94, SYSTEMATICFAULT ANALYSIS USING BUS IMPEDANCE MATRIX. 363, ‘The prefault bus voltage is usually assumed to be 1.0 per unit, and we therefore obtain from (9.6) the following approximate formula for the short-circuit caps or the short-circuit MVA. soc = 52 va On) Xe 9.4 SYSTEMATIC FAULT ANALYSIS USING BUS IMPEDANCE MATRIX ‘The network reduction used in the preceding example is not efficient and is not applicable to large networks. In this section a more general fault circuit analysis using nodal method is obtained. We see that by utilizing the elements of the bus impedance matrix, the fault current as well as the bus voltages during fault are readily and easily calculated. Consider a typical bus of an n-bus power system network as shown in Fig- ure 9.8. The system is assumed to be operating under balanced condition and a per phase circuit model is used, Each machine is represented by a constant voltage source behind proper reactances which may be X/, X4, or Xq. Transmission lines are represented by their equivalent 1 model and all impedances are expressed in per unit on a common MVA base. A balanced three-phase fault is to be applied at bus k through a fault impedance Zy. The prefault bus voltages are obtained from the power flow solution and are represented by the column vector FIGURE 98. Atypical bs of a power system,364 9, BALANCED FAULT v0) Vius(0) = | Vi(0) 98) Val) As already mentioned, short circuit currents are so much larger than the steady. state values that we may neglect the latter. However, a good approximation is to represent the bus load by a constant impedance evaluated at the prefault bus volt age, 4b = oe (9.9) i ‘The changes in the network voltage caused by the fault with impedance Zy is ‘equivalent to those caused by the added voltage V4(0) with all other sources short- circuited. Zeroing all voltage sources and representing all components and loads by their appropriate impedances, we obtain the Thévenin’s circuit shown in Figure 9.9. The bus voltage changes caused by the fault in this circuit are represented by the column vector AY AVins = | AV 9.10) AVa ft ay! ‘A typieal bus of a power system, "pon194, SYSTEMATIC FAULT ANALYSIS USING BUS IMPEDANCE MATRIX. 365 From Thévenin’s theorem bus voltages during the fault are obtained by superposi- tion of the prefault bus voltages and the changes in the bus voltages given by Vius(F) = Vius(0) + AVius (9.11) In Section 6.2, we obtained the node-voltage equation for an n-bus network. The injected bus currents are expressed in terms of the bus voltages (with bus 0 as reference), ic., Tous = YourVivs @.12) where Ipus is the bus current vector entering the bus and Y>,,, is the bus admittance matrix. The diagonal element of each bus is the sum of admittances connected to itie, Dw j#i (9.13) i ‘The off-diagonal element is equal to the negative of the admittance between the buses, ie, ¥iy = Yin = vy (9.14) where ys (lower case) is the actual admittance of the line i-j refer to Section 6.2. In the Thévenin’s circuit of Figure 9.9, current entering every bus is zero except at the faulted bus. Since the current at faulted bus is leaving the bus, it is taken as a negative current entering bus k. Thus the nodal equation applied to the For more details ‘Thévenin’s circuit in Figure 9.9 becomes 0 va Vik 7 Min AY —h(F) in ot Ua o> Yin | | AVE (9.15) 0 Yor + Yok 7 Yan | | AVa or Tous(F) = Yous Vous (9.16) Solving for A Vous, we have AVinus = ZousTous(F) 17)366 9. BALANCED FAULT where Zeus = Yay is known as the bus impedance matrix. Substituting (9.17) into (9.11), the bus voltage vector during the fault becomes Vins (0) + Ziuslius(F) 9.18) Writing the above matrix equation i WF)] [uO] fn Vous (F) terms of its elements, we have Zn Zin 0 vetP) | = | va(0) | + | Zan o> Zee > Zin |] CF) | 2.19) 0 wtP)} Evin] [za Since we have only one single nonzero element in the current vector, the kth equa- tion in (9.19) becomes Ve(P) = Vi(0) ~ Zeln(F) (9.20) Also from the Thévenin’s circuit shown in Figure 9.9, we have Vi(F) = Z/h(F) 21) For bolted fault, Zy = 0 and Vi(F') = 0. Substituting for V,(F) from (9.21) into (9.20) and solving for the fault current, we get UP) = 7 (0.22) ‘Thus fora fault at bus k we need only the Zxx element of the bus impedance matrix. This element is indeed the Thévenin’s impedance as viewed from the faulted bus. Also, writing the ith equation in (9.19) in terms of its element, we have VCP) = (0) — intel) 02) ‘Substituting for Jj,(F), bus voltage during the fault at bus i becomes Zi F) = ¥i(0) = 5 2 9.24 WA) = WO) — ZAR V0) 02H With the knowledge of bus voltages during the fault, we can calculate the fault current in all the lines. For the line connecting buses i and j with impedance 2, the short circuit current in this line (defined positive in the direction i —» 5) is Ty(e) = WB) 2 (9.25)94, SYSTEMATICFAULT ANALYSIS USING BUS IMPEDANCE MATRIX. 367 ‘We note that with the knowledge of the bus impedance matrix, the fault cur- rent and bus voltages during the fault are readily obtained for any faulted bus in the network. This method is very simple and practical. Thus, all fault calculations are formulated in the bus frame of reference using bus impedance matrix Zyus ‘One way to find Zius is to formulate Ys, matrix for the system and then find its inverse. The matrix inversion for a large power system with a large number of buses is not feasible. A computationally attractive and efficient method for finding Zs mattix is “building” or “assembling” the impedance matrix by adding one network element at a time. In effect, this is an indirect matrix inversion of the bus admittance matrix. The algorithm for building the bus impedance matrix is described in the next section. Example 9.2 A three-phase fault with a fault impedance Zy = 30.16 per unit occurs at bus 3 in the network of Example 9.1. Using the bus impedance matrix method, compute the fault current, the bus voltages, and the line currents during the fault, In this example the bus impedance matrix is obtained by finding the inverse of the bus admittance matrix. In the next section, we describe an efficient method of finding the bus impedance matrix by the method of building algorithm. To find the bus admittance matrix, the Thévenin’s circuit in Figure 9.2(b) is redrawn with impedances converted to admittances as shown in Figure 9.10. The ith diagonal element of the bus admittance matrix is the sum of all admittances connected to bus i, and the ijth off-diagonal element is the negative of the admit- tance between buses i and j. Referring to Figure 9.10, the bus admittance matrix by inspection is 78.75 j125 725 Your =| 71.25 -§6.25 52.5 25 j25 — 75.0 Using MATLAB inverse function inv, the bus impedance matrix is obtained 40.16 50.08 70.12 Zine = | 50.08 70.24 70.16 50.12 50.16 50.34 From (9.22), for a fault at bus 3 with fault impedance Zp = 0.16 per unit, the fault current is Yl) _ 10__ 2.0 po WP) = Zo 4Z, ~ j03i + 7016 ~368 9, BALANCED FAULT FIGURE 9.10 ‘The admittance diagram for system of Figure 9.2 (b). From (9.23), bus voltages during the fault are Vi(F) = Vi(0) ~ Zisfs(F) = 1.0 ~ (70.12)(—72.0) = 0.76 pu Va(F) = Va(0) ~ Zaals(F) 0 — (j0.16)(—J2.0) = 0.68 pu Va(F) = Va(0) — Z3a13(F) = 1.0 — (j0.34)(—32.0) = 0.32 pu From (9.25), the short circuit currents in the lines are. Va(F) ~ Va(F) 228 The results are exactly the same as the values found in Example 9.1(a). ‘The reader is encouraged to repeat the above calculations for fault at buses 2 and 1, and com- Pare the results with those obtained from parts (b) and (c) in Example 9.1. Note that the values of the diagonal elements in the bus impedance matrix are the same as the Thévenin's impedances found in Example 9.1, thus eliminating95. ALGORITHM FOR FORMATION OF THE BUS IMPEDANCE MATRIX. 369 the repeated need for network reduction for each fault location. Furthermore, the off-diagonal elements are utilized in (9.24) to obtain bus voltages during the fault. ‘Therefore, the bus impedance matrix method is an indispensable tool for fault stud- ies. 9.5 ALGORITHM FOR FORMATION OF THE BUS IMPEDANCE MATRIX Before we present the building algorithm for the bus impedance matrix, a few def- initions from the discipline of the graph theory are introduced. The graph of a network describes the geometrical structure of the network. The graph consists of redrawing the network, with a line representing each element of the network. The graph of the network for Figure 9.2(a) before the fault application is shown in Fig- ture 9.11(a). The buses are represented by nodes ot vertices and impedances by ) FIGURE 9.11 Graph, a selected tree, and a cotre for the network of Figure 9:20). line segments called elements or edges. A tree of a connected graph is a connected subgraph connecting all the nodes without forming a loop. The elements of a tree are called branches. In general, a graph contains multiple trees. The number of branches in any selected tree denoted by bis always one less than the nodes, ie., ben-1 (9.26) where n is the number of nodes including the reference node 0. Once a tree for a ‘graph has been defined, the remaining elements are referred to as links. The cotlec- tion of links is called a cotree. If e is the total number of elements in a graph, the number of links in a cotree is lse-b an41 (9.27)370 9. BALANCED FAULT A loop that contains one link is called a basic Joop. The number of basic loops is unique; it equals the number of links and is the number of independent loop equations. A cut set is a minimal set of branches that, when cut, divides the graph into two connected subgraphs. A fundamental cut set is a cut set that contains only one branch, The number of fundamental cut sets is unique; it equals the number of branches and is the number of independent node equations. Figure 9.11(b) shows a tree of a graph with the tree branches highlighted by heavy lines and the cotree links by dashed lines, ‘The bus impedance matrix can be built up starting with a single element and the process is continued until all nodes and elements are included. Let us assume that Zius matrix exists for a partial network having m buses and a reference bus 0 as shown in Figure 9.12, Reference FIGURE 9.12 Partial network ‘The corresponding network equation for this partial network is Vous = Ziuslous 0:28) For an n-bus system, m buses are included in the network and Zpus is of order ‘m x m, We shall add one element at a time from the remaining portion of the network until all elements are included. The added element may be a branch or a link described as follows. ADDITION OF A BRANCH ‘When the added element is a branch, a new bus is added to the partial network creating a new row and a column, and the new bus impedance matrix is of order (m +1) x (m+ 1). Let us add a branch with impedance zpg from an existing bus to anew bus q as shown in Figure 9.13(a). The network equation becomes95. ALGORITHM FOR FORMATION OF THE BUS IMPEDANCE MATRIX. 371 1 1 2 12 Partial Pp q Partial TP network -}—*————* network -—* Fis | im Ts Lim 0 0 q Reference Reference @ ) FIGURE 913 [ado ofa branch pa % Zn Ba Zap Zim Zag vs Za ea oni Vp |=] Zp. Zon oe Spy Zo Zon (929) Vn | | Sm Zma - Zp + Bim Zing Vel [Zn Zao Lap + Zam Zep ‘The addition of branch does not affect the original matrix, but requires the calcu- lation of the elements in the q row and column, Since the elements of the power system network are linear and bilateral, Zs = Zig, for g = 1,... 7. First, let us compute the elements Z,: for i = 1,...,m and i # q (i.e excluding diagonal element Zqq). To calculate these elements we will apply a cur- rent source of 1 per unit at the ith bus, ic., J; = 1 pu, and keep remaining buses open-circuited, ie., I {mand k # i. From (9.29), we get (9.30) From Figure 9.13(@) Va = Vo tm 31)372 9. BALANCED FAULT Where tpy is the voltage across the added branch with impedance zg, and is given by pa = pai 0.32) Since added element p-g is a branch, pg = 0, thus upg = 0 and (9.31) reduces to Zu = 2p To calculate the diagonal element Zyg, we will inject a current source of 1 per unit at the qth bus, ic, fy = 1 pu, and keep other buses open-circuited, From (9.29), we have sm i¢g (9.33) Vy= Zag (9.34) Since at the qth bus, the injected current flows from the bus g towards the bus p, ‘pg = Iq = =1. Hence, (9.32) reduces to pq = —Epq (935) Subs ting for vpg in (9.31), we got Ve = Vp+ Zpq (936) Now, since from (9.30) for i = q, Vy = Zqq and Vp = Zpq, (9.36) becomes Zag = Lg + 29g 937) Ifnode pis the reference node as shown in Figure 9.13(b), Vp = 0 and we obtain Zui = yp = sam ik (9.38) From (9.37), the diagonal element becomes 2oa = 99 (9.39) ADDITION OF A LINK ‘When the added element is a cotree link between the bus p and q, no new bus is created. The dimension of the Zu, matrix remains the same bat all the elements are required to be calculated. Let us add a link with impedance zpq between two existing buses p and q as shown in Figure 9.14(a). If Zz is the current through the ‘added link in the direction shown in Figure 9.14(a), we have Zpqle = Vp —Vq (9.40)95, ALGORITHM FOR FORMATION OF THE BUS IMPEDANCE MATRIX. 373 1 cH Partial | i network |—4P we LT, jaa |_§ Reference @ FIGURE 9.14 ‘Addition of a link pg. or network es Vy—Vo+2pale=0 aly ‘The added link modifies the old current Jp to (Jp — Je) and the old current I, to (q+ Ie) as shown in Figure 9.14(a), and the network equation becomes Vie Zuht Zah+ Zah+ + Van = Zrii +Zip(Ip ~ Te) + ZaglIq + e+ + + tZpp(Ip ~ Ie) + ZpalIg + Le) + + ZaplIp ~ Te) + ZaglIq + Te) + +2 mp(Ip — Te) + Za Iq + Te) -* ‘Substituting for Vp and V, from (9.42) into (9.41) results in (Za — Zp ++ (Zam — Zpmn\len + (2pq + Zpp + Zaq ~ 2Zpq)le = 0 +Zimlm +2 pmlm. ++ +Zomim +Z mmm + (Zap — Zpp)p + +++ + (Zaa ~ Zpa)lq oo + (9.42) (9.43)374 9, BALANCED FAULT Equations in (9.42) plus (9.43) result in m +1 simultaneous equations, which is written in matrix form as v Zuo Zap Zig > Zim Ze] fh Y Zp. > Bye Zpq ++ Lym Zoe | | Ip Va f=] Za s+ Zap Zag ++ Zam Zqe || Iq | 9.44) Vn 2mm Zrrm Ze | | Im 0 Za Zim Ze} | Te where (9.45) and Zee = 29q + Zpp + Zqq — 22 pq (9.46) Now the link current I; can be eliminated. Equation (9.44) can be partitioned and rewritten in compact form as old [Yer] =| Zo 32 | iw] 04 where AZ=[ Ze + Zoe Zee 2me \" (9.48) Expanding (9.47), we get Vous = Zi Tous + AZIe (9.49) and O= AZT Iyus + Zeele (9.50) or Ie SE ne (9.51) ‘Substituting from (9.51) for Zz in (9.49), we have Vine = [zeit - ABA? (9.52) Zu195, ALGORITHM FOR FORMATION OF THE BUS IMPEDANCE MATRIX. 375 ot Vous = Zins Tous (9.53) where AZAZT ew old Bee = Big — (9.54) Note that (9.54) reduces the matrix to its original size. The reason for this is that wwe have not added a new node but only linked two existing nodes. “The bus impedance matrix can be constructed with addition of branches and links in any sequence. However, itis best to select a tree that contains the elements connected to the reference node. If more than one element is connected between a given node and the reference node, only one element can be selected as a branch placing other elements in the cotree. The step-by-step procedure for building the ‘bus impedance matrix which takes us from a given bus impedance matrix Zits, to anew Zpcis summarized below. Rule \ddition of a Tree Branch to the Reference Start with the branches connected to the reference node. Addition of a branch zgo between a new node q and the reference node 0 to the given Zit, matrix of order (im x m), results in the ZY matrix of order (m+ 1) x (m+ 1). From the results of (9.38) and (9.39), we have Zy + Zim 0 gewn| i 0 0 9.55) ms 0 Zane 0 _ 0 0 20 ‘This matrix is diagonal with the impedance values of the branches on the diagonal. Rule 2: Addition of a Tree Branch from a New Bus to an Old Bus Continue with the remaining branches of the tree connecting a new node to the existing node. Addition of a branch zp between a new node q and the existing node p to the given Z¢!d, matrix of order (m x m), results in the Ze” matrix of order (m + 1) x (m+ 1). From the results of (9.33) and (9.37), we have Zu Zip Zim Ap a ee 056 Zang + Zum Zp Up + Bam mv FF376 9, BALANCED FAULT Rule 3: Addition of a Cotree Link between two existing Buses, When a link with impedance zy is added between two existing nodes p and q, we augment the Zl4 matrix with a new row and a new column, and from (9.44) and (9.45) we have Ir Lap Bag Lam ag Zap Zp Boe Dias = Zp Zoq Zima 2p 2g * Za. Zp Zap—Zpp Zag Zon * where Zee = tq + Lop + Zqq — 22 pq 9.58) ‘The new row and column is eliminated using the relation in (9.54), which is re- peated here Azazt Lys = Zig — (9.59) a = is — (59) and AZ is defined as Zaq— Zip =| 27 2p az=| Poe (0.60) Zina ~ ‘When bus q is the reference bus, Z; = Zig 0 (for i = 1,m), and (9.57) reduces Zp + Zim Zip 61) Zana, —an‘95, ALGORITHM FOR FORMATION OF THE BUS IMPEDANCE MATRIX 377 where Zee = 299 + Zpp, and (9.62) The algorithm to construct the Zpus matrix by adding one element at a time can be used to remove lines or generators from the network. The procedure is identical to that of adding elements, except that the removed element is considered ‘as negative impedance, in order to cancel the effect of the element. Based on the above algorithm, two functions named Zbus = zbuild(zdata) and Zbus = zbuild(linedata, gendata, yload) are developed for the formation of the bus impedance matrix. These functions are described in Section 9.6. Before demonstrating this program, for the sake of better understanding the building algo- rithm, we shall demonstrate the hand calculation procedure for the simple three-bus network of Example 9.1. Example 9.3 Construct the bus impedance matrix for the network in Example 9.1. The one-line impedance diagram is shown in Figure 9.15(a). FIGURE 9.15 Impedance diagram of Example 9.1 and a proper tee. ‘The elements connected to the reference node are included in the proper tree as shown in Figure 9.15(b). We start with those branches of the tree connected to the reference node. Add branch 1, 219 = 0.2 between node q = 1 and reference3789, BALANCED FAULT node 0. According to rule 1, we have 2) = Zy = a Next, add branch 2, 299 = j0.4 between node q = 2 and reference node 0 2 -[% 0 ]_[ 902 0 mw 0 Ze Jl 0 joa Note that the off-diagonal elements of the bus impedance matrix are zero. This is because there is no connection between these buses other than to the reference. In this example, there are no more branches from a new bus to the reference. We con- tinue with the remaining branches of the tree. Add branch 3, z1 = 70.4 between the new node q = 3 and the existing node p = 1. According to rule 2, we get @ [2 42 Zn j02 0 502 2Q2.=| 21 22 Zr |=| 0 joa 0 By Zo Anta 502 0 506 All tree branches are in place. We now proceed with the links. Add link 4, 212 = 40.8 between node q = 2 and node p = 1. From (9.57), we have Zu Zn fay Ean 2un 2) Zam Bay in — in ca Zag Za 2, 2-2 taba Yay- Bra Zan 02 0 502 -702 0 04 0 504 50.2 0 706 —j0.2 50.2 50.4 -j0.2 Zyq From (9.58) Zsa = 21a Zur + Za — 22g = 50.8 + 50.2 + 0.4 — 250) = j1.4 and AZAZT 1 [790.2 aq} i04 | [-s02 jo4 -jo2 Za ja we 402 PA —702] ~30.08714 50.1428 —j0,05714 0.02857 —30.05714 0.02857 70.02857 ~j0.05714 0.0285795, ALGORITHM FOR FORMATION OF THE BUS IMPEDANCE MATRIX, 379) From (9.59), the new bus impedance matrix is j02 0 502 0.02857 —j0.05714 0.02857 z= | 0 jo4 0 |-| -j0.05714 70.11428 —j0.05714 702° 0 506 0.02857 —j0.05714 0.02857 0.17143 j0.08714 0.17143 = | 0.05714 5028571 70.05714 0.17143 j0.05714 50.57143 Finally, we add link 5, 223 (0.4 between node q = 3 and node p = 2. From (0.57), we have ay 22 43 Zis— Zia 7. | % %m =a tas In us 2s 20, 23 xa Zan 21 ~ 2 232-2 Zsg— Fas Zan 0.17143 0.05714 0.17143 0.11429 _ | s005714 50.28571 j0.08714 ~30.22857 = | jo.17143 0.05714 5.57143 70.51429 50.1429 —70.22857 70.51429 Zu From (9.58) Zana 209+ Zant Zs — 222g = j0.4+§0.28571 + J0.57143 —2( 0.05714) = j1.14 and T 0.11429 Bear. 1 | —jo2nse7 | [j0.11429 -jo22867 jo.s1420] a d 50.51429 j0.01143 ~70.02286 0.05143 = | -y0.02286 0.04571 —70.10286 0.05143 —70.10286 70.23143 From (9.59), the new bus impedance matrix is ‘j0.17143 0.05714 j0.17143] f j0.01143 —70.02286 70.5143 Lous = [ms 15714 0.28571 50. wna] [= “70.0286 0.04571 pi 50.17143 30.05714 j0.57143] | 70.05143 —70.10286 0.23143 40.16 50.08 40.12 - [ae oat 21| 5012 70.16 70.34 This is the desired bus impedance matrix Zius, Which is the same as the one ob- tained by inverting the Yius matrix in Example 9.2.380 9, BALANCED FAULT Example 9.4 ‘The bus impedance matrix for the network shown in Figure 9.16 is found to be FIGURE 9.16 Impedance diagram for Example 94, 0.183 0.078 70.141 Zeus = | 90.078 50.148 70.106 30.141 50.1068 50.267 The line between buses 1 and 3 with impedance Z3 = j0.56 is removed by the simultaneous opening of breakers at both ends of the line. Determine the new bus impedance matrix. ‘The removal of an element is equivalent to connecting a link having an impedance equal to the negated value of the original impedance. Therefore, we add link 2; —30.56 between node q = 3 and node p = 1. From (9.57), we have Zn Za Za is Zn lne=| 22 Zn Zn Zea Za a Za 2a Ya Zan Zar @- 2 Z2-2i2 ss - Lis 2 Thus, we get 40.183 0.078 0.141 ~j0.042 Zine =| 30:078 0148 70.106 70.028 ms | 50.141 70.106 70.267 50.126 30.042 50.028 0.126 Zag96, ZBUILD ANDSYMFAULT PROGRAMS 381 From (9.58) Zua= 213+ Zit 253-2213 = —J0.56 + j0.183 + j0.267—2(j0.141) = and o 30.042 ear = =pyq | 10.028 | [70.042 50.028 70.126 } ad 70.302 | 0.196 “79.0030 —J0.0020 —70.0090 —j0.0045 50.0030 sn | 30.0135 30.0090 —0.0405 From (9.59), the new bus impedance matrix is 0.183 0.078 70.141 .0045 70.0030 0.0135 Zins [ 70.078 50.148, ‘uo | -| “40.0030 —70.0020 —. sc 50-141 30.106 70.267 30.0135 —30.0090 —30.0405 40.1875 0.0750 40.1275 = [® 0750 70.1500 za | {70.1275 0.1150 70.3075 9.6 ZBUILD AND SYMFAULT PROGRAMS ‘Two functions are developed for the formation of the bus impedance matrix. One function is named Zbus = zbuild(2data), where the argument zdata is an ¢ x 4 matrix containing the impedance data of an e-element network. Columns 1 and 2 are the element bus numbers and columns 3 and 4 contain the element resistance ‘and reactance, respectively, in per unit, Bus number 0 to generator buses contain generator impedances. These may be the subtransient, transient, or synchronous re- actances. Also, any other shunt impedances such as capacitors and load impedance to ground (bus 0) may be included in this matrix. ‘The other function for the formation of the bus impedance matrix is zbus = zbuildpi(linedata, gendata, yload), which is compatible with the power flow programs. The first argument linedata is consistent with the data required for the power flow solution. Columns 1 and 2 are the line bus numbers. Columns 3 through 5 contain line resistance, reactance, and one-half of the total line charging suscep- tance in per unit on the specified MVA base. The last column is for the transformer tap setting; for lines, 1 must be entered in this column. The lines may be entered in any sequence or order. The generator reactances are not included in the linedata of the power flow program and must be specified separately as required by the gen- data in the second argument. gendata is an ng x 4 matrix, where each row contains