The PSIngo or also PSI Intelligent Grid Operator - who is that actually? As the name suggests, this is an intelligent control unit for the electrical grid. In the flexQgrid project, PSIngo sits in a small red box installed in the secondary substation. PSIngo continuously monitors the current grid status and calls up the red light phase in the event of a congestion occurring (see grid traffic light concept). Although attempts are made to prevent this in the yellow traffic light phase - inaccurate forecasts or incidents will never be completely prevented.
How is the grid status monitored?
The most important basis for a grid state estimate are always real measured values. In the higher grid levels, we classically have measured values for all nodes in the grid, but these are subject to a certain measurement error and do not necessarily match each other. Therefore, an algorithm is used to search for a "common" grid state for which the individual states must deviate as little as possible from their measured value.
However, we are now in the distribution grid. For our field test, measuring devices are used in the secondary substations that provide current values for current and voltage for all feeders and phases at regular intervals. Furthermore, smart meters are installed in some households, and we can access current measurements here as well. This is good, but we are far from receiving measured values for ALL nodes.
But how can I find out if the voltage at the end of the string is too high when there is a lot of power being fed in?
To do this, we have to generate so-called "pseudo-measured values". We take all the information we know, i.e. current measured values, irradiation sensors, historical data, master data, etc. and estimate the power for each unmeasured node. In the simplest case, for example, we would distribute the measured power at the beginning of the line equally to all unmeasured nodes behind it. In this way, we generate an over-determined grid and can use the algorithm described above to determine the state of the entire grid. This generation of the pseudo-measured values is obviously a critical point for the quality of the estimation, which is why we also check the here. In Figure 1 you can see by way of example that only some of the nodes are measured by means of sensors and pseudo-measurement values have to be generated for the rest.
Figure 1: Example topology for a state estimation with generation of pseudo-measurements.
What happens when a congestion is detected?
If the state estimate shows, for example, that the voltage at one end of the string exceeds the permitted tolerances, PSIngo can intervene. In the field test, it has access to a large number of flexible systems for this purpose - partly "directly" via aggregators, partly aggregated via intelligent building energy management (GEMS). PSIngo determines the actuators that have the greatest effect on the detected congestion, i.e. are electrically close to it. Setpoints are then sent to reduce the systems just enough to eliminate the congestion. If the situation relaxes again, this is also reflected in the grid status estimate and the plants can be released again step by step.
The grid is thus quickly brought back below its permitted operating limits in the event of limit violations and the regulated energy (whether from the charging station or the PV system) is reduced to a minimum. We are curious to see how much PSIngo will have to do in the field test!