Scribd
Upload
696 views9 pages

Dynamic Simulation of Gas Turbine Generating Unit

Gas-turbine generators are often connected to weak supply systems. Disturbances arising from fault or load fluctuation could cause these generators to become unstable. A fullscale site test on a 13 MW gas turbine generator has been conducted.

Uploaded by

alinaghib
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
696 views9 pages

Dynamic Simulation of Gas Turbine Generating Unit

Gas-turbine generators are often connected to weak supply systems. Disturbances arising from fault or load fluctuation could cause these generators to become unstable. A fullscale site test on a 13 MW gas turbine generator has been conducted.

Uploaded by

alinaghib
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 9
TT Dynamic simulation of gas-turbine generating unit WW, Hung, PRO, CEng, MIEE ‘Absract: Gasturbine generators are often com ‘ected to weak supply stems or even sed in io Inted operation. Sytem disturbances arising from fault or load Actuation could cause these gener flocs to become unstable Thee i an increasing ‘Semand for an acurate model of such a system, to enable the system feponse (o be investigated tnd improvement to the associated control system to be made. A mathematial model describing the dynamic bebaviout of gas-rbine generator St, ficiently accurate forthe above application and Jet eficient in computer usage presented. A Tul. fcale site test on a 1) MW gasurbine generator has been conducted and the recorded results ate aly in good agent ith the computed 1 introduction Synchronous machine stability studies have been a subject of intrest for many years. Much ofthe work pre ‘ented has been based on stan of hydro-turbine ener- Sting sets Emphasis in this paper is placed on dynamic Sablity studies of gas-turbine generating unit. ‘Acrotype gasturbine engines ave been widely adopted as prime candidates for electrical power genet- ‘tion. The fully automatic startup capably and the fst fun-up characteristic of such eagines have made them particularly suitable for peak-load lopping and standby pote supply purposes. Many engines of ths ype have eon installed in Targe thermal and nuclear power tions for black-tart applications. n terms of zz, weight land adaptability to a wide range of fuels (rom natural fis to crude or residual ofthe gas turbine is far fuperior to other forms of generation The gas turbine has lent tel directly to the oil and gas industries, not ‘only aban deal candidat for eectrical power generation, Bur also for gus compression and injection and crude of pumping. ‘nlite steam-turbine or hydrovclectric generators gas turbine generator ate commonly conneced to small bet. ‘Works, or even used in isolated operation such asin ll elds in desert areas and offshore installations. Such system are lable to become unstable afer a severe ‘stem disturbance, The relatively small rotating inert of this type of gas turbine forther aggravates this problem “The increasing we of large induction motor drives ‘with ratings up to about 10 MW and the general require ‘ment for diect-online starting ofthese motors make it Paper 403 (PI, PO eine ns 191 ‘he ator ih he System Teal Bunch, Noon Gd (Compuy ple Nato! Gn Howe Sune Set London, Usd Kino ‘xsential that the posible interaction berween the ga: forbine generators and the large indction motors be ‘early defined to avoid any future operational problems. ‘On offshore installations, thyristor switching employ {nga borst-fring technique is often wed for trace heating fpplications. The switching power could be in the MW ‘ange and this bas been Known to caus atraction prob- lems with pas-turbine generators. Shor-busts of power ‘demand of above 2 MW, at regular intervals during the en ~ oer a power on connection to the system. The tripping of the fenerator cieut breaker, owing to the opening of the bock-load switch, causes a stepedown change of output demand resulting ina fast cutback of fol onload reje= ton, The effect of this switching arrangement on the ga ‘turbine response isevident in the test result given “The final control function voltage Vo, as show in Fig 3, is directed tothe increased jump and rate limit circuit, whore Ve is modified, i required, to ensure the rate of Increase of power demand to the gat turbine ie within 8 This han fect on the demand or tion of power to secure the safety ofthe gu trbine Sheraton, The utpet wolage H ot fs eet ten ‘compared with the comprestr discharge preseure (CDP) feedback voltage Vy, with the eror signal being ample fed and fed tothe sr fol controller (AFC) which acs ‘an interface between the eletronic governor and the fel Valve package ‘The acceleration and. deceleration fuel schedule limits discussed are imposed on the governor demand voltage in the AFC, if required, before entering the throte control amplifer ‘a a throttle. demand voltage Fro This vllage is then compared with the throtle angle feedback voltage Vat the throttle amplife. The output ‘controls the fel valve actuator and hence the engine fel ‘ow by adjusting the fue thro valve opening 8 33 Mathematical rpresentation of governor contro! system [A detailed model of the electonichyéraulic governor ‘stem shown in Fig 3 waa developed by deriving tans: fer function from the corresponding electronic circuits fand the measured hydraulic. response. characteistcs ‘Details ofthe control cuits and hydraulic systems and their mathematical derivations are given in Reference 1 ‘with the corresponding transfer fuetion Block diagram cr eres a a tint et OP een forthe complete gas trbine and governor contrl system shown in Fig. 4 The valuce adopted for each of the parameters inclding gan constants C, bas seting volt ges B, time constants and mit are given the ‘Appendix. 4 Generator and axctestion system 41 Generator ‘The stationary two-asis representation is adopted for the ‘gncrator model The model equations ae arranged in ‘Satespace form with five frst order iflerential equa tons forthe generator currents, ie, ope forthe Held, two for the armature and two for the damper windings, and to forthe rotor dynamics, 4.2. Excitation system Various types of excitation system are used with gas ‘The elimination of commutators slip ringr and. any sociated brushgear lads to an excitation sytem which Ss reliable, compact, requires minimum attendance end, ‘most important of all is fee fom electrical sparking ‘These characteristics age capecilly favourable i gat turbine applications, since such Uni ‘aminated with inflammable gas “The excitation system model employed i ofthe bru less type consisting of a plot exciter, a main excite, voltage from 3 zenerstabiled reference seting The resulting eror signal is amplified to control the Bring . Taal wats) 0,4 Sia fon gran f ahr orm pulses of a half-controlled thyristor bridge atthe power ‘utpur stage, which in turn contols the excitation to the ‘A mathematical represeatation of the AVR_ was derived sing the component model approach [1]. To ‘adopt a universal approach tothe represeaation of exe {ation system, the model was modified and coupled with fn exciter model 10 conform with the IEEE Type 2 format [6, 7], Detal of the model derivations are given in Reference 1 with the excitation data adopted for the present studies given in the Appendix. Such a computer fepresatation of excitation systems” has been widely ‘Adopted and has formed a consistent frame of erence ‘whereby manufactures can respond fo a users request for excitation system dats 5 Overall ester mods ‘The mathematical models developed for the generator, the gas turbine and the associated excitation and gover- for control systems are coupled to form a complete gas {urbine generator model “The difretial equations describing the transient behaviour of the gasturbine/geneator uni are exprested im state variable form. The approach provides 8 stem lic means of assembling the system equations ina form ‘which can conveniently be solved by 2 digital computer “sing: numeral integration methods. The predictor corrector routine developed by Gear [8] was chosen for the present studies. ‘Deas of the system equations and ther derivations ae shown in Reference The data adopted are given in Reference {and the Appendix. 6 Comparison of test and computed rorults ‘A fullscal site test was conducted on a typical 13 MW ‘asturbine generator unit operating in isolated. mode fnta't load tank Various fete such as governor and ‘AVR step-esponse ters and load acceptance and ojo tiom teste were eased out and the test results (1) ae in ‘ood correlation with the computed results. Some of ‘hese els are ceused and presented 6.1 Excitation step-response tess Stepresponee tere on the excitation system were con- ‘acted with the generator operating at the rated spect fand on open eifeuit The teat revults are unavailable ‘eeause of technical dificules on site These are pial routine commissioning tests, so results obtained (fom 2 similar set using an identical type of AVR are considered tobe adequate for the present excitation system/ ‘enerator mode! validation purpose. The coresponding {enerator and) excilation system data, given in the Appendix, were adopted for these excitation step response test simulations. Another dataset given in the ‘Appendix was used forthe simulation of load secepeance and rejection tests and fault studies, ‘A 10% sep increase inthe AVR reference voltage was sade, Ths resulted in the AVR output increasing ts fel forcing level which was sustained for about O18 s before reducing to almost zero output at 035 (Fig. 5). This short-bust of AVR output improved the response time and reduced the overshoot ofthe generator output voltage Ta spite ofthe responsive action ofthe AVR in sup- pressing its output the generator volage was slow Fesponding to the 10% sep reduction of volage demand i 6, This could be improved if a flly-controle thy: Fistor bridge had been wwed to allow for an inverting AVR output i. 8 Gerry 10% sepdow demand tr age 62 Losd rejection tests ‘With the generator ‘operating at the rated speed, the outage vollage was adjusted to generate the requited level of power to the load tank. The load was rejected by Opening the generator citui-breaker afer the gat {rbine was thermally stabilised. Load rejection test at various load levels ie, 3 5 and 75 MW, were conducted and the resulting transients (1 are generally im good Agreement with the computer simulation rele The ‘corded results oblained fom the 3 MW load rejection {est was found to ive the best insight into the various limiting and contol Rinctions of the governor and is therefore presented (Fig. 7, The complex and nonlinear governor control actions and the gar turbine dynamic Fesponse arising fom such a disturbance are described. Guertin repo 2S drt ‘The opening of the circuit breaker, owing tothe opet- ation ofthe block:toad switch inthe power forbine speed Control circuit as described, caused 2 Hep reduction of the power turbine speed demand vollage Fay. This switching. action (Fig. 7) resulted in am inital. step reduction of ¥e from 3.510 28V. The subsequent decay (of ¥,ser about 09s isthe result of the power turbine ‘speed droop contro characteristic. "The eft of the increased jump and rate limit creut ‘on the control voltage Ves evident in the response char- ‘acters ofits Outpt voltage Y,, Te initial reduction of Ye was immediately followed bythe reduction of ¥ inde cating that no. control restriction was imposed by the ‘Grell The subsequent recovery rato of We exceeded the ‘ate limit setting ofthe eireut giving rise to a ramping characterise of ts output vollage b, daring the period between about? ands “The operation f the deceleration fuel schedule, incor. porated tn the air fuel controler (AFC) i evident in the Tesponse characteristic of Vrp shown in Fig. 7. Prior to the load rejection, the gaturbine engine was operating under steady-state condition and the compressor di ‘charge pressure demand voltage Veo atthe input ofthe AFC (Fig #) was in control, ie, the AFC output voltage Veg followed the input valage Yeap ‘Om load rejection, the rapid reduction ofthe demand voltage V, resulted in Vege to reduce to 8 evel which ould cause Mame extinction problem. The deceleration fuel schedule contol limit voltage Voc (Fig. 4) took over ‘control, ie, V9 = Voce ad ied the tial drop of Vyp to about 89 V. The subsequent slow decay of Vip ws caused by the gradual redetion of Face which ir’ nction of the decaying enpine compressor discharge Dresure Py. The vollage Voge regained contol after bout 03 & ‘Despite the complex governor control actions, the ‘comptter simulation results are in good agreement with the measured gas turbine and governor responses. tis unlikely tata simplified model would provide the same degree of accuracy, 63. Loed acceptance test ‘On the ? MW load acceptance tet, the response charac- testes ofthe gas turbine, as shown in Fig & as dom nated by the ramp rate setting ofthe increased jump and fate limit circuit The rate of lncease of output power ftom the gas turbine was thus limited and no acceleration ‘power was available until 35 safer the Load was appliod AS the power turbine speed fecovered, the speed over” ‘nor output voltage Tyr, caused bythe governor contol ‘ction, was educed and regained coatrol atthe Teast gate fesulting in the rapid. reduction of Yo. The demand ‘voltage V; was reduced leading tothe rapid reduction of feel flow to the gas turbine. This limits the subsequent power turbine sped overshoot. ‘Since a wate tak was used for loading the generator the effective resistance ofthe load was dependent om the water temperature, pority and turbulence in the tank. “The load inthe computer simulation wat based on & constant resistance and this could be the major reason for the small discrepancy between the measured and ‘mulated tansent power turbine sped. 64 Goneratorvoltge responses on stop-lo0d changes. “The generator and excitation system responses during the load acceptance and rejection tesa were not recorded ‘ovate of technical difculies.The corresponding gener~ eal > f h i i i i + r 3 3 y Tet + wT I, Peenanttne [ry tsa maa 10. Manel enna AV Raposo 2 MW gating a responses, appeared tobe incorrect because of the abnor- taal phenomenon, Le, the ini exponential decay of ‘Voltage on load rection and exponential ie of vllage ‘on load application. This phesomenon is possible on the ‘cceptance or rejection of unity or leading power factor Toad (9]. A similar phenomenon was observed in test Fess obtained from a 22 MW gasturbine generator Fig 10) 85 Three-phaso short-circuit fault “Te chance of conducting a three-phase short-cut test ‘Ona gar turbine generators remote Iti therfore essen- tal to develop an sovurate model to study its dynamic ‘behaviour under severe system fault coaditons "The gu-turbine generator was assumed to be con- rected fo an finite system through a stepup trans: former and’ transmission ine, with a total impedance of {06 +) 048) pu on 100 MW. A sjmmeaial three-phase Shore-icuit fat was assumed to be applied to the hih- ‘voltage terminals of the generator transformer, with the generator at rated load and power factor, for a period of ors, “The computed maximum rotor angle swing as shown in Fig 11 is about 160". The generator i table and the ‘ulation is wel-damped. The prodicted transient fesponses of the AVR output volage, generator terminal tnd fed circuit voltages and currents and gas turbine ‘tiput torque are also shown T's important to note Rom Fig. 12 that the intrven- tion of the speed governor control action BY the increased jump and rate Hmit creuit results in 2 fast ‘eduction but slow in recovery of ga turbine power. The fet effet (Fig. 12) ithe sigaifeant reduction of as turbine output torque immediatly afer the fault dis- turbance. The recovery time i governed by the increased Jump and rte imi setting. If the generator is connected to a small system, the phenomenon could cause the system frequency to fall Tending to the complete collapse of the system on under frequency. An optimised increased rate limit setting (0 Improve the gat turbine power recovery rate without compromising the safety operation ofthe engine is there- fore recommended. I is alo important to implement an tffective load shedding scheme to limit the fling ofthe fymem frequency to an acceptable level by disconnecting the requited amouat of non-essential loads “The fault clearance time of 0.37 ss ertcal Is shown in Fig. 13 that an extended fault duration of 639 results ina los of generator stability, Unlike steam-turbinegen- tratog the governor control ation of gs turbine can hve asgnifcant effet, as shown in Fig 1, on the trae ality ofthe ga turbine generator 7 Conclusion [A detailed computer model simulating the dynamic {esponse characteristic oft gas-lurbine gonerato uni is presented. Thi theoretical work has been substantiated by extensive test results, The high level of agreement fvideat between test and computed reslls gives con ‘Bdence in both the formulation of the theory and the ‘adaption ofthe modelling technique. It encourages the “pplication of this computer model to. inverigate the

You might also like