Process optimization is of utmost importance for a correct management within any industrial field and especially within the energy and carbon intensive ones. Increasingly stringent regulations aiming at reducing the environmental impact force the companies to lower their CO2 emissions while preserving the economical sustainability of their production processes. Therefore the need arises to face these challenges by formulating optimization problems involving multiple objectives, which correspond to different and often conflicting requirements. In the present work the process gases network of a real integrated steelwork is analyzed and modeled to the aim of formulating an optimization problem considering two conflicting objectives. The multi-objective optimization problem is firstly faced through linear programming (LP) and solved by exploiting the a constraint method. As an alternative, an innovative evolutionary algorithm is proposed, which presents the advantage of allowing for a more flexible expression of the process model and constraints as well as an easier integration of different modules. The two approaches are discussed and compared. The results shown that surely multi-objective optimization is surely a viable solution, as CO2 emissions can be reduced by more than 4% and profit increased up to 20% compared to the presented base cases in normal operating conditions. The savings are even higher if off-design plant operation conditions subsist. The LP formulation provides faster generation of analytically optimum solutions, which lead to higher savings. However, in the generation of intermediate trade-offs and complete exploration of the solution search space the evolutionary approach is more flexible and more suitable to this application in the long-term. (C) 2014 Elsevier Ltd. All rights reserved.