Prognosis of product take-back for enhanced remanufacturing

Abstract

Remanufacturing reduces final wastes to sinks, extraction of virgin materials and pollution from production processes by reinstating products taken back by end-users to satisfy part of overall demand. Product returns are delayed and possibly limited in periods of fast growth and excessive in the aftermath. Varying growth/demand and volatile take back by consumers and industrial end-users introduce uncertainty, regarding quantity and quality of returns. As remanufacturing expands, escalating competition for acquisition of high quality returns exacerbates uncertainty. Production planning and control for efficient remanufacturing depends on reliable prediction of quantity and quality of returns. A method is developed for prognosis of product return quantity and quality grades, as reflected by vintage flows. It is anchored on a law relating stock and end-of-life level, under random losses and arbitrary end-of-life distribution. Efficacy is tested via a model that describes stock and flows in reuse/remanufacturing, allowing for varying demand, random stock losses, random product returns with time-varying distributions and time-varying utilisation of product returns. Realisations are obtained by Marko-chain Monte-Carlo simulation. Inherently integral in nature, using scaled data and founded on rigorous balances, the method enables prognosis of returns and age-vintage flows, under realistic conditions, including unknown nonlinearities and non-stationarities. It features improved performance (mean absolute error less than one half) compared to leading methods in-use that employ black-box models with error-driven parameter adaptation (e.g. regression). Efficacy is particularly high at crucial peaks and lows (shortage or surplus periods) enabling resourceful planning of acquisition and inventory control of product returns towards sustainability.