Industry: how to reduce energy costs resorting to AI?

Thinking about Energy is much more than presenting statistics on energy dependency, electricity production and consumption, liquid fuel transactions or even CO2 emissions.Energy is Economy. As Professor Ferreira Dias Jr. used to say, "electrify to industrialize". If in the 50s of the last century access to electricity was the great lever for socio-economic development, today’s energy challenges are much different.

Nowadays, Europe has structural weaknesses in terms of energy and competitiveness. The EU is the largest energy importer in the world and its external energy dependence is of approximately 400 billion euros per year.

According to data provided by the Portuguese Energy Government Directorate (DGEG) electricity and natural gas consumption by industry is, respectively, around 35% and 70% of the total national electricity consumption. This energy demand is mostly derived from the needs of industry sectors such as Pulp and Paper, Petroleum Products, Mineral Products, Metallurgy, Food, Chemical, Plastics and Textile.

The importance of energy for the industrial sector is irrefutable, both from a national and business perspective. In fact, on average, energy costs amount to 50% of the industry GVA (2018 data) [1]. The relative impact of the cost of energy on the competitiveness of businesses has aggravated in recent years, a straight forward conclusion if one considers that, as of January 2020, costs of electricity[2] and natural gas[3] increased by 250% and 1000%, respectively.

In this context, how can an industrial director guarantee a seamless production process in a cost-efficient way? The traditional answer would be to resort to the technological reconversion of inefficient equipment, replacing said equipment by assets with improved energy consumption performance. However, in my opinion, this is not the adequate response to the current challenge, one that requires speed, financial discipline and a holistic vision of the energy balance of an industrial unit.

Sensing, advanced data analytics and artificial intelligence (AI) models may offer a roadmap for reducing energy costs without impacting production output and further providing a cleantech investment prioritization according to the prediction of economic savings.

One can identify many examples of AI implementation opportunities with a view to improve industrial energy efficiency. For example, Unilever implemented an AI and digital twins project in eight industrial units[4]. This project (whose implementation lasted just over a month) allowed energy savings of around three million dollars (in only one factory) and increased productivity by 3%, through the monitoring of process variables and optimization of the use of equipment.

Similar results can be found in other industries and sectors. In fact, process optimization can generate up to 25% energy savings in the Pulp & Paper industry and, more generally, digital systems and AI could lead to 15% energy efficiency in industrial units[5].

We are considering, specifically, very low investment projects (a sensor costs 30 cents and a cloud database costs less than 10 euros/month), implemented in a short period of time (less than eight weeks) which, however, have may have a very significant economic (and environmental) impact.

The current scenario - in which the cost of energy affects the competitiveness of various industries - challenges us to move beyond the traditional framework and the conventional set of solutions. All things considered, it is not difficult to anticipate that only industrial units managing to implement quick and cost-efficient approaches (with significant impact in the short term) will be able to stand out from their competitors.

[1] Banco de Portugal; [2] OMIE; [3] MIBGAS; [4] Kearney; [5] SET PLAN

(Article originally published by Jornal Económico)