My research project
The use of Data Analytics for Sustainable Product Development
Sustainability is becoming a top priority in any industry. However, difficult decisions are part of the sustainability challenge due to its inherent complexity. A complexity that is manifested as divergent priorities, conflicting targets and metrics. This cause decision-makers to struggle to see the opportunity to drive more sustainable businesses aligned with profitable product and services.
The research focuses on giving business decision-makers a process and the related tools to simplify such complexity and make more informed and better decisions. The process is built to be a Sustainability Assessment to perform at the early stage of new products and services R&D to lead to its commercialisation. The focus element of the analysis is the product life-cycle considered as part of the broader socio-technical system made by the firm, the market, society, and the environment. This system has been defined as the Product-market system. The scope of the assessment is to understand both negative and positive impacts that products and services and related business decisions cause on the external system.
The Shared Value creation process and the Multi-Capitals approach have been chosen as frameworks to assess the net-positive gives potentially resulting from sustainable business decisions through the adoption of the Product-market system framework. Value is defined as the capacity to satisfy human needs, in accord with Max-Neef's definition of human needs, but the research extends this definition to include non-human needs in terms of societal and environmental needs. We argue that businesses should target more objective non-human needs alongside customers' and stakeholders' needs to active Sustainability. We called this the "society's needs-oriented" approach as a step further the so far prominent "customers' wants-oriented" approach. The aim of the research is to give business decision-makers the right tools for targeting needs in common between customers, other stakeholders, society, and the environment with new business solutions.
The project partner is Costain Group Plc.
The developed tools in short
Depending on the current development phase, a set of different tools have been developed:
- During the early stages, the Product-Market System Compassⓒ and the Sustainable Product Canvasⓒ can help to identify the business context with the Shared Value creation lens. In this phase, it is keen to involve stakeholders to record their needs and environmental and social experts that can advocate for natural assets and social needs. The Product-market system framework and related workshop can drive critical thinking to question the product to achieve a more sustainable business offer.
- After the decision cotext is clear, quantitative and qualitative data need to be collected to describe the framed system. Life-Cycle Assessment (E-LCA), Life-Cycle Costing (LCC), and Social Lyfe-Cycle Assessment (S-LCA) can be used to quantify negative impacts and costs caused by the product or service. A novel Multi-Criteria Decision Analysis (MCDA) has been developed as part of the research to map the priorities of stakeholders and academic experts and associate those with the LCA figures.
- The data gathered in the previous step are then integrated into a unified Hierarchical Bayesian Neural Network (H-BNN) to analyse the expected satisfaction of the different actors and groups. This makes the Life Cycle impacts more relevant to the specific context, allows future scenario analysis, and enables assessing similarities and synergies between clients' and stakeholders' categories, social goals, and natural assets. For more details about the analysis, see the Research page.
Current case study
Decarbonisation for Road Transport - How can your organisation's Values drive the change?
A mapping assessment of actors' needs across South East Wales in association with Costain and the University of Surrey
Hydrogen and Battery Electric vehicles have a great potential to reduce road transport emissions, but many challenges are pushing down the upscale of these technologies. Many actors have a role to play and understand their priorities is keen to accelerate the transition.
We are currently investigating stakeholders' needs regarding low-carbon emissions alternatives for road transport. If you think your organisation has a role to play in reducing the carbon emissions of road transportation, we would like to hear your view via this questionnaire. The insight will feed future strategic plans. We would like to share the results of the analysis with the participants of the study and the public.
The use of Data Analytics for Sustainable Product Development
The project partner is Costain Group Plc.
The relevance of Sustainability, Social Value, and Multi-Capitals approaches is increasing in the infrastructure sector. Multi-Capitals approaches, such as the Value Toolkit by the Construction Innovation Hub, assess the value delivered by a project with wider and more relevant criteria than the simpler economic cost-benefit analysis. They require the involvement of broader stakeholders in doing so. Every voice needs to be heard to assess potential impacts and benefits on each stakeholder category affected by the project and thus propose a better solution. Quantitative decision support methods can help increase objectivity, reduce decision biases, and understand the reason behind the suggested outcome. Furthermore, they can potentially combine stakeholder qualitative judgments with quantitative figures, such as environmental or operational performances. However, quantifying stakeholder opinions, impacts and priorities is still a challenge in practice.
The Multi-Criteria Decision Analysis (MCDA) approaches available in the literature fail to meet the robustness and the simplicity required by the industry. One of the main limits of standard MCDA analysis is the amount of data required that translates into very time-demanding surveys for experts or stakeholders to answer, which causes survey fatigue, low accuracy, and low response rate. Moreover, the decision process of available methods is often difficult to explain, hiding the decision path followed by the different responders and so the justification of the decision suggested.
Our Solution – Hierarchical Bayesian Neural Network approach for MCDA
The PhD focuses on developing a standard procedure to support the business decision-making to embed strong Sustainability in business decisions since the early stage of a project. The Sustainable Product Development (SPD) process includes a novel MCDA method based on state-of-the-art machine learning methods to solve the aforementioned issues. The research focused on giving business decision-makers a robust and easy-to-apply method to collect data insight from broader stakeholders following the "wisdom of crowd" approach. As a citizen science approach, the method is flexible to work with a small group of experts or a broader study with the public.
The main use of the developed method is to quantify groups priorities with a survey that requires an exceptionally low time to be completed. The approach is Hierarchical, which means grouping responders to cross-pollinate insight from each group up to the overall scale. For example, responders can be distinguished based on organisations, job title or even geographical relevance and pinpoint their different Value Profiles. This is a unique benefit of this approach, which enables obtaining a relevant aggregate picture without losing insight into each stakeholder's category.
The survey burden has been reduced thanks to the use of machine learning, in particular the Bayesian neural network methods. Those methods are more robust at missing data than the current MCDA in the literature and enable dividing the amount of data to collect on multiple responders, reducing the survey burden of each responder. The hierarchical approach also helps in this direction, best using the small dataset gathered from each responder. Finally, Bayesian inference enables a good fit of the parameter with Small Data thanks to the Markow Chain Monte Carlo method while measuring the uncertainty of each parameter, also a valuable source of insight for decision-makers.
A second and more in-depth potential application of the method enables the modelling of more complex systems merging stakeholders' and experts' feedbacks with qualitative assessments, such as Environmental Life-Cycle Assessment (E-LCA), Social Life-Cycle Assessment (S-LCA) and Life-Cycle Costing (LCC). This can be used to assess the impacts calculated by LCA analysis in the context that is relevant for the specific application. For example, the LCA environmental impacts can be analysed in conjunction with the Value Profiles of the relevant stakeholders to have a picture of the project more relevant in the specific context, allowing future scenario analysis, and enables assessing similarities and synergies between clients and stakeholders' categories, social goals, and natural assets. This potential application will be tested with the Decarbonisation of Road Transport case study.
Meeting industry challenges
To answer these questions, it is necessary a clear, holistic view of the firm's decisions and activities' impact pathway on the environment, society, and the market.
The challenge is that a firm's impact pathway is not simple to describe and monitor over time. It is a complex tree of cause-consequence relationships within it. The firm affects clients, stakeholders, and natural assets, and those affect each other over time. This generates a complex network of relationships. The challenge for the industry is to define clear priorities and best actions to deliver value to the three spheres without sacrificing any one of them.
There is a need in the industry for tools and procedures that are able to analyse and simplify such complex behaviour. The current decision-making processes rarely aim to systematically investigate the complex nature of the relationships existing between the firm, the market, the environment and society. With better tools, a firm can make insightful decisions in investing in specific product development and initiate a positive cascade effect on the whole external system.
Meeting academic challenges
Life Cycle Assessment is a powerful and well-known procedure for sustainability assessment, but it does not apply to many business decisions and social impacts. A Sustainability Assessment for business models needs to be able to assess the impact of intangible factors, such as, for example, brand reputation, competitor strengths, or social wellbeing are. Moreover, it needs to take into account uncertainty and problematic preferences.
For these reasons, the Hierarchical Bayesian Neural Network (H-BNN) will be proposed and tested as a novel approach for the Multi-Criteria Decision Analysis (MCDA) for complex system analysis. The firm's impact pathway will be described via an influence diagram obtained by experts' and stakeholders' judgements combined, thanks to the HBNN. The developed SPD will identify the factors to prioritise within the impact pathway, run scenario analysis, and compare alternative solutions. For estimating Sustainability, the three Capitals approach has been chosen for comparing the business models' impacts. Finally, the SPD will be developed according to the Bellagio STAMP principles to adopt a strong sustainability perspective.
As the sustainability plan is getting more and more important on both national and international levels, it is happening in the infrastructure sector as well.
Data-driven companies are already disrupting their market by offering cheaper and more efficient services to customers with higher proximity. The next step is to do the same to tackle social and environmental issues. Targeting the right customer, it is possible to engineer the business to deliver desired values as co-benefits. It is necessary to have a clear and holistic understanding of the interaction between the company and the system within its acts. Big Data analytics help to have a deeper understanding of the system's interactions, and the appropriate Data strategy might become part of each effective Sustainability strategy in the future.
There is a big opportunity in the need for tools for sustainability assessment that can deal with business decisions, uncertainty and intangible factors.
The research has the scope to develop the procedure and the tools to guide this transition. The SABM will be adaptive to the business environment, the market and the future scenarios in which the customer acts. By analysing these variables, Costain's consultancy can guide customers in their sustainability transitions with digital solutions. The SABM will be tested on projects running in Costain and then included in the consultancy offer for Sustainability.
Many existing technological solutions are ready to transform more sustainable infrastructures, but there is still the challenge to embed Sustainability into common business practices and to tackle the inertia of the business-as-usual to solve the global crisis of our time.
Summary and current findings
The enterprises have played a key role in shaping contemporary society and so will be in the future. Many successful firms have used data-driven decisions to shape their business and disrupt their market. This has raised the profits; however, nowadays, enterprises are called more sustainable with urgency. Making more sustainable business decisions in the early stage of a project would be highly valuable for the infrastructure industry. Data analytics can help in this challenge. However, there is no standard procedure to assess the business impacts on society yet.
The decisions taken internally by a firm can have long term impacts on the system within its act. Those impacts are difficult to predict due to the complex relationships between the firm, environment and society. The firms will have key roles in tackling or harming global issues, and their decision-makers will decide which direction to take. Those decision-makers need to be in the best position to make insightful decisions to run their business in more sustainable ways. The research aims to understand how to represent the complex system enterprise-environment-society and then simplify the representation for decision-makers, giving them the digital tools necessary to make better-informed decisions to reduce impacts and drive more sustainable outcomes.
Anna Bozano, Alessandra Schiaffino, Alessandra Spessa, Francesca Valeriani, Raimondo Mancinelli, Vanna Micheli, Diego Dolcetta JIMD Reports. 2020; 1– 9. https://doi.org/10.1002/jmd2.12100
Lesch‐Nyhan disease (LND; OMIM 300322), caused by virtually absent hypoxanthine‐guanine phosphoribosyltransferase activity, in its classic form is characterised by hyperuricemia, variable cognitive impairment, severe motor disorder and a characteristic behavioural disorder (Lesch‐Nyhan Behavior, LNB), typically described as self‐injurious behavior (SIB) and “self‐mutilation.” This work focuses on the latter aspect with the aim of exploring and broadening it. [...] We are proposing a wider LNB description, beyond the classical Self‐injurious behavior (SIB), stating that it is widespread and pervasive, involving every facet of the patients' life. Caregivers and operators should be aware that they might face different LNBs, and have to recognize them to find the better way to manage patients.
Ameduri, S., Ciminello, M., Dimino, I., Catignani, A., & Mancinelli, R. (2019). Archive of Mechanical Engineering.
An optimal sensor placement methodology is implemented and herein proposed for SHM model-assisted design and analysis purposes. The kernel of this approach analysis is a genetic-based algorithm providing the sensor network layout by optimizing the probability of detection (PoD) function while, in this preliminary phase, a classic strain energy approach is adopted as well established damage detection criteria. The layout of the sensor network is assessed with respect to its own capability of detection, parameterized through the PoD. A distributed fiber optic strain sensor is adopted in order to get dense information of the structural strain field. The overall methodology includes an original user-friendly graphical interface (GUI) that reduces the time-to-design costs needs. The proposed methodology is preliminarily validated for isotropic and anisotropic elements.