Electric grid hardening and resiliency: Part I, resiliency and safety

State of Louisiana is among the most industry populated states in USA and it provides the means for transferring approximately 25% of the US energy needs to the rest of the nation.

Go to the profile of Parviz Rastgoufard
Jul 25, 2017
0
0
Upvote 0 Comment

Authors: Parviz Rastgoufard; Ittiphong Leevongwat; Rastin Rastgoufard

Abstract

Part I of the article summarizes research findings used for electric grid resiliency and safety. Resiliency and safety aspects of the grid, appearing as two sections of Part I are based on development of algorithms using grid topology in steady-state mode of operation and provide information for designing a more robust, efficient system while ensuring safety. Planning for robust grids using dynamic models and real-time response of equipment to system disturbances is addressed in Part II as an independent complementary sequel to Part I. In this article, grid resiliency is defined by reduction of the load down-time in Gulf Coast states, and it is improved by implementing the results from studying the system response to impact of tropical storms on the transmission system. Grid safety is measured by correctly identifying fault locations on transmission lines in Gulf Coast states and by providing guidelines for safely deenergising the faulted transmission lines. While the authors' main focus in Part I is on the use of steady-state system topology and modelling, the authors will use system dynamics and real-time modelling and simulation of Independent Pole Operation addressing dynamic aspects of robust system planning for power flow improvement in Part II of the article. The work presented in Part I and Part II of the article is performed at the Entergy-UNO Power and Energy Research Laboratory (PERL) at the University of New Orleans.

Introduction

State of Louisiana is among the most industry populated states in USA and it provides the means for transferring approximately 25% of the US energy needs to the rest of the nation. The energy transfer encompasses collaboration and working of several industries including chemical, petrochemical, pulp and paper, and food processing industrial plants (IPs) all of which depend on use of electricity in their daily operations. While a few of these IPs have their own generating capability, they all rely on continuous and resilient flow of electricity from the national/local electric grid for their electricity consumption and production of goods. There are several electric utilities and municipalities that provide the bulk of electricity for the industry in Louisiana. Reliable and safe operation of electric utilities, not only impacts residents of these states, but also has an impact on the rest of the nation. Due to geographical location of Gulf Coast states, the residence and the industries of these states experience annual tropical storms measured by wind speeds of 75–150 mph some causing electricity interruption and hence, discontinuity of flow of goods produced by the industries to the nation. To reduce the downtime of industries caused by interruption of electricity in response to impact of tropical storms, the researchers at PERL have developed algorithms which are used in conducting numerous simulations to aid IPs in preparing for a potential plant shutdown more accurately. Currently shutdowns during a storm are planned by use of experience and by following tropical storm trajectories that may be too close to the shoreline. The developed algorithms may be used to prepare potential shutdowns in an optimal fashion reducing the overall downtime of IPs in any given geographical area.

Rastgoufard et al. [1] present an overview of the proposed algorithms and cites the relevant literature in modelling and assessments of impact of tropical storms on power system reliability. While modelling of tropical storms and their integration with power system models vary, application of storm models has been seen in a simulation of small-scale power system test cases and lacks utilisation of real-size power system topology. The authors of [2,3] simulate the IEEE Reliability Test System to test their algorithms based on fuzzy logics. The methodology followed in this investigation is based on use of real-size power system topology and actual geographical locations of transmission line poles.

Useful in study of Grid Resiliency, the developed algorithms may be used to quantify the impact of potential tropical storms on separation of IPs in particular, or any node (bus) in the transmission system in general, from the rest of the system. Quantitative knowledge of continuity or discontinuity of electricity to IPs will allow the plant managers to determine the probability of failure/success for the IP operation and to more accurately, rather than randomly, determine the IP's downtime for a specific tropical storm. Accurate determination of an individual downtime for a specific IP corresponding to a specific tropical storm and then using the individual downtimes to determine an average aggregate downtime of IPs located in a geographical area in the Gulf Coast region may be used to numerically measure the impact of specific tropical storms on discontinuity of electricity, and hence, the flow of goods from the region. Use of the results of this investigation in determination of economic losses caused by tropical storms and hurricanes is of vital interest to the researchers and the plant operators in the Gulf Coast region, and has gained substantial importance after hurricane Katrina. As reported in [4], economic losses caused by hurricane Katrina were approximately $96 billion impacting 300,000 homes [5].

As a result of experiencing tropical storms and hurricanes, the electric grid moves away from its normal operating conditions losing the overall system integrity and splitting into several subsystems referred to as ‘islands’. A power grid can be considered as one big ‘island’ where all of the generating units are connected and supplying power to loads through power lines at the highest level of designed reliability. When tropical storms make landfalls, wind gusts cause flying debris and downed trees to trip power lines. Grid islanding is referred to as an event when tripped lines disconnect and separate a smaller portion of the grid from the rest of the grid. If the size of the island is large enough and includes sufficient generating units which are connected to loads within the island, the island may operate on its own as a working subsystem of the entire grid. IPs and residential clients within an island could continue to receive power delivered to them during tropical storms, however at a lower level of reliability. Occasionally an IP which is equipped with co-generation facility may be operational on its own as an island without connection to the electric grid. Determination of islands plays an important part of our strategy for determining weak (or soft) locations in the grid that need hardening.

Once the weak locations in the system are identified, they need to be hardened or strengthened by adding or upgrading transmission lines and towers, transformers, protection system equipment, and generating units in order to maintain reliability of the grid, even in the case of abnormal operation during storm seasons. The proposed methodology is also useful in aiding system planners to assess their strategy for long-term expansion plans and for determining short-term decisions for retiring old generating units and installations of new equipment such as flexible AC transmission system (FACTS) devices.

Mathematical models to assess and to manage economic disasters are outlined in [4] with introduction of ‘resiliency matrix’ and dynamic inoperability input–output model based on Leontief input–output model [6,7]. Elements of resiliency matrix in [4] may be related to the results from simulating Gulf Coast region electric grid, the probability of continuity of electricity to IPs, and the size of system islanding and outflow of goods.

Evaluation of hurricane impact on failure rate of transmission lines using fuzzy logic expert systems is covered in [3]. The authors in [3] use the IEEE Reliability Test System and partition the system into two parts along the 138 and 230 kV voltage levels rather than testing their proposed methodology on a real-size power system. A relatively large size power system is used by the authors of [8] by proposing outage metrics in IEEE 859-bus transmission systems [9].

In this article, we will present the results of applying the proposed methodology in determining impact of 76 real and hypothetical simulated hurricane models on transmission system and continuity of electricity to IPs in the Gulf Coast region. The real-size power system topology consisting of 50,000 buses is then integrated with system connectivity information, system historical wind impact data, and the tropical storm model generated by Hurrtrak [10]. Using a real-size power system topology, the proposed methodology, and the developed algorithms three real-size power system case studies which we refer to as industrial plant (IP); IP-A, IP-B, and IP-C differentiated by their actual longitude and latitude are reported in the subsequent sections.

Grid resiliency

An objective of our investigation is to determine the impact of seasonal hurricanes and tropical storms on the security and delivery of electricity in Gulf Coast states’ electric grid to industrial customers producing goods for the region and the rest of the nation. The Gulf Coast in general and the state of Louisiana in particular include a relatively high number of IPs that are connected to the electric grid, and continuity of electricity to these IPs is of vital importance in flow of energy from Gulf Coast states to the rest of the nation. Fig.  shows a map of Louisiana Electric Grid, which is a portion of the eastern grid consists of 55,000 buses, 52,000 transmission lines, 18,000 transformers, and 8000 generators. Our methodology in the Planning phase of hardening the system consists of the following steps:

  • Step 1: Identifying and creating a database from actual tropical storms and their characteristics including their velocity and trajectory which occurred over a period of 50 years.
  • Step 2: Determining latitude and longitude of IPs in the Gulf Coast region including their products and services.
  • Step 3: Creating connectivity topology of the actual electric grid of the Gulf Coast region including transmission lines ranging from 69 to 500 kV.
  • Step 4: Developing algorithms that take storm characteristics, system connectivity, and grid information and result in quantitative measurements of the impact of tropical storms and hurricanes on continuity of electricity to any specific IP characterised by its longitude and latitude, and
  • Step 5: Identifying weak locations in the system that need to be hardened in order to achieve a more resilient system.

Fig 1: Louisiana electric grid

The proposed methodology determines ‘islanding’ of a portion of the electric grid from the rest of the grid that may result from the impact of the simulated tropical storms and would provide sufficient information to plant managers for continuing or halting their plant operation in time prior to the landing of tropical storms or hurricanes. This article includes a summary of simulation results of 50 historical and 26 hypothetical tropical storms on selected IPs in the geographical area. To display the usefulness and practicality of the developed algorithms on analysis of a real-size power system, we will include results of three case studies in detail and only outline results of three additional case studies in the state of Louisiana. All six case studies include actual IPs with masked identities for confidentiality reasons. Decision on hardening and resiliency is made by simulating numerous actual and hypothetical storms and identifying their impact on the grid with the intention of ranking system buses from weakest to strongest. The end result will identify weak regions in the system which need to be hardened. Further work on use of the developed algorithms in hardening and development of resilient transmission system in Gulf Coast states will be reported in future publications.

To begin the procedure, we first find a set of storms whose effects we wish to examine. For this study, we selected 50 storms in a period of 60 years, and we shifted three of these actual storms to obtain several hypothetical storms. Table  1 includes a list of 50 actual storms including their name, wind speed, longitude and latitude of their shore landing.

NUMBERNAMEWIND SPEED, MPHLONGITUDE, ° ELATITUDE, ° N
0ALICIA30−90.527.3
1ALLISON30−93.731
2ANDREW65−91.630.5
3BABE30−91.330.2
4BAKER75−88.129.4
5BARBARA35−92.329.4
6BERTHA60−92.329
7BERYL30−91.630.3
8BETSY65−91.830.8
9BILL50−90.330.4
10BONNIE70−92.928.2
11BRENDA35−91.631.1
12CANDY60−97.228.3
13CARMEN75−92.130
14CHANTAL70−93.528.7
15CHRIS40−93.430.8
16CINDY65−94.429.8
17CLAUDETTE40−93.930.3
18DANIELLE30−92.428.8
19DEBBIE30−88.930
20DELIA60−9428.4
21DENNIS70−87.531.3
22EDITH60−91.630.5
23EDOUARD50−92.228.7
24ELENA60−90.431
25ELLA60−93.226.6
26ELOISE95−88.527.3
27ESTHER45−90.530.4
28ETHEL45−8930.7
29FELICE55−92.228.8
30FERN25−90.329.9
31FLORENCE30−90.730.7
32GEORGES85−8930.4
33GUSTAV80−91.529.9
34HANNA45−88.930.1
35HERMINE25−90.830.5
36HILDA60−91.230.2
37HUMBERTO20−91.531.8
38IKE95−93.828.2
39IRENE45−87.929.3
40ISIDORE55−90.429.7
41IVAN115−88.129.3
42JUAN60−91.930.3
43KATRINA110−89.630.2
44LILI65−92.330.2
45MATTHEW30−90.829.9
46OPAL110−87.829.2
47RITA105−93.229.1
48storm-115−9632
49storm-540−91.130.3

Table 1: Fifty historical hurricanes

Information displayed in Table and similar information including storm trajectories and their corresponding temporal information are included in a database for use in conjunction with grid data that includes actual location of poles and transmission line voltage levels.

According to Saffir–Simpson hurricane wind scale, entries of Table show that there are five Category-1 of wind speeds in the range of 74–95 mph, three Category-2 of wind speeds in the range of 96–110 mph, and one Category-3 with the wind speed in the range of 111–129 mph hurricanes. The remaining 41 entries of Table 1 correspond to tropical storms which are not identified as hurricanes using Saffir–Simpson categories. However, we find that tropical storms of lower sustained winds than Category-1 hurricanes could cause damage to the electric grid as highlighted by studying impact of Alicia (entry 0 of Table 1) and Beryl (entry 7 of Table 1) in [1]. In addition to the wind speed, tropical storm trajectory and the location of its landfall have the most impact on the electric grid and the flow of electricity to different geographical locations in general and to specific IPs in particular. This point is thoroughly studied in determining weak and candidate hardening locations in the system.

To fulfil the objectives of Step 2 of the proposed methodology, we have populated our database to include the latitude and the longitude of all IPs as well as their raw material suppliers and service providers which are connected to the Gulf Coast electric grid. In addition to electricity, IPs rely on receiving certain raw materials and services necessary for producing their portion of the overall industry output. Discontinuity of electricity is detrimental to not only the IPs, but also the raw material suppliers and service providers of IPs. The unavailability of raw materials and services will cause discontinuity in flow of goods from affected IPs. In determining candidate hardening locations and system resiliency, we consider raw material suppliers and service providers which are connected to the grid in addition to the IPs.

Step 3 of the methodology includes study of the grid one-line diagram of the Gulf Coast region and in more details the one-line diagram of the study area. A typical one-line diagram of an area in the Gulf Coast region is depicted by Fig 2. One-line diagrams provide us with two important sets of information – the list of all buses (or nodes) and the list of all the lines (or edges) in the network – for building a graph of a power network and for the connectivity graph of the Gulf Coast electric grid.

Fig 2: Power system one-line diagram

Transformer and load information available at each node is not considered in building of the system connectivity graph and is addressed along with load flow studies in the detailed round of system modelling, simulation, and analysis. A sample of a simple connectivity graph providing node and edge information is depicted in Fig 3. As an example, in a case study represented in the results section of the article, node 0 may be connected to node 7 by a 230 kV line and the connection is through two intermediate nodes 2 and 4.

Fig 3: Graph with ten nodes and ten edges

Fig shows the same graph with same nodes, but missing several edges representing the lines that are tripped by storms for the case studies. Compared to Fig , the 230 kV line between nodes 0 and 7 in Fig is no longer in service due to a missing edge connecting nodes 2–4.

Fig 4: Graph with missing edges

A connectivity graph that includes 55,000 buses and 52,000 transmission lines is developed for our study.

Step 4 objectives are satisfied with the development of algorithms that take storm and grid topology as input and produce line tripping – loss of graph edges – due to impact of simulated tropical storms. The algorithms examine the impact of a storm on the transmission system by traversing outward, away from a node – a bus connected to the IP under study – moving from station to station along transmission lines that are projected by the model to remain in service following the storm and its simulated trajectory. The algorithms continue their outward search until an out of service – tripped – line is reached indicating one boundary of the IP that is disconnected from the grid. When all disconnected boundaries of the plant are identified, the search stops and the IP are designated as an ‘islanded’ node (bus, or plant).

Fig shows a track map of U.S. landfalling major hurricanes during 2001–2010 [11]. The trajectory of 50 historical hurricanes with time-stamped locations and intensities was collected for our investigation.

Fig 5: National hurricane centre's deadliest, costliest, and most intense United States tropical cyclones

Step 5 of our investigation is devoted to performing extensive simulations of the electric grid and finding islanding scenarios of all IPs in response to impact of all simulated tropical storms. The result of the extensive simulations is the determination of weak geographical locations in the Gulf Coast grid that require hardening and hence resulting in improvement of system resiliency.

Planning for improved resiliency

We present results of electric grid robustness by studying three IPs connected to the grid. The plants, although masked for confidentiality, are electrically connected to the Louisiana electric grid. Steps 1–4 of the methodology is applied to all three IPs designated as IP-A, IP-B, and IP-C. Step 5 of the methodology is performed by analysing the results similar to the presented case studies results when performed for all nodes in the study area. In Fig 6, the square represents the location of an IP. Circles depict the locations of the point along each hurricane's trajectory that was the closest to the coastline.

Fig 6: Historical storms landfall: industrial A

Historical hurricanes would have caused tight islanding in four cases (shown in red). In one case, the island might have been large enough to contain generation (yellow). There would not have been islanding in the other cases (green). Fig indicates that out of the simulated tropical storm trajectories of various wind speeds, only five of them caused the IP IP-A to island and be separated from the electric grid. These five islands have a different degree of connectivity and are differentiated by outward number of edges selected as 12 based on comparison with IP-A collected historical data. When a number of outward edges are larger than 12, IP-A is capable of staying in service and continue operation by using internal plant electricity generation or by using electric utility generation that is within the island. As indicated by the size of the circles in Fig , tropical storm 43 – indicated on the yellow circle – has wind speed of 95 mph or higher (entry 43 of Table corresponding to Katrina). However, tropical storm 8 (entry 8 in Table  1corresponding to Betsy) with speed of around 65 mph causes a smaller, hence tighter island with connectivity degree of smaller than 12 around IP-A. While wind speed of hurricane Betsy of 65 is smaller than that of Katrina at 110 mph, Betsy causes a tighter island due to its trajectory relative to latitude and longitude of IP-A and its landfall location at the coast. Furthermore, Betsy crosses certain transmission lines of which their tripping would impose a harder constraint on flow of electricity to IP-A. Comparison of simulation results from Betsy with Katrina by IP-A management reveals different levels of preparation prior to occurrence of such tropical storms.

We next focus on presentation of simulation results for IP-B case study. Fig depicts four red circles indicating islands of connectivity degree of 12 or smaller and two yellow islands of connectivity degree of 12 and higher. Katrina – entry 43 of Table – causes a tighter island for IP-B than for IP-A, however the most severe condition for IP-B is caused by Betsy of entry 8 of Table .

Fig 7: Historical storms landfall: industrial B

Fig depicts the results for IP-C with four red and one yellow coloured islands. We note that hurricane Hilda in entry 36 of Table with wind speed of 60 mph causes an islanding of connectivity degree of larger than 12 for IP-C. Results for IP-D and IP-E are shown in Figs and 10, respectively. While results for IP-F are not presented in this study, Table refers to this location for comparative analysis with IPs IP-A to IP-E.

Fig 8: Historical storms landfall: industrial C

Fig 9: Historical storms landfall: location D

Fig 10: Historical storms landfall: location E

HURRICANESTUDY LOCATION
IDNAMEABCDEF
2ANDREW142
8BETSY9000
10BONNIE5
13CARMEN121100
33GUSTAV9900
36HILDA1214155
37HUMBERTO7
38IKE41
42JUAN
43KATRINA13030
44LILI26
47RITA41
total hurricanes565346
islanding score55481830137
hardening index1183.6103.31.2

Table 2: Impact of historical hurricanes by study location

The entries in Table correspond to tropical storms tabulated in Table with respect to IPs IP-A to IP-F. Hurricane Andrew, for example causes islands with connectivity edges of 14 and 2 for IP-B and IP-F, respectively, while the remaining 4 IPs (IP-A, IP-C, IP-D, and IP-E) are shown as empty entries in the table indicating that Hurricane Andrew does not cause the 4 IPs to the island. An empty entry simply indicates no islanding of the studied IP by the corresponding hurricane. Hurricane Juan while having the same wind speed of 60 mph as for hurricane Hilda, it will cause no islanding for the IPs in the study. Hurricanes Betsy, Carmen, Gustav, Hilda, and Katrina have similar islanding effect on IP-A, IP-B, IP-C, and IP-F. IP-F will be islanded with no connectivity edge.

The last three entries of Table correspond to the number of hurricanes causing islanding of an IP, islanding score relating the number of edges of the islanded plants, and the hardening index that is used for relative ranking of weak locations in the transmission system.

Hardening index used in the proposed methodology is the average of islanding size per affecting hurricane and can be calculated using the following equation:

A location with a smaller hardening index is a weaker location in the system which requires hardening by adding or upgrading transmission lines and towers, transformers, protection system equipment, and generating units.

According to Table 2, relative comparison of studied IPs shows that the geographical area that includes IP-F needs further hardening consideration and is a candidate location for further load flow studies and more detailed modelling, simulation, and analysis using real-time simulators (covered in Part II of the article).

The purpose of Table is to show part of our methodology and is not conclusive as it does not contain all studied hurricanes and IPs in the region. Also, there are more metrics which relate to the results of load flow studies and real-time simulation of hardened transmission system scenarios.

Operating for improved resiliency

One aspect of electric grid resiliency is to work on the tripped transmission lines due to tropical storms and to safely reconnect them to the system. The reconnection requires de-energising the tripped lines at proper locations and in a timely fashion – the faster the reconnection, the shorter the downtime and hence improved resiliency for the entire transmission system and the region.

Personnel protective grounding of overhead distribution and transmission lines is a required safety practice involving field personnel using grounding wires to de-energise lines prior to working on the lines [12]. As part of our hardening study of the grid, we have created an automatic process that determines the possible fault currents at all locations of transmission lines to help guide field personnel through selection of grounding wires that are appropriate for the level of the fault currents at corresponding locations.

Basic criteria for safe grounding practices and personal protective grounds include six criteria [13] first of which relates to maximum personal safety while working on de-energised high-voltage equipment and stating that de-energised lines shall be considered energised until protective grounds are installed. While Hanaffil et al. [14] conclude that transmission line methods are disadvantageous and are limited to certain frequencies for injected currents with fast rise times [15], our methodology is based on transmission line models and using existing software tools suitable for steady-state analysis such as advanced systems for power engineering (ASPEN) and to a certain degree electro magnetic transients program (EMTP). Methods that find fault locations and their corresponding fault current may be used to find not only the proximity to fault locations on the tripped lines, but also to aid the maintenance crew in equipping themselves with proper grounding connectors corresponding to estimated conservative fault magnitudes. While several such methods are available to planners and the maintenance crew, they mostly lack the necessary versatility for integration of new transmission system topology in the model, and hence, in analysis of the system. The automated method proposed in [12] presents the right tool using impedance-based modelling as well as proper computer software using python programs.

As power systems are upgraded in the planning phase, their topology changes, and as a result the fault currents at different locations of the lines change. The automated algorithm developed for our study uses updates available at planning stage and from the tripped lines information after the impact of tropical storms are determined to provide a safer environment for faster field personnel response. ASPEN OneLiner, a short-circuit analysis program, was used to perform sliding fault analysis – a series of short circuits placed at incremental distances between the two endpoints of a transmission line resulting in a current profile that shows the fault current level as a function of distance along the line. Automation of safety grounding fault current studies will allow for more accurate analysis of grounding requirements as changes to system topology are made. This will provide values used by field personnel that are more accurate. Our methodology in the Operating phase of improving system resiliency consists of the following steps:

  • Step 1: Updating any new system topology to the database prior to occurrence of tropical storms in storm seasons.
  • Step 2: Applying ASPEN to determine all substations where fault currents are above a pre-determined (nominal) value, say 40 kA.
  • Step 3: Performing sliding faults along all lines leaving the high fault current substations.
  • Step 4: Determining breakpoints along all lines where fault currents fall below the nominal value for single line-to-ground faults (1LG).
  • Step 5: Determining breakpoints along all lines where fault currents fall below the nominal value for three phase-to-ground faults (3LG).
  • Step 6: Ranking de-energising tripped lines for reconnection according to islanding and IP customers’ priorities, and
  • Step 7: Dispatching field personnel for fault clearance and line reconnection according to Step 6.

Certain parts of the operating methodology are applied to a real size electric grid and reported in the subsequent sections of this article. For application of planning and/or operating methodologies for improved system hardening and resiliency reported in this paper, we use steady-state models and appropriate simulation tools such as ASPEN. The more in depth modelling, simulation, analysis, and design requiring real-time computing capabilities are covered in the accompanying Part II article.

Modelling of transmission lines for inclusion in the automated algorithm is based on traditional sending–receiving end buses represented by π-model transmission lines. Fig 11 illustrates two breakpoints of a transmission line between two substations represented by Bus A and Bus B.

Fig 11: High current breakpoints of a transmission line

Step 1 objectives are satisfied by updating all added, deleted, or upgraded/downgraded transmission lines in the system during the year preceding the annual storm seasons.

Step 2 of the methodology uses ASPEN to determine all buses with currents above the nominal value. In our study, the nominal current value is selected at 40 kA, although the user may choose a different value suitable for the transmission system under study. An additional tool of the program will compare the previous safety grounding output with the current one and alert the user if any changes occurred since the previous run – a very useful feature of the program aiding in more robust design and improving personal safety.

Step 3 objectives are met by use of system model depicted by Fig 11 and applying sliding faults along selected lines connected to high current substation buses. Electric utilities perform short-circuit analysis on models that they have built in ASPEN. One type of analysis practiced by certain electric utilities involves sliding faults. A sliding fault is a series of short circuits placed at incremental distances between the two endpoints of a transmission line. The fault current is recorded at each increment, and the final result of a sliding fault on a transmission line is a ‘current profile’ that shows the fault current level as a function of distance along the line. The current profile is useful to electric utilities’ field engineers. When field personnel want to work on a transmission line, they need to ground the line in order to be safe, but the type of grounding cable depends on the fault current level at that location. The fault study data is exported to a program, listing all high fault current substations and the breakpoints on the lines leaving their respective substations.

Steps 4 and 5 correspond to applying 1LG and 3LG faults at selected lines, respectively. The automated algorithm is capable of handling other types of faults. In these steps, breakpoints similar to Fig 11 are determined and the fault locations are marked and recorded for relaying to the field crew. The breakpoints on each line leaving a high fault current substation will be compared with a distance/structure list. The program will output the first structure number where fault current will fall below 40 kA. The breakpoints on each line will be evaluated for when fault current falls above or below the threshold fault current value.

Steps 6 and 7 are devoted to ranking lines according to customer priorities and relaying the information to management for dispatching the crew for line reconnection. In case of total system loss (blackouts) and the need for black-start, Steps 6 and 7 are followed according to the system restoration plan and not the industrial customer priorities.

To facilitate and address the electric utility's yearly manual updating and improving system robustness and personal safety, the following assumptions were made in developing the methodology and its corresponding program [12]:

  • It is assumed that a distance/structure list of all lines in the system is known.
  • ASPEN produces valid fault analysis results.
  • The electric utility model is fairly accurate and severe irregularities in the model might affect the program behaviour.
  • The fault current along a transmission line is less than that of one of the line's end points.

We next present some results obtained from simulating a test system using the proposed methodology, developed algorithm, and its corresponding software program. The proposed methodology is applied to analysis of a 10-bus test system which is an equivalent of a portion of a transmission network in the Gulf Coast region. Fig 12 shows the one-line diagram of the test system.

Fig 12: 10-Bus test system equivalent of a portion of Gulf Coast grid

The 10 buses are connected through 11 transmission lines and 2 transformers. There are four generators in the system. Buses I, J, and H are 115 kV and the other buses are 230 kV. Table  lists fault locations and corresponding three-phase fault currents and single-phase-to-ground fault currents of the transmission line from Buses J to H after performing sliding faults on the system. The fault locations on the line are represented here as an increment of 5% of the total distance from Buses J (i.e. 0%) to H (i.e. 100%) [12]. The program can be configured to perform sliding faults at a smaller or larger increment.

FAULT LOCATION3PH FAULT CURRENT, KA1LG FAULT CURRENT, KA
0 (Bus J)15.3110.67
515.7511.02
1016.2311.41
1516.7611.83
2017.3312.30
2517.9512.82
3018.6413.40
3519.4014.05
4020.2514.77
4521.1915.60
5022.2416.54
5523.4317.63
6024.7818.90
6526.3320.40
7028.1222.18
7530.2124.36
8032.6927.05
8535.6730.49
9039.3135.01
9543.8641.23
100 (Bus H)49.7150.28

Table 3: Fault currents on transmission lines J and H

The threshold between high fault current and low fault current is 40 kA in this study. Locations of a transmission line that are expected to experience a fault current at or above the threshold require the use of a large gauge wire (i.e. small American Wire Gauge (AWG) number) for protective grounding that has higher ampacity. Table shows that the locations within the 10% distance from Bus H require a smaller gauge number for the grounding wire. The breakpoint location on lines J and H is at about 91% of the distance of the line.

Performing sliding faults on the 10-bus system reveals the breakpoint locations crossing the current threshold for all 11 transmission lines, as illustrated in Fig 13. The breakpoint locations are indicated in Fig 13 as a percent of the total distance from the closest bus of a line. For example, the locations within 10% from Bus D on lines A–D have fault currents at or above the 40-kA threshold and are highlighted in red colour. All locations along one of the lines between Buses A and F are above the threshold. All locations along lines I and J are below the threshold and are highlighted in blue colour.

Fig 13: Above-threshold locations of the 10-bus equivalent test system

The diagram of breakpoint locations as shown in Fig 13 will be updated as the operation of the system is changed (e.g. generator shutdowns) or if the system topology is modified during the year preceding the tropical storms. This allows an engineer to compare the change in breakpoint locations within the system. If the breakpoint locations change, the grounding wire for different locations in the system can be chosen appropriately by the field personnel.

Automatic determination of fault current breakpoint locations in a fraction of the time done by engineers manually allows a much faster determination of fault locations and field personnel and reducing the repair time substantially. While the Planning methodology provides the means for identifying weak geographical locations for potential hardening of the system, the Operating methodology aids the ground crew to de-energise the tripped lines very efficiently and safely and results in reducing the system downtime and improving overall system resiliency. As noted, the proposed operating methodology is used in conjunction with any utility restoration plan or theoretical power system restoration methodologies such as the classical IEEE task force methods described in [16,17] or newer approaches documented in [18,19].

In hardening and resiliency studies of electric grid, addition of components such as transmission lines and relays which exhibit fast dynamic in the range of microseconds may require real-time simulation tools. Part of our overall investigations at PERL is devoted to real-time modelling and simulation of electromagnetic transients. These types of studies that take into account transmission lines, relays, SVC, and modelling for real time simulation of generators, transformers, and independent pole operation are documented in the accompanying Part II of the article.

References

  1. Rastgoufard R. Leevongwat I. Rastgoufard P.: ‘Impact of hurricanes on gulf coast electric grid islanding of industrial plants’. 2015 Power Systems Conf., March 2015.
  2. Liu Y. Singh C.: ‘Evaluation of hurricane impact on failure rate of transmission lines using fuzzy expert system’. 15th Int. Conf. on Intelligent System Applications to Power Systems (ISAP 09), 8–12 November 2009, pp. 1–6.
  3. Liu Y. Singh C.: ‘A methodology for evaluation of hurricane impact on composite power system reliability’, IEEE Trans. Power Syst., 2011, 26, (1), pp. 145–152 (doi: 10.1109/TPWRS.2010.2050219 .
  4. Akhtar R. Santos J. R.: ‘Risk analysis of hurricane disruptions on workforce and interdependent regional sectors’. Proc. of the 2013 IEEE Systems and Information Engineering Design Symp., University of Virginia, 26 April 2013.
  5. Garber M. Unger L. White J. et al.: ‘Hurricane Katrina's effects on industry employment and wages’, Month. Labor Rev., 2006, 129, (8), pp. 22–39.
  6. Leontief W. W.: ‘ Input-output economics’, Sci. Am., 1951, 185, (4), pp. 15–21 (doi: 10.1038/scientificamerican1051-15).
  7. Leontief W. W.: ‘The Structure of American Economy, 1919–1939: An empirical application of equilibrium analysis’ (Oxford University Press, 1951), 2Rev Ed.
  8. Krishnamurthy V. Kwasinski A.: ‘ Characterization of power system outages caused by hurricanes through localized intensity indices’ (IEEE, 2013).
  9. ‘IEEE Standard Terms for Reporting and Analyzing Outage Occurrences and Outage States of Electrical Transmission Facilities’, IEEE Std 859-1987.
  10. PC Weather Products Inc: ‘Tropical Cyclone Analysis & Tracking Software Products: Hurrtrak’, http://www.pcwp.com/products.html, accessed 7/25/2015.
  11. National Hurricane Center: ‘Past Track Maps of Major U.S. Landfalls 2001–2010’. NHC Data Archive. National Weather Service – National Oceanic and Atmospheric Administration. http://www.nhc.noaa.gov/data/#tracks_us. Accessed 7/25/2015.
  12. Rastgoufard R. Leevongwat I. Rastgoufard P.: ‘Automatic determination of fault current breakpoint locations for personnel protective grounding of distribution and transmission lines’. 2015 Power Systems Conf., March 2015.
  13. ‘Reclamation, Facilities Instructions, Standards, and Techniques: Personal protective grounding for electric power facilities and power lines’, U.S. Department of Interior, July 2005 .
  14. Hanaffil F. Hoon Siew W. Timoshkin I.: ‘Grounding grid safety evaluation under lightning current’. PIERS Proc., 12–15 August 2013.
  15. Grcev L.: ‘Time and frequency-dependent lightning surge characteristics of grounding electrodes’, IEEE Trans. Power Deliv., 2009, 24, (4), pp. 2186–2196 (doi: 10.1109/TPWRD.2009.2027511).
  16. Adibi M. Clelland P. Fink L.: ‘Power system restoration – a task force report’, IEEE Trans. Power Syst., 1987, 2, (2), pp. 271–277 (doi: 10.1109/TPWRS.1987.4335118) .
  17. Adibi M.: ‘Special considerations in power system restoration’, IEEE Trans. Power Syst., 1992, 7, (4), pp. 1419–1427 (doi: 10.1109/59.207363).
  18. Ross H. B. Zhu N. Giri J. et al.: ‘An AGC implementation for system islanding and restoration conditions’, IEEE Trans. Power Syst., 1994, 9, (3), pp. 1399–1410 (doi: 10.1109/59.336124).
  19. Kostic T. Cherkaoui R. Germond A. et al.: ‘Decision aid function for restoration of transmission power systems: conceptual design and real time considerations’, IEEE Trans. Power Syst., 1998, 13, (3), pp. 923–929 (doi: 10.1109/59.708813).
Go to the profile of Parviz Rastgoufard

Parviz Rastgoufard

Entergy Endowed Chair for Power Systems Engineering, University of New Orleans

No comments yet.