BEGIN NEW DATA CASE C BENCHMARK DC-7 C Test of 200-km line energization example from proposed 1975 IEEE PES C Winter Meeting paper by Tripathy et al. The far end of the line has C an open-circuited transformer, which the Indians modeled as a parallel C connection of a saturable reactor and a capacitor. For the reactor, a C true nonlinearity (Type-93) has been placed in phase "a", and pseudo- C nonlinear ones (Type-98 modeling) are used for phases "b" and "c". C The reactor of phase "a" is actually driven onto the final segment C above the point (i=55.572, PSI=624) on step number 185 at t= .004625 C seconds, so this represents a real stress test for the Type-93 element C 23 July 1983 ---- remove Type-98 modeling (replace with Type-93). PEAK VOLTAGE MONITOR PRINTED NUMBER WIDTH, 13, 2, { Full precision on each of 8 columns of printout .000025 .020 60. 60. 1 1 1 1 -1 5 5 10 10 100 20 C The following series capacitors represent 20% compensation of the C positive-sequence series line reactance at 60 Hz, at each end: GENA SWA 89352. GENB SWB GENA SWA GENC SWC GENA SWA RECA TRANA GENA SWA 1 RECB TRANB GENA SWA RECC TRANC GENA SWA TRANA GNDA 7.5398 1 GNDA .0001 TRANB 7.5398 TRANC 7.5398 C The following piecewise-linear magnetization curve was constructed C from Tripathy's arc-tangent function (evaluation at regular intervals): 93TRANA NAME 1st NL .559 300. 1 0.0 0.0 .5590 300. .9344 400. 1.2555 450. 1.8057 500. 3.0251 550. 4.9429 580. 8.4609 600. 13.092 610. 28.847 620. 55.572 624. 753.46 628. 9999 C ------------- Following original, pseudo-nonlinear modeling upgraded: C 98TRANB NAME 2nd NL .559 300. 1 C .5590 300. C .9344 400. C 1.2555 450. C 1.8057 500. C 3.0251 550. C 4.9429 580. C 8.4609 600. C 13.092 610. C 28.847 620. C 55.572 624. C 753.46 628. C 9999 C BRANCH NAME:3rd NL C 98TRANC TRANB .559 300. 1 C ------------- Preceding, original modeling was improved by replacement C on 23 July 1986 when multi-phase, Type-93, nonlinear C inductor logic was finally perfected. Instead of Type-98 C modeling in phases "b" and "c", we copy the original, C true nonlinear modeling. Replacement cards follow: BRANCH NAME:2nd NL ! { Hold lower case of this 2nd, but not following 3rd 93TRANB TRANA .559 300. 1 BRANCH NAME:3rd NL 93TRANC TRANA .559 300. 1 C Although Tripathy's line was untransposed, we make the continuously- C transposed assumption here for simplicity, it will be noted: -1SENDA RECA .137681.07755.6806124.27 -2SENDB RECB .03009.450289.5000124.27 -3SENDC RECC BLANK card ends all branch cards SWA SENDA 1.0 SWB SENDB 1.0 SWC SENDC 1.0 BLANK card ends all switch cards 14GENA 188000. 60. 0.0 -1. 14GENB 188000. 60. 120. -1. 14GENC 188000. 60. -120. -1. BLANK card ending source cards C Total network loss P-loss by summing injections = 0.000000000000E+00 C Step Time GENA TRANA RECA TRANA TRANB C TERRA TERRA C *** Switch "SWA " to "SENDA " closed after 0.00000000E+00 sec. C *** Switch "SWB " to "SENDB " closed after 0.00000000E+00 sec. C *** Switch "SWC " to "SENDC " closed after 0.00000000E+00 sec. C 0 0.0 188000. 0.0 0.0 0.0 0.0 C 1 .25E-4 187991.6504 .930228E-37 .930004E-37 -.57418E-39 0.0 C 2 .5E-4 187966.6022 .482578E-37 .482013E-37 .545732E-44 -.24529E-53 C 3 .75E-4 187924.8578 0.0 0.0 0.0 0.0 C 4 .1E-3 187866.4209 0.0 0.0 0.0 0.0 C 5 .125E-3 187791.2965 0.0 0.0 0.0 0.0 C 10 .25E-3 187165.6493 0.0 0.0 0.0 0.0 C 20 .5E-3 184670.0031 0.0 0.0 0.0 0.0 C 30 .75E-3 180535.2129 368529.9098 368561.0118 .0356766432 -.018316651 GENA TRANA RECA BLANK card ending output variable requests (just node voltages, here) C Last step begins: 800 .02 58095.19494 -59199.4152 -58796.2914 1.081609222 C Last step continued: .299848177 -.932282601 17.5862178 16.50460858 C Variable max : 188000. 513558.5344 514002.0191 749.9628668 3.237548678 C Times of max : 0.0 .0184 .0184 .00465 .01645 C Variable min : -187999.072 -498079.391 -497500.143 -513.064439 -2505.93213 C Times of min : .008325 .0077 .0077 .011675 .0073 C ------------------------------------------------------------------------- C To see how much improvement the 3-phase nonlinearity gives, compare these C values with the original solution with pseudo-nonlinear L(i) in "b", "c": C ------------------------------------------------------------------------- C Last step begins: 800 .02 58095.19494 -50098.9838 -49701.5785 1.086096522 C Last step continued: .3006006893 -.931914393 19.37948907 18.29339255 C Variable max : 188000. 520082.5127 520520.0418 749.9593109 3.304226242 C Times of max : 0.0 .0184 .0184 .00465 .01645 C Variable min : -187999.072 -498611.424 -498032.337 -516.239708 -2590.47657 C Times of min : .008325 .0077 .0077 .011675 .0073 C Plots: { Axis limits: (-5.107, 7.760) { Axis limits: (-0.443, 7.760) PRINTER PLOT { Axis limits: (-5.177, 7.760) 194 2. 0.0 20. RECA TRANA CURRENT 1 SMOOTH { Axis limits: (-0.439, 7.760) 194 .1 4.5 5.0 RECA TRANA CURRENT IN AMPS BLANK card terminating plot cards BEGIN NEW DATA CASE C 2nd of 4 subcases illustrates user-supplied FORTRAN to provide for smooth C modeling of magnetic saturation. Special logic presently is built into C only 2 modules: "INNONL" to bypass normal error checks on characteristic C (since parameters for hyperbolic tangent curve are inputted this way) and C to store these parameters in raw form in List-10 vectors CCHAR and VCHAR, C and second, "SOLVNL" to calculate "i" and di/dv for Newton's method C (as part of coupled nonlinear solution mixed with any other elements). A C single block of some 6 lines is involved in "INNONL" (see the check for C input text: ABUFF(33:39) .NE. 'FORTRAN'). A type-93 branch is being C used, although others pass through the same logic, so there is no magic C about this. Remembrance of this special component by "SOLVNL" is due C to numerical flag -333777 of first point of characteristic. Different C models could have different values in "SOLVNL", note --- no other code C is affected. The hyperbolic tangent saturation curve actually requires C only 2 cards of (x, y) characteristic, but a 3rd is required because of C existing Type-93 logic that extends the final segment by a factor of 1000 C (this would overlay the 2nd point, otherwise). There are 2 reactors, and C these are coupled via the network, resulting in a 2 x 2 Netwon solution C within "SOLVNL". But there is a 2nd Newton solution that might not be C noticed --- to find the current given the flux (or voltage). This is one C dimensional. As for changes to "SOLVNL" for user-supplied FORTRAN, a C single block is involved, below the check for flag -333777. Finally, C data for the conventional Type-93 element was copied from DC-4. C 25 August 2001, add the following test of table dumping and restoring C half way through the simulation. Answers are not affected. Note the C definition precedes the last miscellaneous data card (a requirement if C TSTALL is being used, as here): $DEPOSIT, TSTALL=-0.5 { Negative TSTALL ===> experimental dump/restore of tables PEAK VOLTAGE MONITOR, 3, { Request network extrema of both node & branch voltage PRINTED NUMBER WIDTH, 13, 2, { Request maximum precision (for 8 output columns) .020 4.0 1 1 1 1 1 -1 5 5 20 20 GEN NODE 100. { Create unknown-voltage "NODE" for coupling NODE TRAN 5.0 5.E4 1 93TRAN .005 30. 1 0.0 0.0 .005 30. .01 40. .02 45. .10 50. 5.0 100. 9999 TRAN 1000. { Need to damp hash within previous NL element NODE XXXX 500. { Current-limiting, phase-shifting resistor C User-supplied fortran follows. This is a regular Type-93 NL inductor until C the time-step loop. Note 3-card characteristic, followed by "9999" bound. C The characteristic parameter usage is: PSI = a * tanh ( b * i ) + c * i. 93XXXX .08 35.0 3 -333777. 35.FORTRAN { -333777 = flag; a = peak iron flux 20. 5.0FORTRAN { b = current mult; c = linear series L 1.0 1.0FORTRAN FLUX { Dummy 3rd card to protect card 2 9999 { End of user-supplied fortran (see request in cols. 33-39) BLANK card ending program branch cards. BLANK card terminating program switch cards (none, for this case) 14GEN 70. .1591549 -1. BLANK card terminating program source cards. C Total network loss P-loss by summing injections = 4.465360492400E+00 C GEN 70. 70. .12758172835428 .13968171687261 4.4653604924 C 0.0 0.0 -.0568672543615 -24.0240608 1.9903539026515 C ---- Initial flux of coil "TRAN " to " " = 2.86434431E+00 C ---- Initial flux of coil "XXXX " to " " = 3.08335629E+01 XXXX TRAN { Selective node voltage outputs: voltages across NL reactors C Step Time XXXX TRAN TRAN XXXX NODE C TERRA TERRA TRAN C 0 0.0 22.00346954 56.62762238 .4773907E-3 .0704767153 .0571050131 C 1 .02 22.4849728 56.58915579 .6660853E-3 .0696258393 .0572552411 C 2 .04 20.71249379 56.51611741 .8545941E-3 .0724907376 .0573707115 BLANK card ending output requests C 200 4.0 8.696586066 -17.9578249 -.058915918 -.077940538 -.076873743 C Variable max : 22.4849728 56.62762238 .0944331073 .1005485993 .0955171556 C Times of max : .02 0.0 1.4 .34 1.32 C Variable min : -40.5895565 -56.4586274 -.058915918 -.103607476 -.076873743 C Times of min : 2.48 3.0 4.0 3.4 4.0 PRINTER PLOT 143 .5 0.0 4.0 XXXX TRAN { Axis limits: (-5.646, 5.663) 193 .5 0.0 4.0 XXXX TRAN { Axis limits: (-1.036, 1.005) BLANK card ending all plot cards BEGIN NEW DATA CASE C 3rd of 4 subcases is modification of 2nd as 1st described in January, C 1998, newsletter story. Orlando Hevia of Sante Fe, Argentina, suggests C use of SINH rather than TANH for modeling of some magnetic saturation. C The numerical flag -333777 of TANH becomes -444777 for SINH use. PEAK VOLTAGE MONITOR, 3, { Request network extrema of both node & branch voltage PRINTED NUMBER WIDTH, 13, 2, { Request maximum precision (for 8 output columns) .020 4.0 1 1 1 1 1 -1 5 5 20 20 GEN NODE 100. { Create unknown-voltage "NODE" for coupling NODE TRAN 5.0 5.E4 1 93TRAN .005 30. 1 0.0 0.0 .005 30. .01 40. .02 45. .10 50. 5.0 100. 9999 TRAN 1000. { Need to damp hash within previous NL element NODE XXXX 500. { Current-limiting, phase-shifting resistor C User-supplied fortran follows. This is a regular Type-93 NL inductor until C the time-step loop. Note 3-card characteristic, followed by "9999" bound. C The characteristic parameter usage is: i = a * sinh ( b * psi ) + c * psi 93XXXX .08 35.0 1 -444777. 4.E-5FORTRAN { -444777 = flag; a = 1st param .16 6.E-5FORTRAN { b = 2nd of 3; c = 3rd of 3 params 1.0 1.0FORTRAN { Dummy third card to protect card 2 9999 { End of user-supplied fortran (see request in cols. 33-39) BLANK card ending program branch cards. BLANK card terminating program switch cards (none, for this case) 14GEN 70. .1591549 -1. BLANK card terminating program source cards. XXXX TRAN { Selective node voltage outputs: voltages across NL reactors C 25 August 2001, add the following test of table dumping and restoring C on step number . Answers are not affected. Note the preceding subcase C used TSTALL, so had to be defined early. Not so if integer ISTDMP is C directly defined. This can be done at anwhere that will be read prior C to entry into the dT loop: $DEPOSIT, ISTDMP=80 { Time step # for experimental dump/restore of tables BLANK card ending output requests PRINTER PLOT 143 .5 0.0 4.0 XXXX TRAN { Axis limits: (-5.646, 5.663) 193 .5 0.0 4.0 XXXX TRAN { Axis limits: (-1.036, 1.005) BLANK card ending all plot cards BEGIN NEW DATA CASE C 4th of 4 subcases is modification of 3rd as 1st described in January, C 1998, newsletter story. Orlando Hevia of Sante Fe, Argentina, suggests C use of PSI = A * i**B + C * i for modeling of some magnetic saturation. C The numerical flag -333777 of TANH becomes -555777 for i**B use. C Parameters are picked so solution to this 4th subcase approximates the C the solution to the 3rd. Printer plots have similar shapes and numbers. C 25 August 2001, cancel tests of table dumping and restoring (see two C preceding subcases). Note only TSTALL need be cancelled since ISTDMP C is cancelled automatically after each use. But TSTALL is not. TSTALL C will remain in effect for all later subcases, if not cancelled. $DEPOSIT, TSTALL=0.0 { Cancel experimental dumping and restoring of tables AUTO NAME { Toggle binary NMAUTO of STARTUP that controls branch/switch naming C The preceding is added 25 October 2002. It decreases List 7 from 9 to 3 as C the 6 network branches (there are no switches) no longer are being named. PEAK VOLTAGE MONITOR, 3, { Request network extrema of both node & branch voltage PRINTED NUMBER WIDTH, 13, 2, { Request maximum precision (for 8 output columns) .020 4.0 1 1 1 1 1 -1 5 5 20 20 GEN NODE 100. { Create unknown-voltage "NODE" for coupling NODE TRAN 5.0 5.E4 1 93TRAN .005 30. 1 0.0 0.0 .005 30. .01 40. .02 45. .10 50. 5.0 100. 9999 TRAN 1000. { Need to damp hash within previous NL element NODE XXXX 500. { Current-limiting, phase-shifting resistor C User-supplied fortran follows. This is a regular Type-93 NL inductor until C the time-step loop. Note 3-card characteristic, followed by "9999" bound. C The characteristic parameter usage is: psi = a * i**B + c * i 93XXXX .08 35.0 1 -555777. 100.FORTRAN { -555777 = flag; a = 1st param .22 0.0FORTRAN { b = 2nd of 3; c = 3rd of 3 params 1.0 1.0FORTRAN { Dummy third card to protect card 2 9999 { End of user-supplied fortran (see request in cols. 33-39) BLANK card ending program branch cards. BLANK card terminating program switch cards (none, for this case) 14GEN 70. .1591549 -1. BLANK card terminating program source cards. XXXX TRAN { Selective node voltage outputs: voltages across NL reactors BLANK card ending output requests PRINTER PLOT 143 .5 0.0 4.0 XXXX TRAN { Axis limits: (-5.646, 5.663) 193 .5 0.0 4.0 XXXX TRAN { Axis limits: (-1.036, 1.005) BLANK card ending all plot cards BEGIN NEW DATA CASE BLANK