Image: viboo

viboo: When the algo­rithm turns the thermostat

The Empa spin-off “viboo” has devel­oped a self-learn­ing algo­rithm to con­trol the indoor cli­mate. This allows build­ings to be cooled or heat­ed in advance, sav­ing around one-third of ener­gy. Fol­low­ing suc­cess­ful exper­i­ments in NEST, the research and inno­va­tion build­ing of Empa and Eawag, the first pilot projects are now being imple­ment­ed with indus­tri­al partners.

Con­ven­tion­al ther­mostats, which are installed in many res­i­den­tial build­ings today, only react when the tem­per­a­ture falls below or exceeds a cer­tain thresh­old. The reac­tion is there­fore always too late and only then abrupt and vio­lent, since the desired tem­per­a­ture is to be reached again as quick­ly as pos­si­ble. This costs ener­gy and ulti­mate­ly mon­ey. The solu­tion: a ther­mo­stat that looks into the future and reg­u­lates the tem­per­a­ture in advance.

Look­ing ahead to save energy

A spin-off of Empa, which is cur­rent­ly in the start-up phase, pro­vides a glimpse into the crys­tal ball. The com­pa­ny calls itself “viboo,” which stands for “viable intel­li­gent build­ing oper­a­tion opti­miza­tion.” In the course of their research at Empa, founders Felix Bün­ning and Ben­jamin Huber have devel­oped an algo­rithm that allows build­ings to be con­trolled pre­dic­tive­ly. The amaz­ing thing is that only two weeks’ worth of build­ing data, such as valve posi­tions and room tem­per­a­ture mea­sure­ments, are need­ed to cre­ate a mod­el of the build­ing. In com­bi­na­tion with pre­dic­tions of the local out­door tem­per­a­ture and glob­al solar radi­a­tion, the algo­rithm then inde­pen­dent­ly cal­cu­lates the ide­al ener­gy input to heat or cool the build­ing up to twelve hours in advance. This elim­i­nates hec­tic, reac­tive con­trols, which requires sig­nif­i­cant­ly less energy.
“The poten­tial is enor­mous. Our exper­i­ments at NEST have shown that this approach can achieve ener­gy sav­ings of between 26 and 49 per­cent,” explains Felix Bün­ning. The researchers used the Research and Inno­va­tion Build­ing on the Empa cam­pus to test and fur­ther devel­op the algo­rithm in a real-world envi­ron­ment. The field tests gen­er­at­ed inter­est from indus­try, and the researchers real­ized that their algo­rithm should be tak­en from research to market.
The moti­va­tion of the two sci­en­tists is not finan­cial, but rather social. “An enor­mous amount of ener­gy is used world­wide for heat­ing and cool­ing build­ings. This is one of the rea­sons why the build­ing sec­tor accounts for a large pro­por­tion of glob­al CO2 emis­sions. With our algo­rithm, we want to help as many house­holds as pos­si­ble to save ener­gy and thus make our con­tri­bu­tion to reduc­ing these emis­sions,” says Bün­ning. An ini­tial fund­ing ini­tia­tive has already rec­og­nized that this project makes sense. In Novem­ber 2021, “Ven­ture Kick” award­ed “viboo” a grant of 40,000 Swiss francs. In addi­tion, Bün­ning is sup­port­ed by a “BRIDGE Proof of Con­cept Fel­low­ship” from the Swiss Nation­al Sci­ence Foun­da­tion and Innosuisse.

Pilot project for smart control

“We are cur­rent­ly focus­ing on man­u­fac­tur­ers of ther­mostats for res­i­den­tial build­ings. Many of these com­pa­nies already have smart ther­mostats in their port­fo­lio. By means of a cloud con­nec­tion, we can inte­grate our algo­rithm into them,” explains Felix Bün­ning. A first part­ner is the inter­na­tion­al­ly active ther­mo­stat man­u­fac­tur­er Dan­foss. Togeth­er with the com­pa­ny, “viboo” is now imple­ment­ing the first pilot project in a con­ven­tion­al build­ing. In the process, the ther­mostats in Empa’s admin­is­tra­tion build­ing will be replaced by smart ther­mostats run­ning the “viboo” algo­rithm. Based on the indoor cli­mate data, this first cre­ates the build­ing mod­el. After that, the algo­rithm takes over the con­trol of the heat­ing for four months. The cen­tral ques­tion is: How effi­cient­ly does the algo­rithm reg­u­late com­pared to the stan­dard solu­tion? This will pro­vide infor­ma­tion on how high its poten­tial is in con­ven­tion­al, old­er buildings.
In addi­tion to the pilot project, how­ev­er, talks are already under­way with oth­er indus­try part­ners to explore fur­ther poten­tial appli­ca­tions. “viboo” will, for exam­ple, inte­grate the algo­rithm direct­ly into the build­ing automa­tion sys­tem in a new build­ing in Zurich, opti­miz­ing con­trol through­out the office building.