3.3 Diagnosis of electrical networks
3.3.2 Fault detection and isolation for the one feeder radial
In this case the fault isolation part of the diagnosis is based on the voltage dierence principle introduced in Section 3.3.1. The structural decomposition of the network is also utilized during the diagnosis. Dierent cases of the di-agnosis (for multiple and single load subnetworks) are distinguished according to the number and locations of the illegal loads. The notation of single and multiple load subnetworks introduced in Section 3.2.2 are used here.
The fault detection method is the same as it is described in Section 3.3.1.
Fault isolation in multiple load subnetworks
The rst case is when there are one or more illegal loads and they are located in multiple load subnetworks. Then the fault isolation method can be applied separately to each multiple load subnetwork. The advantage of the decomposition is that the fault isolation method can be applied to the subnetworks in parallel and the complex problem can be traced back to a simpler one.
There is only one minor modication in the computation of the voltage dierences and the illegal currents in this case. After the decomposition there is no feeder at the start of the subnetwork but the junction node of the branch acts as a pseudo-feeder. Therefore, the voltage of the feeder is substituted by the voltage of the junction node at the start of the branch. Let it be denoted byUB. If the illegal load is the rst one, then the voltage dierence of the rst load can be computed by:
∆U1 = ∆UB−IillRw1
The problem is that the measurement points are at the loads. Therefore the voltage of the junction node (UB) is not measured therefore ∆UB cannot be computed. Without knowing the value of ∆UB the illegal current at the rst load cannot be computed. Therefore it is impossible to localize the illegal load with this method if it is the rst load of the subnetwork. In this case the method presented in the next section can be applied to that load.
Fault isolation in single load subnetworks
It might happen that the illegal load is in a single load subnetwork. The problem is that the single load subnetworks contain only one load therefore the voltage dierence cannot be compared to another load in the same subnetwork.
In this case the decomposition based method cannot be applied. Instead, the original network is used in the diagnosis.
In this case the voltage dierence of the single load is compared to the nearest load in both directions in the network. (It is assumed that these loads are not illegal loads). The steps of the fault isolation method in this case are the following:
1. The single load is denoted by Ls and the voltage dierence at this load is ∆ULs.
2. Go back to the previous junction from the single load. This node is the predecessor of the single load in the directed graph.
3. Find the nearest load from that node backwards in the graph. This load is denoted by Lb and the voltage dierence at this load is ∆ULb.
4. Find the nearest load from that load forward in the graph. This load is denoted by Lf and the voltage dierence at this load is ∆ULf.
5. If ∆ULb >∆ULs <∆ULf then Ls is considered as an illegal load.
The exact value of the illegal current can be computed in this case if there are no additional illegal loads in the network. The magnitude of the illegal current is ∆ULbR−∆Uwbs Ls, where Rwbs is the resistance of the wire between the start of the branch of theLb load and the single load. If there are more illegal loads, then the computed current may not be accurate because it may be aected by the currents of the other illegal loads.
Diagnostic algorithm for one feeder radial networks
The diagnostic algorithm of a one feeder radial network combines the de-tection of the illegal loads and the localization of them whether they are in multiple or single load subnetworks. There are two main parts of the diagnosis method.
First the illegal loads in multiple load subnetworks are localized using the method described in Section 3.3.2. Then the currents of the found illegal loads are updated with the currents of the illegal loads, and the network is simulated with the updated currents. In these steps all illegal loads in the multiple load subnetworks are discovered.
If the criterion of the detection is still satised after the simulation then there are remaining illegal loads in the network. They should be in the single load subnetworks. These illegal loads are localized in the second part of the algorithm using the method described in Section 3.3.2.
The steps of the complete diagnostic method are the following.
1. Decompose the radial feeder network into single and multiple load
sub-0.03
Figure 3.10: A simple one feeder radial network and its decomposition.
3. Detect the illegal loads using Equation (3.7). If the inequality is true then continue with analyzing the voltage dierences.
4. Compute the dierence between the measured and the nominal voltages.
5. If the dierence is outside of the minimum/maximum bounds then con-tinue with the localization of the illegal loads.
6. Find illegal loads in multiple load subnetworks.
7. Update the currents of the found illegal users with the computed illegal current.
8. Compute the nominal voltages and the minimum/maximum bounds us-ing the updated currents. Attach the computed values to the subnet-works. The network does not need to be decomposed again.
9. Repeat Steps 3-5.
10. Find the illegal loads in single load subnetworks.
A simple one feeder radial network example
The diagnostic method is illustrated on a simple example. The network can be seen in Figure 3.10a. It is the same network as in Section 3.2.2. The decomposition of the network can be seen in Figure 3.10b. The resistances of the wires can be seen on the edges of the graph. The uncertainty of the resistances is 2%, the current and voltage measurement error is 0.2%. The measured currents and voltages of the feeder and the loads are in Table 3.1 in the rows I˜and U˜.
Table 3.1: Current and voltage values of the example.
F1 L1 L2 L3 L4 L5 L6 L7 L8 L9 L10
I˜[A] 63 5 4 8 8 2 5 5 7 8 8
U˜[V] 250 244.44 243.04 242.56 243.99 242.53 242.08 242.02 241.60 246.99 246.83 U[V] 250 244.78 243.58 243.10 244.34 242.85 242.40 242.34 241.92 247.08 246.92
∆U[V] 0 -0.34 -0.54 -0.54 -0.35 -0.32 -0.32 -0.32 -0.32 -0.09 -0.09
The steps of the diagnostic algorithm, the used and computed data and the results of the steps in case of the example can be seen in Table 3.2. It can be seen, that after the rst part of the diagnosis (Steps 1-6) one illegal load (L2) is found in a multiple load subnetwork with illegal current of 2A. After that the network is updated and simulated with this current (Steps 7-8). The computed voltages after Step 8 can be seen in Table 3.3. In the second part of the diagnosis (Steps 9-10) an other illegal load (L4) is found in a single load subnetwork with illegal current of 1A. In conclusion our diagnostic method successfully detected and localized these loads, and determined the magnitude of the illegal currents too.
A more complex case study can be found in Appendix A where the di-agnostic method is demonstrated on a benchmark example. The case study shows the application of the decomposition and diagnostic algorithms in an almost real-world case, with simulated measurement errors.