Comprehensive Design of an Earthing System for Broadlands Hydropower Station Switchyard

Broadlands Hydropower Project (BHP) harnesses the last hydropower potential of Kehelgamu-Maskeli Oya rivers with an installed capacity of 35 MW and 126 GWh annual energy generation. Excavation of the powerhouse switchyard site exposed to a bedrock formation with highly weathered granitic gneiss beneath a thin layer of scum soil top, resulting in an irregular high soil resistivity profile. Therefore, the main purpose of this research is to design a safe and effective grounding system for the switchyard of BHP, which can carry fault current into the ground without exceeding tolerable ground potential rise, ensuring the desired operation of protective & control devices so that not to endanger human & equipment. Owing to the nature of non-uniform high soil resistivity and limited land space for extension, this has become a great challenge. Research used two approaches, guidelines of conventional IEEE 80-2000 standards and Finite Element Method (FEM). Initially, resistivity measurement was conducted covering the entire area of concern. A soil model was prepared using orthodox horizontally stratified two-layer soil model using Sunde’s graphical technique based on measured data. Then the grounding grid was designed adhering to guidelines given in IEEE 80:2000 standard and observed high overall grid resistance, eventually exceeding the tolerable step and touch potential levels. Thereafter a soil model was prepared based on FEM which facilitates plot of accurate and smooth surface voltage distribution over the entire switchyard area. Applying fault current to these discrete finite elements and based on the first principle of Kirchhoff’s current distribution balance, the localized voltage distribution has been developed for the entire area and plotted using a self-developed MATLAB computer program. FEM model can trace the points where the touch and step potentials exceed safe limits in two-dimensional stratified grid, estimation of voltage gradients at boundary areas, which all are unable to track using conventional IEEE method. Accuracy of the model can further be increased by reducing the size of the soil element. Finally, several sensitivity studies were conducted so as to optimize the BHP switchyard grid design ensuring safe grid operation.


Broadlands
Hydropower Project (BHP) harnesses the last hydropower potential of Laxapana complex located in the downstream of Kelani river basin. The main dam and weir sites of the BHP is located in Pallewaththa and Polpitiya areas respectively in the Central province, whereas the power station site is located in Kalukohuthenna, Kithulgala area, belonging to Sabaragamuwa province. Powerhouse site is bordering Kelani river from North, forest reservation from East and South, and Kartaranoya from West. The total working area is constraint to about 1.6 ha. According to the basic design layout, about 34% of land covered under the powerhouse construction and the land extent of the switchyard approximated to 20% of the total land use. Tailrace construction occupies an estimated 46% of land [2]. Excavation of powerhouse and the tailrace went down to the bedrock where the earth mat laid beneath concrete and the switchyard consists mainly of weathered rock/soil mix geology. earth resistance. Designing of low resistive earthing system in BHP powerhouse and tailrace structures was found extremely difficult where excavations were done down to bed rock having high soil resistivity, and further grid expansion was not possible owing to geological constraints. Also, no appreciable buffer zone was possible for powerhouse construction area in which public roads are located adjacent to earth mat boundaries.
The performance analysis of an earthing system essentially consists of determination of earth resistance and earth surface potential distribution owing to the dissipated current during an earth fault [1]. In here, BHP generation switchyard would be connected to the national grid at 132 kV level and hence counts for high fault currents. Therefore, proper design of an earthing system is a must requirement to safeguard against human and property damages.

Objectives
Primary objectives of this research study are to:  Conduct, soil resistivity survey for the proposed switchyard area and prepare a soil resistivity profile both under wet condition as well as dry condition.  Model, soil resistivity using: o Conventional Sunde's method o Distributed parameter method run on basic principles  Design, an earthing system based on IEEE empirical formulas.  Develop, discrete earth design model based on FEM using basic principles.  Evaluate, surface potential distribution, touch potential distribution, step potential distribution and grid resistance.  Conduct, sensitivity studies.

Literature Survey
In a ground fault condition, the flow of current to earth will produce potential gradients within and around the earth rod or mat. Unless proper precautions are taken in design, the maximum potential gradients along the earth's surface may be of sufficient magnitude during ground fault conditions endangering a person in the area. A person can experience an electric shock as a result of this potential gradient, and it refers to as step potential. Moreover, dangerous voltages may develop between grounded structures or equipment frames and the nearby earth. This can also make an electric current to flow when a person touches both the structure and earth at the same time and this refers to as the touch potential [3]. A proper earthing system should be capable of avoiding risky step and touch potentials and be able to safeguard human and machine under worse possible electric fault conditions. Hence sound design of a safe grid substation earthing system should mainly consist of:

Methodology
Design methodologies of the research work consist of two main categories and two subcategories and they are as follows: The most commonly used soil resistivity models are the uniform soil model and the twolayer soil model. A more accurate representation of the actual soil conditions can be obtained using two-layer model [5]. Under discrete soil resistivity modelling, the area that has been taken into consideration for the earthing system design should be thoroughly investigated for 3D resistivity profiling. Shorter the distance between points of measurement, higher the depth of penetration leading to higher accuracy of the model. Using measured apparent soil resistivity values of top layer and layers up to 8 m depth, conceptual framework is formulated. Thereafter, desired discrete point characteristics are calculated using mathematical interpolation using MATLAB coding. A cloud of points with respective resistivity values for each element are calculated for the entire area of study.  Under FEM, earth can be represented by a pure resistance [5] and hence 3D soil region of switchyard is modelled as a combination of large number of small pure resistive elements connected with other elements through their intermediate surfaces. If the soil element happens to be the top layer, it is connected with 5 other elements surrounding it, otherwise it is connected with 6 other elements (except boundary elements) as shown in Figure 2. Also, grounding conductors and rods are divided into small elements and these conductive elements are considered as voltage elements because of ground potential rise. Owing to these voltage elements, leakage currents passed to the other soil elements through their interconnected surfaces, hence leakage current distribution throughout the region is according to Kirchhoff law.
As a result of the leakage currents distribution, surface voltage distribution builds up. Surface voltage distribution is the main figure of the earthing system design process because it can be used to evaluate other critical design parameters such as step voltage and touch voltage.
Further analysis of surface voltage distribution can be extended by changing conductor space and rods of earthing systems until it satisfies the step, touch and mesh voltage criteria as given in IEEE standards. A simplified working algorithm of the MATLAB programme is as attached in Appendix B, Figure B1. Table 1 shows the measured resistivity values using Wenner method for about 10 points distributed over Broadlands switchyard.   Table 1 and relevant probe spaces, Figure 3 was plotted. Thereafter, top layer resistivity value (ρ1) and bottom layer resistivity value (ρ2) were selected by using Figure 4 and further process of modelling of two-layer soil model was prepared as described below [5].

Discrete Soil Resistivity Model
Actual 3D image of the BHP switchyard can be modelled using interpolation technique and measured resistivity values. Resistivity survey was conducted to collect the resistivity value of 50mx45m BHP switchyard. Using the surface resistivity values, soil resistivity profile of the surface of BHP switchyard can be plotted as shown in Figures 5 and 6, respectively.   [5].
Hence calculated tolerable touch voltages and step voltages for particular conditions of 50 kg and 70 kg persons respectively are as given in Table 2.  [5]. Calculated single phase to earth fault current was 12.6kA.

Calculation of split factor
To calculate the split factor, remote contribution and the local contribution of the fault current are required. According to IEEE 80-2000 standard, four different types of contributions such as 100% remote, 25% local 75% remote, 50% local 50% remote, 75% local 25% remote are discussed, and graphical analyse method is used for this calculation. Earth resistance of a transmission line tower earthing system was taken as 15Ω. Finally, worse case value of 56% was selected as final split factor [5].
Calculation of decrement factor Decrement factor Df, was derived to take into account the effect of DC current offset. Table 10 given in the IEEE 80-2000 standard was used to calculate the decrement factor. X/R ratio and fault duration were taken as 10 and 0.5s respectively for this calculation and based on which the selected decrement factor was 1.026.

Calculation of grid current
Using the calculated values of split factor, single phase to ground fault current and decrement factor, computed grid current was found to be 7.24 kA.

Calculation of grid resistance
To calculate the grid resistance, equation 52 of IEEE 80-2000 standard can be used. For this purpose, the area of the grid, soil resistivity and total conductor length are required. Total conductor length is the sum of horizontal conductor lengths and total earth rod lengths. This depends on the space factor of the horizontal conductors. For this calculation, value of soil resistivity and area of BHP switchyard were required and those were taken as 3799 Ωm and 2250 m 2 , respectively. Calculated grid resistance for 5 m x 5 m mesh size was 37.98 Ω and resultant ground potential rise was found to be 274975 V. According to IEEE standard, this value is not satisfying the step and touch voltage criteria mentioned in the regulation. Hence possible alternative methods should be analysed to reduce the grid resistance [12].

Development of a mathematical model to calculate surface voltage distribution
According to the given description of the FEM modelling under methodology section, leakage current owing to fault current passes throughout the entire region as shown below in Figure 7. Here, 2D layer of soil region was considered only for explanation purpose only. Kirchhoff's current and voltage law were applied for each and every small soil element in 2D soil layer considered to be small resistances connected to each other as shown below.
Applying Kirchhoff's current law for the elements connected to node 11: Hence, unknown voltages of these equations can be found by solving this matrix.

5.2.2
Earthing system of BHP switchyard using FEM model Under simulation of FEM modelling, surface voltage distribution of earthing system, which consists of 3 m in length of 38 earth rods along the perimeter and earth conductors of 5 m x 5 m mesh, was simulated applying grid current of 7239 A. For this simulation, MATLAB coding was used.
Since the actual earthing system is spread over 45 m x 50 m area, resistivity values of 60 m x 55 m x 8 m soil region were used for discrete soil resistivity modelling, because the effect of the boundary region can be emphasized by extending the soil region. Figure 8 shows the surface voltage distribution of the FEM simulation with cubic element size 1 x 1 x 1 m. Smaller element size produces more accurate results, but it requires high computer RAM capacity. Variation of the model accuracy with the element size is discussed under Appendix C. According to model verification, cubic element size of 0.053 m x 0.053 m x 0.053 m produces the best result of FEM modelling.
FEM Model simulated ground potential rise (GPR) of the earth grid is 57730 V (Calculated GPR using the conventional method was 274975 V), which is much higher than the tolerable touch voltage of 8090 V. Grid resistance 7.974 Ω was calculated by dividing GPR from grid current. Since the simulated grid resistance and GPR is higher than the IEEE safe limits, initial design of earth grid was not recommended. Touch voltage distribution also violates the tolerable limits specified by IEEE regulation as shown in Figures 8 and 9.

Sensitivity Analysis
It was observed as per the results obtained in the previous sections that grid resistance of the switchyard was not falling within the acceptable range with conventional earthing designs. Owing to the limitation of the grid area for expansion, further reduction of the grid earth resistance was also not possible. Therefore, following alternatives were considered to reduce the grid resistance to a standard acceptable value [6][13].

Connecting another grid in parallel which
has a low resistance value [14]. 2. Changing the driven rod diameter or its cross section [7]. 3. Changing the earthing grid conductor cross section and its length [7]. 4. Refilling the switchyard area with low resistivity soil [8].
6.1 Connecting the concrete encased tailrace grid Concrete, being hygroscopic, attracts moisture. Concrete block buried in soil behaves as a semiconducting medium with a resistivity of 30-90 Ωm. This can be taken as better grounding medium rather than higher resistivity soil region in earthing design [5] [13]. Typical values used for the computation of parallel grid resistance were: 5 m rod length, 120 Ωm wet concrete resistivity, 0.2 m buried depth (h) and 5 m length rods buried around perimeter.
Grid resistance and total conductor length variation against switchyard area are as shown in Figures 10 & 11, respectively. Using these graphs, mesh size of 12.5 m x 10 m was selected as the most economical mesh size. Thereafter the combined grid resistance is computed as described below. As can be seen, it would reach the desired value after increasing the tailrace area up to 3250 m 2 [13]

Calculation of Ground Potential Rise
Ground potential rise depends on the passing grid current through the earth grid. With the combined grid arrangement, larger portion of the grid current passes through the satellite grid and lesser part of the same would pass through the main grid. Calculated ground potential rise was found to be 7596 V which is lower than the tolerable touch voltage 8090 V. Also, it can be seen that the design value is satisfying the IEEE standard condition when the surface layer material of ρs=45000 Ωm is used since the tolerable touch voltage of switchyard with 10,000 Ωm surface material is 2,155 V.  Figure 12 shows the variation of grid resistance against the rod diameter as per the calculation performed according to the Schwarz's equation. It can be seen from the graph that the earth resistance doesn't show any significant variation on rod diameter size. Hence, change in rod diameter does not influence much in reducing the earth resistance [7].

Figure 13 -Variation of grid resistance with mesh conductor cross section
As can be seen from Figure 13, selection of earthing conductor having a large cross section area doesn't give any considerable reduction in earth resistance. Hence, this alternative too cannot be used as a possible option to reduce grid resistance. Total conductor length directly depends on the mesh size of the earth grid and the potential distribution of the switchyard depends on the mesh size. However, when the mesh size is reduced, potential distribution of the switchyard tends to distribute smoothly even though the earth resistance is not reduced by a considerable amount. Variation of grid resistances with mesh conductor length for various mesh sizes is shown in Figure 14. As can be seen, the reduction of grid resistance below 1 Ω is not possible by decreasing the mesh sizes [7].

Soil treatment by backfilling the contact surface of the electrodes
Under this option, critical resistance region [9][10] of the grounding system was replaced by 90 -110 Ωm low resistivity material [11]. Critical resistance region of 56 m x 51 m x 3.8 m is shown in the following Figure 15.

Figure 15 -Critical resistance region of the BHP switchyard
After replacing soil in critical resistance region, following earth resistance values were obtained using Schwarz's equations [7]. To maintain the tolerable higher touch potential value of 8090 V, surface layer material of asphalt with ρs=45000 Ωm has to be used to satisfy the IEEE requirement.

m mesh
After changing the mesh size, GPR of the earth grid was found to be 4932 V, which is lower than the tolerable touch voltage of 8090 V. Accuracy of results of the simulation can be increased with the selection of smaller element size. However, the value exceeds the tolerable touch value since the difference between GPR and the tolerable touch value is relatively high. Comparison of GPR values between discussed alternative solutions are illustrated in Table 3.  Figure  19 shows that the maximum potential is lower than the tolerable touch potential 8090 V.

Figure 19 -Touch potential distribution after 3.8 m soil layer replacement
Step potential distribution of a person when moves along the X-direction and Y-direction are shown in Figures 20 and 21, respectively. It can be observed that the maximum step potential is lower than the tolerable step voltage 31869 V.

Validation of FEM Model
Accuracy or the sensitivity level of analytical capabilities of the parameters such as surface voltage distribution, touch voltage distribution, step voltage distribution, GPR calculation and grid resistance (Rg) calculation depends on the size of the finite soil element. Minimizing of soil element size tremendously increases the requirement of computer RAM capacity. As such, the minimum requirement for 1mx1mx1m element size simulation requires minimum of 8 GB RAM capacity. Therefore, better accuracy can be reached by reducing the element size as described in Appendix C, Figure C1 & Figure C2 respectively.

Research observations in summary
This research study focused on the design and assessment of switchyard grounding system of Broadlands hydropower station. Initially, two soil models were prepared based on the soil resistivity survey results of the area concerned. The first model was based on the guidelines of IEEE 80:2000 standards (Sunde's graphical method) and the second was a discrete soil resistivity model. Observed apparent resistivity value of soil from the former method was 3,799Ωm. Based on the results of soil resistivity survey, the design of earth grid was conducted for the whole switchyard area using two techniques based on, namely, Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 Step 7 Step 8 Step 9 Step 10 Step 11 Step 12 Apparent resistance given by an earth rod buried in the soil is calculated by the following equation But in reality, discrete FEM model treats all the soil and conductor parts as elements and therefore limited satisfactory is acceptable in fair comparison for both methods. The basis was prepared on sphere of influence concept to be used in FEM and the basic model comparison concept is illustrated below.

Figure C1 -Conversion of Critical resistance region of single rod to FEM model
The practical satisfactory values for the elements of discrete FEM model were identified as 1 m, 0.5 m, 0.25 m and 0.2 m. Calculation of smaller element capacities was curtailed due to computer RAM capacity. Therefore, comparison had been done in equal flat form by making all parameters kept the same for both cases. The results are given below in Figure C2.

Figure C2 -Comparison of Equation Vs Model calculated Rg values
It was clearly evident that results of the two models follow a similar pattern. Also, smaller the element size, less error generated between the two methods. This was a significant result to validate the discrete model and, as a result, mathematical analysis conducted to further ascertain this phenomenon. Best fit line for the given points were well matched to 3 rd order polynomial claiming for best theoretical regression fit line. According to the forecast, optimum results will reach around element size of 0.053 m or 53 mm.