This article presents a mathematically rigorous 1 unification of the pipe roughness identification problem of water 2 distribution networks considering all Reynolds regimes, i.e., 3 laminar, turbulent, as well as transitional flow regimes. Although 4 the identification procedure is based on steady-state hydraulic 5 network equations, the identified roughness parameters are also 6 key for dynamic models that can be used for model-based 7 controller and observer designs. While a three-cycle network 8 simulation example serves to illustrate the presented problem 9 formulation and solution in an extensive manner, the application 10 on a real-world drinking water network is in focus. In addition, 11 vital aspects, such as topology simplifications of the underlying 12 network and the importance of the generation and selection of 13 independent measurement sets, are addressed. We apply root-14 finding methods instead of methods based on optimization and 15 thereby show that the pipe roughness identification problem may 16 actually be applicable to identify as well as locate leakages. 17 Index Terms-Laminar flow, pipe roughness identification, 18 roughness calibration, transitional flow, turbulent flow, water 19 distribution networks. 20 I. INTRODUCTION 21 D RINKING water networks are of utmost importance 22 for society all over the world. However, the share of 23 nonrevenue water, i.e., water that never reaches a registered 24 consumer, accounts for 25%-50% of the total amount of 25 supplied water when put into a global measure (and up to 75% 26 in the emerging markets) as stated by the International Water 27 Association [1]. Consequently, the potential contributions of 28 control system technologies to water distribution systems are 29 manifold and cover, e.g., the coordinated control of distributed 30 actuators (valves and pumps), and the detection of leakages in 31 the water networks [2]. 32 The basis for advanced techniques is usually formed by 33 mathematical modeling of the underlying distribution network. 34 In the literature, two classical modeling approaches are widely 35 used.