Image: Empa/ETH

Local pow­er grids made to measure

In our high­ly dig­i­tal­ized world, it is easy to for­get that elec­tric­i­ty is not only respon­si­ble for the glow­ing dis­plays of our numer­ous gad­gets, but also makes healthy, clean liv­ing spaces or even access to edu­ca­tion pos­si­ble in the first place in many parts of the world. Many devel­op­ing coun­tries are stuck in a vicious cir­cle of pover­ty with their low elec­tri­fi­ca­tion rates. With­out light­ing in the home, there is a lack of oppor­tu­ni­ties for val­ue-added off-farm work. Chil­dren can no longer do their home­work or learn to read in the evening. In addi­tion, there are health prob­lems often caused by smok­ing fire­places in the house or sooty kerosene lamps.

Access to clean ener­gy is gen­er­al­ly seen as a step­ping stone to gen­er­at­ing a high­er income and thus escap­ing pover­ty. This is why it has been iden­ti­fied as one of the 17 UN Sus­tain­able Devel­op­ment Goals. Fol­low­ing on from this goal, Cristi­na Dominguez, a doc­tor­al stu­dent at ETH Zurich’s Insti­tute of Build­ing Physics and Empa’s Urban Ener­gy Sys­tems Lab, is devel­op­ing a com­put­er mod­el that will pro­vide project devel­op­ers in rur­al areas with esti­mates of house­hold elec­tric­i­ty require­ments. This should enable accu­rate and thus sus­tain­able plan­ning of the pow­er grid.

One mod­el for many regions

This is because elec­tri­fi­ca­tion projects in devel­op­ing coun­tries often fail because hard­ly any reli­able data is avail­able to deter­mine the needs of the often wide­ly scat­tered house­holds. Data col­lec­tion in par­tic­u­lar is a major cost item that makes project devel­op­ers hes­i­tant to invest there. If a pow­er grid is then planned too large, for exam­ple, this is passed on to the elec­tric­i­ty prices, mak­ing elec­tric­i­ty unaf­ford­able for the poor pop­u­la­tion. Ulti­mate­ly, elec­tric­i­ty grids need to be tai­lored to ben­e­fit peo­ple in the long term, on the one hand, and to offer devel­op­ers an attrac­tive and real­is­tic invest­ment oppor­tu­ni­ty, on the other.

For the data col­lec­tion, Dominguez chose an area in sub-Saha­ran Africa, the region with the low­est elec­tri­fi­ca­tion rate in the world: “In addi­tion to the polit­i­cal prob­lems, the areas here are extreme­ly sparse­ly pop­u­lat­ed and the small set­tle­ments are very wide­ly scat­tered. This makes elec­tri­fi­ca­tion much more dif­fi­cult — and of course more expen­sive,” says Dominguez. As part of her doc­tor­al the­sis, she deter­mined the ener­gy use and demand of around 250 house­holds in east­ern Kenya. In order to even­tu­al­ly make their mod­el applic­a­ble world­wide, they are sup­port­ing research insti­tutes in Guatemala and Pak­istan to pro­vide her with equiv­a­lent data sets from these countries.

Think­ing along devel­op­men­tal thrust

Dominguez’s field­work in Kenya col­lect­ed data from house­holds with­out access to elec­tric­i­ty and those that had been con­nect­ed to an elec­tric­i­ty grid with­in the last six years. It was not only con­cerned with record­ing exist­ing ener­gy sources and their demand, but also the change in use fol­low­ing the com­ple­tion of elec­tri­fi­ca­tion. In addi­tion, the Empa researcher used diaries in which the res­i­dents record­ed the activ­i­ties they car­ried out through­out the day in order to gain a bet­ter under­stand­ing of peo­ple’s dai­ly lives and needs, and to antic­i­pate changes that would set in after elec­tri­fi­ca­tion and would then be reflect­ed in the demand for elec­tric­i­ty. In Kenya, for exam­ple, kerosene is an impor­tant source of ener­gy to light the dark mud huts. In order to get the kerosene, often longer march­es must be cov­ered to the deal­er. Time that could per­haps be invest­ed in val­ue-adding work at home in the future — if a pow­er source were available.

And once the pow­er sup­ply is avail­able, peo­ple begin to adjust their behav­iour accord­ing­ly; they acquire elec­tri­cal appli­ances such as tele­vi­sions, and pow­er con­sump­tion increas­es accord­ing­ly. But how long can the pow­er grid con­tin­ue to func­tion if demand con­tin­ues to rise? These are exact­ly the dynam­ics Dominguez wants to bring into her mod­el: “In our on-site sur­veys, we asked peo­ple what appli­ances they would buy after the first year or sec­ond year with elec­tric­i­ty. We then matched this with house­holds that had already gone through this process.” Through this, Dominguez want­ed to find out how peo­ple would use ener­gy when it was avail­able to them. Dominguez knows from her research that engi­neers are often unable to assess this cor­rect­ly: “There are major prej­u­dices here, which often result in pow­er sup­ply sys­tems being designed too large.”

Tai­lor-made pow­er grids

To make accu­rate pre­dic­tions and iden­ti­fy con­sump­tion dynam­ics, Dominguez applies “machine learn­ing” algo­rithms and “data min­ing” tech­niques. To build the mod­els, the researcher com­bines glob­al datasets from orga­ni­za­tions such as the World Bank with data from project devel­op­ment com­pa­nies so that she can include addi­tion­al con­sump­tion pat­terns such as sea­son­al vari­a­tions. These are then val­i­dat­ed for the three focal regions using field data from Kenya, Pak­istan and Guatemala. Mini-grid com­pa­nies have also pro­vid­ed their elec­tric­i­ty con­sump­tion data in return for the oppor­tu­ni­ty to test their mod­el against local conditions.

Cristi­na Dominguez’s approach high­lights the prob­lems faced by devel­op­ing coun­tries with hard­ly any infra­struc­ture: Although tech­ni­cal pos­si­bil­i­ties for elec­tri­fi­ca­tion are avail­able and have also become cheap­er with solar tech­nol­o­gy, invest­ments must be made care­ful­ly in a weak eco­nom­ic envi­ron­ment. Oth­er­wise, there is a risk of over-indebt­ed­ness on the part of the elec­tric­i­ty users and, in the worst case, on the part of the oper­at­ing com­pa­nies — all in all, a risk of exac­er­bat­ing pover­ty and a dis­in­cen­tive for oth­ers to invest in these areas. In Dominguez’s com­put­er mod­el lies the poten­tial to over­come at least one hur­dle to elec­tri­fi­ca­tion and thus pro­vide the impe­tus for a path out of poverty.