BEGIN NEW DATA CASE C BENCHMARK DC-48 C Demonstration of random switch opening: A Monte Carlo ("STATISTICS") study C only with random opening rather than the more common closing. The network C is purely resistive by design, in order that switch current is known to C have a current zero at 0.625 seconds. There are 3 parallel, independent C networks radially branched off of generator GEN to nodes LOAD, LOAD1, C and LOAD2. All 3 involve statistical opening. If a switch opens early C (before .625 sec), then negative peak of -100 at t = 3/4 will be kept C for extrema. If not (nearly half the time, since the mean of the switches C shorting LOAD, LOAD1, and LOAD2 to ground is t = 0.6, which is very C close to the current zero), the opening is delayed until t = .875, and C then the peak is +30.9 at the end time t = 0.9 seconds. Why? This is C because frequency equal to 2 Hz and use of BEGIN PEAK VALUE SEARCH at C 0.525 means that we ignore the positive peak at 0.5 sec, and we never C make the next one at 1.0 sec since TMAX = 0.9 sec. The negative peak C at 0.750 sec is seen only for the early opening at 0.625 sec. As for the C randomness, we have mixed random closing and opening as an illustration C of this new capability that first worked on 9 July 1987. All the actual C switchings of interest (three) use opening, whereas the reference switch C (one of two components of the opening time) uses random closing. Because C of use of built-in random numbers (data value XMAXMX is negative), all C computers should produce identical results. Confirmation of the actual C opening times can be deduced from the resultant peak voltage: C 1) For opening at 0.63 sec, source voltage appears at this time, and it C lasts through the end of the simulation. The negative peak occurs C at time 0.75 sec, and this shows up in the printout (the extrema C will have a value of -100.0 volts). C 2) For opening at 0.875 sec, source voltage appears at this time, and C only last through the positive rise toward the peak at 1.0 sec. But C this rise is prematurely ended at time TMAX = 0.90 sec, when the C voltage is 30.901699 volts, and this will be the extrema. C Not all computers require that variable NSEED of columns 73-80 of the C STATISTICS miscellaneous data card be punched, meaning use of the fixed C seed. But Apollo does require this, since built-in random numbers are no C different than installation-dependent ones: the VAX-11 algorithm is used. PRINTED NUMBER WIDTH, 18, 3, { Extra precision and separation for 3 variables CHANGE PRINTOUT FREQUENCY 5 5 BEGIN PEAK VALUE SEARCH, 0.525, { Ignore 1st cycle, which has time less than 1/2 .01 .90 1 1 0 0 1 1 5 C ISW ITEST IDIST IMAX IDICE KSTOUT NSEED 1 1 0 0 1 1 GEN LOAD 1.0 { 1st of 3 identical resistors hung from source GEN LOAD1 1.0 { 2nd of 3 identical resistors hung from source GEN LOAD2 1.0 { 3rd of 3 identical resistors hung from source DUMY 1.0 { Connect independent Monte Carlo switch node BLANK card ending branch cards DUMY .40 .06 STATISTICS { Reference C Preceding switch does nothing for network; but it is referenced by the two C dependent switches of LOAD1 and LOAD2. Following LOAD is independent: LOAD .60 0.10 3333.STATISTICS { Independent LOAD1 .20 0.08 3333.STATISTICSDUMY { Dependent C Mustafa's STAT*.DAT within \TEPCO introduces current margin Ie to open: C < n 1>< n 2>< Tclose >< Ie >< type > C GEN LOAD .0051 .0028868 3. 3333.STATISTICS 1 C Make a comparable addition here, to the LOAD2 switch: C LOAD2 .20 0.08 3333.STATISTICSDUMY { Dependent C LOAD2 .20 0.08 40. 3333.STATISTICSDUMY { Dependent LOAD2 .20 0.08 45. 3333.STATISTICSDUMY { Dependent C The preceding Ie = 40 advanced the LOAD2 switch-opening time slightly: C With zero or blank: Open switch "LOAD2 " to " " after 6.30000000E-01 sec. C With Ie = 40 amps : Open switch "LOAD2 " to " " after 6.00000000E-01 sec. C But this was just base case output. Extrema were not affected. Adding five C amps (40 becomes 45) affected the extrema, so data is left this way. BLANK card terminating switch cards 14GEN 100. 2.0 -1. BLANK card terminating sources LOAD LOAD1 LOAD2 GEN C Step Time LOAD LOAD1 LOAD2 GEN C C *** Phasor I(0) = 1.0000000E+02 Switch "LOAD " to " " closed C *** Phasor I(0) = 1.0000000E+02 Switch "LOAD1 " to " " closed C *** Phasor I(0) = 1.0000000E+02 Switch "LOAD2 " to " " closed C 0 0.0 0.0 0.0 0.0 100. C 1 .01 0.0 0.0 0.0 99.211470131448 C 2 .02 0.0 0.0 0.0 96.858316112863 BLANK card ending requests for program outputs (just node voltages, here) C *** Open switch "LOAD2 " to " " after 6.00000000E-01 sec. C 60 0.6 0.0 0.0 0.0 30.901699437494 C *** Open switch "LOAD " to " " after 6.30000000E-01 sec. C *** Open switch "LOAD1 " to " " after 6.30000000E-01 sec. C 65 .65 -30.9016994375 -30.9016994375 -30.9016994375 -30.9016994375 C 70 0.7 -80.9016994375 -80.9016994375 -80.9016994375 -80.9016994375 C 75 .75 -100. -100. -100. -100. C 80 0.8 -80.90169943749 -80.90169943749 -80.90169943749 -80.90169943749 C 85 .85 -30.90169943749 -30.90169943749 -30.90169943749 -30.90169943749 C 90 0.9 30.901699437495 30.901699437495 30.901699437495 30.901699437495 C Variable maxima: 30.901699437495 30.901699437495 30.901699437495 92.977648588825 C Times of maxima: 0.9 0.9 0.9 .53 C Variable minima: -100. -100. -100. -100. C Times of minima: .75 .75 .75 .75 PRINTER PLOT BLANK card ending non-existent plot cards for the base case C The following documents extrema for the 3rd shot as found on 22 March 2007: C Random switching times for energization number 3 : C 1 3.3795450E-01 2 4.3170384E-01 3 6.2506590E-01 4 6.5220506E-01 C -100. -100. -100. -100. C Times of maxima : .75 .75 .75 .75 C This is using Ie = 45 amps. Using an older version of ATP with this new C data should produce the following different and wrong answer: C -100. -100. 30.901699437495 -100. C Times of maxima : .75 .75 0.9 .75 C How old? A month or more. The wrong answer corresponds to use of the C current margin Ie = 0 with either the old or the new program. As C pointed out by Prof. Mustafa Kizilcay about a month ago using other data, C ATP seemed to be ignoring current margin for the 2nd and later shots. In C response, WSM proposed avoidance by making TENERG of STARTUP negative C (simply append a minus sign). This was prior to correction of code, of C course. That 3rd output variable is the voltage at node LOAD2 --- C the node of the switch that has current chopping. STATISTICS DATA 1.0500001 0.0 LOAD LOAD1 LOAD2 C Note about change to preceding. The LOAD2 tabulation continues to span C compartments 7-20. However, compartment 7 used to have 2 but now it has C only one energization. Where did one of them move? To final compartment C 20 which increased from 3 to 4. The change of summary statistics is big: C Old solution : Mean = 7.15001430E-01 ... Variance = 1.26750507E-01 C New solution : Mean = 8.45001690E-01 ... Variance = 8.45003380E-02 C This is for "Grouped data." The "Ungrouped data" are comparable. C Add "FIND" usage with "DISK" output to demonstrate that Ie of the LOAD2 C switch is not erased on 17 March 2007. Prior to correction of DICTAB, C this was the case according to a trouble report from Orlando Hevia. FIND { Enter interactive search for which shot and which variable gave extremum LOAD LOAD1 { Search these 3 node voltages for extremum EXCLUDE { Repeat the preceding search after 1st excluding the just-found shot DISK /LIST /FULL { Create disk file of data for deterministic simulation of EXCLUDEd shot QUIT { Done locating extrema, so exit the "FIND" alternative C Historical note. Upon learning that we had corrected Ie treatment for Prof. C Mustafa Kizilcay, Orlando Hevia reminded the workers in Portland that he had C submitted a comparable complaint earlier. In his mail archive, he found a C message to BPA having time stamp Fri, 28 Jan 2005 17:37:39 -0300. Without C doubt, Orlando's complaint was the same, fundamentally. Why it was not C handled at the time remains unclear. In that an additional problem with C DISK output was noted, Orlando's complaint was more complete. How did C Orlando first note the phenomenon? He explained : "The first users that C discovered a problem with Ie and 3333 switches are a couple from Uruguay .." BLANK card ending requests for statistical tabulation BEGIN NEW DATA CASE C 2nd of 5 subcases verify that STATISTICS (Monte Carlo) can involve TACS. C The data comes from BPA's Dan Goldsworthy. Between August of 1993 and C May ?? of 1994 (see the July, 1994, newsletter), this was not possible. C Data is not at all realistic. TMAX is artificially small, etc.. LINEAR BIAS USAGE { Reference angle will vary linearly rather than randomly PRINTED NUMBER WIDTH, 11, 1, { Restore default precision for dt-loop columns C $STARTUP, dc37star.dat { Use disk file for re-initialization at this point C This is a Monte Carlo case, for which TENERG is critical. But this C parameter is initialized only in the STARTUP file. For this 2nd or C later data case, we reinitialize via the preceding $STARTUP request CHANGE PRINTOUT FREQUENCY 5 5 20 20 .000100 .020 60. 0.0 1 1 0 0 1 0 0 0 3 1 2 1 1 TACS HYBRID CASE 90PCBA 90A1 88PS5A = ABS(PCBA - A1) 93STATA 88PS2A 58+PLUS1 1000. 0.0 1.0 STATA 88PS4A 56+PS2A 1. -1.0 480. 0.0 480. 8.0 0.0 1000.0 0.0 9999. 88PS6A = (PS5A - PS4A) .AND. STATA 88CLSA 62+PLUS1 PS6A 33PS5A PS2A PS4A PS6A 33STATA 33CLSA BLANK CARD ENDING TACS DATA CASE SENDA PCBA 10.0 5.0 A1 10.0 15. DUMMY 1. STATA 1. BLANK CARD ENDING BRANCH CARDS DUMMY .001 .000001 STATISTICS 0 76STATA .005 .0015 STATISTICSDUMMY 0 13PCBA A1 CLSA 13 BLANK CARD ENDING SWITCH CARDS 14SENDA 188.0 60. 0.0 -1. BLANK CARD ENDING SOURCES PCBA A1 C Step Time PCBA PCBA A1 PCBA TACS C A1 A1 PS5A C 0 0.0 188. 188. 0.0 0.0 0.0 C 1 .1E-3 187.866421 187.866421 0.0 0.0 187.866421 C 2 .2E-3 187.465873 187.465873 0.0 0.0 187.465873 C 3 .3E-3 186.798926 186.798926 0.0 0.0 186.798926 BLANK CARD ENDs NODE VOLTAGE OUTPUTS C 120 .012 -35.227687 -35.227687 0.0 0.0 35.2276871 C Switch "PCBA " to "A1 " closing after 1.32000000E-02 sec. C 140 .014 0.0 70.7309673 70.7309673 .964122164 0.0 C 160 .016 0.0 113.976248 113.976248 4.51879549 0.0 C 180 .018 0.0 88.9755078 88.9755078 6.91674682 0.0 C 200 .02 0.0 12.7036093 12.7036093 6.17355737 0.0 C Variable maxima : 188. 188. 114.548243 7.06193095 187.985156 C Times of maxima : 0.0 0.0 .0163 .0186 .0083 C Variable minima : -187.98516 -187.98516 0.0 0.0 0.0 C Times of minima : .0083 .0083 0.0 0.0 0.0 BLANK CARD ending plot C Random switching times for energization number 1 : C 1 6.5564603E-03 2 9.7966634E-03 C ==== Table dumping for all subsequent restorations. Time [sec] = 9.0000000E-4 C Switch "PCBA " to "A1 " closing after 1.52000000E-02 sec. C 120. 188. 188. 125.885042 6.21833213 187.985156 10.25 480. C Times of maxima : 0.0 0.0 .016 .0189 .0083 .02 .1E-3 C C MAIN20 dumps OVER12 dice seed SEEDSV = 5.7295779513082323E+01 C Statistical distribution of peak voltage at node "A1 ". The base voltag C Interval voltage voltage in Frequency Cumulative C number in per unit physical units (density) frequency C 1 0.0500000 0.94165000E+01 2 2 C 2 0.1000000 0.18833000E+02 0 2 C < < Etc. (omit the zero rows) > > C 14 0.7000000 0.13183100E+03 1 3 C Summary of preceding table follows: Grouped data Ungrouped data C Mean = 2.41666667E-01 2.22809327E-01 C Variance = 1.40833333E-01 1.48931988E-01 C Standard deviation = 3.75277675E-01 3.85917075E-01 0 188.33 A1 { One and only request for statistical tabulation, of A1 C Column positioning of "NO SWITCH PLOTS" on blank terminator is arbitrary: BLANK card ending statistical tabulation requests --- NO SWITCH PLOTS --- BEGIN NEW DATA CASE C 3rd of 5 subcases verify that STATISTICS (Monte Carlo) works with MODELS C The case is similar to subcase 2, with TACS section converted to a model C Data is not at all realistic. TMAX is artificially small, etc.. LINEAR BIAS USAGE { Reference angle will vary linearly rather than randomly PRINTED NUMBER WIDTH, 11, 1, { Restore default precision for dt-loop columns CHANGE PRINTOUT FREQUENCY 5 5 20 20 .000100 .020 60. 0.0 1 1 0 0 1 0 0 0 3 1 2 1 1 MODELS INPUT pcba { v(PCBA) } -- voltage at node PCBA a1 { v(A1) } -- voltage at node A1 stata { switch(STATA) } -- switch status at node STATA OUTPUT clsa -- used to contol switch PCBA-A1 MODEL subcase3 INPUT dv -- voltage across controlled switch sw_status -- switch status of type-76 switch OUTPUT sw_control VAR sw_control, dvi, ps2, ps4, ps6, ps2x, ps2i FUNCTION ptlst POINTLIST (-1E6,480),(0,480),(8,0),(1E6,0) INIT sw_control :=0 histdef(ps6):=0 integral(ps2x):=0 histdef(ps2x) :=0 ENDINIT EXEC ps2x := sw_status * 1000 ps2 := integral(ps2x) -- corresponds to TACS type-58 ps4 := ptlst(ps2) -- corresponds to TACS type-56 ps6 := (dv-ps4) AND sw_status IF ps6>0 AND prevval(ps6)<=0 THEN -- corresponds to TACS type-62 sw_control := 1 ENDIF ENDEXEC ENDMODEL USE subcase3 AS sub INPUT dv:=abs(pcba-a1) -- voltage across controlled switch sw_status:=stata -- status of type-76 switch OUTPUT clsa:=sw_control -- control signal to controlled switch ENDUSE RECORD sub.dv AS dv sub.ps2 AS ps2 sub.ps4 AS ps4 sub.ps6 AS ps6 stata AS stata clsa AS clsa ENDMODELS SENDA PCBA 10.0 5.0 A1 10.0 15. DUMMY 1. STATA 1. BLANK CARD ENDING BRANCH CARDS DUMMY .001 .000001 STATISTICS 0 76STATA .005 .0015 STATISTICSDUMMY 0 13PCBA A1 CLSA 13 BLANK CARD ENDING SWITCH CARDS 14SENDA 188.0 60. 0.0 -1. BLANK CARD ENDING SOURCES PCBA A1 C Step Time PCBA PCBA A1 PCBA TACS TACS TACS TACS TACS TACS C A1 A1 DV PS2 PS4 PS6 STATA CLSA C 0 0.0 188. 188. 0.0 0.0 188. 0.0 480. 0.0 0.0 0.0 C 1 .1E-3 187.866421 187.866421 0.0 0.0 187.866421 0.0 480. 0.0 0.0 0.0 C 2 .2E-3 187.465873 187.465873 0.0 0.0 187.465873 0.0 480. 0.0 0.0 0.0 C 3 .3E-3 186.798926 186.798926 0.0 0.0 186.798926 0.0 480. 0.0 0.0 0.0 BLANK CARD ENDs NODE VOLTAGE OUTPUTS C 120 .012 -35.227687 -35.227687 0.0 0.0 35.2276871 6.05 117. 0.0 1.0 0.0 C Switch "PCBA " to "A1 " closing after 1.32000000E-02 sec. C 140 .014 0.0 70.7309673 70.7309673 .964122164 0.0 8.05 0.0 0.0 1.0 1.0 C 160 .016 0.0 113.976248 113.976248 4.51879549 0.0 10.05 0.0 0.0 1.0 1.0 C 180 .018 0.0 88.9755078 88.9755078 6.91674682 0.0 12.05 0.0 0.0 1.0 1.0 C 200 .02 0.0 12.7036093 12.7036093 6.17355737 0.0 14.05 0.0 0.0 1.0 1.0 C Variable maxima : 188. 188. 114.548243 7.06193095 188. 14.05 480. 1.0 1.0 1.0 C Times of maxima : 0.0 0.0 .0163 .0186 0.0 .02 0.0 .0132 .006 .0132 C Variable minima : -187.98516 -187.98516 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 C Times of minima : .0083 .0083 0.0 0.0 .0133 0.0 .014 0.0 0.0 0.0 BLANK CARD ending plot C Random switching times for energization number 1 : C 1 6.5564603E-03 2 9.7966634E-03 C ==== Table dumping for all subsequent restorations. Time [sec] = 9.00000000E-04 C Switch "PCBA " to "A1 " closing after 1.52000000E-02 sec. C 120. 188. 188. 125.885042 6.21833213 188. 10.25 480. 1.0 1.0 1.0 C Times of maxima : 0.0 0.0 .016 .0189 0.0 .02 0.0 .0152 .0098 .0152 C C MAIN20 dumps OVER12 dice seed SEEDSV = 5.7295779513082323E+01 C Statistical distribution of peak voltage at node "A1 ". The base voltage for per unit printout is V-base = 1.88330000E+02 C Interval voltage voltage in Frequency Cumulative Per cent C number in per unit physical units (density) frequency .GE. current value C 1 0.0500000 0.94165000E+01 2 2 33.333333 C 2 0.1000000 0.18833000E+02 0 2 33.333333 C < < Etc. (omit the zero rows) > > C 14 0.7000000 0.13183100E+03 1 3 0.000000 C Summary of preceding table follows: Grouped data Ungrouped data C Mean = 2.41666667E-01 2.22809327E-01 C Variance = 1.40833333E-01 1.48931988E-01 C Standard deviation = 3.75277675E-01 3.85917075E-01 C 0 188.33 A1 { One and only request for statistical tabulation, of A1 BLANK card ending statistical tabulation requests --- NO SWITCH PLOTS --- BEGIN NEW DATA CASE C 4th of 5 subcases is unrelated to other subcases other than that it C involves Monte Carlo simulation. It is added 27 November 2002 to C demonstrate that KIZILCAY F-DEPENDENT (KFD) modeling is compatible C with STATISTICS (and/or SYSTEMATIC or START AGAIN, although only C STATISTICS is demonstrated). As first reported by Ricardo Tenorio of C ABB in Vasteras, Sweden, there was incompatibility (execution would C hang after the 1st energization had begun) prior to correction on C 16 November 2002. Ricardo's initial complaint was about minimum dT (see C the January, 2003, newsletter), but shortly thereafter, privately, he C reported inability to use STATISTICS switches with KFD data. The data C used here is the same as will be found in the 7th subcase of DC-23. C Step size dT is plenty big, so there is no question of instability C caused by roundoff error. A switch has been added in series with the C voltage source, and the associated mean and standard deviation are C arbitrarily chosen to give a nice distribution of 4 energizations. C Finally, the sinusoidal source has been restored to 60 Hz (in DC-23, C 50 Hz was used in order that the period be a rounder number), and the C angle has been rotated so the voltage upon closure is small. I.e., C the closures are artificially soft, intentionally, so the base case C plot has no big spike at 4 msec. PRINTED NUMBER WIDTH, 13, 2, { Request maximum precision (for 8 output columns) .000100 .040 5 1 1 0 1 -1 4 C ISW ITEST IDIST IMAX IDICE KSTOUT NSEED 1 1 0 0 1 { KSTOUT is blank, not 0! } 0 60 20 100 100 ELEM 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 GEN ELEM .004 .001 STATISTICS 1 C Preceding column-80 punch really is not needed. It was added simply to C demonstrate that current output of the KFD branches agrees with it. BLANK card ending switch cards (none) 14GEN 187794.214 60. 20. -1. BLANK card ending source cards GEN C First 1 output variables are electric-network voltage differences (upper voltage minus lower voltage); C Next 2 output variables are branch currents (flowing from the upper node to the lower node); C Step Time GEN GEN ELEM C ELEM TERRA C 0 0.0 176468.8371 0.0 0.0 C 5 .5E-3 161307.6987 0.0 0.0 C 10 .1E-2 140432.1546 0.0 0.0 C 15 .0015 114581.7315 0.0 0.0 C 20 .002 84672.19336 0.0 0.0 C 25 .0025 51763.10058 0.0 0.0 C 30 .003 17020.27416 0.0 0.0 C 35 .0035 -18325.504 0.0 0.0 C 40 .004 -53022.092 0.0 0.0 C *** Close switch "GEN " to "ELEM " after 4.10000000E-03 sec. C 45 .0045 -85840.3459 -1534.79533 -1534.79533 C 50 .005 -115617.663 -2541.85468 -2541.85468 BLANK card ending node voltage output requests CALCOMP PLOT C 400 .04 -180519.385 -3271.23717 -3271.23717 C Variable maxima : 187793.4818 4132.085734 4132.085734 C Times of maxima : .0324 .0344 .0344 C Variable minima : -187793.482 -5009.38242 -5009.38242 C Times of minima : .0074 .0089 .0089 C Max>Name1 Name2 Name3 Name4 16-byte title Y-axis label 194 4. 0.0 40.-6.E34.E3ELEM dT = 100 usec H(s) current BLANK card ending base-case plot cards -2 4000. ELEM { -2 ==> tatistical tabulation of branch current C The preceding base current of 4K amps is close to the peak value of the C base-case (no-variance) solution. It results in per unit values of a C little more than 1. Using default compartment size of .05, this results C in reasonable compartment numbers, and a pleasing dispersion of results: C Statistical distribution of peak current for branch "ELEM " to " ". Base current for per unit output = 4.00000000E+03 C Interval current current in Frequency Cumulative Per cent C number in per unit physical units (density) frequency .GE. current value C 22 1.1000000 4.40000000E+03 0 0 100.000000 C 23 1.1500000 4.60000000E+03 1 1 75.000000 C 24 1.2000000 4.80000000E+03 0 1 75.000000 C 25 1.2500000 5.00000000E+03 1 2 50.000000 C 26 1.3000000 5.20000000E+03 0 2 50.000000 C 27 1.3500000 5.40000000E+03 1 3 25.000000 C 28 1.4000000 5.60000000E+03 1 4 .000000 C Summary of preceding table follows: Grouped data Ungrouped data C Mean = 1.26250000E+00 1.26057789E+00 C Variance = 1.22916667E-02 1.35490825E-02 C Standard deviation = 1.10867789E-01 1.16400526E-01 BLANK card ending statistical tabulation requests BEGIN NEW DATA CASE C 5th of 5 subcases is a modified version of the 1st. The FIND command C is added as earlier illustrated in DC-24. That already illustrates the C use of DISK to create deterministic data SHOT*.DAT corresponding to C an EXCLUDE-d shot. But what DC-24 does not illustrate is RESIMULATE C to simulate the just-created data. Also missing was random opening and C use of the optional /MEMSAV switch. So, provide these here. This C data subset is added 6 October 2002. Until 27 November 2002, it was C the 4th subcase. Note it always must be last. PRINTED NUMBER WIDTH, 18, 3, { Extra precision and separation for 3 variables CHANGE PRINTOUT FREQUENCY 5 5 BEGIN PEAK VALUE SEARCH, 0.525, { Ignore 1st cycle, which has time less than 1/2 .01 .90 C MEMSAV Note value 0 is keyed 1 1 0 0 1 0 1 2 C ISW ITEST IDIST IMAX IDICE KSTOUT NSEED 1 1 0 0 1 1 GEN LOAD 1.0 { 1st of 3 identical resistors hung from source GEN LOAD1 1.0 { 2nd of 3 identical resistors hung from source GEN LOAD2 1.0 { 3rd of 3 identical resistors hung from source DUMY 1.0 { Connect independent Monte Carlo switch node BLANK card ending branch cards DUMY .40 .06 STATISTICS { Reference C Preceding switch does nothing for network; but it is referenced by the two C dependent switches of LOAD1 and LOAD2. Following LOAD is independent: LOAD .60 0.10 3333.STATISTICS { Independent LOAD1 .20 0.08 3333.STATISTICSDUMY { Dependent LOAD2 .20 0.08 3333.STATISTICSDUMY { Dependent BLANK card terminating switch cards 14GEN 100. 2.0 -1. BLANK card terminating sources LOAD LOAD1 LOAD2 GEN BLANK card ending requests for program outputs (just node voltages, here) PRINTER PLOT $CLOSE, UNIT=4 STATUS=KEEP { Disconnect base-case .PL4 file for TPPLOT use $OPEN, UNIT=4 STATUS=C-like FILE=dummy.pl4 ! { Connect C-like scratch file BLANK card ending non-existent plot cards for the base case STATISTICS DATA 1.0500001 0.0 LOAD FIND { Enter interactive search for which shot and which variable gave extremum C Normally the user does not know which shot he wants, so he first will use C EXCLUDE to find the worst transients, and then he wants data for this. C See DC-24 for this technique. But it also is possible for the user manually C to name the shot of interest. This is what is done here: C EXCLUDE { Repeat the preceding search after 1st excluding the just-found shot C < Shot # Cols 25-32 are I8 data for shot # of interest DISK /LIST /MEMSAV 1 { Create determinstic disk file for shot # 1 RESIMULATE { Abort this data case. Begin with SHOT0001.DAT just created by DISK C Any data card following RESIMULATE should be ignored. So, this is a good C place to add miscellaneous comments. Begin these with mention of /OPEN C --- another flag of the DISK subcommand that is not being illustrated. C Because there is no /OPEN qualifier, $OPEN and $CLOSE cards will be C converted to comment cards. If /OPEN were added, they would be passed C through without modification. Another unused feature is /FULL (see DC-24 C for an illustration). Lack of /FULL means that switch times will be C optimally encoded but with a leading blank byte. This provides separation C from what precedes the number. Yes, one digit is being given up, but it is C highly unlikely that this will make any difference, considering the DELTAT C discretization of use. Use of /MEMSAV has no practical effect in this C case because the value keyed on the integer miscellaneous data card is zero. C This is deliberate. Normally, the field would be left blank, but if this C were the case, one could not be sure that the /MEMSAV flag was working. C The user will note that the zero is passed through, demonstrating opeation. C A final item: For the record, request cards that always are destroyed C (missing in the DISK output) are: C OMIT BASE CASE LINEAR BIAS USAGE C TRULY RANDOM NUMBERS MODELS PROCESSES EXTREMA C REPEATABLE RANDOM NUMBERS USER SUPPLIED SWITCH TIMES C RANDOM NUMBER GENERATOR SEED C 30 December 2003, add /CYCLE qualifier to continue to have T-open = 0.0 C As Orlando Hevia observed in E-mail of that day, this normally is not the C engineering choice. Normally, a switch that closes should remain closed, C and DICTAB is being modified to do this. T-close = 1.E+9 will be seen on C the modified switch card. To restore 0.0, add /CYCLE (which will allow C the switch to cycle open again (the status at time zero). BEGIN NEW DATA CASE BLANK