Fault Analysis and Diagnosis of Aeroengine Fuel Metering Device

Transcription

1 016 Inernaional Conference on Mechanical, Conrol, Elecric, Mecharonics, Informaion and Compuer (MCEMIC 016) ISBN: Faul Analysis and Diagnosis of Aeroengine Fuel Meering Device Kai Yin, Guo-long Jiang, Guo-jian Liu and Ying-qing Guo ABSTRACT Based on he failure analysis of he Hydro-Mechanical Uni (HMU) of aeroengine fuel conrol sysem, boh of daa-based and model-based mehods are proposed o diagnose he ypical fauls of he Fuel Meering Device (FMD). Firsly, he AMESim model of FMD was esablished. Then is faul ree was se up and simulaion analysis for wo ypical fauls was conduced. Finally, wo approaches were applied o hese fauls diagnosis and simulaion resuls show heir effeciveness. KEYWORDS Aeroengine conrol sysem; Hydro-Mechanical Uni; Failure analysis; Faul diagnosis. INTRODUCTION Full auhoriy digial elecronic conrol (FADEC) are widely used in he modern aeroengine [1-3]. As is acuaing device, he behavior of Hydro-Mechanical Uni (HMU) affecs aeroengine s reliabiliy, securiy and performance direcly [4-6].Therefore, he research of HMU s condiion monioring and faul diagnosis has imporan significance. The HMU faul diagnosis is based on he daa colleced by sensors insalled in fuel conrol sysem, using he faul diagnosis mehod, realizes he faul deecion and predicion of impending failure for is main pars [7]. As he key par of HMU, he Fuel Meering Device (FMD) regulaes fuel supply o he combusion chamber, according o he command of he elecronic conroller, hrough anchging flow area under he condiion of consan difference pressure. In his Kai Yin, Guo-jian Liu, AVIC Commercial Aircraf Engine Corporaion Limied, Shanghai, China. Guo-long Jiang, Ying-Qing Guo, School of Power and Energy, Norhwesern Polyechnical Universiy, Xi an, China. 8

2 paper, wih a high bypass raio urbofan engine as he objec, ypical fauls of is FMD were diagnosed for example. The FMD s AMESim model, as well as he faul ree, were build and analyzed. Two mehods of daa-based and model-based were applied o he FMD s faul diagnosis. MODELING, FAULT ANALYSIS AND SIMULATIONS Modeling and Faul Analysis FMD comprises a Differenial-Pressure-Conroller (DPC), a Fuel-Meering- Valve (FMV), and a Bypass-Valve. The FMD s simulaion model based on AMESim is shown as Fig.1. In order o research faul impacs on fuel conrol performance, he elecronic PI conroller was included so ha a closed loop for he fuel conrol was esablished. Figure 1. FMD simulaion model based on AMESim. According o some failure daa in realiy, valves clamping sagnaion and nozzle s clogging are wo ypical kinds of fauls in FMD. The faul ree of he fuel flow conrol abnormiy is shown as Fig.. Fuel Flow Conrol Abnomaly + Bypass Valve faul FMV loop faul DPC faul Bypass Valve sagnai on Servovalve faul + FMV faul Head Sensor faul + Noz zle clogging Coil disconn ec FMV sagnai on Figure. The faul ree of fuel flow conrol abnormiy. Faul Simulaion The valve sagnaion is simulaed by adding fricion. The analysis of valve sagnaion by simulaion can be generally divided ino four seps. Firsly he minimum required force ha could make he valve sagnaion is found and he influence of he sagnaion on he fuel sysem is analyzed. Secondly he condiions, causes and he influence of he crawling phenomenon on he fuel sysem are analyzed. Then he sysem oscillaions caused by he valve sagnaion if possible is analyzed. Finally he influence of general dynamic fricion (no sagnaion, crawling, oscillaion) on he performance of he fuel sysem is analyzed. 9

3 When given he signal of displacemen (sep from 10.mm o 11.mm), he faul condiions of valve sagnaion and valve crawling is shown in Table 1 and Table. "Sagnaion force" refers o he minimum required force ha will make he sagnaion; "Possibiliy" refers o he relaive possibiliy of he fauls; "Impacion" refers o he change of seady-sae error percenage in he fuel flow conrol. "Criical Force" refers o he minimum difference beween saic fricion and dynamic fricion ha will make he crawling phenomenon. There is no sric boundaries beween crawling phenomenon and general dynamic fricion, so he "Criical Force" in Table is for reference only. Table 1. The faul condiions of valve sagnaion. Pars Sagnaion force[n] Possibiliy Impacion [%] Head sensor 3.4 Larger 4. Bypass valve 65.3 Smaller 54.3 Fuel meering valve Middle 100 The main valve of servovalve 18.3 Larger 100 Table. The faul condiions of valve crawling phenomenon. Pars Criical Force[N] Possibiliy Head sensor 0.5 Large Bypass valve 15 Middle FMV No Smaller The main valve of he servovalve 0.1 Larger Faul simulaion daa shows: here is a greaer impac on he sysem when he valve is saged; he crawling phenomenon causes no only he seady-sae error, bu also poor dynamic performance; muli-sage amplificaion mechanism may occur selfoscillaion because of sagnaion; general dynamic fricion (no sagnaion, no crawling and no self-oscillaion) affecs he response speed of he fuel sysem. The simulaion mehod of nozzle clogging is similar o he approach menioned above. DIAGNOSTIC STRATEGIES Diagnosic Sraegies Design Daa-based and model-based are boh maure and classical diagnosic mehods in he sudy of faul diagnosis. To diagnose faul of he fuel meering device, he paper combine he daa-based and model-based. The mehod of daa-based or model-based is shown as Fig. 3. The paper assumes ha he sensors are working properly and he faul occurs alone. The AMESim simulaion model is used in place of he real sysem for faul diagnosis. The fuel meering device conains he following sensors: a pressure sensor in fron of he FMV, a pressure sensor afer FMV, he FMV s displacemen sensor. According o he sensor daa, i is easy o ge he fuel flow of he FMV. The nex secion will use hese sensors daa o diagnose. 30

4 Conrol Signal Fuel Meering Device Daa processing Faul diagnosis Model or daa Figure 3. The principle of faul diagnosis. Daa-based Faul Diagnosis Sraegy Firsly, some specific daa is obained as diagnosic daa by simulaion; secondly, he fuel sysem can be programmed o a specific acion in order o deermine he faul componen and furherly assess componen healh indicaors and sysem healh indicaors according comparing he diagnosic daa and he dae in he specific acion. The mehod was used in he diagnosis of saic fricion for he FMV as an example. There is a shor ime for valve o remain saionary when exising saic fricion. We can design an appropriae conroller for he FMV in order o make he pressure of drive oil change as Fig. 4. According o he sar-up ime of he FMV driven by he oil, we can calculae he saic fricion. Figure 4. The force curve of he FMV`s drive oil. Diagnosis of dynamic fricion for he FMV and diagnosis of nozzle clogging for he FMV servovalve are boh similar o he diagnosis of saic fricion. Model-based Faul Diagnosis Sraegy Firsly, he simplified ransfer funcion of he FMV loop is obained. Secondly, he same signal is used o conrol he real fuel meering device and he ransfer funcion, according comparing he daa from sensors and he dae from he ransfer funcion, i is possible o assess componen healh indicaors and sysem healh indicaors. In order o compare he daa from sensors and he dae from he ransfer funcion, he mean square of flow rae difference, he mean square funcion of pressure deviaion, he average value of flow rae difference are defined as follows: 1 r e q [q -q ] d (1) 1 r e p [ p - p ] d () 1 q [qr -q e ] d (3) 31

5 q r represens he fuel flow of fuel meering device (acual fuel flow), represens he DPC s oupu pressure (acual differenial pressure), p r represens he fuel flow in he ransfer funcion model (esimaed fuel flow), p r represens he DPC s oupu pressure (esimaed differenial pressure).when ends o infiniy, hese funcions show he saisical significance. q reflecs he faul degree of he fuel meering device, p reflecs he faul degree of he DPC, he higher hese parameers are, he more serious he fauls are. p e is used o disinguish he difference beween he faul caused by he servovalve s clogging and he faul caused by he fricion of he FMV. I is possible o disinguish four ypical fauls according o he above faul parameers. According o a large number of simulaion resuls, hese ypical fauls diagnosis daa are summarized as Table 3. p q q [ L/min] Sae [bar ] [(L/min) ] Normal sae (0,0.05) (0,) (-1,1) Head sensor Un sagnaion (0.05, + ) (, + ) The FMV sagnaion (0,0.05) (, + ) (-1,1) Lef nozzle clogging of he (0,0.05) (, + ) (1, + ) servovalve Righ nozzle clogging of he servovalve (0,0.05) (, + ) ( +,-1) The value areas of hese parameers in Table 3 shows he value of corresponding funcion when ends o limi infiniy, he "Un" in able is on behalf of uncerainy. p Normal sae: he limi of q is abou 0.0, less han 0.05; he limi of is q abou 1.35, less han ; he limi of is abou 0, he absolue valve is less han 1. The diagnosic funcion s limis in Table 3 are no sricly deermined. CONCLUSIONS Two faul diagnosis mehods are inroduced in his paper and applied o he fuel meering device, which can be popularized and exended o oher pars in HMU. The AMESim model of he fuel meering device was esablished a he beginning of he paper; hen he faul ree was proposed o analyze hese fauls ha may happen and wo kinds of ypical fauls in he faul ree were simulaed; finally, wo mehods were applied o diagnosis hese ypical fauls. The AMESim model is physics-based so ha i is suiable o analysis he saic and dynamic characerisics of each componen par for faul simulaion. The "daa-based" mehod can be used o analysis he fauls quaniaively and he "model-based" mehod can be used o analysis he fauls quaniaively. Simulaion resuls show ha boh hese wo faul diagnosis approaches are effecive. REFERENCES 1. Bai F. Mehods of Scheduling and Condiion Based Maineance Decision Making in Civil Aero Engine Flee [D]. Nanjing Universiy of Aeronauics and Asronauics PhD Thesis, Kong X.T. Research on Online Healh Diagnosis Technology for Turbofan Engines [D]. Nanjing Universiy of Aeronauics and Asronauics Maser Thesis, 01. 3

6 3. [3]Sun J.G. Prospecs of he Aeroengine Conrol Developmen in he Early Time of he 1s Cenury [J]. Journal of Aerospace Power, 001(): [4]Mercer C.R., Simon D.L., Huner G.W.,e.al. Fundamenal Technology Developmen for Gas-Turbine Engine Healh Managemen [R]. NASA , [5]Jiang C.H. Digial Conrol Sysem Design of an Aeroengine [J]. Aeroengine, 005(1): [6]Ashok J. Modelling of Fligh Conrol Hydraulic Acuaors Considering Real Sysem Effecs [J], Modeling and Simulaion Technologies Conference and Exhibi, 005, 3(11): [7]Huang D., Huang X. Mixed Faul Predicion Based on Kalman Filering and Exper Sysem [J]. Compuer Simulaion, 005():