Future HVDC grids – yes please, but what is the price tag?

Co-author: Philipp Härtel

Offshore HVDC grids will be a crucial part of the future energy system – Knowing what they cost is a crucial part of planning them.

Offshore wind power needs an offshore HVDC grid

The European Green Deal and the European Commission’s dedicated Offshore Renewable Energy Strategy envision more than 300 GW of offshore wind parks in Europe, which creates the need for substantial investments in offshore electric power transmission infrastructure. With Europe’s enormous plans for the Energiewende in general and offshore wind power in the North Sea in particular, the construction of an HVDC grid in the North Sea seems unavoidable.

The ongoing large-scale deployment of offshore wind parks creates the need for substantial investments in Voltage Source Converter (VSC) based High Voltage Direct Current (HVDC) transmission infrastructure. VSC HVDC technology will play a major role in this development due to its applicability for long-distance subsea cable transmission and its capability to form isolated offshore Alternating Current (AC) wind power collection grids.

Such offshore long-distance electric power transmission infrastructure is expensive. Just to give you an idea: The cost of the electrical equipment that brings the offshore renewable energy from the wind turbines to the shore is in the same range as the cost of the wind turbines themselves. The future offshore grid will need tens of billions of Euros of investments in HVDC transmission infrastructure. For instance, Dutch-German transmission system operator TenneT has contracted 30 G€ for such connections in the North Sea. With numbers this large, it is evident that these investments need to be planned and optimised as well as possible, as any relatively small change will result in substantial economic consequences in absolute terms. Ultimately, it will impact your electricity bill.

Grid expansion planning

Grid expansion planning is vital in transforming energy systems, especially as potential trade-offs exist with other energy carrier networks, such as hydrogen networks. Grid expansion planning, based on state-of-the-art energy system models and the best available data, is performed to find the best way to build the future offshore HVDC transmission grid.

Meanwhile, grid expansion planning is not a discipline where a final answer can be found: the context changes, political plans change, technology evolves, future market predictions get updated and mathematical methods improve. Providing a scientific knowledge base for the development of the future energy system is and always will be a continuous effort. Planning and optimising the North Sea super grid has been ongoing for more than a decade already, and it will continue for a long time to come.

The importance of cost estimations

Grid expansion planning is challenging because it involves many uncertainties. What will wind turbines cost in 2040? Will there be a large hydrogen market, and at what price will hydrogen be traded? Moreover, another crucial question is the main focus here: What are the costs of building HVDC transmission infrastructure?

Investment models for transmission expansion planning studies always need cost estimates for infrastructure. Accurate CAPital EXpenditure (CAPEX) cost estimates of HVDC infrastructure are essential in grid expansion planning. They are necessary for determining the costs of different grid expansion options that must be compared with their benefits. The cost estimations have a large impact on the grid expansion optimisation outcome.

If transmission infrastructure is assumed to be cheap, it will be optimal to build a lot of it. Assuming it is costly, the optimal solution will be only to build it where necessary and invest in other technologies like battery storage systems in the other cases.

The challenge of cost estimations

So, while it is easy and intuitive to understand the relevance of the cost assumptions, it is unclear why it would be so difficult to establish what HVDC technology actually costs. There are many existing HVDC projects already, so one would think we can check their cost! Yes we can, but… we must recall that G€-scale HVDC projects are not a commodity. High-quality input data on the cost of VSC HVDC systems are challenging to obtain.

The companies able to manufacture these systems can be counted on the fingers of one hand. Each system is unique and custom-made; there is no standardised product where prices can be compared. Both the manufacturers and the customers (mostly transmission system operators) are quite silent when it comes to the cost/price of these projects. Even in cases where costs are published in a press release, they are often roughly rounded numbers without specifying what they include. Cost breakdowns are generally unavailable to the public. Prices differ a lot (e.g. factor of two) between HVDC projects that seem similar. Price levels change over time, as a market bottleneck can improve the technology suppliers’ negotiation position and quickly drive prices up.

SINTEF and Fraunhofer join forces

Back in 2016, Philipp Härtel from Fraunhofer IEE in Germany spent three months at NTNU/SINTEF as a guest researcher through the IRPwind mobility programme. During discussions with other researchers in Trondheim, a crucial knowledge gap in the field of offshore grid transmission planning was identified: The quality of the cost model input data.

Previous work has shown that cost parameter sets for estimating the cost of HVDC transmission projects are subject to large uncertainties. It was found that earlier studies often based their HVDC cost assumptions on a single source, without the possibility of knowing if that source is actually representative, or whether it is a cheap or expensive exemption.

Research on grid expansion planning is often mostly focused on improving the energy system models. However, even though valid models are used in a study, the uncertainty of the cost input data for VSC HVDC components poses a significant problem for the validity of the results. Without trustworthy cost input data, the best model will not deliver meaningful results. In order to address this challenge and knowledge gap, a collaborative research effort was established between myself, Til Kristian Vrana, from SINTEF Energi, and Philipp.

What we have done so far

A mixed-integer linear uniform cost model for estimating HVDC transmission infrastructure investment cost has been defined. The cost model captures the cost of converters and cables, and the potential additional cost for converter deployment at sea. The mixed-integer linear approach yields significant benefits for long-term, large-scale transmission expansion planning problems and the optimisation algorithms solving them, as computation time and convergence face severe challenges when more complex cost models are applied. That said, continuous linear cost models, i.e., without integer components, would be even more advantageous from a computational perspective. However, such simple models fail to reflect that HVDC systems, in reality, are almost always built with high power ratings because the per-MW-cost is much higher for low-power projects. The mixed-integer linear approach, therefore, presents a viable trade-off between accuracy and the computational burden.

The approach for parameterising the cost model was based on the idea that creating the “average” of various  poor-quality data will remove some of the disturbance. This average HVDC cost parameter set was published in a review article in 2017.

The work was then extended. Based on published cost data of real VSC HVDC projects and the average cost parameter set, all the available information was condensed into a more accurate cost estimation data set. This improved cost parameter set was developed by parameter fitting that minimises cost estimation errors of real HVDC projects without deviating too much from the average cost parameter set, using particle swarm optimisation. The parameter fitting employs a logarithmic error metric and distinguishes three project categories, Back-To-Back (B2B) systems, InTerConnector (ITC) cables, and Offshore Wind Connections (OWC).

The activity led to an optimised cost parameter set, which delivered the best possible cost estimations with regards to the available cost data. This can serve as a more solid basis for future transmission investment studies, increasing the validity of the findings. This work was published in another peer reviewed journal article in 2018, that laid a profound basis for grid expansion planning, providing much better input data than what was used before.

Ongoing work

In 2022, the activity was reactivated, mainly for one reason: Many new HVDC projects have been planned, contracted, or constructed since 2018, which can be used to feed the parameter fitting optimisation. More input data means less noise, which in turn means better cost parameter fitting. So, our plan was to do an “easy” exercise: Collect all the new data and rerun the parameter fitting to obtain an updated cost parameter set.

The reality, however, looked different, and while working on the topic, several possible improvements for the cost model and the parameter fitting methodology were identified. The two main aspects are the cost model itself and the way to consider overhead costs for real HVDC projects that are taken as references.

The cost model has been updated and it is now parameterised with nine parameters, including also underground cables for a more accurate inclusion of onshore extensions of offshore HVDC networks. The old model was submarine cable centric (with focus on offshore grids), and onshore extensions with underground cables were only included in a simplified way. The new model now includes the underground cables in the same way as submarine cables, with two additional dedicated parameters.

The parameter fitting methodology, which identifies the cost model parameters, has also been updated to consider different overhead costs, which increases the precision of processing cost data from real HVDC projects for parameter fitting, and, therefore, the accuracy of the cost parameter set. A systematic approach for considering overhead costs has been developed, which attempts to categorise published cost numbers from press releases into five defined investment cost levels. Of these five, the third level is the CAPEX cost, which is the main interest here; it is what the cost model is trying to estimate, so the parameter fitting needs to be based on the CAPEX cost of reference HVDC projects. In case other cost levels than the CAPEX cost are supplied by the press release, these cost levels are then “converted” to CAPEX by adding or subtracting overhead cost. In case the Owner’s project costs are published, the financing overhead (financing, insurance, risk premium, etc.) needs to be subtracted to identify the CAPEX of the project. In case the contracted costs for the HVDC converter and cable are published, the overhead cost for auxiliary equipment, land acquisition, project managements etc.. needs to be added to derive the CAPEX cost of the project.

This systematic consideration of the overhead cost reduces the disturbances caused by different press releases publishing cost figures at different investment cost levels. An article describing the new updated cost model and the overhead consideration approach has been accepted at the EEM Conference and will be presented there.

The way forward

And now, the final step in the process is ongoing: applying the improved parameter fitting methodology on an updated data base (with all the new projects) to fit the parameters of the improved cost model. Once this is concluded, an article will be submitted to the journal Electric Power System Research.

This new upcoming cost parameter set will be the one to use. It will be coming soon, so stay tuned!