BEGIN NEW DATA CASE C BENCHMARK DC-23 C 1st of 2 parts: Simple TACS-only test including steady-state ac solution. C 1st of 8 total subcases. Following 2 TACS data subcases, there will be 5 C data subcases that illustrate Kizilcay frequency dependence (add May, 93). 0.1 1.0 1 1 1 1 1 TACS STAND ALONE RES1=0 +1/1+S -1/S RES2=0 +1/1+S +S/1+S -MIXER 11/1+S +MIXER 1. 1. 1. 11/S +S/1+S 1. 0. 1. 1S/1+S +MIXER 0. 1. 1. 1. MIXER +DC +AC 11DC 1.0 14AC 1. -90. 33S/1+S 1/S 1/1+S RES1=0RES2=0MIXER DC AC C Next 8 output variables belong to TACS (with "TACS" an internally-added C Step Time TACS TACS TACS TACS TACS TACS C S/1+S 1/S 1/1+S RES1=0 RES2=0 MIXER C 0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 C 1 0.1 .952380952 .047619048 .047619048 0.0 0.0 1.0 C 2 0.2 .861678005 .138321995 .138321995 -.2776E-16 -.111E-15 1.0 BLANK card ending all TACS data C 10 1.0 .386918466 .613081534 .613081534 -.3331E-15 -.4441E-15 1.0 C Variable max: .952380952 .613081534 .613081534 0.0 0.0 1.0 C Times of max: 0.1 1.0 1.0 0.0 0.0 0.1 C Variable min: 0.0 0.0 0.0 -.3331E-15 -.4441E-15 0.0 C Times of min: 0.0 0.0 0.0 1.0 1.0 0.0 PRINTER PLOT 143 .2 0.0 1.0 S/1+S 1/S DC { Axis limits: (0.000, 1.000) BLANK card ending plot cards BEGIN NEW DATA CASE C 2nd of 8 cases: Simple TACS-only test of supplemental TACS device type 66 ABSOLUTE TACS DIMENSIONS 10 40 50 20 20 200 2000 100 C TEST FOR DEVICE 66 .0001 .03 1 1 1 1 1 -1 5 5 20 20 TACS STAND ALONE DUM +ZERO 11DC1 1.0 11DC2 1.0 -1.0 14AC1 1.0 50.0 14AC2 1.0 50.0 -90. -1.0 14AC3 1.0 100.0 -1.0 88DV66A 66+AC1 +DC1 50. 88DV66B 66+AC2 +DC2 50. 88DV66C 66+AC2 -DC2 50. 88DV66D 66+AC2 +DC1 50. 88DV66E 66+AC1 +DC2 50. 88DV66F 66+AC2 +AC3 +DC2 50. 33DC1 DC2 AC1 AC2 DV66A DV66B DV66C DV66D DV66E DV66F C Next 10 output variables belong to TACS (with "TACS" an internally-added u C Step Time TACS TACS TACS TACS TACS TACS C DC1 DC2 AC1 AC2 DV66A DV66B C 0 0.0 0.0 1.0 0.0 .6123E-16 0.0 1.22474487 C 1 .1E-3 1.0 1.0 .99950656 .031410759 .141386465 1.22474487 C 2 .2E-3 1.0 1.0 .998026728 .06279052 .199876678 1.22474487 BLANK card ending all TACS data C 300 .03 1.0 1.0 -1. .39893E-13 1.22474487 1.22474487 C Variable maxima : 1.0 1.0 1.0 1.0 1.22474487 1.22474487 C Times of maxima : .1E-3 0.0 .02 .005 .0282 .03 C Variable minima : 0.0 1.0 -1. -1. 0.0 1.22474487 C Times of minima : 0.0 0.0 .01 .015 0.0 .0246 PRINTER PLOT 144 3. 0.0 30. DV66F DV66A { Axis limits: (0.000, 1.414) BLANK card ending plot cards BEGIN NEW DATA CASE C 3rd of 8 subcases begins the illustration and verification of Kizilcay C frequency dependence, named after Dr. Mustafa Kizilcay of Lahmeyer C International in Frankfurt, Germany. His doctoral dissertation at C the University of Hannover was dominated by such modeling, although C that work may have been restricted to Laplace transformation (possible C alternate use of Z transformation is believed to be newer). In either C case, TACS-like transfer functions are allowed not in TACS, but rather C in the electric network. As subcases 3 through 6 are added to this C standard test case on 8 May 1993, the most easily available reference C for the average EMTP user would seem to be an 11-page contribution C entitled "A New Branch in the ATP-EMTP: High-Order, Linear Admittance C Model." This was published by LEC in its international journal titled C EMTP News (see Vol. 6, No. 1, March 1993, pp. 19-29). In WordPerfect C format, the associated disk file soon should be available from Prof. C Bruce Mork's Fargo server of E-mail fame. Details should be found in C the April, 1993, newsletter. C Ordinary RLC branch representation C Compare the results of this subcase with that of the following one. C The circuit is driven by a voltage step (type-11 source). C Series R-L-C branch has R = 0.05 ohm, L = 1 mH, C = 5 mF C and the resistive load has RL = 0.05 ohm C DELTAT TMAX XOPT COPT .000500 .050 0. 0. C IOUT IPLOT IDOUBL KSSOUT MAXOUT 1 1 1 1 1 -1 5 5 10 10 RES 0.05 GEN RES 0.05 1.0 5000. 1 BLANK card -- ending electrical branches -- BLANK card -- ending switches -- 11GEN 10. BLANK card -- ending sources -- C Step Time RES GEN GEN C RES C 0 0.0 0.0 0.0 0.0 C 1 .5E-3 .120481928 10. 2.40963855 C 2 .1E-2 .349833067 10. 6.99666134 1 C 100 .05 -.00286701 10. -.05734015 C Variable maxima : .943967142 10. 18.8793428 C Times of maxima : .0035 .5E-3 .0035 C Variable minima : -.66356661 0.0 -13.271332 C Times of minima : .0105 0.0 .0105 BLANK card -- ending plot cards -- BEGIN NEW DATA CASE C 4th of 8 subcases is to be compared with the preceding 3rd. The two C solutions should be identical. The same RLC circuit is represented C here using a second order rational function as an admittance in the C S-domain of Laplace transformation. This provides validation of C Kizilcay frequency dependence for a very simple case where we have C an alternate means of computing the transient. C DELTAT TMAX XOPT COPT .000500 .050 0. 0. C IOUT IPLOT IDOUBL KSSOUT MAXOUT 1 1 1 1 1 -1 5 5 10 10 RES 0.05 C -- preceding R branch defines node names and output option C <....>: dummy value! GEN RES 99. 1 C The following illustrates F95 ability to size Kizilcay frequency-dependent C tables. This documents the format. It is legal for F77, however, only C because the 3 numbers shown equal the fixed, default dimensions. If any C one of these numbers is changed for F77 use, an error termination should C result. Any blank data field is converted automatically to the default C value shown below. Change made 20 October 2000: C MAXHAN MAXORD MAXTOT C KIZILCAY F-DEPENDENT LIMITS 30 25 1000 C The preceding has been moved downward 2 subcases. We want the first use C of KFD not to have this declaration because lack of it caused trouble for C F95 Lahey ATP prior to a correction to HANNFD on 19 November 2002. C ---Request-------> < Order><------ Gain --------> | KIZILCAY F-DEPENDENT 2 1.0 S-transform C -coeff. of numerator--><-coeff. of denominator-> 0.0 1.0 0.005 2.5E-4 0.0 5.0E-6 BLANK card -- ending electrical branches -- BLANK card -- ending switches -- 11GEN 10. BLANK card -- ending sources -- C Step Time RES GEN GEN C RES C 0 0.0 0.0 0.0 0.0 C 1 .5E-3 .120481928 10. 2.40963855 C 2 .1E-2 .349833067 10. 6.99666134 1 C 100 .05 -.00286701 10. -.05734015 C Variable maxima : .943967142 10. 18.8793428 C Times of maxima : .0035 .5E-3 .0035 C Variable minima : -.66356661 0.0 -13.271332 C Times of minima : .0105 0.0 .0105 BLANK card -- ending plot cards -- BEGIN NEW DATA CASE C 5th of 8 subcases continues illustration and verification of Kizilcay C frequency dependence. Whereas subcases 3 and 4 considered only step C excitation of the time-step loop beginning with 0 initial conditions, C the concluding two cases will involve a continuation of the sinusoidal C steady state. Here, we first establish a standard of comparison by C using only tradition EMTP modeling. C Ordinary RLC branch representation C Compare the results of this subcase with that of the following subcase. C The circuit is driven by a sinusoidal voltage source. C Series R-L-C: R = 0.05 ohm, L = 1 mH, C = 5 mF C and resistive load, RL = 0.05 ohm C DELTAT TMAX XOPT COPT .000500 .050 0. 0. C IOUT IPLOT IDOUBL KSSOUT MAXOUT 1 1 1 1 1 -1 5 5 10 10 RES 0.05 GEN RES 0.05 1.0 5000. 1 BLANK card -- ending electrical branches -- BLANK card -- ending switches -- 14GEN 10. 50. -1. BLANK card -- ending sources -- C Step Time RES GEN GEN C RES C 0 0.0 .438670455 10. 8.7734091 C 1 .5E-3 .212452642 9.87688341 4.24905284 C 2 .1E-2 -.01899859 9.51056516 -.3799719 1 C 100 .05 -.44323561 -10. -8.8647122 C Variable maxima : 1.48971876 10. 29.7943752 C Times of maxima : .016 0.0 .016 C Variable minima : -1.491383 -10. -29.827659 C Times of minima : .026 .01 .026 BLANK card -- ending plot cards -- BEGIN NEW DATA CASE C 6th of 8 subcases continues illustration of Kizilcay F-dependence. C This shows KIZILCAY F-DEPENDENT modelling of the series RLC C circuit in the Z-domain. The RLC circuit is represented by a second order C rational function of admittance type in Z-transform. The coefficients are C obtained by applying the bilinear transform to the rational function in C Laplace domain. C DELTAT TMAX XOPT COPT 0.0005 0.05 50. C IOUT IPLOT IDOUBL KSSOUT MAXOUT ICAT 10 1 1 1 1 2 RES 0.05 C -- preceding R branch defines node names and output option C <....>: dummy value! GEN RES 99. 1 C The following illustrates F95 ability to size Kizilcay frequency-dependent C tables. This documents the format. It is legal for F77, however, only C because the 3 numbers shown equal the fixed, default dimensions. If any C one of these numbers is changed for F77 use, an error termination should C result. Any blank data field is converted automatically to the default C value shown below. Change made 19 November 2002: C MAXHAN MAXORD MAXTOT KIZILCAY F-DEPENDENT LIMITS 30 25 1000 C ---Request-------> < Order><------ Gain --------> | KIZILCAY F-DEPENDENT 2 1.0 Z-transform C -coeff. of numerator--><-coeff. of denominator-> 10.0 41.0 0.0 -79.0 -10.0 40.0 BLANK card -- ending electrical branches -- BLANK card -- ending switches -- 14GEN 10. 50. -1. BLANK card -- ending sources -- C Step Time RES GEN GEN C RES C 0 0.0 .443567421 10. 8.87134842 C 10 .005 -1.4216487 -.315E-14 -28.432974 1 C 100 .05 -.44356742 -10. -8.8713484 C Variable maxima : 1.48913814 10. 29.7827628 C Times of maxima : .036 0.0 .016 C Variable minima : -1.4891381 -10. -29.782763 C Times of minima : .026 .01 .026 BLANK card -- ending plot cards -- BEGIN NEW DATA CASE C 7th of 8 subcases illustrates higher-order Kizilcay F-dependence. C It is a single-phase subset of the 3-phase data named KFDHEV.DAT C that was received from Orlando Hevia attached to E-mail dated C 23 November 2002. But Orlando said that he received it from Ricardo C Ricardo Tenorio of ABB in Vasteras, Sweden. It seems this is a low- C or intermediate-order approximation that was "derived from ATP C simulations, although the system represented is a real one" (see C E-mail dated 2002-12-11, 7:39). Mr. Tenorio was the first to report C stability problems of KFD for high order and small dT as first C mentioned in the January, 2003, newsletter. Well, that high-order C data is proprietary (secret), so can not be disclosed. But the C general phenomenon can be illustrated using the lower-order C approximation of this subcase. Consider what happens if dT differs C from the 2 microseconds (usec) of the miscellaneous data card: C dT = 3 usec would be perfectly sinusoidal. That final cycle [60, 80] msec C has maximum current = 4117 (at 61.06 msec) and min = -4112 (at 71.07 C msec). This is very symmetrical and balanced. Also, the phasor solution C has 4138, which seems plenty close. So all is well at dT = 3 usec using C 64 bits of Salford precision. But 2 usec differs. It adjusts to sinusoid C during 1st cycle. The phasor solution is not too different at 4138. But C the dT loop is quite different. That final cycle [60, 80] msec has maximum C current of only 3272 (at 61.24 msec) and min = -3322 (at 71.28). Yet, C even with the obvious error (too small amplitude), stability seems good. C Both 1.9 and 2.1 usec do _not_ adjust, but rather become highly distorted C over the first 40 msec (1st of 2 plots). The second plot, over [40, 80], C clearly is diverging. The amplitude of the current grows exponentially. C So, on both sides of dT = 2.0 usec there is instability whereas dT = 2.0 C seems perfectly stable & sinusoidal at 80 msec. All of this is using 64 C bits of Salford EMTP precision. Using 128-bit Lahey, there is no hint of C distortion anywhere. The graph for dT = 3.0 usec seems identical to 1.9 C or 2.0 usec. Conclusion: for 64 bits, dT = 3 usec is close to the C stability limits of the KFD model. C Addition on 10 December 2002: Having tested Watcom, GNU Mingw32, and C 64-bit F95 Lahey, it can be reported that all behave comparably, but C differently. Each has different roundoff, so numbers are different. C But the graphs look similar. So, be advised that one can not easily C compare any two of the .LIS files using Mike Albert's FC. Human C intelligence is required. But for larger time steps, comparisons are C perfect. To see this, consider the 4th subcase of DC-48. This uses C the same data, but a larger step size of dT = 100 usec. FC indicates C perfect agreement in this case. PRINTED NUMBER WIDTH, 13, 2, { Request maximum precision (for 8 output columns) BEGIN PEAK VALUE SEARCH .060 { Compare max & min over just 1 cycle POWER FREQUENCY 50. { Nicer period than using Hevia's 60 Hz .0000020 .080 1 7 1 1 1 -1 5 5 20 20 100 100 1000 1000 C ONLY KFD ELEMENT IS HERE: THE CASE IS UNSTABLE FOR SMALL DELTAT C NETWORK EQUIVALENT BY KFD C <-BUS1<-BUS2<-BUS3<-BUS4<----R<----L<----C SOURCA 99. 1 C PUNCH-OUT FILE GENERATED BY ARMAFIT (NODA SETUP) KIZILCAY F-DEPENDENT 6 1.00000E+000 S 2.55564999999999990E-002 1.00000000000000000E+000 2.58174256593059220E-006 1.26190169827623520E-003 3.44528114161255420E-008 6.56757488674167030E-007 2.16578732571550020E-012 6.22037676357393080E-010 7.77318622992575270E-015 1.10717058895704090E-013 2.25856143523397890E-019 5.18376352266273340E-017 3.19918022205034290E-022 4.41171918010679540E-021 BLANK card ending branch cards BLANK card ending switch cards (none) 14SOURCA 187794.214 50. 0. -1. C 14SOURCA 187794.214 60. 90. { -1. BLANK card ending source cards SOURCA BLANK card ending node voltage output requests CALCOMP PLOT { Switch to screen from printer plot of a preceding subcase 194 4. 0.0 40. SOURCA dT = 2.0 usec H(s) current 194 4. 40. 80. SOURCA 2nd 40 msec H(s) current BLANK card ending plot cards BEGIN NEW DATA CASE C 8th of 8 subcases continues illustration of Kizilcay F-dependence. C The following illustrates F95 ability to erase Kizilcay frequency-dependent C tables. This documents the format. For F77, no storage actually is closed, C however. For F77, the declaration has no effect on memory. 0.0005 0.05 1 1 C MAXHAN MAXORD MAXTOT KIZILCAY F-DEPENDENT LIMITS -1 { Special 1st of 3 integers ==> close DUMMY 1.0 { Dummy branch will confirm branch data input $STOP { Since done showing cancellation of Kizilcay tables, terminate execution BEGIN NEW DATA CASE BLANK