“Research is relatively good at asking ‘How should we design the system before something happens?’. It is less developed in answering ‘How should the system adapt in real time while the disruption is happening?’ and ‘How should it learn and evolve afterwards?’. These two gaps create some particularly exciting opportunities for future research.”
Today we are joined by NUSC expert, Dr An Dang, to talk to us about her recently published paper, ‘Resilience in closed-loop supply chain: a systematic literature review’. Co authored by fellow NUSC expert Dr Quang (James) Huy Duong and NUSC co-director Professor Li Zhou. This paper reviewed nearly 200 studies across a 20 year period to develop a framework to understanding risks in closed-loop supply chains.
Their findings revealed significant gaps in existing research. While the literature is rich in strategies for anticipating disruptions, far less attention has been given to how organisations should respond while disruptions are unfolding or how they should learn from them afterwards.
Perhaps most strikingly, the review uncovered an overwhelming preference for model-based research over data-driven approaches. “Of the 194 studies we reviewed, only one study clearly used a data-driven machine-learning approach to analyse reverse-logistics risks,” Dr Dang explains.
At a time when organisations generate unprecedented volumes of operational data, this finding points towards a major opportunity for future research and practical innovation.
Business Insights
Dr Dang urges businesses to consider resilience not as a ‘single emergency plan’, but as ‘a connected portfolio of capabilities covering what happens before, during and after a disruption’. More precisely, Dr Dang elaborates:
“A practical starting point for businesses is to map both the forward and reverse parts of the supply chain and identify the most critical facilities, partners and flows. They should then ask: what could go wrong, how quickly would we detect it, what alternatives do we have and how would we learn afterwards?
Businesses should also test whether their plans work under realistic disruption scenarios. Having a contingency plan on paper is not the same as having the data, decision rights, alternative partners and operational capacity needed to activate it.”
Understanding Resilience in Complex Supply Chains
Closed-loop supply chains are significantly more complex than traditional supply chains. Beyond moving products from suppliers to customers, they must also manage returns, recovery, recycling, remanufacturing and disposal.
As Dr Dang explains:
“A complex system is a network in which connected parts interact, adapt and create outcomes that no single part controls.”
A disruption affecting one part of the system can quickly ripple across suppliers, manufacturers, customers, recyclers, waste handlers and regulators. Understanding and managing these interconnected risks is therefore central to building resilience.
One approach frequently used by researchers is game theory, which examines how organisations make decisions when outcomes depend not only on their own actions, but also on the actions of others.
This is particularly relevant in closed-loop supply chains, where resilience often requires cooperation between independent organisations with competing priorities. During a disruption, one partner may need to absorb additional costs, carry excess inventory or accept reduced margins to support the wider network.
Game-theoretic models help researchers explore how mechanisms such as profit-sharing, cost-sharing and risk-sharing arrangements can encourage collaboration.
“A technically optimal solution may fail if the organisations involved have no incentive to participate,” says Dr Dang. “Game theory helps identify arrangements in which both individual organisations and the overall supply chain can benefit.”
Which Industries are Leading the Way?
While the review did not formally compare industries, several sectors stood out for their sophisticated approaches to resilience.
The automotive and electronics sectors appear to demonstrate some of the most developed portfolios of mitigation strategies. These industries commonly combine robust network design and situational awareness with redundancy, flexibility, security and visibility. This makes sense as their supply chains are usually international, highly interconnected and dependent on specialised components. A shortage or facility disruption can therefore affect several tiers of the network very quickly. This has encouraged the development of backup sourcing, flexible capacity, alternative collection or recovery channels and more sophisticated network-design models.
Healthcare and hazardous-waste supply chains also demonstrate important resilience practices, particularly because failure can create immediate safety and regulatory consequences. Healthcare systems combine flexibility with contamination-aware design, while hazardous-waste systems place strong emphasis on visibility, secure handling, compliance and knowledge management.
Therefore, I would say that automotive and electronics appear particularly advanced in structural and operational mitigation, while healthcare and hazardous-waste systems are especially strong where safety, traceability and regulatory control are concerned.
As Dr Dang notes, different industries prioritise different resilience capabilities depending on the consequences of failure.
From Social Listening to Supply Chain Research
Dr Dang’s interest in supply-chain resilience grew out of an unexpected place: marketing.
Before entering academia, she worked in social listening, analysing social media data to understand customer experiences and perceptions.
During the COVID-19 pandemic, however, she observed that many customer frustrations stemmed from issues beyond marketing.
“Products were unavailable, deliveries were delayed, demand changed unexpectedly and businesses struggled to respond. These were fundamentally supply-chain problems.”
The experience sparked a deeper interest in how organisations anticipate, manage and recover from disruptions.
Closed-loop supply chains proved particularly fascinating because they introduce additional uncertainty. Organisations must not only manage the forward flow of products but also contend with unpredictable product returns, varying product conditions and a much broader network of stakeholders.
Despite their increasing importance within circular economy initiatives, their resilience remains comparatively underexplored.
“That made it both an important and a relatively underdeveloped area to investigate.”
What’s next for Dr An Dang?
The direction I am most interested in pursuing is the use of new data sources and data-driven methods to support real-time and adaptive resilience. The review shows that the field remains heavily dependent on optimisation models using predefined assumptions and scenarios. These methods are useful for designing robust systems, but real supply chains continuously generate data and evolve in ways that may not fit those predefined assumptions.
I would therefore like to examine how sources such as social-media discussions, customer complaints, incident reports, recall data and sensor information can be combined to identify emerging risks and support faster decisions.
This also connects closely with my current research on product recalls. Product recalls are a clear example of a disruption involving both forward and reverse flows: firms must stop or correct the forward movement of products while simultaneously locating, collecting, repairing, replacing or disposing of products already in the market. Accordingly, my research is moving from understanding how product recalls occur and are managed towards developing methods that help organisations detect warning signals earlier and adapt while events are unfolding.
My ideal study would be a large-scale, longitudinal digital representation of several closed-loop supply chains across industries. Rather than examining individual risks independently, the study would model how disruptions interact and cascade through both forward and reverse flows.
It would combine operational data from manufacturers, suppliers, logistics providers, collectors, recyclers, retailers and regulators. It would also incorporate real-time information from sensors, transport systems, product returns, maintenance records, customer complaints, recalls and external events.
Most importantly, the study would work with practitioners to test whether the proposed interventions genuinely improve outcomes during real disruptions, rather than only improving performance within a theoretical model.
Methodology Corner: A Literature Review That Grew Alongside the Researcher
Behind every published paper lies a story about the research process itself.
For Dr Dang, one of the biggest challenges was not analysing nearly 200 studies—it was managing the information over a project that evolved across several years. The review began early in her PhD as part of a broader exploration of resilience in closed-loop supply chains. When her doctoral research later narrowed to product recalls, the review was temporarily set aside. Returning to it years later required updating the literature and incorporating a substantial body of new research. As Dr Dang highlights, “I had also developed considerably as a researcher since the beginning of my PhD.
As her expertise grew, so did her ability to identify key contributions, extract relevant insights and refine the emerging framework. The experience also reinforced the importance of rigorous data management. “When I returned to the project several years later, I had to spend weeks reconstructing the earlier process.”
Her advice for early-career researchers is simple but invaluable: maintain a detailed research log, record decisions clearly and write notes with your future self in mind. Just as resilient supply chains must adapt to changing circumstances, research projects must be designed to evolve.
“A systematic review often evolves. Categories may be revised, new papers may need to be added and reviewers may ask for additional analyses. A clear and scalable dataset makes those adjustments much easier.”
Her final piece of advice?
“Treat the review as an iterative learning process. You become more efficient at reading and coding papers over time, but you should periodically return to earlier studies to ensure they have been assessed consistently with the more developed framework.”
Cover image photo by Tiger Lily from Pexels; Generative AI (CoPilot) was to suggest improvements on earlier drafts.
Discover more from Networks and Urban Systems Centre (NUSC)
Subscribe to get the latest posts sent to your email.
