angle-left Algorithms to prevent energy wastage
INNOVATION TUNGSTENO

Algorithms to prevent energy wastage

Energy effiency has its greatest allies in technologies such as Big Data or Artificial Intelligence. Credit: Israel Palacio.

FRANCESCO RODELLA | Tungsteno

Achieving energy efficiency is one of the great challenges facing society, the cities of the future and, of course, the economic sectors. The rise of technologies such as machine learning, Big Data and Artificial Intelligence, represents the opportunity to find new allies in the fight against energy wastage and to learn how to use energy better.

The concept of energy efficiency "is an umbrella term that can cover many fronts," says Luis Hernández, a professor at the University of Valladolid. One aspect refers to the efficiency "within the network infrastructure itself," but one can also talk about energy efficiency in terms of construction or mobility, he adds.

In the first case, an example of a technological intervention is the deployment of sensors so that homes and infrastructure "work in a more efficient way,'' he explains. The same idea, he adds, can be applied to achieving "much more efficient urban transport routes than before." There is also the fact that an electric vehicle can be "highly efficient" in comparison with internal combustion, such as engines based on an otto cycle or diesel, says the researcher. Here we discover some novel applications in the field of energy.

The smart grid model

Currently, the main Spanish electricity companies are working to provide networks with "smart" elements, a transformation that seeks to optimise energy production and consumption, as well as making them more sustainable. One of the first steps has been to replace traditional electricity meters with digital meters capable of detecting consumption in real time. However, the updating of the meter pool in Spain had to be completed before December 31, 2018.

Hernandez says that normally the priority of the companies is the installation of sensors, both actuators and measuring, to automate the processes of distribution and consumption. Next there is the need to handle all the data that is generated "from this new hardware implemented in the end customers, be they producers, consumers or both [in reference to the self-consumption model]".

It is in this second phase, the professor continues, when algorithms can be especially useful. "Having elements of artificial intelligence presents the possibility of carrying out processes more automatically and perhaps faster," he says. "The data analysis tools will allow these companies to make more efficient use of the information and even operate in the network itself."

Electricity companies are implementing "smart grid" models introducing digital meters and sensors to automate the distribution and consumption of energy. Credit: SSE.

Solutions for small business

Solutions based on the use of algorithms are spreading to more and more sectors. The Institute of Knowledge Engineering of Madrid (IIC), for example, has developed a system for the Satel Iberia company that offers small and medium-sized service companies a way of knowing "specifically what they spend more or less energy on," as explained in a statement from the institute.

The formula is designed to allow shops, offices, restaurants or supermarkets to know the sub-consumption of each activity almost without installing physical meters. To do so, the algorithm is based on the monitoring of a sample of energy expenditure and uses parameters such as "surface area, coordinates and hours of establishments" for its calculations, adds the IIC note.

Big data to improve the energy efficiency of disadvantaged neighbourhoods

Big Data can also be used to detect forms of energy inefficiency such as those due to the architectural characteristics of a particular site. Last year, researchers from the Polytechnic University of Madrid and CSIC published a new model that shows how the application of this technology can help identify the points that demand more energy in vulnerable neighbourhoods, and thus understand what pieces of infrastructure need upgrading in order to improve the situation.

The system analyses the information coming from the open cadastral databases and calculates the energy losses based on different urban parameters (such as dimensions, orientation, shape of buildings and layout) and construction parameters (such as the quality of roofs, facades, walls and floors). Fernando Martín-Consuegra, principal investigator of the study, explains that his group focuses on social housing without "any kind of energy efficiency measure" and with "a very high need for rehabilitation," but it is a model that "can be applied in any neighbourhood."

Google uses automatic learning algorithms to reduce the energy consumption needed to cool their data centres. Credit: Connie Zhou/ Google.

The technology giants’ answer to energy inefficiency

The daily transmission of millions of bytes on the Internet requires the use of more and more energy. And major technology companies, which need huge amounts of power to launch and maintain their services, are among the operators most concerned about how to achieve greater efficiency and sustainability, according to Bloomberg.

Google is one of the most committed to finding solutions. This tech giant says, for example, that when building its data centres it used efficient cooling techniques and intelligent temperature and light control systems to limit energy losses. In this way, since the spring of 2014, its centres have used 50% less energy than the industry average.

The company also explains that the integration of automatic learning algorithms allowed the firm to reduce the energy consumption needed to cool their data centres by between 15% and 40%. In 2018, Google went one step further by using a machine learning model that managed to improve the efficiency of some of its wind farms in the US by "up to 20%", which "collectively generate as much electricity as is needed by a medium-sized city." The efficiency of the system, says the company, is based on the fact that its neural networks predict the production of wind power "36 hours before the actual generation."

· — —
Tungsteno is a journalism laboratory to scan the essence of innovation. Devised by Materia Publicaciones Científicas for Sacyr’s blog.

También podría gustarte

FIND US ON FACEBOOK

INNOVATION VIDEOS

CONTACT SACYR

Paseo de la Castellana, 83-85.
28046 - Madrid (Spain)
Phone: +34 91 545 50 00

www.sacyr.com