Investigation into the transmission system modeling for the effective assessment of voltage unbalance due to AC railway operation: Evaluation using on-site measurement data

AC railway is well known for introducing large asymmetrical loads to power systems. Lately, the increasing number of traction load connections has raised concerns for transmission system operators (TSOs) regarding the balanced and secure operation of the grid. The quantification of voltage unbalance (VU) contribution by traction loads is important in relation to VU management, both for TSOs and railway operators. As voltage unbalance propagates through the electric network, the assessment of the railway impact requires the analysis of the grid with all traction substations (TSSs), in a common study. The effectiveness of this kind of studies depends both on the use of accurate load profiles and the appropriate modeling of the electric network. In relation to the latter, the grid model should be able to calculate accurately the unbalance propagation from the railway feeding nodes through the network.
To this direction, the work presented here investigates the importance of modeling the network’s inherent asymmetries, when estimating the unbalance contribution from loads. To support the analysis on-site measurement were performed in the Danish grid and the results are used to evaluate the grid model calculations.
In transmission grids, where the trains are connected, untransposed lines are the main source of network’s inherent unbalance. The question is, how necessary is the accurate representation of line models when calculating the unbalance contribution from loads connections? Although this study refers to unbalanced loads in general, it finds great application in electric railway as this is the dominant unbalanced customer connected to the transmission level. For the analysis, the grid model of the western Danish transmission system is used and power flow calculations are performed in DigSILENT PowerFactory. For several unbalanced load connections in the grid, the effect of modeling or not the line couplings is investigated under several scenarios. The results showed that the actual level of unbalance differs significantly between the cases of considering balanced or unbalanced lines. However, when calculating only the contribution from the loads the modeling of line couplings becomes not significant.
For the evaluation of the grid model calculations, on-site measurement data were used. More specifically, voltage unbalance propagation measurements conducted in a small area of the Danish grid. The measurements carried out during a period test activities of the Danish railway operator in a newly built TSS. In particular, several short-circuit tests from the catenary to the ground in the new TSS were conducted, inducing a high unbalanced loading at the TSS connection node. The corresponding unbalance variations, and therefore the short-circuit contribution to unbalance was measured in three HV nodes. Eventually, the comparison between the measurement data and the load flow results show very good agreement, validating the model performance to calculate unbalance contribution using the balanced representation of the grid.