Tri-generation systems are utility systems which produce heat, power and cooling simultaneously. Use of tri-generation systems in industrial sites reduces the importation of power and improves local power reliability; at the same time, their inherently higher efficiency also reduces environmental impacts. However, interdependencies among process units in tri-generation plants can lead to vulnerability to cascading failures. Process units may become non-functional during the course of operations as a result of planned or unplanned stoppages. This issue is normally handled by installing additional capacity to the process units based on heuristics. However, such heuristics may not be able to address complex decisions pertaining to the installation of multiple units to provide redundancy, and may result in excessive capital and/or maintenance costs. In this work, a systematic approach for the grassroots design of a reliable BTS (biomass-based tri-generation system) considering equipment redundancy is presented. Chance-constrained programming and k-out-of-m system modelling are used to develop a multi-period optimisation model for a generic BTS. Two case studies are then solved to illustrate this modelling approach. (C) 2015 Elsevier Ltd. All rights reserved.