Leaks in pipe networks

Leak detection in water-distribution systems can be accomplished by solving an inverse problem using measurements of pressure and/or flow. The problem is formulated with equivalent orifice areas of possible leaks as the unknowns. Minimization of the difference between measured and calculated heads produces a solution for the areas. The quality of the result depends on number and location of the measurements. A sensitivity matrix is key to deciding where to make measurements. Both location and magnitude of leaks are sensitive to the quantity and quality of pressure measurements and to how well the pipe friction parameters are known. The overdetermined problem (more measurements than suspected leaks) gives the best results, but some information can be derived from the underdetermined problem. The variance of leak areas, based on the quality of system characteristics and pressure data, indicates the likely accuracy of the results. The method will not substitute for more traditional leak surveys but can serve as a guide and supplement.