Recycle, resell, reboot: Researchers promote laptop remanufacturing

As laptops and notebook computers become outdated as operating systems and software advance, the hardware needed to support their myriad features becomes more demanding. As such, there is a growing problem of electronic waste (e-waste) from this particular sector as consumers replace effectively redundant devices with new ones. Millions of devices are discarded annually, so finding effective strategies to manage their environmental and social impact has become a global priority.

Research in the International Journal of Operational Research has focused on remanufacturing, the process of disassembling used devices, restoring or replacing, and even upgrading components, and rebuilding them to near-new condition. This approach not only reduces waste and conserves raw materials but also offers wider benefits such as creating jobs and making technology more accessible to lower-income users, given that a refurbished laptop may well be a lot cheaper than the latest model, but should in many cases be perfectly capable of running those advanced operating systems and software.

One of the big challenges in remanufacturing is deciding what to do with individual laptop components at end-of-life. These components vary widely in quality, and their potential for reuse is far from straightforward. High-quality parts are typically worth the cost of repair, but lower-quality ones present more difficult economic decisions. Yet, discarding them outright leads to the loss of materials that could otherwise be salvaged.

There is a lot of uncertainty in the various processes. So, the team has developed a sophisticated decision-making framework using a multi-period nonlinear integer programming model. This is a mathematical model designed to determine, over several time periods, the most cost-effective and resource-efficient use for each component, whether through reuse, conditional repair, or disposal.

The approach uses advanced approximation techniques, metaheuristics, essentially a sophisticated form of trial-and-error to reduce the computing resources needed for the assessment. Metaheuristics can test, virtually, possible solutions and identify high-quality outcomes without having to examine every possibility in full detail.

Two such trial-and-error algorithms were auditioned for the job: Discrete Particle Swarm Optimization (DPSO) and Genetic Algorithm (GA). DPSO is inspired by the behavior of birds, fish, or insect flocks, shoals, and swarms. GA is based on natural selection and allows improved solutions to evolve over successive generations. Both were integrated into a decision support tool built in Microsoft Excel using Visual Basic. This choice should enable usability for business practitioners without expertise in complex mathematics.

The team found that on smaller problems, the algorithms produced answers that were close to the mathematically proven optimal ones. For larger-scale cases, GA proved more reliable, while DPSO occasionally settled on less effective outcomes. An important finding from the tests was that repair costs are a major factor in the profitability of remanufactured laptops. This highlights the need for systems that can respond flexibly to changing economic conditions.