Assessing Physical Climate Risks for the European Bank for Reconstruction and Development's Power Generation Project Investment Portfolio

Research output: Working paperResearch


  • Tianyi Luo
  • Lihuan Zhou
  • James Falzon
  • Cheng, Yan
  • Giulia Christianson
  • Yili Wu
  • Amir Habchi
This paper pilot tests the method on EBRD’s thermal power and hydropower generation portfolio (i.e., excluding solar and wind). WRI conducted physical climate risk assessments under two climate scenarios, Representative Concentration Pathways (RCPs) 4.5 and 8.5, representing potential climate impacts ranging from moderate to the most extreme. We assessed the direct impact on power plants, and measured risks in potential electricity generation losses. Many of the pilot findings are useful not only to EBRD but also to the broader investor community. The lessons learned throughout the pilot also shed light on why and how existing climate-related disclosure guidelines could be improved and what needs to be done for more widespread and faster adoption of comprehensive climate disclosure.

Key Findings
At the portfolio level, EBRD could expect thermal and hydro plants in its portfolio to experience a 7.0 terawatt-hour (TWh) (or 3.3 percent) loss in annual generation by 2030 relative to its baseline level under a moderate climate change scenario. As shown in Figure ES-2, we estimate that, for projects in EBRD’s current power portfolio, physical climate risk–driven annual average generation losses are about 0.76 TWh (or 0.4 percent of total annual generation) for the baseline period. Losses could grow to 7.0 TWh (or 3.3 percent) under RCP4.5, and to 7.4 TWh (or 3.5 percent) under RCP8.5 by 2030.

Intensified water stress levels and rising water temperature are, by far, the two biggest drivers of increased generation losses associated with physical climate hazards for the thermal assets in EBRD’s portfolio. Water stress and water temperature account for the largest increases in generation losses, followed by air temperature and drought-induced water shortages.

EBRD’s current hydro portfolio is likely to see a reduction in overall annual generation due to reduced precipitation and increased upstream water consumption, but some hydro plants in its portfolio could expect surpluses in generation. We estimate a 222 gigawatt-hour (GWh) reduction in annual average hydro generation by 2030 under RCP4.5 and a 148 GWh reduction under RCP8.5. The primary reasons for reduced power generation in the future are decreased precipitation and increased water consumption in hydro plants’ upstream watersheds. Rising air temperature also increases evapotranspiration though it can also increase glacier melt and water availability in the short-to-midterm. Some of the hydro plants could expect surpluses in generation due to increased local precipitation.

The combined cycle gas turbine (CCGT) plants in EBRD’s portfolio are substantially more climate resilient than the coal-fired power plants. As shown in Figure ES-3(a), a typical CCGT plant in EBRD’s portfolio has roughly the same percentage loss in generation as that of a typical coal-fired power plant for the baseline period. However, by 2030, coal-fired power plants are projected to see their percentage generation losses increase to 1.7 percent under RCP4.5 and 2.1 percent under RCP8.5, more than twice as high as that of CCGT plants.

We estimate that, while both recirculating and dry cooling systems will experience reduced efficiency driven by climate change, once-through cooling systems are by far the most affected today and will continue to be in the future. Cooling technologies also contribute to the resilience of thermal plants to climate change. As Figure ES-3(b) illustrates, a typical once through cooled power plant has a much higher percentage loss compared with plants cooled with recirculating or dry systems, for both the baseline and future periods.
Original languageEnglish
PublisherWorld Resources Institute
Number of pages48
Publication statusPublished - 9 Dec 2021

Number of downloads are based on statistics from Google Scholar and

No data available

ID: 286570777