Multiobjective design of Work-In-Process buffer for scheduling repetitive building projects V. González a,b, ⁎, L.F. Alarcón a , K. Molenaar c a Department of Construction, Engineering and Management, Pontificia Universidad Católica de Chile, Santiago, Chile b Construction Engineering School, Universidad de Valparaíso, Valparaíso, Chile c Department of Civil, Environmental, and Architectural Engineering, University of Colorado, Boulder, CO, USA ABSTRACT ARTICLE INFO Article history: Accepted 20 May 2008 Keywords: Buffering Evolutionary Strategies Lean production Multiple objective analysis Simulation–Optimization models Work-In-Process Variability in production is one of the largest factors that negatively impacts construction project performance. A common construction practice to protect production systems from variability is the use of buffers (Bf). Construction practitioners and researchers have proposed buffering approaches for different production situations, but these approaches have faced practical limitations in their application. A multiobjective analytic model (MAM) is proposed to develop a graphical solution for the design of Work-In-Process (WIP) Bf in order to overcome these practical limitations to Bf application, being demonstrated through the scheduling of repetitive building projects. Multiobjective analytic modeling is based on Simulation–Optimization (SO) modeling and Pareto Fronts concepts. Simulation–Optimization framework uses Evolutionary Strategies (ES) as the optimization search approach, which allows for the design of optimum WIP Bf sizes by optimizing different project objectives (e.g., project cost, time and productivity). The framework is tested and validated on two repetitive building projects. The SO framework is then generalized through Pareto Front concepts, allowing for the development of the MAM as nomographs for practical use. The application advantages of the MAM are shown through a project scheduling example. Results demonstrate project performance improvements and a more efficient and practical design of WIP Bf. Additionally, production strategies based on WIP Bf and lean production principles in construction are discussed. © 2008 Elsevier B.V. All rights reserved. 1. Introduction Variability in production is one of the largest factors that negatively impacts construction project performance. It can induce dynamic and unexpected conditions, unsteadying project objectives and obscuring the means to achieve them. To understand the effect of variability on production processes, Hopp and Spearman [1] distinguished two kinds of variability in manufacturing systems: 1) the time process of a task and 2) the arrival of jobs or workflow at a workstation. Koskela [2] proposes a similar classification to variability in construction systems, where the processes duration and the flow of preconditions for executing construc- tion processes (e.g., space, equipment, workers, component and materials, among others) are understood as variable production phenomena. From a practical standpoint, construction practitioners everyday observe this behavior in the project environment through varying production rates, labor productivity, schedule control, cost control, etc. Several researchers have shown that variability is a well-known problem in construction projects, which leads to a general deteriora- tion of project performance on dimensions such as: cycle time [3–7], labor productivity [8,9], project cost [10], planning efficiency [11,12], among others. A way to deal with variability impacts in production systems is through the use of buffers (Bf). By using a Bf, a production process can be isolated from the environment as well as the processes depending on it [2]. Buffers can circumvent the loss of throughput, wasted capacity, inflated cycle times, larger inventory levels, long lead times, and poor customer service by shielding a production system against variability [1]. Hopp and Spearman [1] define three generic types of Bf for manufacturing, which can be applied in construction as: 1. Inventory: In-excess stock of raw materials, Work in Process (WIP) and finished goods, categorized according their position and purposes in the supply chain [13]. 2. Capacity: Allocation of labor, plants and equipment capacity in excess so that they can absorb actual production demand problems [14]. 3. Time: Reserves in schedules as contingencies used to compensate for adverse effects of variability. Float in a schedule is analogous to a Bf for time since it protects critical path from time variation in non- critical activities. Theoretically, the analysis of Bf in this paper is based on lean production principles. Lean production is a management philosophy focused on adding value from raw materials to finished product. It allows avoiding, eliminating and/or decreasing waste from this so-called value stream. Among this waste, production variability decreasing is a central point within the lean philosophy from a system standpoint [15]. Lean Automation in Construction 18 (2009) 95–108 ⁎ Corresponding author. Mailing address: Vicuña Mackenna 4860, Macúl, Santiago, Chile. Campus San Joaquín, Edificio San Agustín, 3rd Floor. Postal Code: 7820436. Tel.: +56 2 3544245/3544244; fax: +56 2 3544806. E-mail address: vagonzag@uc.cl (V. González). 0926-5805/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.autcon.2008.05.005 Contents lists available at ScienceDirect Automation in Construction journal homepage: www.elsevier.com/locate/autcon