The procedure used to determine the type of instrument and mechanism used to measure various required variables is called sensor network design.
This problem has been addressed by several researchers .
As the design of a sensor network for GTNs is not the main concern of this work, we develop here some heuristic rules for sensor placement from the accuracy point of view.
To study the effect of different measurements, two cases are distinguished: redundant and non-redundant cases.
In the non-redundant case, noise corrupted measurements of pressure and temperature, or mass flow and temperature are provided as inputs in our approach, and the flows, pressures and temperatures are ...view middle of the document...
Estimated supply and demand mass flow rates, and delivery pressure obtained using the proposed method for case studies 1 to 3 are compared against their corresponding true values in Figs. 6- to 8. In case study 1, one pressure and one temperature sensor is located at node N3. Figures 6 to 8 show that the estimated supply mass flow and delivery pressure for this case are following their corresponding true values significantly better than the reconstructed demand flow.
This is because of the fact that no flow measurement is considered in this case, and only pressure measurements are provided.
As a result, the estimated demand flow takes some times to converge to the true value during transitions. It is also noted that the error of estimated demand flow at node N3 is affected mostly by abrupt transients induced by demand flow fluctuations rather than pressure variations in the upstream.
It is also noted that if the pressure sensor is closer to supply node, the initial convergence time becomes shorter.
Moving the pressure sensor towards upstream node N2 (case study 2), deteriorates the estimated states, especially demand flow. Addition of another pressure sensor at node N2 (case study 3) does not seriously affect the accuracy of estimations.
Total line-pack of the pipes can be seen in Fig. 9 for case studies 1 to 3. The effect of estimated pressure and temperatures and the pressure sensor location on this parameter, ignoring initial convergence rate is insignificant.
In case studies 4 to 7, shown in Figs. 10-12, the effect of mass flow rate sensor is analyzed.
As in the cases of the pressure sensor placement, one can see that flow meters are preferred to be placed near the outlet of the pipelines.
It can be seen that the use of pressure sensors, as opposed to mass flow meters, has a positive...