After an exposure to energy systems i.e. working on policies and data in energy industry, I decided to confuse my life’s purpose more and go broader. I decided to apply my learning for a bigger scope – understanding the impacts of these industries to the environment. In short, I joined the research under industrial ecology.
This transition was challenging, but rewarding at many levels because it meant the perspective had to be changed – in work and methodology styles (which solutions to propose and how), in thinking styles (which questions to ask), and especially in stakeholder relations (who matters, when and why).
Why we do industrial ecology.
Basics of Industrial Ecology
This transition from energy to environmental field, meant changing the viewpoint from an industry system to a broader industry-and-environment system. My first question was how to quantify these exact industry trade-offs for environmental benefits. One of the major goal of Industrial Ecology (IE) is to calculate the exact consequences (or impacts) of every small process (right from the metal extraction, i.e. mining to the disposal of the final product used by a consumer) and material used in the industry. The different fields in industrial ecology (try to) encompass every industry functioning worldwide e.g. one of the database, called exiobase 3, includes 163 industries across 49 global regions. So yes, a lot of matrices to deal with. The environmental impacts themselves are usually divided into many categories, for example, there is greenhouse gas emissions, which is quite heavily reported, but then there are also many Eco-toxicity indicators which are useful to understand water or soil pollution.
Challenges and Methods of the field
As a data-scientist the first thing I could peculiarly note were the depressed and wandering researchers – asking “where did you get the data from?” Because that is one of the hardest part in this field: to get RELIABLE and high-resolution (spatial-temporal) data. Obviously, there are many databases available internationally (sometimes regional/national databases are more detailed though) and there are many petty battles on which one is better. But a constant effort is made in the field to gather or update this data via numerous discussions and partnered research with the industry stakeholders. It is mostly a tricky business because the industries are not very open with their data, and most of the data has to be aggregated, probabilistically estimated and/or computed from many industry process assumptions. This means the methods used in the industrial ecology are majorly complemented with statistical tools ranging from data preprocessing to advanced Monte Carlo simulations.
But at the end, which methods are core to the industrial ecology? Especially in the field of environmental engineering, the methods of Life Cycle Assessment (LCA) and Material Flow Analysis (MFA) are something, which you will hear almost on a daily basis. The name giveth their purpose. There are many other methods e.g. input-output analysis, which aims at understanding the supply chain processes and associating exact impacts of these processes. As can be clearly noted that the industrial ecology entails a lot of simulation and modeling, and thus the big data analysis methods like Machine Learning have become quite common in recent studies e.g. to use GIS data and aggregating/ classifying the data based on the land-use development over years. Having said that, as much as this field of IE has immense expertise in environmental tools and methods, there is a lack of data science experts, which makes some complicated big data tasks difficult to be solved on everyday basis. For example, the problems can be as small as connecting a database to the server or as big as merging heavily complicated GIS and neural network models’ results together and aligning all the outputs of the databases together.
Image courtesy: ecoinvent.org (life cycle assessment)
Finally, the purpose of the industrial ecology is to make industrial processes ‘ecological’ or at least more sustainable. And this includes a lot of stakeholder interaction, multiple group discussions, interviews, surveys, workshops with producers to consumers. As it is difficult to interact with all the stakeholders, there are other simulation tools and collaborative disciplines like Agent Based Modelling (simulating actions of many ‘agents’ in a community, a bottom-up approach to understand impacts caused by individual and societal actions), and social sciences/ economics, which have made industrial ecology more holistic.
Multidisciplinary aspect and the research network
This multidisciplinary aspect of the industrial ecology makes it more distributed in expertise all around the world. This is not just limited to different industries, but also spread across international schools (e.g. NTNU, Yale, Tsinghua, etc.) and many environmental fields (biodiversity, air pollution, land-use changes, etc.). Personally, I became aware of these universities and groups via multiple events: There are many conferences in this field, but the generally ‘big’ conferences are: Gordon Research Conference in Industrial Ecology and International Society of Industrial Ecology conferences. But then there are more specific ones, e.g. Socio-Economic Metabolism section of the International Society of Industrial Ecology (ISIE-SEM) or Food Life Cycle Assessment (Food LCA) conference, Society of Environmental Toxicology and Chemistry (SETAC), and so on.
IE exposes you to many other fields and industries in depth as well – I stepped into the building industry through my project, not just in the energy systems (primarily heating), but also in the material and process of the construction sector of this industry. Additionally it exposed me to better network in the building industry e.g. the recent conference I was part of was CISBAT (Climate resilient Cities Energy Efficiency & Renewables in Digital Era), which is not an industrial ecology conference per se, but it invites many researches in the climate resilient cities’ design.
Transdisciplinarity of my project (doodling during a workshop)
What is ahead
Looking it from a data-science perspective once again, as much it is my personal need to collaborate further with Computer science (CS) researchers, it is also something important for the field of industrial ecology I believe. IE needs more support from CS methods; but also it needs setup of protocols like better documentation and opening of databases. There have been definitely good progresses in that direction for better data-management protocols in IE. But all these efforts are only useful, if we bring in maximum validation of these data-models with the industry stakeholders – to provide more robust models.
Following up on stakeholders, the bigger purpose and advantage of IE is to interact with stakeholders in industry for actual policy implementation. Due to this stark communication gap in research and implementation, sometimes it becomes quite difficult to let the result become an actual decision. So the aspect of interdisciplinary research needs to be definitely encouraged, right from PhD students to the grant providers. Thus, more soft skill training and workshop to stakeholder discussion management is a must. Because only with these ‘real-world’ collaborations the real ecological industries can be achieved, and all the PhD self-doubts would be well paid-off.