Development of a computer system for the analysis of the operational characteristics of dynamicaly regulated channels: The case of Mornos aqueduct

The management of the aqueduct of Mornos (and by extension of the total water supply system of Athens) until 1996 was done by experienced executives of EYDAP guided by their experience. Empirical management means the manual adjustment of the aqueduct's valves after visual surveillance of the measurements and evaluation of the current situation based on demand-abstraction forecasts. Since then, the company has launched a plan to modernize the operation of the canal through the formulation of detailed operating rules that would help optimize its operation, while maintaining the security of the system. One operating parameter of the channel is the flow factor Cd of the gates mounted on the regulators L. Objective of this paper is to capture the ratio Cd in relation to the ratio a/H1 for the regulator L10 of the aqueduct. Better mapping of the distribution of Cd values can ensure better handling of the channel through existing remote-control systems. This, for the most part, means better regulation of the aqueduct's supply and, by extension, the possibility of implementing a more automated management system. The above process is elaborated through a set of processes that run in a programming environment (MATLAB). In the first stage, a data cleaning and validation process is performed and then all intervals of the time series in which no change in the opening of the gate has been observed for a given period of time are isolated. A value of Cd is calculated for each interval, resulting in a cloud of values for all intervals. The above procedures are performed multiple times through a loop alternating the duration of the interval. In order to evaluate the results and analyze the influence of certain factors on the deviation of the Cd value from the theoretical values and on the general dispersion of the cloud of Cd values, the above procedure is performed multiple times while isolating one or more factors at a time. Additional parameters are then introduced into the calibration process, such as the spillway coefficient and the local loss factor that is used into the spillway hydraulic load calculation. Thus, a global calibration process of all the above-mentioned parameters is performed either through a sequential calibration model or through simultaneous calibration through a calibration process of random combinations of episodes (Monte Carlo).