{"id":969,"date":"2023-10-09T08:33:32","date_gmt":"2023-10-09T07:33:32","guid":{"rendered":"https:\/\/circeular.org\/?p=969"},"modified":"2024-04-26T13:15:24","modified_gmt":"2024-04-26T12:15:24","slug":"how-can-machine-learning-help-prioritize-building-energy-retrofits","status":"publish","type":"post","link":"https:\/\/circeular.org\/how-can-machine-learning-help-prioritize-building-energy-retrofits\/","title":{"rendered":"How can machine learning help prioritize building energy retrofits?"},"content":{"rendered":"

[et_pb_section fb_built=”1″ _builder_version=”4.20.2″ _module_preset=”default” global_colors_info=”{}”][et_pb_row _builder_version=”4.20.2″ _module_preset=”default” global_colors_info=”{}”][et_pb_column type=”4_4″ _builder_version=”4.20.2″ _module_preset=”default” global_colors_info=”{}”][et_pb_text _builder_version=”4.22.2″ _module_preset=”default” hover_enabled=”0″ global_colors_info=”{}” sticky_enabled=”0″]<\/p>\n

To reach the 1.5\u00b0C goal, the current annual renovation rates of buildings in the EU ranging from 0.4% – 1.2% must be increased to 5% – ideally prioritizing buildings with the highest energy saving potential. To identify such buildings, the year of construction is very helpful, as it is a proxy for thermal insulation, ventilation rate, or glazing ratios and is therefore a key factor in modeling the energy consumption of buildings.<\/span>\u00a0<\/span><\/p>\n

In a recent study, our partner Technische Universit\u00e4t Berlin (Sustainability Economics of Human Settlements)<\/strong><\/a> – TUB looked at 25 million buildings in France, Spain and the Netherlands and tried to estimate the construction year and retrofit need. The overarching goal was to assess if machine learning methods can facilitate the identification of retrofit candidates at scale. It has been found out that energy savings from large-scale retrofits may be improved by more than 50% when the machine learning estimates of the year of construction are used to target old, energy-inefficient buildings<\/strong>. Furthermore, the approach may help to identify regional clusters of buildings in need of refurbishment and focus policy efforts and funding allocations.<\/span>\u00a0<\/span><\/p>\n

While TUB faced many challenges due to data inconsistencies and regional differences in urban form that currently limit the generalizability of this approach and need to be addressed before rolling it out in practice, the study demonstrates the overall potential of spatially explicit methods to improve the scalability and granularity of climate solutions and provide geographically differentiated policy advice<\/strong>.<\/span>\u00a0<\/span><\/p>\n

By Florian Nachtigall – TUB<\/em><\/p>\n

[\/et_pb_text][et_pb_button button_text=”Read the publication” _builder_version=”4.22.2″ _module_preset=”default” button_url=”https:\/\/doi.org\/10.1016\/j.compenvurbsys.2023.102010″ hover_enabled=”0″ sticky_enabled=”0″ button_alignment=”center”][\/et_pb_button][et_pb_text _builder_version=”4.22.2″ _module_preset=”default” text_font_size=”11px” text_line_height=”1.3em” hover_enabled=”0″ global_colors_info=”{}” sticky_enabled=”0″]<\/p>\n

Figure<\/strong><\/em>:\u00a0Results overview of a targeted prioritization of buildings with a heating demand above 150 kWh\/m2a. (A) Energy savings per m2 of floor space for different levels of knowledge of the year of construction: ground truth knowledge (left), predicted year according to different cross-validation approaches (middle), and no knowledge (right). (B) Regional heterogeneity in the share of buildings in need of retrofitting in France.\u00a0(C) Illustration of accuracy in determining retrofit needs exemplified by the city of Valence, France.<\/p>\n

[\/et_pb_text][\/et_pb_column][\/et_pb_row][\/et_pb_section]<\/p>\n","protected":false},"excerpt":{"rendered":"

In a recent study, our partner Technische Universit\u00e4t Berlin (Sustainability Economics of Human Settlements) – TUB looked at 25 million buildings in France, Spain and the Netherlands and tried to estimate the construction year and retrofit need. The overarching goal was to assess if machine learning methods can facilitate the identification of retrofit candidates at scale.<\/p>\n","protected":false},"author":52,"featured_media":978,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_et_pb_use_builder":"on","_et_pb_old_content":"","_et_gb_content_width":"","inline_featured_image":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[16,9],"tags":[],"class_list":["post-969","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-news","category-news"],"jetpack_featured_media_url":"https:\/\/circeular.org\/wp-content\/uploads\/sites\/21\/2023\/10\/News-week40-TUB.png","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/posts\/969","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/users\/52"}],"replies":[{"embeddable":true,"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/comments?post=969"}],"version-history":[{"count":8,"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/posts\/969\/revisions"}],"predecessor-version":[{"id":983,"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/posts\/969\/revisions\/983"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/media\/978"}],"wp:attachment":[{"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/media?parent=969"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/categories?post=969"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/circeular.org\/wp-json\/wp\/v2\/tags?post=969"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}