeks after EPR 3 month report: Strange result for NO4 – disclaimer mention a bug for those weeks. Could you explain how to use these results and if the simulations would be re-run?

What about the arbitrage possibilities in intraday results?

Will you adjust simulation results for changed water values?

We saw some high price differences in SE1-SE2 which then the result was fixed after some weeks. Will you rerun those EDDs prior to the fix?

Were you surprised by the consumer loss in south Sweden?

What are the optimization variables in addition to the capacities Cn and Cn,o?

How do capacities Cn and Cn,o relate to the constraints? Are they included in the RAM term of cross-border CNECs?

Remedial Actions (RA), how are they included in the capacity calculation process?

Domain Validation – what is it?

What would happen if you are forced to use intuitive FB?

Will there be overloads in FB?

Can you have negative capacity under FB?

In regard to GSK strategies; what are the choices to be made by the TSOs?

Is the GSK system prone to gaming?

Will the operational expertise and experience be part of the FB capacity calculation process?

Would it not be preferable to countertrade or redispatch and have cross-border constraints only?

How does the ACER recommendation on internal constraints affect the Nordic FB methodology?

Are stability/dynamic issues simulated in offline grid models?

What will happen with unforeseen outages of e.g. power plants?

Will the PTDF matrix be stable for all 24 hours?

To what extent are the exchanges with Germany and the Baltics influenced by a Nordic FB?

There are currently no phase shifters in the Nordics. If such a device is installed, does it fit in the FB methodology?

Under FB, will the capacity on the DC cables deviate from the nominal capacity (e.g. due to the west-coast cut)?

Are losses in the AC grid taken into account in the FB model?

Regarding DC links - how will ramping constraints be handled in FB?

Can you confirm that the FB domain will always be larger than the NTC domain?

How are you able to perform market simulations today?

What setup has been used in the market simulations?

If you change the maximum price at the PXs will the outcome of the optimization change?

What is the optimization that will be used in the FB allocation mechanism?

How will the system price be calculated in FB?

Will the auction structure on the DA be changed?

FB is the better capacity calculation methodology close to real time (ID, balancing). Why start with DA instead of ID?

Is there a risk to end up in a corner of the FB domain after the DA stage, and that the ID trade is stuck?

If you use non-intuitive FB in the DA market, how will this work out on the ID market?

Is there less capacity for the ID after FB is introduced?

Is it true that the optimal solution (in terms of welfare), referred to, holds only for the spot market?

How to split capacity between DA and ID?

Shouldn’t there be the same capacity calculation methodology (CCM) for DA and ID?

The SH information tool shows the border flows based on the Nordic NPs. If you plug in the DA NPs that are determined by the PXs, can you assess the expected ID capacity?

FB ID: when will this be applied?

How often will you recompute the ID capacity?

Are you able to do an ATC extraction from the FB domain (for information purposes)?

Will the shadow prices of the internal CNECs and internal combined dynamic constraints be published?

What is the shadow price?

During the external daily parallel operation, will the FB capacities and simulated market results be published on a daily basis?

Is input expected from the generators to build the D-2 CGM?

Does the FB model in the in the JAO publication tool include the voltage and dynamic limits?

In the data to be published, can you identify from the hourly results if a constraint is internal or cross-border?

What are bidding zones DK1A and NO1A?

Will the PTDFs published in the Nordics contain the names or unique IDs (from day to day) for CNECs or combined dynamic constraints?

Will the common grid model be available?

What about information about transmission outages that are published in UMMs today?

Is there a possibility to have 10-14 days forecast for the PTDFs published?

With regard to transparency – will you publish the same information as on the continent?

What does the LT CCM cover?

MA forecasts are needed, what will be provided to the market participants?

Are you using D-2 information for the DA capacity calculation?

The D-2 CGM as basis for DA capacities - when will the latest updates be applied (i.e. a new wind prognosis)?

What will happen if the parallel runs do not deliver?

Are there no common Nordic grid models used at this moment in time in the operational NTC capacity calculation?

Will there still be countertrade / redispatch under FB?

What is the reference for the work that you are doing and the comparisons that you are making?

Does the CACM refer to a capacity calculation region (CCR) or the Nordic market?

Questions & Answers

Questions 2023-2024

Questions answered in 2024

Answer provided 16 August

Answer provided July

Answer provided July

Answer provided July

Answer provided 8 July

Answer provided 8 July

Answer provided 28 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 18 June

Answer provided 13 June

Answer provided 6 June

Answer provided 6 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 27 June

Answer provided 6 June

Answer provided 6 June

Answer provided 6 June

Answer provided 6 June

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided

Answer provided 20 March

Answer provided 20 March

Answer provided 19 February

Answer provided 19 February

Answer provided 19 February

Answer provided 19 February

Answer provided 19 February

Answer provided 2 February

Answer provided 2 February

Answer provided 2 February

Answer provided 2 February

Answer provided 2 February

Answer provided 29 January

Answer provided 29 January

Answer provided

Answer provided 29 January

Answer provided 29 January

Answer provided 11 January

Answer provided 11 January

General questions answered in 2023

Answer provided 22 November

Answer provided 22 November 2023

1. About the NO2-DK1 situation

2. Regarding FI-SE3