Integrating forward and reverse supply chains has proven to be a challenging task due to the differences in the nature of the activities that make up the forward and reverse flows. With the increasing concerns over environmental degradation, legislative compliance, diminishing supply of raw materials, and consumer demands for eco-friendly products, companies have begun designing the traditional supply chain paths to form a closed loop to facilitate recycling and reuse of product returns. The integration of forward and reverse supply chains is termed as Closed Loop Supply Chains (CLSC). Largely driven by its profitable business propositions, CLSC practices help to achieve both financial success and environmental benefits. Given that the annual estimate of commercial returns in the US is in excess of $100 billion and huge potential for using the value in the returned products, this dissertation develops optimization models for the design of a CLSC network for commercial returns by explicitly considering the various aspects of product returns, refurbishing costs, and customer acceptance of refurbished products.We first consider a four stage CLSC network for a single product incorporating customer acceptance rate for refurbished products. We develop an optimization model, called the Base model, to determine the optimal locations of the facilities and the distribution of flows between the facilities in the CLSC to maximize the total profit and illustrate it using a realistic example. The results from the sensitivity analysis show that the total supply chain profit increases with the increase in refurbishing activity. The changes in product return parameters lead to changes in the optimal network design, implying the need to explicitly consider the uncertainty in these return parameters. We also observe that the refurbishing cost has an effect on the network design decisions implying the need to include it in the model. Quality of product returns vary depending upon how long the product has been in use, nature of defects, shipping damages etc. In this research, we consider four tiers of product returns, based on the quality of returned products and their refurbishing costs. Buying behaviour of the quality-sensitive high-end customers and price-sensitive low-end customers vary with respect to the price of the refurbished products. We extend our Base model considering the four tiers of product returns, their quality and the customer behaviour towards buying refurbished products. It is called the Extended Base model. We show that the inclusion of the tiers of product returns and customer behaviour towards buying refurbished product increase the supply chain profit. Using sensitivity analysis, we show that the quality of the returned products is an important factor in determining the optimal discount rates for the refurbished products. We also show that the network design varies with the customer acceptance rates of the refurbished products. Network design models are intrinsically multi-objective in nature. The Extended Base model is reformulated with two objectives: maximize profit and minimize energy usage in the warehousing facilities and transportation. We develop a bi-criteria MILP model for the problem and illustrate it using a realistic example. An interactive optimization algorithm is proposed to solve the bi-criteria problem. The algorithm poses less cognitive burden on the decision maker and converges faster to the best compromise solution. Product return parameters, customer demands, and customer acceptance of the refurbished products, are unpredictable. We model these uncertainties using a set of discrete scenarios with associated probabilities. A scenario-based robust optimization model is then developed to maximize the expected profit of the supply chain along with minimizing the variability of the profit across the scenarios. Minimizing the variability is necessary to mitigate the risk associated with the uncertain input parameters in the supply chain. The model is illustrated with an example. We provide a comparison between the robust optimal solution and the deterministic design solution. We show that the expected profit and the customer responsiveness are higher in the robust design compared to its deterministic counterpart.