**5. Conclusions**

Throughout the present work, we proved that it is possible to separate the generation technologies into two di fferent sets and to proceed to a double optimization of the generation mix. When we compare the non-pollutant-technology e fficient frontier with the e fficient frontier of the instrumental model, we are able to generate at a lower cost but at a higher risk using only non-pollutant technologies (nuclear, wind, o ffshore wind, hydro, small hydro, and PV).

When analysing the sharing weights in the non-pollutant e fficient frontier, nuclear energy, defending its position as a base-load generation technology, and small hydro participate at their maxima in both the minimum-risk GMV and the minimum-cost GMC portfolios:


Replacing the cost-risk perspective with emission-risk perspective pollutant technologies allows to highlight the important role of biomass and CCS technologies in an e fficient portfolio. Their commercial development is crucial in order to achieve low-carbon emission portfolios.

Oil generation is not included in the power-generation mix in the instrumental model, highlighting its excessive cost and risk. In the emission-risk models, it is only considered when we take out the biomass or when we set upper limits to the participation shares of the technologies. These limits cause the preferred technologies to participate at their maxima in almost every e fficient portfolio.

Solar PV generation takes part only in the e fficient portfolios close to the GMV portfolio. Its participation is needed in order to achieve a highly diversified and lower risk portfolio.

The cross-drawing approach proposed between the pollutant and non-pollutant e fficient frontiers calculated in both cost-risk and emission-risk coordinate axes leads to relevant conclusions:


Finally, drawing an analogy with the CML from CAPM, we presented the CML-A area that could helpful a policy maker design the long-term generation mix in a decarbonisation scenario.

**Author Contributions:** The Contributor Roles Taxonomy (CRediT) of this work is as follows: conceptualization, P.M.-F., F.d.-P. and I.S.; methodology: P.M.-F., F.d.-P. and A.C.-S.; software, P.M.-F.; validation, F.d.-P.; formal analysis, P.M.-F., F.d.-P. and A.C.-S.; investigation: F.d.-P.; resources, F.d.-P. and I.S.; data curation, P.M.-F. and F.d.-P.; writing – original draft preparation, P.M.-F. and F.d.-P.; writing – review and editing, P.M.-F.; visualization, P.M.-F. and F.d.-P.; supervision, A.C.-S. and I.S.; project administration, P.M.-F. and F.d.-P.; funding acquisition: A.C.-S.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.
