In Switzerland like in many countries, many of the populated areas are located close to lakes or rivers, offering a large potential for the use of heat pumps. Although district heating is an obvious solution, adapting the delivery temperature to the most exigent users is detrimental to overall system performance. The best system configuration could avoid this pitfall by relying on a centralized plant of heat pumps with cogeneration, supplemented by decentralized heat pumps for the more demanding users. Using expert knowledge to compose a superconfiguration of all possible components for a district heating network system with both centralized and decentralized heat pumps, a novel methodology is proposed for modeling and optimizing, under known sets of economic, user characteristic and environmental constraint data, both the final configuration resulting from the superconfiguration as well as the configuration’s corresponding component designs. The optimization is accomplished with the help of a cutting-edge operations research tool, a genetic algorithm, while the model itself is based on a total cost objective function which unifies in a single model investment, operation, and pollution costs. These latter costs are (for each of the major pollutants) based on the literature as well as on continuous pollution penalty functions adapted to system emissions and to local immissions ratios. These penalties are included to help guide the choice of configuration and component designs away from regions where pollution limits or undesirable levels of actual emissions in combination with local / global conditions are approached too closely. Results are shown for various user distributions and fuel and electricity prices. A comparison between the results found with and without taking into account pollution is also presented.