Está en la página 1de 60
Cue W i 4 Ny e Pi Y ON ® . > . — J ct) ENMIP News Volume 3, Number 3, September 1990 Editors: D. Van Donmelen W.S.Meyer Assistants: G. Enpereur A, Laeremans In this issue... Information for contributors and subscribers 1 pe 1 2. European EMIP User Group information Pp. 2 : 3. Call for help pee 4. The eprovement to the present synchronous machine code Pp. 3 in UM modelling of ATP (D. Van Domuelen, Y. Yuexin, G. Eapereur) 5. Choice of De-sources for various initializations P. 23 (Y. Yuexin, G. Empereur) 6. Mass driver analysis using ENTP p. 29 (W. Neugebauer) 7. Apple Macintosh versions of ATP Pp. 38 (J.W. Schwartzenberg, Jr.) 8. Salford TPPLOT now scrolis graphics and supports p. 42 Epson Printers and .PCX files ; (WS. Meyer) 9. Sunmary of EWTP NEWS from North America Pp. 51 (W.S. Meyer, Tsu-huei Liu) 10. ATP benchmarks IMB RS/6000 RISC Workstation P. 57-59 (R. Niska, T. Kiriakedis, W.S. Meyer) INFORMATION FOR CONTRIBUTORS AND SUBSCRIBERS EMTP NEWS shall be a continuation from 1988 on of EMTP NEWSLETTER for the users of the versions of the Bonneville Power Administration EMTP and of the Alternative Transients Program (ATP), as distributed by the Leuven EMTP Center. EMTP NEWS is intended as a quarterly means of communication among these users and is made for and by them. In this context, readers are encouraged to contribute ideas suggestions, questions, and to share their experience in usage and modelling. Reviews of EMT related published Papers or books are greatly appreciated as well as advance notices about forthcoming publications in this field In order to simplify the editing, contributions must be typewritten, single-spaced and ready to be photocopied. The format shall therefore be plain white paper, size DIN A4 or 8 1/2 by 11 inches, with 20 mm or 1 inch margina on the right and left side. Selection among contributions is done by the editor, but the responsability of the content of a contribution rests with the author. If important editorial changes are judged necessary, consultation with the authors will be seeked prior to publication. No honorarium will be paid to authors. Please mail your contributions, ideas, suggestions, questions or notices to: Prof. Dr. D. Van Dommelen Katholieke Universiteit te Leuven Elektrotechniach Instituut Kardinaal Mercierlaan, 94 B-3030 Leuven (Heverlee) Belgium ‘ription may be taken at any time during the year, but will always run from January to December of the year in which it is taken. European ENTP User Group Information LEC finally obtained the rights to distribute Salford's PC386 run time library DBOS amongst the supporting members of the User Group. Accordingly, distribution of the PC386 version finally will start at the end of Septenber 1990. The user should not bother about compilers or Linkers (unless extraordinary table sizes would be needed for particular studies). We regret for the delay, but hope that you will be satisfied with this solution. HPGL and postscript support is available for data postprocessing (TPPLOT and PCPLOT). With the same respect, we slowly pick up in generating more flexible code for TPPLOT program, The definition of the interfacing standards kept us busy longer than expected. But results look very promising. Beta releases on Apollo and IBM mainframe are currently in testing. PC versions will be treated before too long. VAX support still seems to be a problem. Also the support of MODELS on VAC, CDC and BM mainframe is cumbersome. Hopefully, most of these problems will be resolved during Scott Meyer's stay at LEC during October. SUN support probably will be abandoned. Call for help The National Thermal Power Corporation would like to be in touch with experts/organizations who have experience in EMTP simulation of AC railway traction systems. Following points are of particular interest: ~ simulation of converters (Rectifier & inverter) with GTO firing - speed control of inverter fed induction motor - traction line simulation with voltage boosters ~ study of overvoltages & harmonics in AC traction system ‘The contact address for people who want to share their knowledge is the following: Indian EMTP User Group Att. Mr. A.K. Tripathy Sr.Ngr/TPE Block 6, 2nd floor BHEL Bhopal-462021 India ‘THE IMPROVEMENT TQ THE PRESENT SYNCHRONOUS MACHINE CODE IN UM MODELLING OF ATE Prof. D. Van Dommelen Yin Yuexin, G. Empereur LEC Leuven, Belgium Abstract ‘This paper presents the UM Modelling development in ATP. It reports some corrections to bugs regarding the simulation of synchronous machines using ATP (version 3). The correction to these bugs is verified in the enclosed examples. 1 Introduction ATP is a strong tool for power engineering problems, able to simulate not only the transient operation of rotating machinery, but also the steady state operation. Since both operations start from the initial conditions, these conditions become very important with respect to further simulation [Ref.1]. in ATP, we can distinguish between automatic and manual initialization. For automatic initialization of synchronous machine, again two different approaches can be considered: ~ specification of terminal voltage (calculated by the user, starting from loading conditions) - load flow option (a more direct way to obtain the above terminal voltage automatically). The automatic initialization option in which only the terminal voltage (for ‘synchronous machine) is necessary has successfully existed for many years. It can be used in the simulation of the balanced network. Although the load flow option (FIX SOURCE) would be easier to use in the context of automatic initialization (since it avoids manual precalculations by the user), unfortunately, this option malfunctions. ‘The manual initialization option is considered favourable to specify the initial conditions in case of unbalanced operation, although J.A. Martinez, G.A. Capolino, G. Bortoni have given some ideas regarding the unbalanced initial conditions using the automatic initialization option [Ref.2]. ‘The per unit system is widely used in power engineering since it gives the direct Comprehension to the machine parameters. In ATP, both the per unit as well as the Physical unit system can be used. 2 WM. Modelling and its interfaces Excitation Power network U. M . net work Mechanical network) Fig. 1: Schematic diagram of UM and its external networks In theory, UM modelling has three basic equations ((1), (2) and (5)) [Ref.3]. Power sid Vamp" “UR pllamp~ 4 amp)/ dt OLN Tramp ay with 0 0 0 tnj-[o0 0-1 o +1 0 Excitation side: ~Vane * “(Reant “FA ame)/Alt 2) The flux linkage is associated with currents Remo" (Lipliame* mp @ Rane ~ Use Hane ne @ Mechanical side: Tam ™ Mplighna~tadme* (Lig~ Lightate? (5) Equations (1) to (5) are used to calculate the phasor solution when automatic initialization is used. In the time-step loop the equations are transformed with the trapezoidal rule into: LA pi Mamn(t) = (t= 1+ LA OD ma (1+ [4 po ]Y amo!) co) Hama (* REE B14 [A CDT ne (+ [Aer ame) ” The meaning of the matrices Ag. Ay.Aw+Aq is irrelevant in this context, but are to be found in [Ref.3]. hp, he are the history terms which are well-known in the trapezoidal rule. To calculate the equations above, two interfaces are available; prediction-based versus compensation-based [Ref.4]. 2.1 Prediction-based interface ‘The prediction-based interface chooses the prediction of the armature flux from the power side as first step of calculation. The other values, voltages, currents and torque are then calculated from this flux. This option is not further taken into account in this text, but probably will be subject for a future report. 2.2 Compensation-based interface This interface adopts the Thevenin equivalent circuit to the external network including electrical and mechanical part (Fig. 2). Every terminal of UM is replaced by its Thevenin equivalent circuit. lume ta fl eee a) = Vg Vv, ia) Yona P Ze 4 v UM. ‘amp zy V, . 08> 1] Oy Tam Th Vt Fig. 2: UM and its Thevenin equivalent circuits ‘The UM terminal is replaced with its Thevenin equivalent circuits v,=(Z2,+02 (8) Ver(ZP lieve oe Var (Ze VR oy with mechanical parameters converted to electrical parameters using following conversion: Vysw ig? Tam 3 Thevenin equivalent circuit on the power side of synchronous machine with sompensation-based interface To calculate the equations above, the Thevenin equivalent parameters must be obtained beforehand. For the power side, the synchronous machine is linked with the infinite bus source Vs (type-14) through the transmission lines (R-L). The single phase circuit and its Thevenin equivalent .circuit are shown in Fig. 3. Al kf Ag Al oS —— ——_ 7 —> a ® vs v “.@ ’ zn 0 0 Fig. 3: Single-phase Thevenin equivalent circuit (power side) For the above single-phase circuit, one can easily derive following differential equation: Vv Rist S ay After applying the trapezoidal rule of integration, one can obtain the following difference equation: vene (ae henev cornistt-an a2) with 24 mistcr~aty=(e- 26 ic an+V,C-an-VE-an Setting i(t) = 0 and V™ = V finally will result in vee Jia-ansre-an-ve-ane1.co (3) mage Zt 4 20a ps2 a So far for the theory. But how about the program code ? After verification of the UM code of ATP, it appears that the manual initialization option uses the external node of the machine (node A in Fig. 3) as the Thevenin equivalent node. Hence, the above theory applies for manual initialization. On the other hand, the automatic initialization option chooses another node, the internal node of the machine as the Thevenin equivalent node at the power side. This implies that for automatic initialization, the leakage inductance Ly, of the synchronous machine is not taken into account in the machine equivalent but included in the external network. In the calculation for automatic initialization, the leakage inductance 1,, is set to zero in the UM equations while the outside transmission line L is augmented with the leakage inductance L,, (Fig. 4). ’ Aa fp aie 2 Ap 4A eae ae th 2 —— a v v Vs y ) eth Sd a 0 Fig.l: Single phase Thevenin circuit at power side with internal node of SM as Thevenin node (automatic initialization) Thevenin parameters thus becom ve ( 1 2a aners-an-VeranrV a0 7 (6) Remark that the power side Thevenin parameters of the automatic initialization option are different from those used in the case of manual initialization. 4 Nowadave troubles in using SM with UM code and suggested improvement In today's synchronous machine benchmarks in ATP, the automatic initialization option is effective only with the input data expressed in physical units (DCNEW - 2). Per unit input data is not suited for this option. On the other hand, the manual initialization option can be used successfully in combination with per unit input data. In combination with the physical unit system, the manual initialization option gives wrong results [Ref.5]. For improvement, two important items should be verified. a) The difference in base value between the transmission line and the internal machine parameters of UM causes the crucial mistake in the automatic initia- lization option using the per unit data system. b) Another important factor to evaluate is finding the reason why the power side Thevenin equivalent network was chosen differently for the manual and automatic initialization option. UM model Fig. 4 probably is responsible for the numerical oscillation in the case of automatic initialization. ‘The examples that will be provided below give both the original results as well as the results after code improvement. 5 Illustrative cases The performance of the synchronous machine model with compensation-based interface will be verified using 4 cases (see Table 1). Case 4 includes two subcases, one giving results before code improvement, another giving final results after code improvement. Furthermore, case 1 and 2 simulate the same system, whereas case 3 and 4 are similarly related. Table 1, Four simalation cases of synchronous machine Physical unit Per unit Automatic 1 4(a) - 4(2) initialization Manual fe 7 initialization Case 1: Automatic initialization (cfr. Fig. 4), data are in physical unit. - The input contains voltage value for nodes BUSA1 and BUSAO. - The program uses BUSA' rather than BUSA1 as Thevenin node. Gase 2: Manual initialization (cfr. Fig. 3), data are in physical unit. ‘The initial values of this case come from the case 1 results when t = 0. - The input thus contains voltage values for nodes BUSA, BUSB1, BUSC1 and BUSAO. Further the currents is, ip and i, are specified as well as the excitation side (ir, vr) and the mechanical side (ig, va, torque). = The program uses BUSA1 (rather than BUSA') as Thevenin node. Comparison: The results of case 1 and case 2 reasonably match. In section 6, we will elaborate on this observation. Gase 3: Manual initialization (cfr. Fig. 3), input data for both the SM as well as the electrical network are in per unit. There is no relation with cases lor 2, The program uses BUSA1 as Thevenin node. Gase 4(1): Automatic initialization (cfr. Fig. 4), data are in per unit - original code was used. The terminal voltage (magnitude and angle) at the node Al of the single-phase Thevenin circuit (Fig. 4) can be obtained from the case 3 results as follows: For t = 0 in case 3, the terminal voltages Vai, Ve1, Ver obtained are: V .4(0) = 1-56469(1) ¥4\(0) = -0,907544(7) a7 Ve,(0) = =1.2571466(7) Because these should be 3-phase symmetric voltages V4(0)*V,cosa V9,(0) = V qcos(a- 120°) «sy V¢,(0) = V,cos(a+ 120°) So, the terminal voltage of the single phase circuit to be specified in case 4 equals: V 4\(0) = 1.657962219.31° w cas) ‘This terminal voltage is the sole initial condition to be specified in the case of automatic initialization. ‘The result of the case 4(1) simulation is wrong: the torque is oscillating, and the torque angle is increasing. Furthermore, attention should be paid to the currents obtained in the first step of the output. The current of BUS-Al to BUS-AO is different from IPA. So are the B-phase and C-phase currents. The factor is almost 100! In section 6, we will elaborate on this observation. Gase 4(2): Automatic initialization, data are in per unit - corrected code is used. 6 Discussion From the results above, case 1 and case 2 almost get the same results. But, the results of case 3 and case 4(1) are entirely different. Only after some modification in the model and program code, case 4 (i.e. case 4(2)) yields the same results as case 3. This can be explained as follows. In equation (6) (i.e. calculation of the armature currents), the third item of this formula can be listed as follows: LM ane) (20) with [4q]°(F 9)(6 5] (6,] being the inverse of the armature resistance. (Fp]=(U3]* 206 plLip)/aty' (21) 10 [U,] being the 3rd-order unit matrix Vnp(t) being the Thevenin voltage (V™) of the power side (only for the compensation-based interface). So far for the general theory. Let us now turn to the numerical values of typical parameters in the different cases: Physical unit Per unit Automatic 912.4 initialization Lip* L = 0.5605 x 10-3 0.151061 vm = 29199.65 vm = 417.6 G.F,v™ = 1.89 x 107 | G,F,V™ = 3.81 x 10 (case 1) (case 4(1)) Manual G, = 648.51 912.4 initialization Fp = 2.755 x 10-4 3.65 x 106 Lig* L = 0.5595 x 10-3 Vs 1,061 x 103 vm = 17650.48 vm 2 1.75 GaF,V™ = 3.15 x 103 | G.FVt = 5,83 x 10-3 (case 2) (case 3) Note: - V™ at second time step - values before code improvement Let us first discuss the per unit situation. Recall that in case of manual initialization, L,, is forced to zero, so Ljp,*L-1.061F-3 (case 3) is nothing but the value for L being equal to 1.061 E-3. From case 4(1), we thus can derive that Lip equals 0.15. This leads to the observation that in case of per unit values eae Hence, the Thevenin voltage in case of manual initialization (formula (13)) largely differs from the Thevenin voltage in case of automatic initialization (formula (15)). Similarly, [Fp] (formula (21)) is much larger in the case of automatic initialization than it is in case of manual initialization. Indeed, in case of automatic initialization [F,] = [V3], because L,, is forced to zero. On the other hand, for the manual initialization, the item “stu! snall. is large, resulting in [Fp] being very To conclude, an error in VT using the automatic initialization option will produce @ larger error than in case of manual initialization, when using per unit input data. Probably this\s the origin for the oscillations observed in the output of case 4(1). It further explains why case (3) results are acceptable. For the physical unit system, other conclusions can be drawn. When comparing cases (1) and (2), it can be concluded that 12/,,. Both have about the same order of magnitude. W Consequently, the Thevenin voltages do not differ much. Also for the physical unit system, (F,] for automatic initialization is larger than [F,] for manual initialization. "However, the difference is not so large as in case of automatic initialization. To conclude, an error in V™" will have less impact to the final result. But still, differences exist. It thus was a challenge to apply the same code corrections to all cases. After applying the code modifications referred to in section 4 (and explained in more detail in section 8), results are improving and converging to the results obtained in the manual initialization option (for both the per unit as well as the physical unit system). Physical unit Per unit Automatic G, = 648.51 Gp = 912.4 initialization F, = 2.755 x 10-4 Fp = 3.65 x 10% Lip* L = 0.5595 x 10-3 | Lip+L = 1.06 x 103 Vm = 17650.48 vm = 1.75 GyF,V™ = 3.15 x 103 | G,F,V™ = 5.83 x 10-3 (case 1 (after code | (case 4 (after code improvement) ) improvement) ) 7 Conclusion UM code in ATP is still not fully tested for all types of rotating machine. It is accepted that UM modelling not only can be universal for all kinds of electrical machine, but also can be flexible to different options so that it can be efficient in particular operation condition. This paper explained how to use the different options (manual versus automatic initialization, per unit versus physical unit system) correctly. After performing the above corrections, the SM code within UM modelling seems to work correctly. On the other hand, it is unclear whether other options of UM modelling still work correctly. Besides the code correction presented here, one should also take care when specifying the sources. This item is subject to a seperate paper. 8 Attachment 1: listing of examples 8.1 CASE 1 awpur FILE Tent of autoastic U.M. initialization using J-phase aymchronous sods ‘The operation is in the ateedy sate-The machine data comer from the BENCHMARK. DeNEW-2 PRINTED NUMBER WIDTH. 13, 2, ( Request ABSOLUTE U.M. DIMENSIONS. 20, 2."50, 100 Pom vnEQUENCY, 60, { Corrects possible S0-Hs declaration of European STARTUP 000200 0.070 sximun prectaton (for 6 output coluans) a auseo dose © sossevevs TRANSMISSION LINES (0.01 OMS AND 1.0 MILLI HENRY) usa ausao ior-2 1.0 2 USB: aUsa0 suSA ausa0 i i BUSEO BUSAI BUSAO i C oseeconees, EMEP CONNECTIVITY REQUIREMENT buses. Tore avers ausro oes 0am. susHsH dora 1 LARK card bounds the Last branch card BUMNK card enda the last electric network switch card co necoe S"EReHtATION SOURCE VOLTAGE (ACTUAL VALUE ET BY 83) Laavara 0.001, 0.00002 Bo Co voceweoes INPINITE BUS SOURCE VOLTAGES Laaveno 3000.0 60.0 0.0, 0 Taavsno 300010600 -30.0 Ne Laauseo 3000.0 600 s420.0 ue © seseosons MECH INPUT TORQUE’ (ACTUAL VALUE SET BY $3) + Laavens 1 0.0000001 0.00008 “10 Co nnnensne= SYNCHRONOUS MACHINE SPECIFICATION 1s om 1 BLAME CARD ENDING CLASS 1 UM OATA Coenccreeee UNL MRCHSTABLE 42am 0.0973 8.0087015 0.00sse50 3030.0 iso uses suse Co ntececcee UCI COMMTARLE ‘usa a 0.001542 0.000895 bust 1 2-ocise2 00005595 use] i o-0010' 00003767 BusFO 1 0.01634 00002075 1 0.07560 0 .0001835 1 BIA card ena’ sll U-n. dace carde ‘BIRWK Gerd ending ll electric network sources BUSAL USBI guscl BUSH BUSHoM DLRWK card ending telective variable output raquest (node voltes here) 194-8. 0.0 96, wer 2008 (plot luates: (-0.913, 1.467) 394 8! 010 36. mcd Gaon (Piet Limite: (0.000: 3.770) 194 8. 0.0 96. oni ea (plot Mates! (10334, 0.718) BLnMK card ending batch-node plot cards 12 OUTPUT FILE Soss.onsess 61 8.2 CASE 2 EOIN NEW DATA cASE ‘Tent of sanual U.M. initialization using 3-phese synchronous node ‘The operation is in the ateaty stave, The eachine data comen fromthe & Benesmann Done? PRINTED NUMBER WIDTH. 13, 2, { Request maximum prectaion (for 8 output coluans) ABSOLUTE U.M. DIMENSIONS, 20, 2, 50. 100 POWER FREQUENCY. 60, { Corrects possible SO-Hs declaration of European STARTUP a Coe ee soe 20 0) 10 1.08 ioe © nnsenseo TARNSMISSION LINES (0.01 OHMS AND 2.0 MILLE HENRY) aus ala0a-n0 Yioe=2.1.0 a ‘uS-8108-208us-A1BuS-n0 1 ‘BUS-c1AU-cOmUS-ALBUS-AO i cg Usscsenees EMEP CONNECTIVITY REQUIREMENT Bus-ro ropes prey aes 1 ‘suum cara bounds the Last Granch card BLANK card enda tha Laat electric network awitch card C neecosnos" EICHTATION SOURCE VOLTAGE ianugcro 3.4824 0.00001 9.0, a0 14nug-n0 3000.0" 60.0. 9.9, a0 Y4pvs-80 300010 60.0 120.0, io Tanus-eo 3000.0 60.0 120.0, io e = MECH INPUT TORQUE Lasvs-ne-i24951.08420.00001 0.0 a0 tom CC nnccrnene Une HACHSTABLE a2 inser 0.0373 376.991118 0.005785 127232036 0,0088850 eon Geet contrat ‘us-at 3 2073.89367 001542 0.008598 US Bi i S.ccise2 00008595. Muse a 00010 :0003767 — Bus-Fo a olorsse 80002075 a 0.07860 60002335 i ian card ende all U.M. date cards BLANK CARD ENDING SoURcES. 26US-A1 2926.7958, daus-a “704.22393 uscet -2342.5356 us-A0 3000.0 Yous-20 -1800.0 2aus=c0 =1500.0 Gaus-ro 3.4824 Yous-rs 401.9430 SHEN 376.981118, {avs-aisus-A0 2073.59367 3aus_sipus-20 -620,90671 Peete era) 3 ies 3 Itetae-6 Saus-nopuanon 24981-8626 Saus-ns 401.9430 BUS-ALBUS-BIBUS-cLBUS-MSBUSMSM (Request for selective node voltage output BIANK card ending all electric network woursee 4 ( Plot Lintea: (-0.923, 1.467) (Cote 110.000, 3.770) (plot Lintter (1.334, 0.718) 15 é rnpur Fin © Tent of U.M. operating as a 3-phase eynchronous nachine, Data ie fully © __ Geplained {nthe BNTP Rule Book, Section 1-63. THAK iz reduced. however Powen FREQUENCY. 60. moon 00 5 2 28100100 ° 10 ° ie ° ro 0.02 1.0610 ‘ onus-m 2.0086 i pus-nowus-r a a BLANK cara ends the leat branch card’ AN card nde (in © je nonexiatent) switch cards Tipus-ro.0:002032, 10 Hiwusce-1 02 421356 60.0 -120.0 Ino Ldmusceo 1l4i42iase 60:0 120.0 ne im ( deginning of UM. dats (Type-19 source) BuaMK card ending clasa-i U.M. 2 uaniBus-m 1 1786.58 (1st card of U.M. machine table 10. 1.350 00.3 1s 1.0 osa767 iase 7 ole. 11,3860 ( tat card of cott table S:001088 8.180 1 -0198a77" Drones © O.180, 3 lolaz7at :00074 8101 308-F0 i “21826 oloist o.0s5 a 10540 oLose a BLANK card ending all ULM. 4 cv" “"total network lous Prloes by sunning injectiona = 2.9999999899328+00 ‘aavs-as 1.56489 { rizet of sany In{tial condition cards for the zmus-p1 -0.3075¢6 (electric network. Since the Ui. Le not a Gnus-ei —1.387466 { part of the phasor solution (see DONEW-1 for 2aus-ao 1.41421356 {fish “a ‘gore sodern probies). synchronous 2aus-n0 -0.70710678 { operation can begin smoothly only if the 2o:7071088 | {atially condieione are manually applied. o-00z081 Fe) f-ALBUB-AO 1.386 3eus-piavs-t0 -0.95877 Shus-eiaus-eo -0.42721 B0S-A0 W1.41422386 308-00 +0.70710678 Bus-co “o-7or10s7@ 3us=70 -0.00208876 woBUs=M1 s1.02 Fine card of node voltage initial conditions Int card of branch current initial conditions mo 0 3bus-m TL0 | tase card of branch current init. condit BUE-AIBUB-B1BUS-C1BUS-H1BUS-HO {Request for selective node voltage output BLMIK cara ending the specification of output variables 193 «1 0.01-0 ose uS-"0 Chtet Lumtea: (-1-010, 0.000) 193 110.0 1.0 Oeel THETA ("ploe Liates: (0-000, 9379) BLANK cara ending ali plot carga ‘BEGIN NEV OATA CASE 16 ourpur 17 8.4 GaSe 41) Test of automatic U.N. initialization operating leachine.oata ts in the per unit Pours PREgUENcY. 60. 0.0010" 9.500, é 2 S-phase eynchronoue oy 103100 ‘obus-Banus-8 i ‘obus-e18us-conus-AIBUS-A0 i Cousscssese ENP CONVECTIVITY AEQUIRENENT 0 suacr0. ta Sxus-rosus-79 iors opis-mi 2.0086 1 pus-ro oor i us-noaus-o 1 i ‘Sia card ende the leat branch card BLA card ende (In this case nonexistent) svitch carde 14sus-rs 0.002091" 0.00001, a0 14sus-ro-1" 1-020, 000e1 Do Tasus-a01.42471356 60 0.0 a0 1asuscg0 1142471986 60 -ti0.0 io tasus-eo 1142471356 60 120.0 20 iow ( deginntng of U.M. data (TyPe-19 source) YP aiaiavea Tiree. (1a card of UL. machine table 1.850 00.3 Ls i.e aes796z127 | 15.31 70 ro) a8 a ( ae card of coll table Sloai086 © 8.450 i Oceo10s6 §=— 8150, 1 D007 = ston a 00132, 8.085 a 90840 036 i BLANC card ending a2 U.M. TANK cera snding EXIF source cards BUS-ALBUS-B1BUS-C1BUS-MIBUS-HO { Request for selective node voltage output BLAYE card ending the specification of output variables 393 -10.01.0 ‘sus-niau8-no (Plot Lintee: (-1.010, 0.000) 393.12 0.0 1.0 tee THETA (Plot Mitta: (0.000, 9.379) BLAM card ending a11 plot cards IEGIN NEW OATA CASE 18 OUTPUT FILE 19 8.5 CASE 4(2) ‘teat of automatic U.N. initialization operating Sachine.oata is in the per unit Powtn FREQUENCY. 60. 0.0010 0.500, 5 5 2 2100200 ~ ante connscriviTy o wus-ao ie © Bus-co Lo. TRANSMISSION LixES ‘onva-aiaue-ao 6.02" 1.0810 a Sepha é menronous Cuascsenees EPP CONNECTIVITY REQUIREMENT =r 1.0) Cuasceenens MECH NETWORK oua-mi 2.0086 Opus-noaus-i oe i BLAM card nda [in this cage nonesiatent) switch carde Lanua-ra 0.002091" 0.00001 ° Tanus-no-11-020-00001 ° tauscnd2.41421386 60, 0.0 ° Tapus-so 241421356 60 -10.0 ° iaaus-eo 141421356 60 120.0 ° om { Beginning of UM. date (Type-i9 source) SLA card ending classi u.m. data Pep anaisusc 1736.38 (tet card of Un. machine table 2.880 0.9 Ls 1.8 1480 ° aesreen27 18031 Bus-FeBus-no eco oe sus-aL 1 (ee card of cof cable Plooiess 8.150, susan 2 Slooiess 8.180 sus-et i oer 8!i01 aus-F0 1 ocost o.055 1 0.0840 0.036 i LANK card ending a2 U.M. data BLANK card ending ENTP source carée BUS-AIBUS-B1SUS-C1AUS-RiBUS-HO (Request for selective node voltage output BLANK card ending the specification of output variables 193 -10.01-0 ‘BUS-M1BUS-RO (Chtot Latte: (1.010, 0,000) 193.11 0.0 1.0 Dee THETAM ('ptoe iimtea: (0.000, 91379) LAN card ending all plot carde EOIN MEW DATA CASE 20 ourPur FILE 2 or ERR Sears SO SES sand Saat eee ne i tee 21 22 References 1 H.W.Dommel, "Electromagnetic Transients Program - EMIP Theory Book", BPA, 1986. 2 J.A.Martinez, G.A.Capolino, G.Bortoni, "Development of a new three-phase syn- chronous machine code using Models", 18th European EMTP Users Group Meeting, May 28-29, 1990, Marseille. 3 Hian K.Lauw and W.Scott Meyer, "Universal Machine Modeling for the Representation of Rotating Electric Machinery in an Electromagnetic Transients Program”, IEEE Trans on PAS, Vol. PAS-101, No. 6, June 1982. 4 D.Shirmohammadi, "Universal Machine Modelling in Electromagnetic Transient Program", Canadian Electrical Association Engineering & operating division Meeting, Montreal, March 25-28, 1985. 5 J.A.Martinez, "Steady state initialization of a three-phase synchronous machine using the UM code", EMTP Summer Course in European users meeting, July 24-28, 1989, Leuven, Belgium. 23 CHOICE OF DC-SOURCES FOR VARIOUS INITLALIZATIONS 1 Introduction Yin Yuexin, G. Empereur LEC Leuven, Belgium Constant-value sources are often used in excitation and mechanical circuits of machines. For the UM model of a synchronous machine, several source types could be used in ATP to represent the excitation voltage and the mechanical torque. In order to represent these sources, two types of function are available, one being the step function (type 11), another being the sinusoidal function (type 14). The distinction is made using a flag value 11 or 14 in the first two columns of the source card [Ref.1]. a. Type 11 source ot Fig. 1: Step function b. Tyne 14 source Fi At | i Fig. 2: Cosinoidal function Source Type 11 is a step function with following peculiarity: 180-3 7(1) 0 120-9 4()=4 (constant) This source type can never be taken into account during steady state (even not if TSTART < 0 would be specified). Source Type 14 is a cosinoidal function of time. The variables set in this card include the amplitude, the frequency, the initial angle, and the starting time (TSTART). If the frequency is specified to be very small (e.g. £ = 0.0001 Hz), the change of the function's value with time is also very small. Thus, the imitation of a DC source can easily be obtained. The source type 14 is the only one taken into account during steady state, provided that TSTART < 0 is specified. 24 2 Sources in IM code of SM Seure in Mechanical part Fig. 3: UM and its connection with sources For representing the infinite bus source voltage, usually the type 14 function at power frequency is necessary. On the other hand, the voltage source representing the excitation source and the current source representing the electrical analogue of the mechanical torque usually are DC sources. Different initialization options imply usage of different types of sources. This will now be discussed for both the automatic and manual initialization options. Automatic initialization option ‘This option uses the steady-state phasor solution to provide the initial values. Because the steady state disregards all sources other than type 14, both the excitation source as well as the source in mechanical part should be of the type 14. Type 11 sources cannot be used. Because one has to emulate DC sources, the frequency of the function should be low (e.g. 0,0001 Hz). Furthermore, TSTART should be set negative, indicating that the source indeed should be taken into account during steady state. b. Manual initialization option All initial values of this option must be set beforehand, by the user. So, usage of a type 11 source is certainly all right to represent sources both in the excitation part as well as in the mechanical part. But, if type 14 sources are used instead of type 11 sources in this option, what will happen ? As we know, setting the starting time TSTART of the type 14 source to a negative value means to activate the source during steady state. On the other hand, a positive value for the starting time refers to a source not being active during steady state. Accordingly, specifying TSTART = DELTAT will lead to the same results as usage of a type 11 source. 25 On the other hand, one can still specify the starting time TSTART of type 14 source to be negative. But this will not produce the same initial values as in the automatic initialization option, as the specified values have priority. Occasionally, even an error might occur during steady state calculations (kill code = 23) [Ref.3]. This can be explained as follows. Usually, the frequency of the type 14 source in the mechanical part is set to a very small value in order to emulate a DC source. If the inductance in the mechanical part (converted from a spring constant) is also very small, that will result in a very small impedance value. It may then underscore the near-zero tolerance value EPSILN (resulting in kill code = 23) during steady state. But this is not a specific machine-problem: the same problem can be obtained in simple networks fed by a current source (see attachment). Two methods can be adopted to cure such problem: 1) decrease the value of EPSLIN in the STARTUP file (or on the first miscellaneous data card) 2) increase the frequency of the type 14 source within the numerical accuracy for DC analogue. Fortunately, the automatic initialization option uses a different procesing routine. So, the above trouble will not appear in case of automatic initia- lization. 3 Conclusion In conclusion, which source is used for a constant-value source as e.g. in the excitation part and mechanical part of a UM machine must be waged against the selected initialization option (manual versus automatic). In case of automatic initialization, type 14 with TSTART negative is used for both the excitation source and sources in mechanical part. In case of manual initialization, both type 11 and type 14 are effective. If a type 14 source is used, the starting time TSTART can be specified either as DELTAT or as a negative value. The former has the same effect as a type 11 source. In case of a negative value, however, one should be careful for the near-zero tolerance value EPSILN. The methods supplied above can be used to avoid this trouble. 4 References 1, ATP Rule Book, LEC, 1990 update. 2. "The improvement to the present synchronous machine code in UM", D. Van Dommelen, Yin Yuexin, G. Empereur, EMTP NEWS, Volume III, Number 3, 1990. 3. "Steady state initialization of a three-phase synchronous machine using the UM code", Juan A. Martinez, EMTP Summer Course, July 24-28, 1989, Leuven. 5 Attachment ANPUL ELLE. CRED = 00001, EPSILN = 1.8-19 -s0otge 018 7 onvaan 5.2489 nusin sus2n aes Onvaan 4.6489 BLANK card bounds the last branch card Sanusin =i 200, -00001, BLAME card ending ali electric network sources SLANE CAD ENDING CUTPUT REQUESTS BLANK CARD ENDING PLOT CARDS. QUIPUT FILE INPUL_EILE € FREQ'= -00001,SFSILN = 1.-08 € 00100 010 1e-os fear ‘onvsim avs24 aes onusze 4.6489 LANE card bounds the Last branch card BLANK card bounds the last awiten cara ianusin "1 200..-00001 BLANK card ending ali electric network sources (Gano Enoine OUTPUT REQUESTS CARD ENDING PLOT CARDS NEV DATA case 26 a7 S+sLL/ caution. staconnacted autnatuork. During the Yomatrin climination for phator vottages, a near-eero dlogonol eleaent for node sausan'= “aciats just prior to reciprocation, 'startacier falton: original apsivany © rr senavoren? CASE 3, {GIN MEY DATA CASE FREQ = .00001,EPSILN = 1, Leavers 3 a0 BLANK card ending ali electric network sources LANK CARD ENDING OUTPUT REQUESTS ANX CARD ENDING PLOT CARDS. BECIM MEW DATA case BLA QUIPUT FILE INPUL_EILE upneg = 00001, EPSILN = 22-13, 000100 010 aes ‘onvain 5.2409 osuezH : sLane card Ljie card bounds the leat switch card Yenusin <1 200. "00001 00001 BLAM card ending ali electric network sources ‘ln cand ENDING OUTPUT REQUESTS Sua QUIPUT FILE sussss 2888882888 i FAQ = 00001, EPSILN = 1.2-08 000100 .010 a.z08 ao der ‘onvsam s.349 ‘obuSIN suszm as ‘onusza ‘4.6489 BIAMK cerd bounds the last branch card BLANK card bounds the lage switch card Lasusin 1200, 00001, 000 BLAM cerd ending ali electric network sources ‘BIANK CARD ENDING OUTPUT REQUESTS TANK EARD ENDING PLOT CARDS. 28 29 Mass Driver Analysis Using EMTP Wendell Neugebauer 586 Middle Line Road Ballston Spa, New York 12020 U.S.A. Telephone : (518) 885 - 6050 Introduction A mass driver is an electromagnetic device which can propel macroscopic particles to high velocities by making use of forces generated by electric currents and magnetic fields. Such devices may find use in the future as means for efficiently launching large payloads from the earth’s surface directly into orbit. Perhaps an even earlier application would be Moon based because of lesser energy requirements and the virtual absence of atmospheric friction. As point of reference, the escape velocity of an object on the earth’s surface is approximately 20 km/sec. A simple form of mass driver is the rail accelerator shown in Fig. 1a at the right where a projectile with a conducting brush is accelerated between two rails by means of the forces created by the current flowing in the rails and in the conducting brush. In this example, the rail current is produced from the discharge of a bank of previously-charged capacitors (Fig. 1b below shows such a sample power supply). The accelerator itself is of the augmented type where the augmenting rails produce an additional magnetic field which helps to propel the projectile away from the source of the current. VBR ‘Augmenting rails —[- Main rail Projectile Muzzle resistor Fig. 1. a) rail accelerator (above); b) power supply for it (below). Lh Tire VBR | Pulse forming inductors as ~~ Crowbar 30 fany different power supplies can be used to energize a mass driver. The capacitor bank 1b was just one example. Another alternative could be based on a storage inductor energized by a homopolar (de) generator, which could be discharged into the rails by means of a very complicated opening switch. Heretofore, the analysis of such power supplies and associated mass drivers has depended on custom computer programs written for each type of drive circuit. ‘The use of EMTP can eliminate the need for custom programs and will allow rapid analysis of the effects of changes in the power supply circuit. The only new task is to represent the electromechanical action of the mass driver by means of TACS or the newer MODELS feature of the EMTP. TACS Representation of Rail Gun @ = ¢ The rails and projectile of Fig. 1a can be Ravg] Laug] Re represented on the electric side of the EMTP_by the circuit shown at the right (see Fig. 2). The electrical network for the driver consists of the final rn values of the rail inductance and resistance together 144] Tyo! with the muzzle resistor which is necessary t0 4 dissipate the remnant energy stored in magnetic fields after the projectile has left the launcher. A snubber at node I1 helps to dampen numerical oscillations. The electromechanical behavior of the mass driver is represented in this circuit by means oon of the various injection currents (more about these eee in succeeding paragraphs). The switch IRAIL to : : : Il is merely used for the measurement of the rail Fig. 2. EMTP modeling of Fig. 1a. current. ‘The philosophy of this model is to represent inductive voltages from the accelerator as voltage drops across inductive elements of the equivalent circuit. Similarly, resistive voltag are represented as voltage drops across the equivalent resistors. For the purpose of illustration ‘we consider the augmenting rails first, and assume that the augmenting rail inductance and resistance are constant in time, The voltage drop Vy, across the augmenting rail in terms of the current Ina is: d - Yous = “gpl Faag* Has + * Eau) + Rang Fa where M is the time-varying mutual inductance between the augmenting and main rails. In this example, M is considered to be directly proportional to the distance between the projectile position and the breech of the launcher. The voltage across the augmenting rail inductance (the time derivative term above) can be forced to appear across L,,, in the network of Figure 2 by means of the equal and opposite injection currents Ij, and Iw at the inductor terminals. To find the value of the injection current, we set d d GU tug? Eat Mt Eyal = 20 Lug* Sut Lug? Ss dt which yields : Fat Lug om and bo -h where the constant of integration is zero as determined from the initial conditions. The resistive voltage drop across the augmenting rail is correctly represented by the rail current flowing through the augmenting rail resistance. In TACS itself, the instantaneous value of Mis computed from the projectile position. The mechanical force on the projectile depends on the square of the rail current which is passed to TACS by means of a measuring switch. The equations of motion are derived from Newtonian mechanics and will not be repeated here, but may be recognized directly in the accompanying, annotated TACS listing of the Appendix. The injection currents 1, Ia, and I, are determined from the back emf produced by the projectile motion and by the ordinary voltage differences caused by currents in passive elements. Again, the secret of correct modeling is to cause resistive voltages to appear across resistive elements and inductive voltages (first derivative voltages) to appear across inductors. ‘The capacitor discharge switches in this example are synchronized with the position of the projectile. During the launch period, the injection current I, is such that the muzzle resistor Rap is effectively open-circuited. The main rail resistance is proportional to the 0.75 power of the projectile position according to a theory developed by Deadrick [2]. The example given here also includes the effects of friction, mass ablation from the rails, arc voltage drop at the projectile brushes, and shock forces due to the presence of gas between the rails. A more direct modeling of the accelerator breech voltage by means of a TACS voltage source was attempted earlier, but this was found to suffer from excessive numerical oscillations. However, such a representation may still be feasible if proper precautions were to be taken. Several zero-order blocks are required in the model to prevent arithmetic loops in TACS. For an EMTP model of the type described here, it is difficult to verify the correctness of the results because field test results are unavailable for a device which has not been built. The next best action is to track the energy balances throughout the acceleration process to verify that energy is neither created nor destroyed by some unknown numerical process. The TACS variable EBAL tracks such unknown energies by taking into account all the known energy dissipations and storages. Only if EBAL is sufficiently small during the entire launch simulation can the model and its time step be considered appropriate. Such a condition is always necessary, but not sufficient, to assure correct results. Ultimately, the model must be verified by field tests. Illustrative EMTP Simulation Mass velocity and rail current as functions of time are shown in Fig. 3 at the top of the next page. Beginning with an injection velocity of 738 m/s, the 2.5-Kgm projectile is accelerated to a final velocity of 2.4 Km/s in approximately 4.5 milliseconds. Note that the rail current reaches a value of nearly 1.7 million amperes. It should be recognized that these particular results have not been verified by field tests. It is the author's opinion, however, that any launch of this nature represents a transient in the true sense of the word, and that EMTP is an ideal tool for studying the action of mass drivers. The author will be glad to send more elaborate information on this model to interested parties. 32 Rail Gun Transients -asooy sor 3 / ™. is fen i sxconss Peet fain, 1 Gr/aiy > ovceocer-00 s:Soc00n-03 5.sc000E-o4 Fig. 3. Mass velocity and rail current as functions of time. Although EMTP data for the rail gun is voluminous (the data case consists of 361 card images), simulation nonetheless is possible using the MS-DOS version of ATP if list sizes are reproportioned. It will run on the standard, executable Salford EMTP dimensioned at 3 times default dimenioning. ‘The author demonstrated this during Prof. Dennis Carroll's EMTP short course in Gainesville, Florida, during March of 1990, and made the data available at that tin. On September 23rd, ATP developers in Portland appended the author's data to BENCHMARK DC-22 as a new 3rd data subcase. See the Appendix for a listing through all TACS data. ‘The use of MODELS (the new control system modeling) rather than TACS (the original, simpler control system modeling) would seem to be the next logical step in the EMTP modeling of mass drivers. In the past, this was not possible because the author's home computer was based on an Intel 8088 microprocessor that supported only the MS-DOS version of ATP. This does not --- and never will -- support MODELS. But a newer personal computer based on the 80386-SX was purchased during August, 1990, and the Salford EMTP, including MODELS, now is running successfully on this more flexible platform. Experimentation continues. Additional Information ‘The following papers give additional background for the analysis of mass drivers and railguns: ae [1]. J. V. Parker, "Effect of Ablation on Plasma Armature Dynamics", presented to DARPA\Service Electromagnetic Propulsion Program Review, Sept. 18-20, 1984. (2). F. J. Deadrick et al. "MAGRAC ~ A Railgun Simulation Program", IEEE Trans. on Magnetics, Vol. MAG-18, No. 1, January 1982, pp. 94-104. Editor’s Note (insert ) : Author Neugebauer is to be commended for his innovative use of EMTP , and his willingness to share his considerable experience with readers around the world. Yet, Mr. Neugebauer is too modest in several respects, beginning with his assumption that the average reader either already understands the equations of rail guns or easily can figure them out by studying comment cards of EMTP data. ‘The subject is important enough to include a derivation of equations as part of the text. Readers can look forward to such a tutorial in the next issue. ‘As author Neugebauer states, there seem to be important practical industrial applications of the technology of this paper. Both the U.S. government and also private industry have interest in such devices, as might be explained further in the next issue. It is hoped that someone familiar with the U.S. space program in Florida could be encouraged to discuss Mr. Neugebauer’s paper in the next issue. Readers may be interested to learn that author Neugebauer recently was awarded a patent associated with mass drivers. "Electromagnetic Launcher with Improved Current Commutating Switch Performance" is the title of U.S. Patent Number 4, 924, 750 which lists the author as the inventor and the General Electric Company of Pittsfield, Massachusetts, as the assignee. Dated 15 May 1990, the Abstract reads as follows: "To improve the performance and life of a charging current-conducting commutating switch in an inductor storage-type electromagnetic launcher, the energy expended in propelling a mass through a breech rail gun section at a high velocity is utilized to generate a current in the breech section rails which flows through the commutating switch in opposition to the inductor charging current. The switch is opened in coordination with the movement of the mass through the launcher with little or no arcing when the current therethrough has been depressed to a minimal level to commutate inductor discharging current into the rails of a longitudinally aligned launch rail gun section of the launcher to accelerate the mass therethrough to a high exit velocity.” ‘The interactive plotting program TPPLOT was used to produce the plot of Fig. 3 as an integral part of the laser-printed page. Since the procedure is new for many readers, details are worth summarizing. First, of course, the simulation created a .PL4 file of plot points because integer miscellaneous data parameter ICAT has value 2. With input data file named MASSDRIV.DAT, the EMTP automatically names the plot file MASSDRIV.PL4 because of the environment of execution (STARTUP variable KTRPL4 carries an extra minus sign). Using TPPLOT, the variables of interest, (TACS, VEL) and (TACS, IRAIL), were selected. Since the rail current in amperes is about 1000 times bigger than the velocity in meters/second, the FACTOR command was used with multipliers (1, 1.E-3). Following acceptance of the screen display, the HPGL command created a disk file representing the plot in Hewlett- Packard Graphics Language (HP-GL). Finally, as the text now being read was being massaged using WordPerfect (software of electronic publishing), the HP-GL disk file was imported as a picture using the Graphics command (Alt - F9). 3u Appendix : TACS Data Listing The electrical network of Figure 2 is modeled with standard EMTP components and will not be described further here. The TACS portion of the input data appears as follows: BEGIN NEW DATA CASE © AUGMENTED RAILGUN (Mass driver) Simulation CAPACITOR BANK DRIVE. 65 CANS OF 65 kJ, 22kV FOR EACH OF i. 8 STAGES, TOTAL INITIAL ENERGY = 33.8 MJ C This is a simulation of a mass driver as energized from a bank of © charged capacitors. The individual switches are timed in © synchronism with the position of the mass along the rails. c The physics of the driver itself are modelled under TACS C using its pseudo FORTRAN equations to implement Newton's laws. ABSOLUTE TACS DIMENSIONS 60 270 300 60 90 ©1250 550 180 1.B-6 5.5E~-3 0 0 500 1 1 ° 1 ° 2 ACS HYBRID © LIST OF INPUT CONSTANTS © RAIL RESISTANCE COEFFICIENT (R. Hawkes method for including skin © effect) with the units ohms/amperes*#0.75 1LRRAILO 5.53E-5 “1. C RAIL INDUCTANCE GRADIENT, H/m LILPRIME 5 765E-6 “1. C PROJECTILE MASS, kg 11MASS 2.500 1. C LENGTH OF RAIL, m LLXRAIL 8.0 “1. C Muzzle discharge resistor, ohms LARDUMP 8.E-3 C final rail inductance, LALRAILF 4.6E-6 “1 C final rail resistance, ohms 11RRAILF 2.63E-4 “1. © augmenting rail inductance, H 11LAUG 4.2E-6 m1. C augmenting rail resistance, ohms 11RAUG 1.0E~4 “1. C Mutual inductance gradient, augmenting to main rail, H/m 11 DMDX .35E-6 C Friction approximation coefficient, fraction of applied force 11FMISC 0.2 C Initial projectile position, m LAXINIT 0. C Projectile initial velocity, m/s LAVINIT 738.0 “1. C Rail mass ablation coefficient, kg/A/V/s L1ALPHA 49.E-9 C Threshold current for for computing effective arc drop L1ITHRES 100000. “1. C Bore diameter, m 35 11 BORE 0.09 © Velocity of sound in the medium within rails, m/s 11VSOUND 346.0 m1. C Coefficient for computing shock force 11GAMMA 1.40 -1. ¢ Ambient pressure, N/m*#2 11PAMB 1.013E5 -1. © positions of mass along the rails where the various switches close 11XA 0.25 -1. 1x2 +50 ; “1. 11x3 1.00 -1: 11x4 1.70 -1. 11Xx5 2.10 -1. 11x6 2.70 -1. 1x7 3.00 “1. c © THIS CONCLUDES THE TACS SOURCES. c ¢ LIST OF EMTP SOURCES c VBREECH FROM EMTP : 90VBR c TRAIL FROM EMTP Q1IRAIL c C --- EMTP NODE VOLTAGES ON 8 CAPACITORS. USED TO TRIGGER CROWBAR DIODES. 9ONODEO1 90NODEO2 90NODEO3 90NODE04 90NODEOS 90NODEO6 30NODEO7 90NODEO8 c EMTP SWITCH STATUS 0 = OPEN 1 = CLOSED c USED TO KEEP CROWBAR DIODES ON ONCE THEY ARE TRIGGERED. 93NODE17 93NODE18 93NODE19 93NODE20 93NODE21 93NODE22 93NODE23 93NODE24 SUPPLEMENTAL DEVICES COMPUTE GRID SIGNALS FOR CROWBAR DIODES C --~ GRID SIGNALS (N1-N8) TURN ON WHEN THE CAPACITOR VOLTAGE IS LESS THAN 0. sani = - NODEO1 B8N2 - NODEO2 88N3 - NODEO3 Bana = - NODEO4 88N5 - NODEOS Bane - NODEO6 88N7 = - NODEO7 Bane = - NODEOS 36 c © SUPPLEMENTAL DEVICES © SIMPLE RAILGUN MECHANICS © COMPUTE MECHANICAL FORCE ON THE PROJECTILE INCL. AUGMENTATION B8FMECH — =.54(1.0-FMISC) *(LPRIME+2 . *DMDX) *ABS (IRAIL) **2-FSHOCK © USE LINEAR MODEL FOR SOLID ARMATURE ARC VOLTAGE DROP B8GNARC _=45.+31.43*TIMEX*1000. © COMPUTE RATE OF MASS ABLATION FROM THE RAILS 88MDOT — =(ALPHA*ABS (IRAIL) *ABS (GNVOLT) ) *FLAG1 C COMPUTE VDOT = PROJECTILE ACCELERATION, INCLUDE TIME DELAY OF ONE © STEP FOR STABILITY VELL +VEL gevDOT —__=( (FMECH-VEL1*MDOT) /MASS1) *FLAG1 COMPUTE MACH NUMBER, PRESSURE RATIO, AND SHOCK FORCE 88MACH — =VEL1/VSOUND 88PR =GAMMA* (GAMMA+1.) /4. *ABS (MACH) **#2+2 88TEMP1 BS ( ( (ABS (MACH) **2*ABS ( (GAMMA+1.) )##2/16.+1.)))*#0.5 88PRATIO = PR + GAMMA * MACH * TEMP] ( WSM splits 1 line into 2 88FSHOCK =PI*BORE**2/4*PRATIO*PAMB c © COMPUTE RESET SIGNAL FOR FIRST LAUNCH ¢ FLAG] IS 1 AS LONG AS PROJECTILE IS IN BARREL € Introduce one time step delay for stability of computation x1 +X 88FLAG1 ‘TIMEX .GT. (2.*DELTAT)) AND. (X1.LE.XRAIL) B8FLAG4 TIMEX .GT. (2.*DELTAT) ) B8FLAGS =NOT(FLAG1) c © COMPUTE MASS1, PROJECTILE PLUS ABLATED RAIL MASS 88MASS1 58+MDOT 1.0 0.0 { Fields C, D, E of preceding card in cols 63-80: 1.0FLAG4 MASS C INTEGRATE VDOT TO GET VELOCITY OF MASS 88VEL 58+VDOT a0 = 0-0 { Fields C, D, E of preceding card in cols 63-80: 1.0FLAG4 VINIT © INTEGRATE VELOCITY TO GET PROJECTILE POSITION asx 58+VEL a0 0.0 { Fields C, D, E of preceding card in cols 63-80: 1. 0FLAG4 XIN © COMPUTE THE INSTANTANEOUS RAIL INDUCTANCE 88LRAIL =LPRIME*ABS (X1) *FLAGI+FLAG5S*LRAILF C COMPUTE THE INSTANTANEOUS RAIL RESISTANCE 88RRAIL _ =FLAG1*RRAILO*ABS (X1) **0.75+FLAGS*RRAILF © COMPUTE INSTANTANEOUS MUTUAL INDUCTANCE, AUGMENTING TO MAIN RAILS gem =DMDX* (FLAG1*X1+FLAGS*XRAIL) c = © CALCULATE THE POWER AND ENERGY DELIVERED TO THE RAILS 88PBR =VBR*IRAIL 88EBR 58+PBR 1.0 0.0 { Fields C, D, E of preceding card in cols 63-80: 1.0FLAG4 ZERO © CALCULATE SHOCK POWER AND ENERGY 88PSHOCK © =FSHOCK*VEL*FLAG1 88ESHOCK58+PSHOCK 0 0-0 { Fields C, D, E of preceding card in cols 63-80: 1.0FLAG4 ZERO © CALCULATE ARC POWER AND ENERGY 88PARC =IRAIL*GNVOLT*FLAG1 SBEARC 58+PARC 70 20r0 { Fields C, D, E of preceding card in cols 63-80: 1. 0FLAG4 ZERO © CALCULATE MIXING POWER AND ENERGY 88PMIX =0.5*VEL**24MDOT*FLAGI 37 B8EMIX 58+PMIX 2200-0 { Fields C, D, E of preceding card in cols 63-80: 1.OFLAG4 ZERO © CALCULATE THE DUMP RESISTOR POWER AND ENERGY 88PMUZ — =RDUMP* (ABS (IRAIL-13A) **2) BBEMUZ 58+PMUZ 1.0 0.0 { Fields C, D, E of preceding card in cols 63-8 1.0FLAGS ZERO © CALCULATE PROJECTILE CHANGE IN KINETIC ENERGY 88DKE =0.5*MASS* (VEL**2-VINIT**2) C CALCULATE ABLATED PLASMA CHANGE IN KINETIC ENERGY : 88PLSMKE =0.5%*(MASS1-MASS) *VEL**2 © CALCULATE INSTANTANEOUS RAIL HEAT POWER AND ENERGY 88HPOWR — =ABS(IRAIL) **2*RRAIL 88HEAT 58+HPOWR 20 20-0 { Fields C, D, E of preceding card in cols 63-8! 1.0FLAG4 ZERO © COMPUTE AUGMENTING RAIL Loss 88PAUG — =IRAIL*IRAIL*RAUG 88EAUG 58+PAUG 1.0 0.0 { Fields C, D, E of preceding card in cols 63-80: 1.0FLAG4 ZERO © CALCULATE FRICTION POWER AND ENERGY 88PFRIC =VEL*0.25*FMISC* (FMECH+FSHOCK) *FLAG1 88EFRIC 58+PFRIC 1.0 0.0 { Fields C, D, E of preceding card in cols 63-80: _1.0FLAG4 ZERO © CALCULATE TRAPPED MAGNETIC ENERGY WITHIN RAIL MATERIAL (ASSUMPTION) BBETRAP —_=3.0*EFRIC COMPUTE ENERGY STORED IN RAIL AND MUTUAL INDUCTANCE BBESTORE =(.5*LRAIL+.5*LAUG+M) *IRAIL*IRAIL © COMPUTE ENERGY BALANCE DYNAMICALLY-should =0 by conserv. of energy geTEMP2 = -DKE-EMUZ-EMIX-EARC-ESHOCK-ESTORE-EAUG 88EBAL 'BR-ETRAP-EFRIC-HEAT-PLSMKE + TEMP2 { WSM splits line c © COMPUTE INJECTION CURRENTS Ti 12s to C THESE CURRENTS EFFECTIVELY REPRESENT THE BACK EMF OF THE MOVING MASS ast1 = (IRAIL* (RRAIL-RRAILF) /RRAILF) *FLAG1 eei2a = (IRAIL* (LRAIL-LRAILF) /LRAILF) *FLAG1+M*IRAIL*FLAG4/LRAILF asiz =I2a-I1 © Compute the effective arc voltage B8GNVOLT =SIGN(IRAIL) *GNARC*(1.-EXP(~-ABS (IRAIL) /ITHRES) ) asI3A (TRAIL* (-RDUMP) +GNVOLT) /RDUMP) *FLAG1 8813 =I3A-12A © COMPUTE INJECTION CURRENT DUE TO MUTUAL EFFECTS asl4r =M*IRAIL/LAUG g8140 141 c € CAPACITOR SWITCHING FLAGS BASED UPON PROJECTILE POSITION 88FLAG11 =X .GT. XA BeFLAGI2 =X .GT. x2 88FLAG13 GT. x3 BeFLAGI4 =X .GT. x4 a8PLAG15 is used to select the file of interest after correct positioning of the colored or highlighted bar using up or down arrow keys [3]. The FORM command supports .PL4 file types of sequential FORMATTED and UNFORMATTED, but not yet C-like. Yet, this is hardly a problem because the Salford EMTP can not yet create such plot files, anyway. But some day it might be able to do so. Following a telephone conversation with David Vallance of Salford, Computer expert David Szymanski has concluded that he probably could provide the missing C-like alternative, but that the author could not because coding could not be restricted to. FORTRAN. The OS command allows for either the execution of a single MS-DOS command from within TPPLOT or the opening of an MS-DOS shell for the subsequent execution of an arbitrary number of MS-DOS commands. Simple commands (e.g., DEL, RENAME, COPY, DIR, and EDLIN) require no special care. But the more interesting and powerful execution of sophisticated DOS programs require additional memory. While visiting BPA offices on September 24th, Mr. Robert Chafer of Salford explained that more memory can be obtained using the COMPACE command of MS-DOS prior to the execution of TPPLOT. For example, the LIST utility will run in 80K bytes, for which the DOS command would be: COMSPACE <’80000'. For editor PCWRITE, between 230K and 250K are required. The PL4 command operates the same way as it did using the MS-DOS version of TPPLOT for which it first was introduced. The program will respond with a display of available. PL4 plot files from which the user can select one by number. This is simpler, faster, and less error-prone than keying names, generally. Or, the user could switch his consideration to a remote directory by sending DIR. The left and right arrow keys function as originally defined for SPY plotting [3]. This involves graphical scrolling, but in quite different fashion than that of Section Ill. No history is involved here. The left and right arrow keys represent requests to regenerate the plot without recourse to any previous results. Il. Special Character Windows of Salford TPPLOT Scrollable windows as used with the Salford EMTP [3] have yet to be added to the Salford TPPLOT, unfortunately. Only a shortage of working time, and the concentration on other improvements (see following sections), have prevented this logical step, which probably is inevitable. No problems with such an implementation are expected. If a big program such as EMTP can handle two windows, it should be obvious that a small Program such as TPPLOT will have no trouble handling one. uy Small, dynamic status windows already have been added to TPPLOT, however. The idea is to monitor the status of operations that are lengthy for plot files with many time steps. There are 4 such different, temporary windows for: 1) TIMESPAN; 2) the loading of RAM from disk with points to be plotted; 3) the search of these RAM-stored points for extrema; and finally, 4) the discarding of visually-redundant points prior to actual pictting. The windows associated with uses numbered 1 and 2 are comparable. In addition to the title line that summarizes the monitoring function that is under way, there are 9 output lines (top to bottom) for the first 9 progress reports. The 10-th and later such output overwrites the Sth using a different color. Output frequency has been given a defautt value of 200 steps, which is reasonable for the author’s 16-MHz computer. Each line shows evolving statistics that include the last-read time-step number and time. Windows for uses numbered 3 and 4 are of the same size but quite different content. Both the extrema search and point discarding ‘occur one curve at a time; so the status report consists of another new line upon the completion of the operation for each new curve. As long as the user does not plot more than 9 curves on the same graph, there is room. The TYPE command is new to TPPLOT, but should be familiar to EMTP SPY users. The purpose is the same: to display a file of the user's choice. The name and syntax of the command were copied from VAX/VMS or MS-DOS. But for Salford TPPLOT, operation is much more sophisticated and useful. A scrollable, full-screen window is opened to display the file of interest using format and controls as already perfected for EMTP SPY [3]. Le., the up and down arrow keys will scroll the window continuously, the Page Up and Page Down keys displace the window 12 lines (paging with 50% overlap), etc. When the user has seen enough, the Esc key will terminate the display, closing the associated window. The HELP and SET DATA commands have a mode of operation that involves scrolling as described in the preceding paragraph. This will be the case for HELP if nothing follows the command (in column 5 onward). Recall that if a user wants instruction applicable just toa single command, that command is to follow HELP separated by at least one blank. This operation for a single, specific command remains unchanged, and it does not involve awindow. itis only the general, unqualified request for assistance that results in a scrollat. display of the entire disk file of instructions arranged in alphabetical order, TPPLOT.HLP. This file has grown to 481 lines, so the ability to move easily within it has become more important. For SET DATA, the LIST subcommand will result in a scrollable display of the file of all data parameter subsets, TPPARAM.DAT. Ill. Graphical Scrolling by Salford TPPLOT Graphical scrolling -~ the restoration of former screen plots by merely touching special keys -~ is another major extension that has been added to Salford TPPLOT. Previously, no graphical history was kept. A new plot was produced on the screen only after erasure of the screen, which destroyed former displays. Now, once a new plot has been painted on the screen, a copy is made in memory as well. While in graphical mode, the user can select among this graphical history using common keys in intuitive fashion. For example, the Page Up and Page Down keys move one graph backward or forward, respectively. These are the graphical paging keys. Also, the Home key will jump to display the first plot and the End key will jump to the last (most recent). Direct movement to any other plot in between also is possible if the user first sets the frame pointer (see below). 45 The BITMAP command of TPPLOT provides character (as opposed to graphical) access to services that support the graphical scrolling of plots. TPPLOT's response will be an inventory of stored graphs followed by a prompt for a subsequent subcommand. Present dimensioning allows a maximum of 99 stored graphs with variable JBIT indicating the current index of interest and NUMBIT the maximum index (slot or position) that has been used. The inventory shows which slots are occupied and which are not much as the FILES command of SPY does for reserved disk files of @ use. The subsequent prompt appears as follows: Send (#, NAME, , COPY, DEL, RENAME) : Ifthe user sends a pounds sign followed by an integer, JBIT will be set equal to that integer. The response to just a carriage return will depend upon JBIT. If JBIT points to an occupied slot, the stored plot will be displayed. If not, there will be an exit from the BITMAP command. The COPY command is used to transfer plots between RAM and disk (where disk storage involves .PCX files as explained in Section IV). The user can request plot transfers either one at a time or in families (a wild card "*" is permitted for decimal serialization). The DEL command will erase plots from one or more slots in RAM, andthe RENAME command allows the movement of a plot from one slot in RAM to another. To enter the BITMAP command while a plot is on the sereen, the user need only touch the Backspace key. Virtual memory is required for each occupied slot of scrollable plots. It is important that the user understand this, since performance degrades noticeably if RAM is inadequate and DBOS must page to the swap file on disk. As graphs accumulate, eventually the swap file will overflow and DBOS will abort the execution of TPPLOT with an error message ("Page memory exhausted at .."). So, it is important for the user to know how much RAM and how big a swap file he has, and operate within these constraints. Note that graphical scrolling works better for users with much RAM, This is to be contrasted with the situation for the Salford EMTP, for which RAM in excess of 2 Mbytes would only affect simulation speed of extremely large networks. Well, those having more than 2 Mbytes of RAM finally will be rewarded! Using VGA graphics and a computer with 4 Mbytes of RAM, it has been observed that there will be use of the swap file on disk after scrollable plots have accumulated to 20 or so. This observation is applicable to various plots made from HHMMSS.PL4 and MMDDYY.PL4. Speed of the COPY subcommand to transfer plots between RAM and disk deserves further comment. It is much faster to copy an old . POX plot into RAM from disk than it is to create anew .PCX file on disk by dumping the plot from RAM. The ratio of the associated two delays seems to be on the order of 10 to one, which seemed curious at first. But, as usual, computer expert Dave Szymanski was able to clarify matters. The disk files provide compressed storage of a bit map of the plot, and it requires a lot of work to create them. An analogy could be made with PKZIP, for which big files or families of files can be decompressed much faster than they can be compressed. Compression programs are working hard, trying many tricks in order to minimize storage. In the case of graphics, Szymanski says that common needs are for compact storage on disk, and fast restoration. A picture typically would only be compressed once, but subsequently might be redisplayed from disk onto the screen hundreds of times. So, it makes sense to put a lot of effort into compression. Well, what are the numbers for TPPLOT graphs? Fora single VGA plot that was made from HHMMSS.PL4 or MMDDYY.PL4, a 20-MHz AST requires about 12 seconds to create the new . PCX file on disk, but only about 1 second for subsequent restoration of this file to RAM. 46 Iv. .PCX Disk Files of PC Paintbrush Fame Disk files of . PCX format, which seem to have been defined by PC Paintbrush [6], can be used to save screen plots on disk as permanent files. The COPY subcommand of the BITMAP command (see Section III) performs the service from character mode, and function key F3 accomplishes the same task from graphic mode for the plot on the screen. If F is used, the desired disk file name must have been previously defined from character mode using the NAME subcommand of the BITMAP command. The file type. PCX carries with it no special powers or privileges for TPPLOT, as far as the author can determine. Any other file type could be used, and operation would be unaffected, it would seem. One would hope that the same flexibility would hold for PC Paintbrush itself. Perhaps Mr. McKay will address this detail in his report (6). Size of .PCX files depends on both the graphical standard being used and also the complexity of the graph being stored. The minimum is provided by simple graphs such as are produced by plotting HHMMSS.PL4 using TPPLOT. For EGA, these typically occupy between 40K and 50K bytes; for VGA, between 50K and 60K bytes; and finally, for Super VGA, between 60K and 70 Kbytes. PKZIP will compress simple . PCX disk files much more, it has been found. For the simple plots of HHMMSS.PL4, PIZIP indicates percentages in the high 70s and low 80s, typically. Recall that 50% means 2:1 compression, so 80% would be 5: 1, which is spectacular. At first Dave Szymanski was surprised by this, but upon reflection, agreed that it makes sense. EMTP graphs are much simpler structurally than are color photographs produced by a camera, Szymanski says that PKZIP is not nearly as effective on color photographs. But EMTP graphs have none of that complexity. They are, in fact, simple line drawings with most pixels unused (the curves, grid, and labeling occupy only a small percentage of screen pixels, remember). WordPerfect publishing software [4] will accept. PCX files as well as HP-GL files, s. .PCX files provide a second possible connection between TPPLOT and the most popular software of electronic publishing. Yet, this is not a recommendation of the. PCX format. Users should be aware of the difference between HP-GL and . PCX, which is fundamental. Whereas HP-GL involves formulas, and can be displayed to the full accuracy of the device being used, this is not the case for . PCX files, which are bit-mapped and colored. When hard copy of the WordPerfect-stored . PCX file is produced on an HP LaserJet Series II printer, the bit-mapped character is readily visible. In Microsoft terminology, bitmaps rather than metafiles are being used [7]. There is adcitional coarseness due to the color (just as on a color monitor, characters are not solid). PC Paintbrush might be usable for artistic editing (tampering with the plot) of a . PCX file that had been produced by TPPLOT. Prof. Dennis Carroll of the University of Florida in Gainesville has been doing such graphical editing using DrawPerfect, the graphical companion to WordPerfect. But DrawPerfect is expensive, whereas PC Paintbrush is cheap [6]. A reduced graphical standard (e.g. EGA rather than VGA) does not make it impossible to load RAM with a higher-resolution .PCX file and display it on the screen. 47 The bottom of the plot, and possibly the right side, simply will be truncated. Salford software seems to ignore those pixels that are numbered higher than the graphical standard being used for display. For example, ifa VGA disk file is loaded into RAM and displayed using EGA, the bottom 27% (pixels 361 through 480) will be missing. Ifa Super VGA disk file were loaded and displayed using EGA, pixels 351 through 600 would be missing at the bottom and pixels 641 through 800 would be missing on the right. Vv. Hard Copy on Epson Dot-Matrix Printers Epson-compatible dot-matrix printers can be used to produce hard copy of Salford EMTP simulation without any extra associated royalty. This is a very important detail. Previously, only Mustafa Kizilcay’s PCPLOT program offered this attraction. While an excellent program, PCPLOT is a PC XT-compatible program that is further restricted by the 640- Kbyte limit of MS-DOS. Salford TPPLOT not only exploits hardware of the 80386 for faster execution, but it also has no limits on table sizes. Now, it offers free hard copy, too. Resolution of Epson hard copy is only 876 pixels vertically by 960 pixels horizontally. Neither the user nor the programmer has any control over this. Most dot-matrix printers sold today could provide higher resolution, but the Salford FORTRAN compiler, FTN77/386, thus far only supports the single, minimum Epson standard. Maybe this is best, since the printer is slow enough as it is!_ For example, the first CALCOMP PLOT of DC-35, after the addition of more labeling (2 lines of the multiline subtitle and the single line of case title), requires about 2: 17 (minutes : seconds) to plot on an Epson FX-100+ dot-matrix printer. Support for HP Laserjet Series Il printers is no where to be found in the disk file of Salford documentation ( GRAPHICS.DOC ). Of course, with the right cartridge, HP-GL or PostScript could be honored, but this is not free for all. Mustafa Kizilcay, now with AEG in Frankfurt, Germany, has an idea that should allow free hard copy by Salford TPPLOT just as he produces today using his own PCPLOT. Look for more about this in the next issue. Meanwhile, use HPGL with publishing software (WordPerfect, MicroSoft Word, etc.). Cell 72 of the text interface vector controls the destination of the Epson plot just as it does the destination of HP-GL (produced by the HPGL command). If this is blank (the default choice), a disk fle named PLOTX.EPS will be created where "X" is an integer between 1 and 89. Subsequent printing is done by the MS-DOS copy command remembering that a binary file is involved (8.g., COPY PLOT3.EPS PRN /B). if the user wants to send his plots directly to a printer port, he need only redefine cell 72 appropriately. For example, "72LPT1" would be added before the blank card ending the appropriate subset of ‘TPPARAM.DAT in order to send the plot to an Epson-compatible printer that was connected to LPT1. The MS-DOS PRINT command has the potential to provide the most efficient way of producing hard copy on Epson-compatible printers. This important idea was suggested by computer expert Dave Szymanski, who observed that the COPY command will hold back the execution of a program whereas the PRINT command will not. Speed is of concern because the mechanical operation of dot-matrix printers usually is much slower than computer creation of the information to be printed. Sending an Epson plot directly to the Printer port clearly is not desirable in most cases because the execution of TPPLOT then would be held up by the slower printer. This is the way PCPLOT works under MS-DOS, 48 and the delay can be aggravating. Of course, the TPPLOT user could send Epson plots to disk, and then COPY them to the printer later. Or, using the multi-tasking DESQview, a separate window could be used for the COPY at any time without inhibiting TPPLOT. But Szymanski had a better idea: why not use the PRINT command of MS-DOS, which is connected to a print queue or spooler? Whereas MS-DOS is not a multitasking operating system in other respects, it does allow printing to go on in the background at the same time a program is being executed, Szymanski observed. Since Salford FORTRAN allows the execution of MS-DOS commands (see the OS command of Section |), TPPLOT itself could spool the Epson plot to the print queue autornatically. Dr. Tsu-huei Liu first confirmed the ‘operation using her Star SG-10 printer connected to her 386-SX computer. But the resulting paper copy was distorted as if noise had broken up what otherwise would be a perfect plot. The distortion has been repeated using a PRINT command via the keyboard of another DOS computer that was connected to a genuine Epson printer. The plot is easily recognizable, but is broken frequently by noise on those "lines" having the most black pixels. Until the printing can be made perfect, implementation has not been automated. But tr OS command allows the TPPLOT user to dispatch his own plot manually (¢.g., OS PRIN: PLOTi2. EPS ). Szymanski has suggested several modifications that might cure the imperfect performance, but none has yet provided a miracle cure. A new versions of DOS would provide more controls than Dr. Liu’s version 3.0, but does not by itself (using default parameters) provide a solution. This was proven using a 25-MHz COMPAQ that ran COMPAQ DOS Version 4.01 (response to the VER command). Research continues. GEOGRAF graphics as used by the MS-DOS version of TPPLOT [2] were not as fiexible as Salford graphics in one important aspect of the support of Epson-compatible dot- matrix printers. Although GEOGRAF could be used to produce a disk file of a plot, that file legally could be plotted only using a computer for which the $75 royalty had been paid to Geocomp Corp. This is because the separate GEOCOMP utility FPLOT would be required. The Satford implementation does not carry this burden. Since nothing more sophisticated than a conventional MS-DOS COPY command is required, any of the 100 million or so MS-DOS computers of the world could plot such a file without royalty if an Epson-compatible dot-matrix printer were available. For the community of EMTP users, this is important progress. Batch-mode EMTP plotting of the CALCOMP PLOT declaration has been extended to Epson-compatible dot-matrix printers using the same general principles as just explained for TPPLOT. Look for details elsewhere in this same issue. VI. Miscellaneous Concluding Observations Source code of the Salford TPPLOT program was air mailed to Leuven on August 30th in response to a request by LEC. Although an 80386-based computer is not yet available in Leuven, one has been ordered, it would seem. The author flies into Brussels on September 30th, so anyone with questions about the Salford EMTP or TPPLOT is advised to contact him there during the following four weeks (through Friday, October 26th). Scalable and rotatable fonts are used for all text of Salford graphics. In the first report about Salford graphics six months ago [1], this was not the case. Of course, not much rotation is desired for EMTP use (at present, just labeling for the vertical axis). But lettering of different sizes is an important advance for higher-resolution screens and Epson-compatible 49 dot-matrix printers. Fonts are selected using integer indices 26, 32, 33, and 35 of SET DATA usage. The user is referred to comments within disk file TPPARAM.DAT for further details. Two integers are required for each usage since there is not only a font number (an integer) but also a floating-point scaling factor. These have been packed in a single integer as can be explained most easily by a simple illustration :_ The number 10425 represents a request for Salford font number 104 with a multiplication of 2.5 times its nominal or default size. Loss of speed is a negative aspect of scalable fonts that should be mentioned. Text no longer is outputted faster than the eye can see to a screen that is in graphics mode. Yet, this is not surprising. Such loss of labeling speed first was observed several years ago using GKS of Sun-3 workstations: letters are painted from left to right at speeds that a quick eye can follow, Well, the loss of character-painting speed is not a serious problem for the typical user, who will be absorbed looking at the unfolding plot, anyway. It is not yet known whether Salford’s exceptional Hershey fonts, which require the use of a different library function, might recover some of the lost speed. The Hershey fonts have been considered, but not yet within the TPPLOT framework that would make comparisons easy. Super VGA (Paradise 600 by 800 pixels) plots now appear perfect on the screen. This was not the case 6 months ago [1]. Previously, a Salford bug made it impossible to write in the bottom 1/4 of the screen, and the use of a single, fixed font (with fixed number of pixels, apparently) meant that lettering was too small. But now, with scalable fonts, and the ability to paint text anywhere on the screen, there no longer is any reason to avoid the highest-resolution (Super VGA) plots if the required hardware exists. The author has such graphics at home, and finally is fully satisfied with its use for graphs of Salford TPPLOT. The floppy disks that are used to distribute Salford EMTP to others within the USA and Canada will carry a copy of both the old and also the new (improved and fortified) Salford TPPLOT. This is because the new TPPLOT.EXE will not operate under the old (Rev. 1.67) Salford DOS extender. Whereas the Salford EMTP is too big to allow the luxury of such duplication, interactive plotting is small, so it is easy to continue to supply both old and new versions. Whereas the Salford EMTP has a switch to exclude (if desired) the use of new features, this is not the case for TPPLOT. Note carefully that the old version of TPPLOT no longer is being maintained, so do not expect the correction of any bugs that might later be discovered. An old executable version simply continues to be distributed because the two high-density disks are far from full. The CURSOR command of TPPLOT has not yet been activated only because other details have seemed to be more important thus far. Salford FORTRAN will return input from @ Microsoft mouse, however, so the software tools would seem to be available for such work. But how many potential users have a microsoft mouse? This might take some thought. Software developers at the University of Salford once again must be congratulated --- this time for quickly offering several important extensions to their FORTRAN compiler. By mail dated 23 March 1990, David Vallance had provided ATP developers in Portland with advance information about planned extensions of FTN77/386 to variable fonts, .PCX files, and hard copy. But the rapidity of delivery was not expected, It is a pleasure to work with a compiler that is being developed with such competence and speed by persons who are reachable (i.e, can be contacted about EMTP needs and/or problems). The author has used many computers over the years, and either he or his associates have contacted many 50 software developers about EMTP problems. None has been as accessible and responsive as David Vallance of Salford. ‘The Glyphix Helverica font [4] was used to publish this paper. Compare the present appearance with that of other articles from the Can/Am user group. For example, all contributions to the June (mislabeled July, which was the actual month of publication) issue used the Tymes Roman font. What is the preferance of readers? The advantage of electronic publishing is that such changes are trivial to make, so it is logical to experiment with other fonts (five were part of the $100 package by SWFTE International for use with WordPerfect). Later, perhaps other articles will illustrate other fonts for readers. Vil. References Used By Preceding Text [1]. James Chun-Heng Chiang and W. Scott Meyer, “Salford EMTP for 80386 has scree graphics and interactivity (SPY)," EMTP News, March, 1990, pages 26-30. [2]. W. Scott Meyer and Tsu-huei Liu, "MS-DOS version of interactive plotting program TPPLOT," EMTP News, June, 1989, pages 39-47. [3]. W. Scott Meyer, "Salford EMTP for 80386 given scrollable SPY and LUNIT6 windows," EMTP News, June, 1990, pages 17-23. [4]. W. Scott Meyer and Tsu-huei Liu, "Summary of EMTP News from North America," EMTP News, June, 1990, pages 9-11. WordPerfect Verson 5.1 and Glyphix fonts of SWFTE International, Ltd., are described on pages 10 and 11. [5]. Tsu-huei Liu and W. Scott Meyer, "HP-GL output of interactive plotting program ‘TPPLOT," EMTP News, March, 1990, pages 33-36. [6]. PC Paintbrush is a trademark of ZSoft Corporation of Atlanta, Georgia, US/ Street (actual as opposed to list) prices for current versions are readily determined by consulting PC Magazine. Page 471 of the issue dated September 11, 1990, contains a full- page advertizement by Computability Consumer Electronics. Entry “PC Paintorush IV" carties a price of $55.95 whereas the entry for "IV+" costs $109.95. Yet, it is not clear that the interested reader has to pay such prices. PC Paintbrush has been bundled with other products for “free” distribution. One such pairing wes with the mouse of Logitech, which has address: 6505 Kaiser Drive; Freemont, California 94555; USA; Phone (415) 795- 8500. Stuart McKay, the pioneer user of Salford EMTP under DESQView (see pages 24-28 of the preceding issue of EMTP News), provided this information. Although Mr. McKay did not buy the hardware, he was able to purchase the associated software separately from a Toronto computer shop (Businessworld) during July of 1990 for a mere $20 Canacian (about $16 American). The only concer is about age (Logitech was distributing version 3.11). Yet, one would hope that the format of disk files has never changed. Mr. McKay volunteered to practice using PC Paintbrush on PLOTXX.PCX files that are produced by Salford TPPLOT, so a disk with several such files was air mailed to him on September 6th. Look for a continuation of this interesting experiment in the next issue. [7]. Charles Petzold, “Metafile support under the OS/2 graphics programming interface," PC Magazine, September 25, 1990, pages 459-465 (the "Environments" section). 51 Summary of EMTP News From North America Co-Chairmen of the Canadian / American EMTP User Group W. Scott Meyer Tsu-huei Liv The Fontaine, Unit 6B 3179 Oak Tree Court 1220 N.E. 17-th Avenue West Linn, Oregon 97068 Portland, Oregon 97232 United States of America Summary : ATP developers in North America have been busy during recent months. Prof. Ned Mohan’s 2-day EMTP short course at the University of Minnesota in Minneapolis during July was the biggest EMTP course ever given (Section I). The new Salford FORTRAN compiler (Rev. 2.40) for the 80386 has significantly extended graphics, both for screen displays and also paper (hard copy). A separate paper details new screen graphics of the interactive plotting program TPPLOT whereas the Salford EMTP itself now offers free plotting on Epson- compatible dot-matrix printers (Section II). More general observations about the Rev. 2.40 Salford software are provided in Section IIT whereas Section IV summarizes EMTP materials recently mailed to Europe. The paper concludes with miscellaneous items of EMTP interest in Section V. I. Mohan’s EMTP Short Course in Minneapolis A total of 63 paying students were exposed to MS-DOS and Salford versions of ATP during the 2-day EMTP short course that both preceded (July 12-14) and followed (July 19-21) the IEEE PES Summer Meeting. This is a reference to Prof. Ned Mohan’s EMTP short course at the University of Minnesota in Minneapolis, about which several observations will be made (the remainder of this section). The amphitheater in which the course was held in the new, underground Civil and Mineral Engineering building provided ideal physical accommodations. With controlled lighting like a theater, and 3 screens behind the speaker to serve 2 conventional overhead projectors and 2 ceiling-mounted computer projectors, the visual environment was exceptional. No loud speaker was used or required, surprisingly. The amphitheater had good acoustics, and background noise did not seem to pose the challenge that it did in Florida. For one thing, no printers were used in Minnesota, and computers were either a quiet, new model (IBM PS/2 Model 50) or were an older model (IBM PC AT) that was enclosed and somewhat muffled below the desks. ‘Two students shared a computer in most cases, rather than each student having his own computer as at previous short courses in Florida and California. This was the key to the large attendance. Each of 4 concentric arcs of stationary counters supported 5 computers and a maximum of 10 students. In fact, paid attendance was only 25 for the first offering, but it was 38 for the second. There seemed to be a clear preference for EMTP education after the IEEE meeting, thereby avoiding a Sunday layover. er Power electronics and rotating machinery were covered more effectively than at other ATP short courses. Of course, it helps to have resident faculty who specialize in these subjects (Prof. Riaz in machinery and Prof. Mohan in power electronics). The course ended on power electronics, with’ Prof. Mohan referring persons having further interest to his John Wiley textbook ("Power Electronics: Converters, Applications, and Design") and his 64 illustrative EMTP data cases ("Computer Exercises for Power Electronics Education"). This was the high point, a natural climax to a successful, new format (2 days) for EMTP education. San Diego, California, is to be the site of the 1991 IEEE PES Summer Meeting, so there naturally has been thought about the possibility of repeating Prof. Mohan’s course there. An appropriate host is understood by all to be the critical prerequisite. Fortunately, an experienced EMTP educator, Prof. Ramon Betancourt of San Diego State University, resides in the area, and is considering the possibility as this paragraph is being frozen for publication. ‘Those who followed EMTP education in California the previous two years will recognize the name (Prof. Betancourt was a major contributor to the Cal Poly course in San Luis Obispo? Tl. Epson Printers Used for Free CALCOMP PLOT ————— The PEN PLOT and SCREEN PLOT declarations now apply to batch-mode (CalComp) plotting of the Salford EMTP just as they have to TP2.EXE of MS-DOS usage for some two years. The default is SCREEN PLOT, so if the user declares nothing, his CALCOMP PLOT graphs will come out on the screen. The PEN PLOT declaration is used to switch this plot to an Epson-compatible dot-matrix printer (or a disk file that could be copied to such a printer at any later time). For an illustration of such usage, see and try the new 2nd subcase of BENCHMARK DC-35. A new disk file named GRAPHICS provides STARTUP-like parameters that are associated with the new graphics. A full description of variables (presently 12 integers for each of the two graphic alternatives) will be found in the READ_ME.DOC file of the program distribution. Like STARTUP, the user must select his GRAPHICS file by copying one o. the 3 alternatives that are provided ( GRAPHICS.EGA, GRAPHICS.VGA, and GRAPHICS.600 ). Readers who are interested in Epson plots of Salford EMTP simulations are referred to the separate paper about TPPLOT in this same issue. The work on TPPLOT was done first, so details are explained there. It would be redundant to repeat such information here even though much of it is applicable. Only important differences have been repeated here. Til. Salford FORTRAN Rev. 2.40 is Important ‘The use of CONFIGDB to alter parameters of DBOS / 386 has been made more convenient by the addition of a menu. Previously (using Rev. 1.67), the user was obliged to step through all parameters even though he might only have wanted to change one. Now, an opening menu allows selective changes. Different roundoff rules for floating-point computation have been observed for the Salford Rev. 2.40 software. Differences are reminiscent of the effect of coprocessor emulation (see the

También podría gustarte