Laboratorial Implementation of Future Intelligent Transmission Network Substation for Smart Grid

Abstract

This paper presents the future intelligent transmission network substation architecture; it identifies the most important design patterns for the purpose of building simplified versions of the communication infrastructure with High-Availability Seamless Redundancy (HSR) and Parallel Redundancy Protocol (PRP) configurations, respectively. The network model used for the laboratory tests is detailed, and details regarding the modelling of current-transformer (CT) saturation are given and discussed. Several intentional communications related to equipment or fibre failures are discussed. The laboratory setup for both HSR and PRP is presented. A real-time digital simulator (RTDS) network model is built and the test processes are presented. Some tests are carried out to test the functionality and interoperability with different fault conditions under the future intelligent transmission network substation architectures’ HSR-only and PRP-only data network configurations. The results are analysed and discussed, and a future intelligent transmission network substation test platform is successfully implemented in the laboratory. The experimental results can provide valuable information for power utilities and manufacturers to optimise substation architecture and their products under different conditions.

1. Introduction

The growing integration of renewable energy resources and new types of loads into the grid has facilitated the development of smart-grid technologies [1,2,3,4]. The risks and costs of replacing substation secondary system equipment (e.g., protection and control devices) or introducing new/digital equipment are high [5,6,7,8]. Therefore, the design and evaluation of digital substation data-flow management systems have become more and more important to reduce failures in equipment. Researchers from the Indian Institute of Technology [9] propose Substation Communication Network (SCN) architecture for the protection and control of IEDs. The costs of replacing substation secondary system equipment (e.g., protection and control devices) or introducing new digital equipment are high based on this architecture. A novel protection and control architecture is described in [10] to minimise costs. However, this architecture is based on virtual devices; a special “all in one” protection and control device is required, and its reliability will become an issue in this architecture. In [11,12], IEC 61850-based substation projects were implemented to test the interoperability of IEDs from different vendors. Both projects developed testing platforms, which follow the Substation Configuration Language (SCL) defined in IEC 61850-6. However, the papers did not emphasise which substation architecture was used for the platforms. In the projects, communication issues still occurred between IEDs from different manufacturers. Although each IED complies with the IEC 61850 standard, the interpretation of the standard from different vendors could be different as ambiguity may still exist.

In order to achieve a highly reliable data communication network, the ring network topology with the High-Availability Seamless Redundancy (HSR) protocol, the star network topology Parallel Redundancy Protocol (PRP) network and two independent parallel star networks are considered.

HSR provides zero recovery time in the face of any single failure in a link or node, robust fault-tolerant networking and loop prevention through duplicate discard [13,14,15]. Ngo and Yang [16] summarise two main drawbacks of HSR: the extra traffic created due to the duplicated copies that are generated and circulated inside the network, and the transmission latency due to the additional delay introduced by each link (transmission and propagation delays) and each node (processing and queuing delays) during the transmission of a frame in an HSR ring. Multicast messages, such as generic object-oriented substation events (GOOSE) or sampled values (SVs) in the International Electrotechnical Commission 61850 standard (IEC61850), create extra traffic, while unicast messages are removed from the ring through their unique destination addresses after they have been received at their destinations. Molina et al. [17] proposed the integration of HSR and OpenFlow technologies to provide redundancy control and traffic prioritisation. As a result, a node can distinguish between critical and noncritical flows, protecting them according to their needs. The effectiveness of the priority-aware network is assessed in terms of throughput efficiency. Gutiérrez-Rivas et al. [18] presented a new challenging approach covering the industrial standard requirements, regarding timing and availability, through the development of mechanisms that make possible the avoidance of a single point of failure in ring topologies; moreover, they addressed the development and results of the High-Availability Seamless Redundancy protocol for WR technology to conform a leading-edge deterministic and reliable ultra-accurate timing system with high-availability data features for industrial facilities, in accordance with IEC 61850 and IEC 62439-3. The aforementioned investigations summarised the drawbacks of HSR and made great improvements on it. This paper intends to set up HSR configuration in the laboratory to implement a digital substation data-flow management system.

PRP provides high availability through two networks. Araujo et al. [19] summarised some of the characteristics of PRP networks as follows. The PRP nodes, called double attached nodes (DANs), are connected to two isolated networks. PRP networks accept single attached nodes (SANs). They may be connected directly to one of the local area networks (LANs) or as a virtual DAN to the two networks through the use of a redundancy box. PRP requires two LANs, and nodes cause some delay due to encoding, decoding and discarding. Urbina et al. [20] proposed a smart sensor architecture which incorporates real-time operation features; the ability to perform local data analysis; high-availability communication interfaces, such as HSR and PRP; interoperability (industrial protocols); and cyber-security. Popovic et al. [21] indicate that PRP is not viable for IP networks that are a key element of emerging mission-critical systems; coupled with diagnostic inability and a lack of security, PRP is unsuitable for reliable data delivery in these IP networks. They presented a transport-layer solution for this limitation: the IP Parallel Redundancy Protocol (iPRP). The aforementioned investigations summarised the characteristics of PRP networks and figured out solutions for the limitations of PRP. This paper also intends to set up PRP configuration in the laboratory to implement the digital substation data-flow management system.

In order to implement the digital substation data-flow management system utilising HSR and PRP configurations, the most important design patterns are identified for the purpose of building simplified versions of the future intelligent transmission network substation communication infrastructure with HSR and PRP configurations, respectively. The network model used for laboratory tests is detailed, while details regarding the modelling of current-transformer (CT) saturation are given and discussed. Several intentional communications related equipment or fibre failures are discussed. The laboratory setup for both HSR and PRP is presented, including the measurement, protection and data communication network employed to carry out the laboratory tests. Some tests are carried out to test functionality and interoperability with different fault conditions under the future intelligent transmission network substation architectures’ HSR-only and PRP-only data network configurations, respectively. The fault conditions include an investigation of the different types of faults applied, as well as intentional equipment or optical fibre failure. The main objectives of this work are to implement the future intelligent transmission network substation architecture test platform in the laboratory with multi-vendor devices and a real-time digital simulator (RTDS). Experimental assessment of the impact of different communication network redundancy architectures on two vendors’ equipment interoperability and functionality. Although different communication network redundancy architectures have been installed and operated in many real-life digital substations, it is prohibited to conduct real-life impact studies of various faulty components in data networks and of fault conditions in the power network regarding the functionality and performance of the installed protection and control devices. This creates a need to set up an experimental test bed to address the above-mentioned impact studies of different faulty components in the data communication network, and fault conditions in the power network, on P&C functionality and performance. For this reason, a virtual digital substation test bed was designed and built. The virtual substation test bed consisted of (i) An RTDS to simulate a virtual digital substation in real time, (ii) two different pieces of vendor protection and control equipment and (iii) a high number of Ethernet switches which are capable of being configured into different communication radiance topologies such as HSR and PRP. This test bed was used to conduct different case studies for the impact of different network redundancies on two vendors’ equipment performance under different faulty component conditions of the network. The main academic knowledge contribution was formulating a data analysis methodology based on the Q-Q method to validate the experiment data and conducting statistical analysis to quantify the impact of different network redundancy topologies on protection and control equipment performance. This paper is organised as follows. Section 2 presents the modelling and test process in the laboratory. Section 3 is the experimental setup of the HSR and PRP configurations. Section 4 presents the results and discussion of the implementation of the HSR and PRP configurations. Section 5 concludes the results.

2. Modelling and Test Process

Figure 1 and Figure 2 are potential PRP-only and HSR-only bays in a simplified future intelligent transmission network substation architecture, respectively. The operation of the future intelligent transmission network can be conceptually divided into two separate smaller networks: one designed according to PRP architecture (Figure 1), and the other designed according to HSR (Figure 2). Testing of the smaller networks from Figure 1 and Figure 2 can cover the most important aspects relevant to the IEC 61850 interoperability of the future intelligent transmission network substation architecture.

2.1. Real-Time Digital Simulator (RTDS) Network Model

For modelling and testing purposes, the future intelligent transmission network digital substation is considered to be located at the beginning of the Rassau feeder. An excerpt from the full RSCAD (RTDS modelling software) network model is given in Figure 3 to indicate the point of connection of the current transformers (CTs), providing current measurements for the laboratory deployment of the future intelligent transmission network substation architecture. Within the RSCAD network model, the voltage transformers (VTs) have the same connection point as the CTs.

However, in order to test the differential protection scheme within future intelligent transmission network substation architecture, there is a need, within the RSCAD model, for a second set of CTs and VTs to be connected at the end of the feeder. This set of CTs and VTs need to feed measurements into the remote-end intelligent electronic device (IED) of the differential protection scheme. The point of connection for the second set of CTs/VTs, together with the feeder’s transmission line model, is presented in Figure 4.

2.2. Modelling CT Saturation

Modelling CT saturation is an integral part of the future intelligent transmission network substation architecture performance evaluation. This is because it represents a physical phenomenon that can detrimentally influence the proper functioning of protection systems. Although CT saturation is a topic already well-documented in the literature [22,23,24], its underlying principles are presented here, as well, for the sake of clarity and completeness. Ideally, the amount of current that flows in the secondary part of a CT is directly proportional to the amount of current that flows in the primary. However, due to electromagnetic phenomena within the CT core, a point of saturation is reached, after which there is no more linear dependence between the primary and the secondary CT currents. This phenomenon occurs within the material of the CT core and can be optimally visualised by plotting the magnetic field density (B) against the magnetic field intensity (H), as shown in Figure 5.

For all CTs, the dependence between B and H is closely related to the dependence between the voltage and current, respectively. The plot in Figure 5 was obtained from the data of the RTDS standard CT model, as given in Figure 6.

The left-most part of the B-H curve from Figure 5 is the linear region, while the right-most part is called the saturation region. The intermediate region between the two is where the knee point of the curve lies. Under normal loading conditions, the operating point of the CT ought to lie within the linear region. Under very high current conditions, e.g., during a fault, there is a chance that the CT operating point will move into the saturation region. For the CT saturation tests, it was necessary to ensure that the CT went into the saturation region. This behaviour was ensured by artificially increasing the initial (under normal loading conditions) operating point of the CT through the use of high burden resistance within the RTDS CT model. Based on the level of saturation of the CT, three saturation test cases can be studied, as indicated in

Table 1. CT saturation test cases.

Test Case	Description	CT Model Parameters
CT-Sat0	Ideal CT or no CT saturation	Rburden = 0.5 ohm; L = 0.035 H
CT-Sat1	Light CT saturation	Rburden = 5 ohm; L = 0.35 H
CT-Sat2	Deep CT saturation	Rburden = 50 ohm; L = 3.5 H (initial operating point close to knee point)

.

2.3. Intentional Equipment or Fibre Failures in the Data Communication Network

2.3.1. HSR Configuration

There were three group scenarios employed to intentionally apply faults within the communication infrastructure, which are shown in

Table 2. Intentional faults applied to communication infrastructure for HSR configuration.

Scenario	Description
A	No faulty device or fibre in the HSR configuration.
B	One optical fibre has failed in the HSR configuration. The redundancy capability of the network is reduced.
C	One AMU from a piece of equipment has failed in the HSR configuration. MP1—differential protection relay blocks their trip function.

:

-

Scenario A: No communication network component failure. In this scenario, the communication network functions at its full redundancy capability.

-

Scenario B: One optical fibre failure in the HSR ring. In this scenario, we assume that one fibre in the HSR ring is broken. As a result, either the sample value packets between the analogue merging unit (AMU) and IED or the GOOSE message between the IED and Circuit Breaker Controller (CBC) is transmitted and received over a longer distance.

-

Scenario C: Equipment AMU failure in the HSR ring. In this scenario, one protection device (either MP1 or MP2) in the ring has been incapacitated. In this test case, the sample values from the Vendor 2 AMU to MP1 (Vendor 1 IED—differential protection scheme) are no longer working. Therefore, the loss of connection between Vendor 2 AMU and MP1 (Vendor 1 IED) at the local end has blocked its pair of differential protection devices at the remote end.

2.3.2. PRP Configuration

Similar to HSR configuration, there are also three scenarios employed to intentionally apply faults within the communication network for PRP configurations, and all scenario groups are summarised in

Table 3. Intentional faults applied to communication infrastructure for PRP configuration.

Scenario	Description
D	No faulty device or fibre in the PRP configuration.
E	One equipment AMU failure, and MP1—differential protection relays block their trip function.
F	Equipment AMU failure + one fibre in PRP network is discounted. The redundancy capability of the network is reduced.

.

3. Experimental Setup

3.1. HSR Configuration Lab Setup

A block diagram of the laboratory HSR configuration setup is shown in Figure 7. The HSR ring consists of Process Bus 1 connecting the Vendor 2 AMU, Vendor 1 IED1 and Vendor 1 DMU1 and Process Bus 2 connecting the Vendor 1 AMU, Vendor 2 IED3 and Vendor 1 DMU2. As the Vendor 1 IED1 is functional as a differential protection device, it is connected to the Vendor 1 IED2 at the remote end. The RTDS is used to model the substation, and all the IEDs and equipment are synchronised by either GPS or by the IEEE1588 master clock.

3.2. PRP Configuration Lab Setup

The block diagram of the laboratory PR configuration setup PRP is shown in Figure 8. The same equipment as in the HSR configuration was used to configure RPR where Process Bus 1 connects the Vendor 2 AMU, Vendor 1 IED1 and Vendor 1 DMU1, and Process Bus 2 connects the Vendor 1 AMU, Vendor 2 IED3 and Vendor 1 DMU2. As the Vendor 1 IED1 is also functional as a differential protection device, it connects to the Vendor 1 IED2 at the remote end. It should be mentioned that Process Buses 1 and 2 are independently configured and the Station Bus is the PRP configuration according to future intelligent transmission network substation architecture. Similarly, the RTDS is used to model the substation, and all the IEDs and equipment are synchronised by either GPS or the IEEE1588 master clock.

4. Results and Discussions

4.1. No-CT Saturation Tests in HSR

The ideal VT and CT produced secondary or no CT saturation (i.e., CTSat0) outputs from the substation RDTS model via voltage and current amplifiers to the AMUs. There are three categories of the direct CTSat0 tests corresponding to the type of fault applied within the RTDS network model: (i) three-phase-to-ground, (ii) single-phase-to-ground and (iii) line-to-line faults.

4.1.1. CTSat0 Test Cases in Category 1: Three-Phase-to-Ground Fault at 50% of the Feeder

Case CTSat0-A1 (Ideal CT, No Faulty Components, 3ph-G Fault at 50% of the Feeder)

This test case regarding the operational status of the HSR lab setup architecture with the consideration of no faulty components (referred to as Scenario A, as described in Table 2 in Section 2.3) was evaluated under three-phase-to-ground faults. A larger number of tests (50 tests) was conducted to assess the IEDs trip times under a three-phase-to-ground fault at the middle (50%) of the Rassau feeder (refer to the feeder in Figure 4). Both the IED1 (local) and IED2 (remote)—Vendor 1 differential protection device MP1, and the IED3—Vendor 2 distance protection device MP2 operated correctly. As MP1 operates faster than MP2, only MP1—protection performance was assessed and discussed. By plotting the number of tests that occurred on MP1’s tripping times, the MP1 tripping time distribution was obtained and is shown in Figure 9a. If the probability of the MP1 tripping time distribution is regarded as a normal distribution, the mean of MP1 tripping times (T_mean) and standard deviation (σ) for the tests can be calculated, and they are 0.0263 s and 0.0044 s, respectively. This predicted MP1 tripping time as a normal distribution is shown in Figure 9b.

By comparing the MP1 trip time distribution in Figure 9a and the predicted MP1 normal distribution in Figure 9b, the MP1 trip time performance can be qualified using a so-called quantile plot, as shown in Figure 10.

As seen in Figure 10, the majority of MP1 trip times are distributed along the predicted standard normal distribution as the linear line. The tripping times closely match the linear line between the MP1 tripping times, with values of 0.0175 s (T_mean of 0.0263 s − 2σ) and 0.0351 s (T_mean of 0.0263 s + 2σ).

R-squared was used as the statistical measure to show how the tests closely match the predicted normal distribution. The formula for R-squared is given by Equation (1),

$$R^2=\frac{\sum \left(Y_{actual}-Y_{predicted}\right){ }^2}{{\displaystyle \sum }{\left(Y_{actual}-Y_{mean}\right)}^2}$$

(1)

where:

-

Y_actual is the actual IED tripping times, expressed a function of the standard deviation (σ)

-

Y_predicted is the best fit for the tested IED tripping times as function of the standard deviation (σ)

-

Y_mean is the mean of the tested IED tripping times.

The calculated R-squared of 0.7965, as shown in Figure 10, shows how the tested MP1 tripping performance closely matches the predicted normal distribution where the predicted normal distribution has a mean of 0.0263 s and a standard deviation σ of 0.0044.

Case CTSat0-B1: (Ideal CT, One Faulty Fibre, 3ph-G Fault at 50% of the Feeder)

This test case regarding the IED performance was evaluated under the HSR lab setup architecture with one optical fibre failure and three-phase-to-ground faults in the middle of the Rassau feeder. In order to test the reliability of the HSR communication architecture, the fibre optic link between red box 1 linked to IED3—Vendor 2 and red box 4 linked DMU1—Vendor 1 was disconnected. In this case, both MP1 and MP2 operated correctly, but either the SV between the IED and AMU or the GOOSE trip signal between the IED and the digital merging unit (DMU) took the longer path in the HSR ring. Since MP1—differential protection is faster than MP2—distance, only the performances of MP1 are assessed and discussed.

The MP1 tripping time tests were repeated 50 times. The MP1 tripping time distribution for case CTSat0-B1 is calculated as shown in Figure 11a. The quantile plot for MP1 trip time performance is shown in Figure 11b.

As seen from Figure 11b, the majority of IED tripping times are within one standard deviation σ. The calculated R-squared of 0.8738 shows that the tested IED tripping performance closely matches the predicted normal distribution with the mean of 0.0278 s and the standard deviation σ of 0.0052.

Case CTSat0-C1 (No CT Saturation, One Faulty AMU, 3ph-G Fault at 50% of the Feeder)

The future intelligent transmission network substation HSR architecture has faulty equipment. Assuming the AMU2—Vendor 2 analogue merging unit in the HSR ring has failed, the IED1—Vendor 1 differential protection device, which is MP1 connected to AMU2, will no longer be working correctly. However, the IED3—Vendor 2 distance main protection device, which is MP2 connected to the AMU1—Vendor 1 merging unit, still operates correctly. In this test case, only the MP2 performances are assessed and discussed.

In this test case, the MP2 tripping time distribution for the faulty equipment was obtained and is plotted in Figure 12a. The quantile plot for MP2 trip time performance is shown in Figure 12b. As can be seen from Figure 12b, the MP2 trip time values follow a heavy-tailed normal distribution. This leads to more fluctuation between the tested trip times and the predicted normal distribution.

Comparisons of Scenarios CTSat0-A1, CTSat0-B1, and CTSat0-C1

The means and standard deviations of the IED tripping times of cases CTSAT0-A1, CTSat0-B1 and CTSat0-C1 were calculated and are listed in

Table 4. Trip times for CTSat0 test case group 1–3-phase-to-ground fault.

Scenario	Description	Mean Trip Time [s]	Std. Deviation [s]
CTSat0-A1	Ideal CT, no faulty comp, 3Ph-G fault	0.02634	0.0044
CTSat0-B1	Light CT Sat, 1 faulty fibre, 3Ph-G fault	0.02779	0.0052
CTSat0-C1	Deep CT Sat, 1 faulty AMU, 3Ph-G fault	0.03612	0.0028

. The corresponding IED trip times under the tests are sorted from the fastest to the slowest in ascending order and the compared results are shown in Figure 13.

As can be seen from both Table 4 and Figure 13, the trip time of test case CTSat0-A1 is the fastest. This is expected because no faulty component in the HSR ring, SV and GOOSE messages among the AMU, IEDs and DMU can select the shortest paths in the HSR ring. As a result, the fastest fault detection and the fastest trip response can be achieved.

In test case CTSat0-B1, a fibre link in the HSR ring is disconnected. SV and GOOSE among the AMUs, IEDS and DMUs may take a longer path. This results in the MP1 trip time increasing by 5.5% in comparison with that in the test case CTSat0-A1. In the case of CTSat0-C1, AMU connected to MP1 is disabled. This results in MP1 being disabled. However, the MP2—distance relay can still operate correctly. Since the MP2—distance protection scheme is slower than the MP1—differential scheme, MP2 in the case of CTSat0-C1 is slower than MP1 in the case of both CTSat-A1 and CTSat0-B1. In summary, the results show that all the tested cases—CTSat0-A1, CTSat0-B1 and CTSat0-C1—in the HSR ring can operate correctly. CTSat0-A1 has the fastest trip time while CTSat0-C1 has the slowest trip time.

4.1.2. CTSat0 Test Cases in Category 2: Single-Phase-to-Ground Fault at 70% of the Feeder

The test cases in this subsection are CTSat,0-A2, CTSat0-B2 and CTSat0-C2 with the consideration of ideal CT; of scenarios A (no fault component) B (one faulty fibre), or C (one faulty AMU); and of a single-phase-to-ground fault at 70% of the Rassau feeder.

Similar to the test cases in the above subsection, both MP1 and MP2 in test cases CTSat0-A2 and CTSat-B2 operated correctly. As MP1 operated faster than MP2, only MP1 in CTSat-A2 and CTSat0-B2 is analysed and discussed. However, in test case CTSat0-C2 where AMU was disabled, only MP2 operated correctly. All the IED tripping time distributions of CTSat0-A2, CTSat0-B2 and CTSat0-C2 and the corresponding quartile plots under single-phase-to-ground fault tests are calculated.

The most useful results are the quantile plots, which display how the tested IED tripping time performances are close to the predicted normal distribution. The quantile plots for test cases CTSat0-A2, CTSat0-B2 and CTSat0-C, as shown in Figure 14, are analysed and discussed.

By comparing the R-squared values of 0.8328 in Figure 14a, 0.7063 in Figure 14b and 0.6005 in 14c, the results show that test case CTSat0-A2, with an R-squared value of 0.8328, has the best match to its predicted normal distribution. Test case CTSat0-C2, with an R-squared value of 0.6005, has the worst match to its predicted normal distribution.

Comparisons of All Scenarios CTSat0-A2, CTSat0-B2 and CTSat0-C2

The means and standard deviations of the trip times for scenarios CTSat0-A2, CTSat0-B2 and CTSat0-C2 for the single-phase-to-ground fault tests are summarised in

Table 5. Trip times for single-phase-to-ground fault.

Scenario	Description	Mean Trip Time [s]	Std. Deviation [s]
CTSat0-A2	Ideal CT, no faulty comp, 1Ph-G fault	0.02556	0.0041
CTSat0-B2	Ideal CT, 1 faulty fibre, 1Ph-G fault	0.02675	0.0055
CTSat0-C2	Ideal CT, 1 faulty AMU, 1Ph-G fault	0.03758	0.0028

. All trip times are sorted from the fastest to the slowest in ascending order, as listed in Figure 15.

As seen from Figure 15, the best IED tripping time is in the case of CTSat0-A2 as there is no faulty component in the HSR ring. Test case CTSat0-B2 has a very close IED tripping performance to case CTSat0-A2. This confirms that the impact of a fibre broken in the HSR ring on IED tripping performance is small. The worst performance is in case CTSat0-C2. This shows that AMU failure will stop MP1 from operating correctly. However, MP2—distance protection in the HSR ring can still operate correctly. This results in a slower IED tripping time. The studies show that the impact of the HSR ring with fibre and equipment failure on MP1 or MP2 tripping time performance is slight.

4.1.3. CTSat0 Test Cases in Category 3: Line-to-Line Fault at End (100%) of the Feeder

The test cases in this subsection are CTSat0-A3, CTSat0-B3 and CTSat0-C3 with the consideration of ideal CT; scenarios A (no fault component), B (one faulty fibre) or C (one faulty AMU); and the line-to-line fault at the end (100%) of the Rassau feeder.

These test cases for line-to-line faults at the end (100%) of feeder have a significant effect on the MP2—distance protection scheme as the fault is outside the 80% Zone 1 setting. In these test cases, only MP2—distance protection Zone 2 can detect the line-to-line fault at the end (100%) of the feeder. The MP2 Zone 2 time setting is at 200 ms, because it does not encroach into Zone 1 in a downstream relay.

Similar to the test cases in categories 1 and 2, both MP1 and MP2 in test cases CTSat0-A3 and CTSat-B3 operated correctly. As MP1 operated faster than MP2, only MP1 in CTSat-A3 and CTSat0-B3 is analysed and discussed. However, in test case CTSat0-C3, where AMU was disabled, only MP2 operated correctly. For the line-to-line fault at the end of the feeder, MP2 Zone 2 showed a delayed trip time performance. All the IED tripping time distributions of CTSat0-A3, CTSat0-B3 and CTSat0-C3 and their corresponding quartile plots under the line-to-line fault at the end of the feeder were calculated.

The MP1 tripping time quantile plots for test cases CTSat0-A3 and CTSat0-B3 and the MP2 tripping time quantile plots for case CTSat0-C3 were analysed and are shown in Figure 16.

As seen from Figure 16a,b, the MP1—differential protection tripping times operate faster than those of MP2—distance protection for both test cases CTSat0-A3 and CTSat0-B3, at about 25 ms. In test case CTSat0-C3, MP1 is disabled due to a faulty AMU, and MP2 still operates correctly. However, MP2 Zone 2 operates to the line-to-line fault at the end of the feeder. As a result, MP2 Zone 2 gives a mean trip time of 246 ms.

Comparisons of All Scenarios: CTSat0-A3, CTSat0-B3 and CTSat0-C3

For all the test cases, CTSat0-A3, CTSat0-B3 and CTSat0-C3, the means and standard deviations of the IED trip time are summarised in

Table 6. Trip times for line-to-line fault.

Scenario	Description	Mean Trip Time [s]	Std. Deviation [s]
CTSat0-A3	Ideal CT, no faulty comp, line-to-line fault	0.02497	0.0035
CTSat0-B3	Ideal CT, 1 faulty fibre, line-to-line fault	0.02580	0.0032
CTSat0-C3	Ideal CT, 1 faulty AMU, line-to-line fault	0.2457	0.0165

. The tripping times are sorted from the fastest to the slowest in ascending order, as indicated in Figure 17. The results show that the impact of faulty fibres in the HSR ring on IED tripping time is slight since MP1 operates correctly. However, the impact of the faulty AMU plus the fault location on MP2 tripping time is significant. The faulty AMU stops MP1 from operating correctly. If the fault location is above 80% of the feeder, MP2 Zone 2 is forced to operate, hence the longer tripping time delay.

4.2. Simulated CT Saturation Tests in HSR

4.2.1. CT Saturation Test Cases in Fault Category 1: Three-Phase-to-Ground Fault at 50% of the Feeder

A large number of tests in category 1 (three-phase-to-ground faults at 50% of the feeder) with the considerations of light CT saturation and deep CT saturation were conducted. The light and deep CT saturation conditions are specified in Table 1.

The test cases with the consideration of CT saturation in fault category 1 include:

• CTSat1-A1 (light CT saturation, no faulty components, 3Ph-G fault at 50% of the feeder);
• CTSat2-A1 (deep CT saturation, no faulty components, 3Ph-G fault at 50% of the feeder);
• CTSat1-B1 (light CT saturation, one faulty fibre, 3Ph-G fault at 50% of the feeder);
• CTSat2-B1 (deep CT saturation, one faulty fibre, 3Ph-G fault at 50% of the feeder);
• CTSat1-C1 (light CT saturation, one faulty AMU, 3Ph-G fault at 50% of the feeder);
• CTSat2-C1 (deep CT saturation, one faulty AMU, 3Ph-G fault at 50% of the feeder).

When conducting test cases CTSat1-A1 and CTSat2-A1, it was found that both MP1 and MP2 operated correctly. As MP1 operates faster than MP2, only MP1 is considered and discussed.

When conducting test cases CTSat1-B1 and CTSat2-B1, it was also found that both MP1 and MP2 operated correctly. Hence, MP1 is considered and discussed.

When conducting test case CTSat1-C1, MP1 stopped working correctly since no sample values could be read from the failed AMU, but MP2 still operated correctly. Hence, only MP2 is considered and discussed.

When conducting test case CTSat2-C1, it was found that both MP1 and MP2 stropped operating correctly. This is due to the impact of the deep CT saturation on both MP1 and MP2. Hence, no results for this test case could be analysed and discussed.

All the IED tripping time distributions of CTSa1-A1, CTSa2-A, CTSat1-B1, CTSat2-B1 and CTSat1-C1 and their corresponding quartile plots under the three-phase-to-ground fault at 50% of the feeder were calculated.

The MP1 tripping time quantile plots for test cases CTSa1-A1, CTSa2-A1, CTSat1-B1 and CTSat2-B1 and the MP2 tripping time quantile plots for case CTSat1-C1 were analysed and are shown in Figure 18.

From the results shown in Figure 18a,b, the impact of CT saturation on MP1 is small if there is no faulty component in the HSR ring. Similarly, as seen from Figure 18c,d, the impact of CT saturation on MP1 is small if one faulty fibre is in the HSR ring. However, the results in Figure 18e show that the impact of CT saturation on MP2 is significant. When CT is saturated, the impedance seen by MP2 is too high; it falsely categorises the fault from Zone 3, which significantly delays the tripping time to an average of 400 ms.

Comparisons of Scenarios CTSat1-A1, CTSat1-B1 and CTSat1-C1

When considering CT saturation at the local end, the trip times of CTSat1-A1, CTSat1-B1 and CTSat1-C1 are sorted from the fastest to the slowest in ascending order, as shown in Figure 19a. It can be seen that the trip times for scenarios CTSat1-A1 and CTSat1-B1 show good trip times, with a mean trip time of 25 ms; this is because MP1 under light CT saturation operates correctly. For CTSat-C1, the impact of even light CT saturation on MP2 is significant. For MP2, the fault impedance in Zone 3 has a significant average tripping time delay of 400 ms.

Comparisons of Scenarios CTSat2-A1 and CTSat2-B1

When considering deep CT saturation (Sat2), the impact of a faulty AMU + deep CT saturation on both MP1 and MP2 for the CTSat2-C1 case is significant as both MP1 and MP2 do not operate correctly. Hence, no results were obtained for the CTSat2-C1 case. As a result, only CTSat2-A1 and CTSat-B1 are considered and compared. The MP1 trip times of CTSat2-A1 and CTSat2-B1 are sorted from the fastest to the slowest in ascending order, as shown in Figure 19b. It can be seen that the trip times in both case CTSat2-A1 and CTSat2-B1 are similar to those in both CT saturation cases (light and deep CT saturation), because MP1—differential protection can operate correctly. The impact of one faulty fibre or communication link on the protection scheme is small.

4.2.2. CT Saturation Test Cases in Fault Category 3: Line-to-Line Fault at End of the Feeder

Similarly, high-voltage (HV) tests in category 3 (line-to-line fault at end of the feeder) with the considerations of either light or deep CT saturations were conducted. The test cases under the light and deep CT saturation conditions are analysed and discussed as shown below:

• CTSat1-A3 (light CT saturation, no faulty components, L-L-G fault at 100% of the feeder);
• CTSat2-A3 (deep CT saturation, no faulty components, L-L-G fault at 100% of the feeder);
• CTSat1-B3 (light CT saturation, one faulty fibre, L-L-G fault at 100% of the feeder);
• CTSat2-B3 (deep CT saturation, one faulty fibre, L-L-G fault at 100% of the feeder);
• CTSat1-C3 (light CT saturation, one faulty AMU, L-L-G fault at 100% of the feeder);
• CTSat2-C3 (deep CT saturation, one faulty AMU, L-L-G fault at 100% of the feeder).

When conducting test cases CTSat1-A3 and CTSat2-A3, both MP1 and MP2 operated correctly. As MP1 operates faster than MP2, only the MP1 tripping time performance was obtained and analysed.

Similarly, when conducting test cases CTSat1-B3 and CTSat2-B3, both MP1 and MP2 operated correctly. Hence, only the MP1 tripping time performance was obtained and analysed.

When conducting test case CTSat1-C3, MP1 did not work correctly due to the disconnected AMU2—Vendor 1; however, MP2 still operated correctly. Hence, only MP2’s performance was obtained and analysed.

When conducting test case CTSat2-C3, it was found that both MP1 and MP2 stropped operating correctly. This is due to the effect of deep CT saturation on both MP1 and MP2. No results for this test case could be obtained and analysed.

The IED tripping time distributions of CTSa1-A3, CTSa2-A3, CTSat1-B3, CTSat2-B3 and CTSat1-C3 and their corresponding quartile plots under the line-to-line fault at the end of the feeder are calculated.

The MP1 tripping time quantile plots for test cases CTSa1-A3, CTSa2-A3, CTSat1-B3 and CTSat2-B3 and the MP2 tripping time quantile plots for case CTSat1-C3 were analysed and are shown in Figure 20.

Comparisons of CTSat1-A3, CTSat1-B3 and CTSat1-C3

When considering CT saturation, the MP1 trip times of CTSat1-A3 and CTSat1-B3 and the MP2 trip time of CTSat1-C3 are sorted from the fastest to the slowest in ascending order, as shown in Figure 21a. It can be seen that the trip times for case CTSat1-A3 and CTSat1-B3 are good, with a trip mean time of less than 26 ms; this is because MP1 under light CT saturation operates correctly. For CTSat-C3, the impact of even light CT saturation on MP2 is significant. For MP2, the fault impedance in Zone 3 has a significant average tripping time delay of 400 ms.

Comparisons of CTSat2-A3, CTSat2-B3

Since the impact of a faulty AMU + deep CT saturation on both MP1 and MP2 for case CTSat2-C3 is significant as both MP1 and MP2 do not operate correctly, no results were obtained for case CTSat2-C3. Hence, only CTSat2-A3 and CTSat2-B3 are considered and compared. When considering deep CT saturation (Sat2), the MP1 trip times of CTSat2-A3 and CTSat2-B3 are sorted from the fastest to the slowest in ascending order, as shown in Figure 21b. The trip time in both case CTSat2-A3 and CTSat2-B3 are similar, as in both cases, for CT saturation at both ends (local and remote), MP1—differential protection can operate correctly.

4.3. Comparison Studies

From the test results provided in Section 4.1 and Section 4.2, all combinations of CT saturation cases and communication fault scenarios were analysed, and are listed in

Table 7. Trip signals obtained for CT saturation tests for all types of faults applied.

	Saturation	Case Sat0: No Saturation	Case Sat1: Lightly CT-Saturated	Case Sat 2: Deeply CT-Saturated
Communication Faults
Scenario A: full architecture		Trip	Trip	Trip
Scenario B: 1 faulty fibre		Trip	Trip	Trip
Scenario C: 1 faulty AMU		Trip	Trip	-

and

Table 8. Trip signals obtained for CT saturation tests for all types of faults applied.

	Saturation	Case Sat0: No Saturation	Case Sat1: Lightly CT-Saturated	Case Sat 2: Deeply CT-Saturated
Communication Faults
Scenario A: full architecture		0.02634	0.02405	0.02394
Scenario B: 1 faulty fibre		0.02779	0.02500	0.02429
Scenario C: 1 faulty AMU		0.02512	0.41187	-

.

As seen from Table 7 and Table 8, the IED in HSR configurations are able to send trip signals for most testing conditions except for scenario C (one faulty AMU) in the deep CT situation. The problem occurs when both CTs are deeply saturated and there is one faulty AMU. One failure of the AMUs disables the differential protection, and deep CT saturation also makes it very hard to detect the faults over such a distance.

The results show that the differential protection scheme in the HSR configuration can operate at about 25 ms in both scenario A and B under ideal CT, and under light and deep CT saturation conditions. The distance protection scheme in the HSR configuration can operate both scenario A and B under ideal CT, under light and deep CT saturation conditions, and can also operate scenario C under ideal CT and light CT saturation. Since the distance algorithm relies on both current and voltage signals, the deep CT saturation affects the current measurement, which results in no trip signal for scenario C under deep CT saturation.

4.3.1. Comparisons of Three-Phase-to-Ground Faults

The IEDs’ function and interoperability performance for three-phase-to-ground faults under three different CT saturation conditions is shown in Figure 22a–c, respectively. An interesting observation can be made from Figure 22a,b, where it can be seen that for Scenarios A1 and B1, the trip times for the ideal CT saturation case are slightly higher. This indicates that the differential protection scheme setup is sensitive enough to detect the faults under slight and deep CT saturations. The increase in burden impedance for simulating CT saturation causes the operating point of the CT to move closer to the saturation zone. Hence, the current measurement at the local and remote ends may be more unbalanced, and there is more spill current between the local and remote ends. This results in faster tripping. Of course, this higher spill current or current imbalance can be desensitised by setting a higher spill current threshold to ensure the dependability of the trip signal. From Figure 22c, it can be seen that the distance relay under the ideal CT situation can operate Zone 1, but under CT saturation, it can only operate Zone 3.

4.3.2. Comparisons of Line-to-Line Ground (L-L-G) Faults

The IEDs function and interoperability performance for line-to-line ground faults under three different CT saturation conditions are shown in Figure 23. As can be seen, the trip times for Scenario A3 and B3 under CT saturation and two different CT saturation cases perform as expected. Similarly, it can be seen that for Scenario A3 and B2, the trip times for the ideal CT saturation case are slightly higher than that for CT saturation cases because different the protection settings are more sensitive on the spill current generated by CT saturations.

It can also be seen from Figure 23c that the distance IED only operates under CT saturation. Distance Zone 2 operates under slight CT saturation, and Zone 3 operates under deep CT saturation.

4.4. No CT Saturation Tests in PRP

The ideal CT secondary outputs from the substation RTDS model were via voltage and current amplifiers, as direct CTSat0 injections into the AMUs. There are three categories of direct CTSat0 tests corresponding to the type of fault applied within the RTDS network model: three-phase-to-ground, single-phase-to-ground and line-to-line faults.

4.4.1. CTSat0 Test Case Category 1: Three-Phase-to-Ground Fault at 50% of the Feeder

Scenario CTSat0-D1 (Ideal CT, No Faulty Components, 3ph-G Fault at 50% of the Feeder)

This scenario regarding the operational status of the PRP configuration is referred to as Scenario D (as described in Table 3 in Section 2.3). A total of 50 tests were conducted to assess the trip times for a three-phase-to-ground fault in the middle of the Rassau feeder, as shown in Figure 4. Firstly, the full PRP architecture (comprising no faulty devices or fibres) was evaluated. In this scenario, both MP1 (differential protection) and MP2 (distance protection) operated correctly. Similarly, as MP1 (differential) operated faster than MP2 (distance), only MP1’s performance is presented and discussed here.

As in Section 4, it is assumed that the behaviour of the projection tripping time follows a normal distribution. Hence, all 50 tests were calculated using normal distribution formulas. A mean for this scenario of 0.0263 and a standard deviation (σ) of 0.0048 were obtained. Additionally, the frequency of the number of tests against the tripping times can be regarded as the IED tripping time distribution, as shown in Figure 24a. The quantile plot of all the trip times against the calculated normal distributions was obtained, as shown in Figure 24b.

If the measured trip times indeed follow a normal distribution, the samples marked with blue crosses in Figure 24b should form a straight line. One can see from Figure 24b that, with the exception of a few high trip times and a few low trip times, all the recorded data follow a normal distribution. Statistically, this is referred to as a “heavy tails” distribution. The “heavy tails” distribution might be a result of uncertainties in communication channel delays associated with data input and output from relays or RTDS. It is important to remember that analogue signals are the output of RTDS gigabit transceiver analogue output (GTAO) cards, while RTDS digital input channels are used to receive the trip signals from the Switch Control Units. The trip time is ultimately computed within the RTDS simulation environment.

Scenario CTSat0-E1 (Ideal CT, One Faulty AMU, 3ph-G Fault at 50% of the Feeder)

In order to verify the reliability of the PRP communication architecture, the Vendor 2 AMU was disconnected and the assessment of the trip times was repeated. This particular disconnection prevents the whole differential protection scheme from operating. This operational state of the PRP architecture is referred to throughout this paper as Scenario E, and is also mentioned in Table 3. The MP2 tripping time distribution for case CTSat0-E1 was calculated as shown in Figure 25a. The probability distribution of the trip time values for this particular scenario (scenario E) was assessed using a quantile plot, as shown in Figure 25b. Except for a few stray samples, all the trip time values follow a normal probability distribution.

Scenario CTSat0-F1: One Faulty AMU and Fibre in the PRP Configuration

The state of the future intelligent transmission network substation architecture was further intentionally worsened for testing purposes by removing the fibre optics cable connecting the main Vendor 1 digital merging unit with Station Bus 1. After this fibre optics cable was disconnected, the Vendor 1 digital merging unit still had, according to the PRP architecture design requirements, a second connection for multicasting the trip signal, i.e., the connection to Station Bus 2. This operational state of the future intelligent transmission network substation architecture is referred to as Scenario F (i.e., a faulty AMU and one faulty fibre), as given in Table 3. For Scenario F, The MP2 tripping time distribution for case CTSat0-F1 was calculated as shown in Figure 26a. The quantile plot against a theoretical normal distribution is presented in Figure 26b. It can be seen that the trip time values for Scenario F also follow a heavy-tailed normal distribution.

The means and standard deviations of the trip times for all the scenarios are given in

Table 9. Trip times for 3-phase-to-ground fault.

Scenario	Description	Mean Trip Time [s]	Std. Deviation [s]
CTSat0-D1	No faulty device in architecture	0.0263	0.048
CTSat0-E1	1 faulty AMU	0.0321	0.070
CTSat0-1F	1 faulty AMU + 1 disconnected fibre connection	0.0307	0.043

, while all the trip time values are sorted and indicated for the all scenarios in Figure 27.

As seen from both Table 9 and Figure 27, the trip time in the case of the full architecture is the fastest. This is to be expected, because within a full architecture, both main protection schemes (differential and distance) can operate; hence, the fastest fault detection and the fastest trip response can be achieved. Following the disconnection of the Vendor 2 AMU in Scenario E, the trip time increases by 20%. This increase is due to the fact that the disconnection that occurs in Scenario E incapacitates the differential protection scheme, hence allowing only the distance protection scheme to operate. The distance protection scheme is slightly slower than the differential scheme, hence the increase in trip time. Following the same logic, one might expect that increasing the number of issues within the communication infrastructure can result in an increase in trip time. However, further removing the fibre connection between the Vendor 1 DMU and Station Bus 1, as part of Scenario F, results in a decrease in trip time when compared to Scenario E. While such behaviour might seem counterintuitive, it is to be expected, since the localised traffic within the communication network is cut in half due to the loss of the optical fibre connection. While such a disconnection means a loss of redundancy, and therefore, a reduction in reliability, it also reduces the level of network traffic and slightly improves communication speed. However, the trade-off between the loss of reliability and the increase in communication speed is not worth it. Tests in Scenario F were performed in order to demonstrate that the PRP communication architecture can still function in the worst-case scenarios.

4.4.2. CTSat0 Test Cases in Category 2: Single-Phase-to-Ground Fault at 70% of the Feeder

Although they are good textbook examples for testing the future intelligent transmission network substation architecture, three-phase faults are quite rare. Hence, the tests for three-phase line-to-ground faults were repeated for single-phase-to-ground faults applied at 70% of the Rassau feeder (see Figure 4). For Scenario D (fully functional PRP architecture), the quantile plots against a theoretical normal distribution are given in Figure 28a. The probability distribution can be interpreted as bi-modal, as there are a significant number of high-value samples that do not follow the straight-line pattern. The quantile plot in Figure 28a looks as if the samples are coming from two different normal distributions, with different means.

For Scenario E, the quantile plot against a normal distribution is presented in Figure 28b. The same bi-modal distribution pattern as in Figure 28a can be identified in Figure 28b. This bimodal pattern can be associated with processing cycles within a protection relay. When a fault is detected, the relay generates a “Protection Start” signal. If there are no operational delays associated with protection coordination, the relay should ideally produce the “Protection Trip” signal instantaneously. However, “Protection Start” and “Protection Trip” can be processed and generated within two different operating cycles of a protection relay. This can explain the two different probability distributions of the trip time, irrespective of whether “Protection Start” and “Protection Trip” belong to the same operational relay cycle or not. The quantile plot for scenario F is shown in Figure 28c, and it exhibits the same bimodal pattern.

For all three scenarios, the means and standard deviations of the trip times in the case of a single-phase-to-ground fault are summarised in

Table 10. Trip times for single-phase-to-ground fault.

Scenario	Description	Mean Trip Time [s]	Std. Deviation [s]
CTSat0-D2	No faulty device in architecture	0.0277	0.0057
CTSat0-E2	1 faulty AMU	0.0355	0.0050
CTSat0-F2	1 faulty AMU + 1 disconnected fibre connection	0.0320	0.0058

, while all the trip times, sorted in ascending order, are indicated in Figure 29.

The same bimodal pattern observed in the quantile plot from Figure 28 can be seen in Figure 29. With only several samples as exceptions, the trip times in all three scenarios (D, E and F) follow the same pattern as in the case of the three-phase-to-ground fault: TripTimeD < TripTimeF < TripTimeE.

4.4.3. CTSat0 Test Cases in Category 3: Line-to-Line Fault at End (100%) of the Feeder

In the case of a the line-to-line fault, the fault was applied at the end of the feeder. This case has significant implications for differential protection. In test cases I and II, the fault was located in the middle and at 70% of the feeder, and therefore, within the reach of the Zone 1 distance protection scheme. The reach of Zone 1 protection was 80% of the feeder length. Hence, for the line-to-line fault at the end of the feeder, the distance protection function will issue a Zone 2 trip. The Zone 2 trip is usually 200 ms delayed, so it does not encroach into Zone 1 of a downstream relay. The theoretical quantile plots against a normal probability distribution can be viewed, respectively, for Scenarios D, E and F in Figure 30.

As can be seen in Figure 30, no bimodal pattern can be detected anymore, as in the case of line-to-ground faults. The trip times for differential protection are quite stable and are distributed around a value of approximately 20 ms. For differential protection, it can be clearly observed from Figure 30b,c that the trip time has gone up to a value over 220 ms. This is because the 200 ms time delay is associated with distance protection trips in Zone 2. Any irregularity in probability distribution at the 20 ms level becomes much more difficult to detect when a constant time delay of 200 ms is superimposed on all the data samples.

For all three scenarios (D, E and F), the means and standard deviations of the trip time probability distributions are summarised in

Table 11. Trip times for line-to-line fault.

Scenario	Description	Mean Trip Time [s]	Std. Deviation [s]
CTSat0-D3	No faulty device in architecture	0.0245	0.0030
CTSat0-E3	1 faulty AMU	0.2263	0.0013
CTSat0-F3	1 faulty AMU + 1 disconnected fibre connection	0.2262	0.0014

. Furthermore, all the sorted trip time values are given in Figure 31. One can see from both Table 11 and Figure 31 that the trip times for Scenarios D and E are almost identical, with Scenario F continuing to be slightly faster than the two. This is associated with a loss in duplicate traffic within the PRP architecture, although there is no question that the increased redundancy and reliability in Scenario E is to be preferred. The pattern—TripTimeD < TripTimeF < TripTimeE—is preserved in this case, as well.

4.5. Tests with CT Saturation

Due to similarities in the protection trip times in the cases of three-phase-to-ground and single-phase-to-ground faults, only the following faults were analysed for CT saturation:

-

Test case IV: three-phase-to-ground fault

-

Test case V: line-to-line fault.

For both test cases IV and V, all the combinations of the CT saturation cases and communication fault scenarios were analysed. As can be seen from

Table 12. Trip signals obtained for CT saturation tests for all types of faults applied.

	Saturation	Case Sat0: No Saturation	Case Sat1: Lightly CT-Saturated	Case Sat 2: Deeply CT-Saturated
Communication Faults
Scenario A: full architecture		Trip	Trip	Trip
Scenario B: 1 comm. fault		Trip	Trip	-
Scenario C: 2 comm. faults		Trip	Trip	-

, a trip signal was obtained for every testing condition except for two. The two problematic situations occur when both CTs are saturated and there are issues within the communication network: Scenario E and F. It is therefore clear that the differential protection scheme employed within the future intelligent transmission network substation, which operates only in Scenario D, is quite robust and can still generate a trip signal under challenging CT saturation conditions.

4.5.1. CT Saturation Test Cases in Fault Category 1: Three-Phase-to-Ground Fault at 50% of the Feeder

In the case of the three-phase-to-ground fault without saturation, 50 tests were performed for each scenario in

Table 13. Mean trip times obtained for CT saturation tests for 3-phase-to-ground faults.

	Saturation	Case Sat0: No Saturation	Case Sat1: Lightly CT-Saturated	Case Sat 2: Deeply CT-Saturated
Communication Faults
Scenario A: full architecture		0.0263 s	0.0240 s	0.0242 s
Scenario B: 1 comm. fault		0.0321 s	0.3757 s	-
Scenario C: 2 comm. faults		0.0306 s	0.3755 s	-

that generated trip signal. The sorted trip times for the no-saturation case, already given previously in Figure 27, are given again here in Figure 32a in order to facilitate the comparison between saturated and non-saturated conditions.

The sorted trip times in the case of light CT saturation are given in Figure 32b. While the trip times for Scenario D continue to have a low value similar to the no-saturation case given in Figure 32a, there is a significant increase in trip time for scenarios E and F.

It is important to remember that in Scenarios E and F, the differential protection scheme is inactive due to the faulty AMU at one end; thus, only the distance and overcurrent protection schemes can operate. In the case of CT saturation, the impedance seen by the relay is quite high, especially due to the modification brought to the CT burden impedance described in Section 2.2; thus, distance protection falsely categorises the fault as stemming from Zone 3. Hence, overcurrent protection is the first to operate for Scenarios B and C under CT saturation conditions. In the interest of completeness, the trip times were once more compared using scenario-based categorisation, as shown in Figure 32c–e for Scenarios D, E and F, respectively.

An interesting observation can be made from Figure 32c, where it can be seen that for Scenario D, the trip times for the ideal CT saturation case are the highest. This is related to the fact that the differential protection scheme is sensitive enough to detect the fault even in the CT saturation case, and also to the way CT saturation was modelled in Section 2.2. The increase in burden impedance causes the operating point of the CT to move closer to the saturation zone. Hence, a current imbalance, which represents the trigger for the differential protections scheme, can be reached faster due to an earlier occurrence of saturation. Of course, such current imbalance still has to surpass the threshold value of the differential protection scheme, in order for the trip signal to be generated. Hence, the differences in trip times are not significant (in the order of ms). The mean trip times for each testing scenario are given in Table 13.

4.5.2. CT Saturation Test Cases in Category 3: Line-to-Line Fault at End of the Feeder

The results for the line-to-line fault from a CT saturation perspective indicate a similar trip time response to that of the future intelligent transmission network substation architecture. Similar to test case IV, the previous results for the ideal CT saturation case were obtained and are indicated again here in Figure 33a. As shown in Figure 33a, for Scenarios E and F, distance protection operates and the differential protection function is blocked. Since the line-to-line fault is applied at the end of the feeder, the relay generates a Zone 2 trip; thus, there is a 200 ms delay between Scenario D and Scenarios E and F.

For case CTSat1 (lightly CT-saturated), the results are shown in Figure 33b. These results are very similar to the equivalent plot for the three-phase-to-ground fault in Figure 33a. In the interest of completeness, the analysis that focused on communication fault scenarios is given for Scenarios D, E and F, respectively.

The mean trip times for all the test scenarios are given in

Table 14. Mean trip times obtained for CT saturation tests for line-to-line Faults.

	Saturation	Case Sat0: No Saturation	Case Sat1: Lightly CT-Saturated	Case Sat 2: Deeply CT-Saturated
Communication Faults
Scenario A: full architecture		0.0245 s	0.0250 s	0.0262 s
Scenario B: 1 comm. fault		0.2263 s	0.3920 s	-
Scenario C: 2 comm. faults		0.2261 s	0.3913 s	-

.

5. Conclusions

In order to analyse the effectiveness of the implemented data network redundancy on the protection and control (P&C) scheme functionality, an experimental testing bed was setup. The assessments of two different redundancy networks, HSR and PRP, and their faulty components’ impact on the P&C scheme’s functionality and performance were conducted. The P&C scheme was based on two main protection and control strategies.

The network faults, with different levels of CT saturations in the RTDS network model, were simulated and fed into the laboratory setup through the use of analogue output cards and amplifiers. The experimental tests for the different faulty components, such as intentional equipment failures in either the HSR or PRP data network Process Bus under different network fault conditions, were conducted. Since each experimental test was very time consuming, 50 tests were selected for each test case scenario based on the assumption for the experimental tests that followed a normal distribution. Then, the R-squared linear regression method was used to statically quantify how close the actual tests were to the assumed normal distribution model. In all the tests, most test case scenarios had an R-squared value of more than 90%. However, the worst test case scenario had an R value of 58%. This explains that approximately 58% of the observed test samples can be explained by the predicted normal distribution model’s inputs.

In the HSR tests, different faulty components and CT saturation levels were considered. For the tests on faulty components in HSR with ideal CT, the comparison results show that the different faulty components in the HSR ring configuration do have an impact on P&C performance. A faulty AMU is more severe than a faulty fibre. However, the faulty components do not affect the P&C functionality or interoperability. For the tests on faulty components in HSR with light CT saturation, the results show that both the faulty component and CT saturation do have an impact on MP1, but MP2 zone 3’s functionality performed correctly. For the tests on faulty components in HSR with deep CT saturation, the results show that faulty fibre + deep CT saturation do not affect the protection scheme, but faulty AMU + deep CT saturation do affect protection scheme, i.e., both MP1 and MP2 fail to operate.

In the PRP tests, various different faulty components and different CT saturation levels were considered. For the tests on faulty components in PRP with ideal CT, the test results show that the protection scheme with no faulty components performs better than those under faulty component conditions. Under faulty component conditions in PRP, the protection scheme selects the best route to trip. For the tests on faulty components in PRP with light CT saturation, the main effect is the faulty AMU, which stops MP1 (differential) from operating and forces MP2 (distance) into backup protection (i.e., zone 2). For the tests on faulty components in PRP with deep CT saturation, the main effect is the faulty AMU, which can stop both MP1 and MP2 in the protection scheme and cause failure to trip.

The experimental results can provide valuable information for power utilities and manufacturers to optimise their substation architecture and products under different conditions. Based on the successful implementation of the future intelligent transmission network substation test, our future work will focus on substation network visualisation and cyber security issues in the substation.