Optimal Ground Grid Design for Large EHV Substations with Auto-Transformer

Xuan Wu, Vinod Simha, and Ronald J. Wellman Station Engineering Design Standards

American Electric Power Gahanna, OH

xwu@aep.com, vsimha@aep.com, and rjwellman@aep.com

Abstract: This paper describes hands-on solid modeling experience of the safe optimal ground grid design (less copper) for a large Extra High Voltage (EHV) substation. The optimized ground grid meets the IEEE-80 standards for safety without compromising on step or touch potential thresholds. The ground grid has been designed in CDEGS using fault current flow concepts, and multiple injection points simulate the actual conditions and fault currents at the substation. The method primarily focuses on substation specific data to perform substation grounding studies rather than utilizing the IEEE-80 curves and assuming fault current values. The method is not only conceptually correct and meets safety requirements, but also leads to significant savings of copper. Thus optimizing the ground grid design for the EHV substation.

Index Terms — circulating current, local source, multiple injection, optimization, remote source, split factor.

ABBREVIATIONS AND ACRONYMS

AEP – American Electric Power, EHV - Extra High Voltage, ASPEN - Advanced Systems for Power Engineering, CDEGS – Current Distribution Electromagnetic Fields, Grounding and Soil Structure Analysis, (3I0) - Total ground fault current, (Ig) - Current flowing into the grid, (In) - Current flowing into the shield wires from the ground grid, (Icir) - Circulating current returning to the neutral of the local transformer from the fault location, 1LG – Single line-to-ground fault condition, SES - Safe Engineering Services & Technologies.

I. INTRODUCTION

This paper describes the modeling of a large EHV substation (Station JF) ground grid using CDEGS software. Technical challenges, commonly made assumptions leading to excessive copper than required and optimization techniques have been discussed. The fault current data obtained in ASPEN is used to identify the contributions made by the remote sources - transmission lines connecting Station JF, and local sources – transformers inside Station JF. The soil model in which the ground grid is installed is determined by soil resistivity tests utilizing the 4-point wenner method.

The ground grid model is imported into CDEGS and the locations of local and remote contributions are identified. A single point injection of 1 Amp in the grid is required to determine the ground grid resistance. The grid impedance is utilized in the FCDIST module to determine the fault current division via the shield wire and the ground grid. The circulating current is determined from ASPEN. The total fault current (3I0) is injected into the grid at the fault location. The shield wire current components and the circulation current component are ejected out of the grid. This allows the grid current to flow into the ground grid and sets a stage to calculate the touch and step potentials on the grid. The touch and step potential values are compared to the threshold limits calculated as per IEEE-80 [1] in the CDEGS software. If the actual touch and step potential values are less than the threshold limits, then the IEEE-80 requirements are satisfied and no additional ground grid conductors are required.

II. FAULT ANALYSIS

In the past thirty years, many publications [2-5] are focused on station fault current distribution and split calculations. The fault analysis is indeed significant as the foundation of station grounding studies.

A. Fault Categories

There are three fault categories that can cause some unbalanced fault currents flowing to the ground which are single line-to-ground fault and double line-to-ground fault. Based upon dividing the three-phase system into the symmetrical components (zero-, positive-, negative-sequence networks), the fault analysis problem for grounding is simplified with only considering the zero-sequence current (I0). Because the value of I0 is proportional to the unbalancing extent of a fault, the single line-to-ground fault is the worst fault category in most cases.

B. Fault Locations

The fault locations are always chosen on the high-side or low-side buses. This is because the incoming lines/feeders and transformer banks are connected to these buses. In other words, the fault current is higher when the fault occurs on a

978-1-4673-8040-9/15/$31.00 ©2015 IEEE

bus. However, it is hard to determine which bus would be the worst-case scenario [5]. In AEP, the fault locations based on all buses are analyzed except for buses with voltage classes lower than 69 kV. For example, in an AEP EHV station there may be five different voltage buses - 765kV, 500kV, 345kV, 138kV and 69kV. In addition, the 34.5kV and 12kV buses are excluded from the fault analysis when they are delta-connected and without a grounding transformer.

C. Fault Components

The basis of grounding fault analysis is the entire ground fault currents (3I0) should return to their sources through either earth or any metallic objects. As mentioned in the previous paragraph, the transformer connection will play a significant role in the fault current distribution. For example, if the fault occurs on the low-side of Δ-Y transformer in a small distribution station, which means the entire fault current will flow back to the transformer neutral through the ground conductors. In other words, only transformer neutral circulating current (Icir) exists in this case.

In AEP large transmission stations, auto-transformer banks are often used. As a result, there are three components for 3I0, which are ground grid current returning to remote sources via earth (Ig), neutral/shield wire current returning to remote sources (In), and transformer neutral circulating current (Icir). This is shown graphically in Fig. 1.

Figure 1. Auto-transformer fault current distribution

D. Fault Analysis Tools and Applications

In AEP, ASPEN OneLiner is widely used by protection & control engineers and station engineers for fault analysis. For a grounding study, faults on specific buses are simulated. As an example, the Fig. 2 shows an ASPEN analysis of Station JF. Information such as 3I0, Icir, and X/R ratio can be acquired from ASPEN.

Figure 2. Example of an ASPEN result report

However, In are calculated utilizing the FCDIST module developed by SES. An example input window of FCDIST is shown in Fig. 3.

Figure 3. Example of a FCDIST input window

With the correct input data and simulation following the FCDIST manual [6], the output file will show the In from Station JF to each specific terminal station as shown in Fig. 4.

Figure 4. Example of a FCDIST output file

E. AEP Proposed Grounding Study Flowchart & Process

Figure 5. Flowchart of AEP Grounding Study

1. Enter and simulate soil resistance measurements using RESAP.

2. Create or import a ground grid in MALZ and use 1-Amp injection point at the selected fault location to calculate the ground grid impedance.

3. Take the ground grid impedance, remote contribution line currents from ASPEN, T-Line information (number and average length of spans, tower footing resistance, circuit configurations, and shield wire size/materials) into FCDIST, and then simulate to get the values of In.

4. Use 3I0, In and Icir as the injection currents at the specific locations in the MALZ module, which means there will be multiple injection points. Typically, 3I0

should be injected into the ground grid, while In and Icir are ejected from the grid to the neutral/shield wires and the transformer banks, respectively.

5. Place a surface profile above the ground grid and calculate the touch and step potentials of each profile point.

6. Calculate the threshold limits for the allowable touch and step potentials with X/R ratio, clearing time and surface material parameters, and compare the calculated values with the allowable values.

a) If the calculated touch potential is lower than 75% of the threshold value, it is suggested to delete some ground conductors.

b) If the calculated touch potential is larger than 75% of the threshold value, it is necessary to add some ground conductors near the high-potential areas to mitigate.

III. CASE STUDY

Station JF is a 765/138 kV EHV station. It consists of two yards, which are the 138 kV yard with dimensions of 281×326 ft and the 765 kV yard with dimensions of 1074×1010 ft. Each yard has its own existing ground grid, which is tied with the other through three underground conductors. The existing ground grid layout is shown in Fig. 6.

Figure 6. Ground Grid Layout

As marked in Fig. 6, “FL” means selected fault location, which is the 138 kV bus. “T1”, “T2”, “T3” represent the 765/500 kV and two 765/138 kV transformer bank locations. “SW” shows the approximate locations that shield wires connect to the ground grid.

Before jumping into the grounding study process, it is necessary to determine the zero-sequence fault current based on the ASPEN result. For a 1LG fault occurring at the 138 kV bus, the ASPEN result is shown in Fig. 2, which has been modeled in the diagram presented in Fig. 7.

Figure 7. Diagram showing worst-case fault current flow

For better clarification, the magnitude 586.8 shown in the red box of Fig. 7 is calculated by multiplying the magnitude 195.6 shown in the red box of Fig. 2 by three and the angles of both are equal to 100. Physically, this specific line and its vector in Fig. 7 indicate the zero-sequence remote fault contribution from terminal station “Wythe” to Station JF. The remaining vectors in Fig. 7 can be determined in the same manner.

The shield wire currents (In) pointed out in green arrows must be calculated from FCDIST with input of all nine zero-sequence remote fault contributions from terminal stations. The results are shown in the red box of Fig. 4. After lumping the vectors with the same voltage class, 765 kV, 500 kV and 138 kV shield wire currents can be obtained and entered into the “Energization” table displayed in Fig. 8.

Figure 8. “Energization” input window and explanation

After the forth and fifth steps, the touch and step potentials are calculated via MALZ with results shown in Fig. 9 and Fig. 10 respectively. Based on the AEP flowchart, this ground grid must be reanalyzed with faults occuring at the 765 kV bus and the 500 kV bus. Based on the analysis, the worst-case scenario is with the fault at the 138 kV bus as shown in Fig. 7 which produces the highest touch and step potentials. The fault current contributing to the worst case scenario may not be the highest fault current at the station.

-800 -300 200 700

X AXIS (FEET)

-1000

-500

0

500

1000

Y A

XIS

(

FE

ET

)

Scalar Potentials/Touch Voltages/Worst Spherical [ID:Jackson Ferry @ f=60.0000 Hz ]

LEGEND

Maximum Value : 549.994 Minimum Value : 1.293

549.99

495.12

440.25

385.38

330.51

275.64

220.77

165.90

111.03

56.16

Figure 9. Touch potential magnitude plot for existing ground grid

-800 -300 200 700

X AXIS (FEET)

-1000

-500

0

500

1000

Y A

XIS

(F

EE

T)

Scalar Potentials/Step Voltages (Spherical)/Worst Spherical [ID:Jackson Ferry @ f=60.0000 Hz ]

LEGEND

Maximum Value : 88.786 Minimum Value : 0.623E-01

88.79

79.91

71.04

62.17

53.30

44.42

35.55

26.68

17.81

8.93

Figure 10. Step potential magnitude plot for existing ground grid

With applying the sixth step, the threshold limits for the allowable touch and step potentials can be calculated which are 870.7 V and 2878.5 V using fifteen cycles (0.25 s) as the fault clearing time, 3000 Ω-m and 4 inches as the resistivity and depth of the surface material, 30.9 as the X/R ratio found from Fig. 2. The maximum allowable values are all larger than the maximum values of touch and step potentials shown in Fig. 9 and Fig. 10. Therefore, there is still some room left for optimizing this ground grid to make it more economical. Hence, some reductions were made for the existing ground conductors in SESCAD, and simulated in MALZ until the touch and step potentials both get close enough to the allowable values. Thus, the safe optimal ground grid is developed and is shown in Fig. 11 and Fig. 12.

-800 -300 200 700

X AXIS (FEET)

-1000

-500

0

500

1000

Y A

XIS

(

FE

ET

)

Scalar Potentials/Touch Voltages/Worst Spherical [ID:Jackson Ferry @ f=60.0000 Hz ]

LEGEND

Maximum Value : 652.946 Minimum Value : 4.157

652.95

588.07

523.19

458.31

393.43

328.55

263.67

198.79

133.91

69.04

Figure 11. Touch potential magnitude plot for redesigned ground grid

-800 -300 200 700

X AXIS (FEET)

-1000

-500

0

500

1000

Y A

XIS

(F

EE

T)

Scalar Potentials/Step Voltages (Spherical)/Worst Spherical [ID:Jackson Ferry @ f=60.0000 Hz ]

LEGEND

Maximum Value : 105.539 Minimum Value : 0.337E-01

105.54

94.99

84.44

73.89

63.34

52.79

42.24

31.69

21.13

10.58

Figure 12. Step potential magnitude plot for redesigned ground grid

The maximum touch potential is 653 V, which is almost equal to the 75% of the allowable touch potential (870.7 V). The step potential is much lower than the threshold (2878.5 V). However, based on the optimization theory, if one constraint meets the threshold, the objective will be optimal. From the optimization point of view and the AEP grounding study process, the design shown in Fig. 11 can be accepted and called “Safe Optimal Ground Grid”.

Assuming this station were designed today using the multiple injection point method, the ground grid would be constructed based on Fig. 12 instead of Fig. 10. The total length of ground conductors would reduce from 41,941 ft to 20,454 ft. Considering the unit cost of material and labor as $15/foot, the cost saving is $15*(41,941-20,454)=$322,300 compared to the single injection point method.

Further, the ground grid at Station JF was analyzed for a transformer outage contingency in addition to the worst case scenario 1LG fault contingency occurring inside the station. For example, if T3 is taken out of service based on a contingency or maintenance, the total fault current (3I0), grid current and the transformer bank circulating current (Icir) are decreased. Therefore, the touch and step potentials calculated without an EHV bank will not exceed the values calculated with an EHV bank.

-800 -300 200 700

X AXIS (FEET)

-1000

-500

0

500

1000

Y A

XIS

(

FE

ET

)

Scalar Potentials/Touch Voltages/Worst Spherical [ID:Jackson Ferry @ f=60.0000 Hz ]

LEGEND

Maximum Value : 622.004 Minimum Value : 0.871

622.00

559.89

497.78

435.66

373.55

311.44

249.32

187.21

125.10

62.98

Figure 14. Touch potential magnitude plot for contingency analysis

The data in Fig. 13 is drawn from ASPEN simulation and both Fig. 13 and Fig. 14 confirm what was described in the previous paragragh. The maximum touch potential without T3 is 622 V, which is smaller than 653 V with it. To be noted, all the other possibilities such as a T1 outage have been studied and none of them exceed 653 V.

IV. CONCLUSION

This paper describes the design of a safe optimal ground grid design for a large EHV substation (Station JF). Conservative approaches such as not calculating the station specific fault current division factor (split factor) or single injection of grid current can lead to the excessive use of copper to design a safe ground grid in large EHV stations. By utilizing the optimization process described in this paper, a safe ground grid designed using less copper has been achieved.

REFERENCES [1] IEEE Guide for Safety in AC Substation Grounding, IEEE Std. 80-

2000. [2] F. P. Dawalibi, “Ground fault current distribution between soil and

neutral conductors,” IEEE Trans. Power App. Syst., vol. 99, no. 2, pp. 452-461, Mar./Apr. 1980.

[3] D. Garrett, J. Mayers, and S. Patel, “Determination of maximum substation grounding system fault current using graphical analysis,” IEEE Trans. Power Del., vol. 2, no. 3, pp. 725-732, Jul. 1987.

[4] Ma, J., Zhao, X., F. P. Dawalibi, “Practical considerations for fault current split computation in power system,” in The International Conference on Electrical Engineering, 2009.

[5] J. Ma and F. P. Dawalibi, "Modern computational methods for the design and analysis of power system grounding," in Proceedings of the 1998 International Conference on Power System Technology, vol. 1, Aug. 1998, pp. 122-126.

[6] CDEGS Software Package, Safe Engineering Services & technologies ltd., Montreal, Quebec, Canada, November 2002.

Figure 13. Diagram of fault current flowing without T3 EHV bank